Add files via upload

arch/arm/net/Makefile

@@ -0,0 +1,3 @@
+# ARM-specific networking code
+
+obj-$(CONFIG_BPF_JIT) += bpf_jit_32.o

arch/arm/net/bpf_jit_32.c

@@ -0,0 +1,1083 @@
+/*
+ * Just-In-Time compiler for BPF filters on 32bit ARM
+ *
+ * Copyright (c) 2011 Mircea Gherzan <mgherzan@gmail.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; version 2 of the License.
+ */
+
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/errno.h>
+#include <linux/filter.h>
+#include <linux/netdevice.h>
+#include <linux/string.h>
+#include <linux/slab.h>
+#include <linux/if_vlan.h>
+
+#include <asm/cacheflush.h>
+#include <asm/hwcap.h>
+#include <asm/opcodes.h>
+
+#include "bpf_jit_32.h"
+
+/*
+ * ABI:
+ *
+ * r0	scratch register
+ * r4	BPF register A
+ * r5	BPF register X
+ * r6	pointer to the skb
+ * r7	skb->data
+ * r8	skb_headlen(skb)
+ */
+
+#define r_scratch	ARM_R0
+/* r1-r3 are (also) used for the unaligned loads on the non-ARMv7 slowpath */
+#define r_off		ARM_R1
+#define r_A		ARM_R4
+#define r_X		ARM_R5
+#define r_skb		ARM_R6
+#define r_skb_data	ARM_R7
+#define r_skb_hl	ARM_R8
+
+#define SCRATCH_SP_OFFSET	0
+#define SCRATCH_OFF(k)		(SCRATCH_SP_OFFSET + 4 * (k))
+
+#define SEEN_MEM		((1 << BPF_MEMWORDS) - 1)
+#define SEEN_MEM_WORD(k)	(1 << (k))
+#define SEEN_X			(1 << BPF_MEMWORDS)
+#define SEEN_CALL		(1 << (BPF_MEMWORDS + 1))
+#define SEEN_SKB		(1 << (BPF_MEMWORDS + 2))
+#define SEEN_DATA		(1 << (BPF_MEMWORDS + 3))
+
+#define FLAG_NEED_X_RESET	(1 << 0)
+#define FLAG_IMM_OVERFLOW	(1 << 1)
+
+struct jit_ctx {
+	const struct bpf_prog *skf;
+	unsigned idx;
+	unsigned prologue_bytes;
+	int ret0_fp_idx;
+	u32 seen;
+	u32 flags;
+	u32 *offsets;
+	u32 *target;
+#if __LINUX_ARM_ARCH__ < 7
+	u16 epilogue_bytes;
+	u16 imm_count;
+	u32 *imms;
+#endif
+};
+
+int bpf_jit_enable __read_mostly;
+
+static inline int call_neg_helper(struct sk_buff *skb, int offset, void *ret,
+		      unsigned int size)
+{
+	void *ptr = bpf_internal_load_pointer_neg_helper(skb, offset, size);
+
+	if (!ptr)
+		return -EFAULT;
+	memcpy(ret, ptr, size);
+	return 0;
+}
+
+static u64 jit_get_skb_b(struct sk_buff *skb, int offset)
+{
+	u8 ret;
+	int err;
+
+	if (offset < 0)
+		err = call_neg_helper(skb, offset, &ret, 1);
+	else
+		err = skb_copy_bits(skb, offset, &ret, 1);
+
+	return (u64)err << 32 | ret;
+}
+
+static u64 jit_get_skb_h(struct sk_buff *skb, int offset)
+{
+	u16 ret;
+	int err;
+
+	if (offset < 0)
+		err = call_neg_helper(skb, offset, &ret, 2);
+	else
+		err = skb_copy_bits(skb, offset, &ret, 2);
+
+	return (u64)err << 32 | ntohs(ret);
+}
+
+static u64 jit_get_skb_w(struct sk_buff *skb, int offset)
+{
+	u32 ret;
+	int err;
+
+	if (offset < 0)
+		err = call_neg_helper(skb, offset, &ret, 4);
+	else
+		err = skb_copy_bits(skb, offset, &ret, 4);
+
+	return (u64)err << 32 | ntohl(ret);
+}
+
+/*
+ * Wrappers which handle both OABI and EABI and assures Thumb2 interworking
+ * (where the assembly routines like __aeabi_uidiv could cause problems).
+ */
+static u32 jit_udiv(u32 dividend, u32 divisor)
+{
+	return dividend / divisor;
+}
+
+static u32 jit_mod(u32 dividend, u32 divisor)
+{
+	return dividend % divisor;
+}
+
+static inline void _emit(int cond, u32 inst, struct jit_ctx *ctx)
+{
+	inst |= (cond << 28);
+	inst = __opcode_to_mem_arm(inst);
+
+	if (ctx->target != NULL)
+		ctx->target[ctx->idx] = inst;
+
+	ctx->idx++;
+}
+
+/*
+ * Emit an instruction that will be executed unconditionally.
+ */
+static inline void emit(u32 inst, struct jit_ctx *ctx)
+{
+	_emit(ARM_COND_AL, inst, ctx);
+}
+
+static u16 saved_regs(struct jit_ctx *ctx)
+{
+	u16 ret = 0;
+
+	if ((ctx->skf->len > 1) ||
+	    (ctx->skf->insns[0].code == (BPF_RET | BPF_A)))
+		ret |= 1 << r_A;
+
+#ifdef CONFIG_FRAME_POINTER
+	ret |= (1 << ARM_FP) | (1 << ARM_IP) | (1 << ARM_LR) | (1 << ARM_PC);
+#else
+	if (ctx->seen & SEEN_CALL)
+		ret |= 1 << ARM_LR;
+#endif
+	if (ctx->seen & (SEEN_DATA | SEEN_SKB))
+		ret |= 1 << r_skb;
+	if (ctx->seen & SEEN_DATA)
+		ret |= (1 << r_skb_data) | (1 << r_skb_hl);
+	if (ctx->seen & SEEN_X)
+		ret |= 1 << r_X;
+
+	return ret;
+}
+
+static inline int mem_words_used(struct jit_ctx *ctx)
+{
+	/* yes, we do waste some stack space IF there are "holes" in the set" */
+	return fls(ctx->seen & SEEN_MEM);
+}
+
+static void jit_fill_hole(void *area, unsigned int size)
+{
+	u32 *ptr;
+	/* We are guaranteed to have aligned memory. */
+	for (ptr = area; size >= sizeof(u32); size -= sizeof(u32))
+		*ptr++ = __opcode_to_mem_arm(ARM_INST_UDF);
+}
+
+static void build_prologue(struct jit_ctx *ctx)
+{
+	u16 reg_set = saved_regs(ctx);
+	u16 off;
+
+#ifdef CONFIG_FRAME_POINTER
+	emit(ARM_MOV_R(ARM_IP, ARM_SP), ctx);
+	emit(ARM_PUSH(reg_set), ctx);
+	emit(ARM_SUB_I(ARM_FP, ARM_IP, 4), ctx);
+#else
+	if (reg_set)
+		emit(ARM_PUSH(reg_set), ctx);
+#endif
+
+	if (ctx->seen & (SEEN_DATA | SEEN_SKB))
+		emit(ARM_MOV_R(r_skb, ARM_R0), ctx);
+
+	if (ctx->seen & SEEN_DATA) {
+		off = offsetof(struct sk_buff, data);
+		emit(ARM_LDR_I(r_skb_data, r_skb, off), ctx);
+		/* headlen = len - data_len */
+		off = offsetof(struct sk_buff, len);
+		emit(ARM_LDR_I(r_skb_hl, r_skb, off), ctx);
+		off = offsetof(struct sk_buff, data_len);
+		emit(ARM_LDR_I(r_scratch, r_skb, off), ctx);
+		emit(ARM_SUB_R(r_skb_hl, r_skb_hl, r_scratch), ctx);
+	}
+
+	if (ctx->flags & FLAG_NEED_X_RESET)
+		emit(ARM_MOV_I(r_X, 0), ctx);
+
+	/* do not leak kernel data to userspace */
+	if (bpf_needs_clear_a(&ctx->skf->insns[0]))
+		emit(ARM_MOV_I(r_A, 0), ctx);
+
+	/* stack space for the BPF_MEM words */
+	if (ctx->seen & SEEN_MEM)
+		emit(ARM_SUB_I(ARM_SP, ARM_SP, mem_words_used(ctx) * 4), ctx);
+}
+
+static void build_epilogue(struct jit_ctx *ctx)
+{
+	u16 reg_set = saved_regs(ctx);
+
+	if (ctx->seen & SEEN_MEM)
+		emit(ARM_ADD_I(ARM_SP, ARM_SP, mem_words_used(ctx) * 4), ctx);
+
+	reg_set &= ~(1 << ARM_LR);
+
+#ifdef CONFIG_FRAME_POINTER
+	/* the first instruction of the prologue was: mov ip, sp */
+	reg_set &= ~(1 << ARM_IP);
+	reg_set |= (1 << ARM_SP);
+	emit(ARM_LDM(ARM_SP, reg_set), ctx);
+#else
+	if (reg_set) {
+		if (ctx->seen & SEEN_CALL)
+			reg_set |= 1 << ARM_PC;
+		emit(ARM_POP(reg_set), ctx);
+	}
+
+	if (!(ctx->seen & SEEN_CALL))
+		emit(ARM_BX(ARM_LR), ctx);
+#endif
+}
+
+static int16_t imm8m(u32 x)
+{
+	u32 rot;
+
+	for (rot = 0; rot < 16; rot++)
+		if ((x & ~ror32(0xff, 2 * rot)) == 0)
+			return rol32(x, 2 * rot) | (rot << 8);
+
+	return -1;
+}
+
+#if __LINUX_ARM_ARCH__ < 7
+
+static u16 imm_offset(u32 k, struct jit_ctx *ctx)
+{
+	unsigned i = 0, offset;
+	u16 imm;
+
+	/* on the "fake" run we just count them (duplicates included) */
+	if (ctx->target == NULL) {
+		ctx->imm_count++;
+		return 0;
+	}
+
+	while ((i < ctx->imm_count) && ctx->imms[i]) {
+		if (ctx->imms[i] == k)
+			break;
+		i++;
+	}
+
+	if (ctx->imms[i] == 0)
+		ctx->imms[i] = k;
+
+	/* constants go just after the epilogue */
+	offset =  ctx->offsets[ctx->skf->len];
+	offset += ctx->prologue_bytes;
+	offset += ctx->epilogue_bytes;
+	offset += i * 4;
+
+	ctx->target[offset / 4] = k;
+
+	/* PC in ARM mode == address of the instruction + 8 */
+	imm = offset - (8 + ctx->idx * 4);
+
+	if (imm & ~0xfff) {
+		/*
+		 * literal pool is too far, signal it into flags. we
+		 * can only detect it on the second pass unfortunately.
+		 */
+		ctx->flags |= FLAG_IMM_OVERFLOW;
+		return 0;
+	}
+
+	return imm;
+}
+
+#endif /* __LINUX_ARM_ARCH__ */
+
+/*
+ * Move an immediate that's not an imm8m to a core register.
+ */
+static inline void emit_mov_i_no8m(int rd, u32 val, struct jit_ctx *ctx)
+{
+#if __LINUX_ARM_ARCH__ < 7
+	emit(ARM_LDR_I(rd, ARM_PC, imm_offset(val, ctx)), ctx);
+#else
+	emit(ARM_MOVW(rd, val & 0xffff), ctx);
+	if (val > 0xffff)
+		emit(ARM_MOVT(rd, val >> 16), ctx);
+#endif
+}
+
+static inline void emit_mov_i(int rd, u32 val, struct jit_ctx *ctx)
+{
+	int imm12 = imm8m(val);
+
+	if (imm12 >= 0)
+		emit(ARM_MOV_I(rd, imm12), ctx);
+	else
+		emit_mov_i_no8m(rd, val, ctx);
+}
+
+#if __LINUX_ARM_ARCH__ < 6
+
+static void emit_load_be32(u8 cond, u8 r_res, u8 r_addr, struct jit_ctx *ctx)
+{
+	_emit(cond, ARM_LDRB_I(ARM_R3, r_addr, 1), ctx);
+	_emit(cond, ARM_LDRB_I(ARM_R1, r_addr, 0), ctx);
+	_emit(cond, ARM_LDRB_I(ARM_R2, r_addr, 3), ctx);
+	_emit(cond, ARM_LSL_I(ARM_R3, ARM_R3, 16), ctx);
+	_emit(cond, ARM_LDRB_I(ARM_R0, r_addr, 2), ctx);
+	_emit(cond, ARM_ORR_S(ARM_R3, ARM_R3, ARM_R1, SRTYPE_LSL, 24), ctx);
+	_emit(cond, ARM_ORR_R(ARM_R3, ARM_R3, ARM_R2), ctx);
+	_emit(cond, ARM_ORR_S(r_res, ARM_R3, ARM_R0, SRTYPE_LSL, 8), ctx);
+}
+
+static void emit_load_be16(u8 cond, u8 r_res, u8 r_addr, struct jit_ctx *ctx)
+{
+	_emit(cond, ARM_LDRB_I(ARM_R1, r_addr, 0), ctx);
+	_emit(cond, ARM_LDRB_I(ARM_R2, r_addr, 1), ctx);
+	_emit(cond, ARM_ORR_S(r_res, ARM_R2, ARM_R1, SRTYPE_LSL, 8), ctx);
+}
+
+static inline void emit_swap16(u8 r_dst, u8 r_src, struct jit_ctx *ctx)
+{
+	/* r_dst = (r_src << 8) | (r_src >> 8) */
+	emit(ARM_LSL_I(ARM_R1, r_src, 8), ctx);
+	emit(ARM_ORR_S(r_dst, ARM_R1, r_src, SRTYPE_LSR, 8), ctx);
+
+	/*
+	 * we need to mask out the bits set in r_dst[23:16] due to
+	 * the first shift instruction.
+	 *
+	 * note that 0x8ff is the encoded immediate 0x00ff0000.
+	 */
+	emit(ARM_BIC_I(r_dst, r_dst, 0x8ff), ctx);
+}
+
+#else  /* ARMv6+ */
+
+static void emit_load_be32(u8 cond, u8 r_res, u8 r_addr, struct jit_ctx *ctx)
+{
+	_emit(cond, ARM_LDR_I(r_res, r_addr, 0), ctx);
+#ifdef __LITTLE_ENDIAN
+	_emit(cond, ARM_REV(r_res, r_res), ctx);
+#endif
+}
+
+static void emit_load_be16(u8 cond, u8 r_res, u8 r_addr, struct jit_ctx *ctx)
+{
+	_emit(cond, ARM_LDRH_I(r_res, r_addr, 0), ctx);
+#ifdef __LITTLE_ENDIAN
+	_emit(cond, ARM_REV16(r_res, r_res), ctx);
+#endif
+}
+
+static inline void emit_swap16(u8 r_dst __maybe_unused,
+			       u8 r_src __maybe_unused,
+			       struct jit_ctx *ctx __maybe_unused)
+{
+#ifdef __LITTLE_ENDIAN
+	emit(ARM_REV16(r_dst, r_src), ctx);
+#endif
+}
+
+#endif /* __LINUX_ARM_ARCH__ < 6 */
+
+
+/* Compute the immediate value for a PC-relative branch. */
+static inline u32 b_imm(unsigned tgt, struct jit_ctx *ctx)
+{
+	u32 imm;
+
+	if (ctx->target == NULL)
+		return 0;
+	/*
+	 * BPF allows only forward jumps and the offset of the target is
+	 * still the one computed during the first pass.
+	 */
+	imm  = ctx->offsets[tgt] + ctx->prologue_bytes - (ctx->idx * 4 + 8);
+
+	return imm >> 2;
+}
+
+#define OP_IMM3(op, r1, r2, imm_val, ctx)				\
+	do {								\
+		imm12 = imm8m(imm_val);					\
+		if (imm12 < 0) {					\
+			emit_mov_i_no8m(r_scratch, imm_val, ctx);	\
+			emit(op ## _R((r1), (r2), r_scratch), ctx);	\
+		} else {						\
+			emit(op ## _I((r1), (r2), imm12), ctx);		\
+		}							\
+	} while (0)
+
+static inline void emit_err_ret(u8 cond, struct jit_ctx *ctx)
+{
+	if (ctx->ret0_fp_idx >= 0) {
+		_emit(cond, ARM_B(b_imm(ctx->ret0_fp_idx, ctx)), ctx);
+		/* NOP to keep the size constant between passes */
+		emit(ARM_MOV_R(ARM_R0, ARM_R0), ctx);
+	} else {
+		_emit(cond, ARM_MOV_I(ARM_R0, 0), ctx);
+		_emit(cond, ARM_B(b_imm(ctx->skf->len, ctx)), ctx);
+	}
+}
+
+static inline void emit_blx_r(u8 tgt_reg, struct jit_ctx *ctx)
+{
+#if __LINUX_ARM_ARCH__ < 5
+	emit(ARM_MOV_R(ARM_LR, ARM_PC), ctx);
+
+	if (elf_hwcap & HWCAP_THUMB)
+		emit(ARM_BX(tgt_reg), ctx);
+	else
+		emit(ARM_MOV_R(ARM_PC, tgt_reg), ctx);
+#else
+	emit(ARM_BLX_R(tgt_reg), ctx);
+#endif
+}
+
+static inline void emit_udivmod(u8 rd, u8 rm, u8 rn, struct jit_ctx *ctx,
+				int bpf_op)
+{
+#if __LINUX_ARM_ARCH__ == 7
+	if (elf_hwcap & HWCAP_IDIVA) {
+		if (bpf_op == BPF_DIV)
+			emit(ARM_UDIV(rd, rm, rn), ctx);
+		else {
+			emit(ARM_UDIV(ARM_R3, rm, rn), ctx);
+			emit(ARM_MLS(rd, rn, ARM_R3, rm), ctx);
+		}
+		return;
+	}
+#endif
+
+	/*
+	 * For BPF_ALU | BPF_DIV | BPF_K instructions, rm is ARM_R4
+	 * (r_A) and rn is ARM_R0 (r_scratch) so load rn first into
+	 * ARM_R1 to avoid accidentally overwriting ARM_R0 with rm
+	 * before using it as a source for ARM_R1.
+	 *
+	 * For BPF_ALU | BPF_DIV | BPF_X rm is ARM_R4 (r_A) and rn is
+	 * ARM_R5 (r_X) so there is no particular register overlap
+	 * issues.
+	 */
+	if (rn != ARM_R1)
+		emit(ARM_MOV_R(ARM_R1, rn), ctx);
+	if (rm != ARM_R0)
+		emit(ARM_MOV_R(ARM_R0, rm), ctx);
+
+	ctx->seen |= SEEN_CALL;
+	emit_mov_i(ARM_R3, bpf_op == BPF_DIV ? (u32)jit_udiv : (u32)jit_mod,
+		   ctx);
+	emit_blx_r(ARM_R3, ctx);
+
+	if (rd != ARM_R0)
+		emit(ARM_MOV_R(rd, ARM_R0), ctx);
+}
+
+static inline void update_on_xread(struct jit_ctx *ctx)
+{
+	if (!(ctx->seen & SEEN_X))
+		ctx->flags |= FLAG_NEED_X_RESET;
+
+	ctx->seen |= SEEN_X;
+}
+
+static int build_body(struct jit_ctx *ctx)
+{
+	void *load_func[] = {jit_get_skb_b, jit_get_skb_h, jit_get_skb_w};
+	const struct bpf_prog *prog = ctx->skf;
+	const struct sock_filter *inst;
+	unsigned i, load_order, off, condt;
+	int imm12;
+	u32 k;
+
+	for (i = 0; i < prog->len; i++) {
+		u16 code;
+
+		inst = &(prog->insns[i]);
+		/* K as an immediate value operand */
+		k = inst->k;
+		code = bpf_anc_helper(inst);
+
+		/* compute offsets only in the fake pass */
+		if (ctx->target == NULL)
+			ctx->offsets[i] = ctx->idx * 4;
+
+		switch (code) {
+		case BPF_LD | BPF_IMM:
+			emit_mov_i(r_A, k, ctx);
+			break;
+		case BPF_LD | BPF_W | BPF_LEN:
+			ctx->seen |= SEEN_SKB;
+			BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, len) != 4);
+			emit(ARM_LDR_I(r_A, r_skb,
+				       offsetof(struct sk_buff, len)), ctx);
+			break;
+		case BPF_LD | BPF_MEM:
+			/* A = scratch[k] */
+			ctx->seen |= SEEN_MEM_WORD(k);
+			emit(ARM_LDR_I(r_A, ARM_SP, SCRATCH_OFF(k)), ctx);
+			break;
+		case BPF_LD | BPF_W | BPF_ABS:
+			load_order = 2;
+			goto load;
+		case BPF_LD | BPF_H | BPF_ABS:
+			load_order = 1;
+			goto load;
+		case BPF_LD | BPF_B | BPF_ABS:
+			load_order = 0;
+load:
+			emit_mov_i(r_off, k, ctx);
+load_common:
+			ctx->seen |= SEEN_DATA | SEEN_CALL;
+
+			if (load_order > 0) {
+				emit(ARM_SUB_I(r_scratch, r_skb_hl,
+					       1 << load_order), ctx);
+				emit(ARM_CMP_R(r_scratch, r_off), ctx);
+				condt = ARM_COND_GE;
+			} else {
+				emit(ARM_CMP_R(r_skb_hl, r_off), ctx);
+				condt = ARM_COND_HI;
+			}
+
+			/*
+			 * test for negative offset, only if we are
+			 * currently scheduled to take the fast
+			 * path. this will update the flags so that
+			 * the slowpath instruction are ignored if the
+			 * offset is negative.
+			 *
+			 * for loard_order == 0 the HI condition will
+			 * make loads at offset 0 take the slow path too.
+			 */
+			_emit(condt, ARM_CMP_I(r_off, 0), ctx);
+
+			_emit(condt, ARM_ADD_R(r_scratch, r_off, r_skb_data),
+			      ctx);
+
+			if (load_order == 0)
+				_emit(condt, ARM_LDRB_I(r_A, r_scratch, 0),
+				      ctx);
+			else if (load_order == 1)
+				emit_load_be16(condt, r_A, r_scratch, ctx);
+			else if (load_order == 2)
+				emit_load_be32(condt, r_A, r_scratch, ctx);
+
+			_emit(condt, ARM_B(b_imm(i + 1, ctx)), ctx);
+
+			/* the slowpath */
+			emit_mov_i(ARM_R3, (u32)load_func[load_order], ctx);
+			emit(ARM_MOV_R(ARM_R0, r_skb), ctx);
+			/* the offset is already in R1 */
+			emit_blx_r(ARM_R3, ctx);
+			/* check the result of skb_copy_bits */
+			emit(ARM_CMP_I(ARM_R1, 0), ctx);
+			emit_err_ret(ARM_COND_NE, ctx);
+			emit(ARM_MOV_R(r_A, ARM_R0), ctx);
+			break;
+		case BPF_LD | BPF_W | BPF_IND:
+			load_order = 2;
+			goto load_ind;
+		case BPF_LD | BPF_H | BPF_IND:
+			load_order = 1;
+			goto load_ind;
+		case BPF_LD | BPF_B | BPF_IND:
+			load_order = 0;
+load_ind:
+			update_on_xread(ctx);
+			OP_IMM3(ARM_ADD, r_off, r_X, k, ctx);
+			goto load_common;
+		case BPF_LDX | BPF_IMM:
+			ctx->seen |= SEEN_X;
+			emit_mov_i(r_X, k, ctx);
+			break;
+		case BPF_LDX | BPF_W | BPF_LEN:
+			ctx->seen |= SEEN_X | SEEN_SKB;
+			emit(ARM_LDR_I(r_X, r_skb,
+				       offsetof(struct sk_buff, len)), ctx);
+			break;
+		case BPF_LDX | BPF_MEM:
+			ctx->seen |= SEEN_X | SEEN_MEM_WORD(k);
+			emit(ARM_LDR_I(r_X, ARM_SP, SCRATCH_OFF(k)), ctx);
+			break;
+		case BPF_LDX | BPF_B | BPF_MSH:
+			/* x = ((*(frame + k)) & 0xf) << 2; */
+			ctx->seen |= SEEN_X | SEEN_DATA | SEEN_CALL;
+			/* the interpreter should deal with the negative K */
+			if ((int)k < 0)
+				return -1;
+			/* offset in r1: we might have to take the slow path */
+			emit_mov_i(r_off, k, ctx);
+			emit(ARM_CMP_R(r_skb_hl, r_off), ctx);
+
+			/* load in r0: common with the slowpath */
+			_emit(ARM_COND_HI, ARM_LDRB_R(ARM_R0, r_skb_data,
+						      ARM_R1), ctx);
+			/*
+			 * emit_mov_i() might generate one or two instructions,
+			 * the same holds for emit_blx_r()
+			 */
+			_emit(ARM_COND_HI, ARM_B(b_imm(i + 1, ctx) - 2), ctx);
+
+			emit(ARM_MOV_R(ARM_R0, r_skb), ctx);
+			/* r_off is r1 */
+			emit_mov_i(ARM_R3, (u32)jit_get_skb_b, ctx);
+			emit_blx_r(ARM_R3, ctx);
+			/* check the return value of skb_copy_bits */
+			emit(ARM_CMP_I(ARM_R1, 0), ctx);
+			emit_err_ret(ARM_COND_NE, ctx);
+
+			emit(ARM_AND_I(r_X, ARM_R0, 0x00f), ctx);
+			emit(ARM_LSL_I(r_X, r_X, 2), ctx);
+			break;
+		case BPF_ST:
+			ctx->seen |= SEEN_MEM_WORD(k);
+			emit(ARM_STR_I(r_A, ARM_SP, SCRATCH_OFF(k)), ctx);
+			break;
+		case BPF_STX:
+			update_on_xread(ctx);
+			ctx->seen |= SEEN_MEM_WORD(k);
+			emit(ARM_STR_I(r_X, ARM_SP, SCRATCH_OFF(k)), ctx);
+			break;
+		case BPF_ALU | BPF_ADD | BPF_K:
+			/* A += K */
+			OP_IMM3(ARM_ADD, r_A, r_A, k, ctx);
+			break;
+		case BPF_ALU | BPF_ADD | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_ADD_R(r_A, r_A, r_X), ctx);
+			break;
+		case BPF_ALU | BPF_SUB | BPF_K:
+			/* A -= K */
+			OP_IMM3(ARM_SUB, r_A, r_A, k, ctx);
+			break;
+		case BPF_ALU | BPF_SUB | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_SUB_R(r_A, r_A, r_X), ctx);
+			break;
+		case BPF_ALU | BPF_MUL | BPF_K:
+			/* A *= K */
+			emit_mov_i(r_scratch, k, ctx);
+			emit(ARM_MUL(r_A, r_A, r_scratch), ctx);
+			break;
+		case BPF_ALU | BPF_MUL | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_MUL(r_A, r_A, r_X), ctx);
+			break;
+		case BPF_ALU | BPF_DIV | BPF_K:
+			if (k == 1)
+				break;
+			emit_mov_i(r_scratch, k, ctx);
+			emit_udivmod(r_A, r_A, r_scratch, ctx, BPF_DIV);
+			break;
+		case BPF_ALU | BPF_DIV | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_CMP_I(r_X, 0), ctx);
+			emit_err_ret(ARM_COND_EQ, ctx);
+			emit_udivmod(r_A, r_A, r_X, ctx, BPF_DIV);
+			break;
+		case BPF_ALU | BPF_MOD | BPF_K:
+			if (k == 1) {
+				emit_mov_i(r_A, 0, ctx);
+				break;
+			}
+			emit_mov_i(r_scratch, k, ctx);
+			emit_udivmod(r_A, r_A, r_scratch, ctx, BPF_MOD);
+			break;
+		case BPF_ALU | BPF_MOD | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_CMP_I(r_X, 0), ctx);
+			emit_err_ret(ARM_COND_EQ, ctx);
+			emit_udivmod(r_A, r_A, r_X, ctx, BPF_MOD);
+			break;
+		case BPF_ALU | BPF_OR | BPF_K:
+			/* A |= K */
+			OP_IMM3(ARM_ORR, r_A, r_A, k, ctx);
+			break;
+		case BPF_ALU | BPF_OR | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_ORR_R(r_A, r_A, r_X), ctx);
+			break;
+		case BPF_ALU | BPF_XOR | BPF_K:
+			/* A ^= K; */
+			OP_IMM3(ARM_EOR, r_A, r_A, k, ctx);
+			break;
+		case BPF_ANC | SKF_AD_ALU_XOR_X:
+		case BPF_ALU | BPF_XOR | BPF_X:
+			/* A ^= X */
+			update_on_xread(ctx);
+			emit(ARM_EOR_R(r_A, r_A, r_X), ctx);
+			break;
+		case BPF_ALU | BPF_AND | BPF_K:
+			/* A &= K */
+			OP_IMM3(ARM_AND, r_A, r_A, k, ctx);
+			break;
+		case BPF_ALU | BPF_AND | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_AND_R(r_A, r_A, r_X), ctx);
+			break;
+		case BPF_ALU | BPF_LSH | BPF_K:
+			if (unlikely(k > 31))
+				return -1;
+			emit(ARM_LSL_I(r_A, r_A, k), ctx);
+			break;
+		case BPF_ALU | BPF_LSH | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_LSL_R(r_A, r_A, r_X), ctx);
+			break;
+		case BPF_ALU | BPF_RSH | BPF_K:
+			if (unlikely(k > 31))
+				return -1;
+			if (k)
+				emit(ARM_LSR_I(r_A, r_A, k), ctx);
+			break;
+		case BPF_ALU | BPF_RSH | BPF_X:
+			update_on_xread(ctx);
+			emit(ARM_LSR_R(r_A, r_A, r_X), ctx);
+			break;
+		case BPF_ALU | BPF_NEG:
+			/* A = -A */
+			emit(ARM_RSB_I(r_A, r_A, 0), ctx);
+			break;
+		case BPF_JMP | BPF_JA:
+			/* pc += K */
+			emit(ARM_B(b_imm(i + k + 1, ctx)), ctx);
+			break;
+		case BPF_JMP | BPF_JEQ | BPF_K:
+			/* pc += (A == K) ? pc->jt : pc->jf */
+			condt  = ARM_COND_EQ;
+			goto cmp_imm;
+		case BPF_JMP | BPF_JGT | BPF_K:
+			/* pc += (A > K) ? pc->jt : pc->jf */
+			condt  = ARM_COND_HI;
+			goto cmp_imm;
+		case BPF_JMP | BPF_JGE | BPF_K:
+			/* pc += (A >= K) ? pc->jt : pc->jf */
+			condt  = ARM_COND_HS;
+cmp_imm:
+			imm12 = imm8m(k);
+			if (imm12 < 0) {
+				emit_mov_i_no8m(r_scratch, k, ctx);
+				emit(ARM_CMP_R(r_A, r_scratch), ctx);
+			} else {
+				emit(ARM_CMP_I(r_A, imm12), ctx);
+			}
+cond_jump:
+			if (inst->jt)
+				_emit(condt, ARM_B(b_imm(i + inst->jt + 1,
+						   ctx)), ctx);
+			if (inst->jf)
+				_emit(condt ^ 1, ARM_B(b_imm(i + inst->jf + 1,
+							     ctx)), ctx);
+			break;
+		case BPF_JMP | BPF_JEQ | BPF_X:
+			/* pc += (A == X) ? pc->jt : pc->jf */
+			condt   = ARM_COND_EQ;
+			goto cmp_x;
+		case BPF_JMP | BPF_JGT | BPF_X:
+			/* pc += (A > X) ? pc->jt : pc->jf */
+			condt   = ARM_COND_HI;
+			goto cmp_x;
+		case BPF_JMP | BPF_JGE | BPF_X:
+			/* pc += (A >= X) ? pc->jt : pc->jf */
+			condt   = ARM_COND_CS;
+cmp_x:
+			update_on_xread(ctx);
+			emit(ARM_CMP_R(r_A, r_X), ctx);
+			goto cond_jump;
+		case BPF_JMP | BPF_JSET | BPF_K:
+			/* pc += (A & K) ? pc->jt : pc->jf */
+			condt  = ARM_COND_NE;
+			/* not set iff all zeroes iff Z==1 iff EQ */
+
+			imm12 = imm8m(k);
+			if (imm12 < 0) {
+				emit_mov_i_no8m(r_scratch, k, ctx);
+				emit(ARM_TST_R(r_A, r_scratch), ctx);
+			} else {
+				emit(ARM_TST_I(r_A, imm12), ctx);
+			}
+			goto cond_jump;
+		case BPF_JMP | BPF_JSET | BPF_X:
+			/* pc += (A & X) ? pc->jt : pc->jf */
+			update_on_xread(ctx);
+			condt  = ARM_COND_NE;
+			emit(ARM_TST_R(r_A, r_X), ctx);
+			goto cond_jump;
+		case BPF_RET | BPF_A:
+			emit(ARM_MOV_R(ARM_R0, r_A), ctx);
+			goto b_epilogue;
+		case BPF_RET | BPF_K:
+			if ((k == 0) && (ctx->ret0_fp_idx < 0))
+				ctx->ret0_fp_idx = i;
+			emit_mov_i(ARM_R0, k, ctx);
+b_epilogue:
+			if (i != ctx->skf->len - 1)
+				emit(ARM_B(b_imm(prog->len, ctx)), ctx);
+			break;
+		case BPF_MISC | BPF_TAX:
+			/* X = A */
+			ctx->seen |= SEEN_X;
+			emit(ARM_MOV_R(r_X, r_A), ctx);
+			break;
+		case BPF_MISC | BPF_TXA:
+			/* A = X */
+			update_on_xread(ctx);
+			emit(ARM_MOV_R(r_A, r_X), ctx);
+			break;
+		case BPF_ANC | SKF_AD_PROTOCOL:
+			/* A = ntohs(skb->protocol) */
+			ctx->seen |= SEEN_SKB;
+			BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff,
+						  protocol) != 2);
+			off = offsetof(struct sk_buff, protocol);
+			emit(ARM_LDRH_I(r_scratch, r_skb, off), ctx);
+			emit_swap16(r_A, r_scratch, ctx);
+			break;
+		case BPF_ANC | SKF_AD_CPU:
+			/* r_scratch = current_thread_info() */
+			OP_IMM3(ARM_BIC, r_scratch, ARM_SP, THREAD_SIZE - 1, ctx);
+			/* A = current_thread_info()->cpu */
+			BUILD_BUG_ON(FIELD_SIZEOF(struct thread_info, cpu) != 4);
+			off = offsetof(struct thread_info, cpu);
+			emit(ARM_LDR_I(r_A, r_scratch, off), ctx);
+			break;
+		case BPF_ANC | SKF_AD_IFINDEX:
+		case BPF_ANC | SKF_AD_HATYPE:
+			/* A = skb->dev->ifindex */
+			/* A = skb->dev->type */
+			ctx->seen |= SEEN_SKB;
+			off = offsetof(struct sk_buff, dev);
+			emit(ARM_LDR_I(r_scratch, r_skb, off), ctx);
+
+			emit(ARM_CMP_I(r_scratch, 0), ctx);
+			emit_err_ret(ARM_COND_EQ, ctx);
+
+			BUILD_BUG_ON(FIELD_SIZEOF(struct net_device,
+						  ifindex) != 4);
+			BUILD_BUG_ON(FIELD_SIZEOF(struct net_device,
+						  type) != 2);
+
+			if (code == (BPF_ANC | SKF_AD_IFINDEX)) {
+				off = offsetof(struct net_device, ifindex);
+				emit(ARM_LDR_I(r_A, r_scratch, off), ctx);
+			} else {
+				/*
+				 * offset of field "type" in "struct
+				 * net_device" is above what can be
+				 * used in the ldrh rd, [rn, #imm]
+				 * instruction, so load the offset in
+				 * a register and use ldrh rd, [rn, rm]
+				 */
+				off = offsetof(struct net_device, type);
+				emit_mov_i(ARM_R3, off, ctx);
+				emit(ARM_LDRH_R(r_A, r_scratch, ARM_R3), ctx);
+			}
+			break;
+		case BPF_ANC | SKF_AD_MARK:
+			ctx->seen |= SEEN_SKB;
+			BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, mark) != 4);
+			off = offsetof(struct sk_buff, mark);
+			emit(ARM_LDR_I(r_A, r_skb, off), ctx);
+			break;
+		case BPF_ANC | SKF_AD_RXHASH:
+			ctx->seen |= SEEN_SKB;
+			BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, hash) != 4);
+			off = offsetof(struct sk_buff, hash);
+			emit(ARM_LDR_I(r_A, r_skb, off), ctx);
+			break;
+		case BPF_ANC | SKF_AD_VLAN_TAG:
+		case BPF_ANC | SKF_AD_VLAN_TAG_PRESENT:
+			ctx->seen |= SEEN_SKB;
+			BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, vlan_tci) != 2);
+			off = offsetof(struct sk_buff, vlan_tci);
+			emit(ARM_LDRH_I(r_A, r_skb, off), ctx);
+			if (code == (BPF_ANC | SKF_AD_VLAN_TAG))
+				OP_IMM3(ARM_AND, r_A, r_A, ~VLAN_TAG_PRESENT, ctx);
+			else {
+				OP_IMM3(ARM_LSR, r_A, r_A, 12, ctx);
+				OP_IMM3(ARM_AND, r_A, r_A, 0x1, ctx);
+			}
+			break;
+		case BPF_ANC | SKF_AD_PKTTYPE:
+			ctx->seen |= SEEN_SKB;
+			BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff,
+						  __pkt_type_offset[0]) != 1);
+			off = PKT_TYPE_OFFSET();
+			emit(ARM_LDRB_I(r_A, r_skb, off), ctx);
+			emit(ARM_AND_I(r_A, r_A, PKT_TYPE_MAX), ctx);
+#ifdef __BIG_ENDIAN_BITFIELD
+			emit(ARM_LSR_I(r_A, r_A, 5), ctx);
+#endif
+			break;
+		case BPF_ANC | SKF_AD_QUEUE:
+			ctx->seen |= SEEN_SKB;
+			BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff,
+						  queue_mapping) != 2);
+			BUILD_BUG_ON(offsetof(struct sk_buff,
+					      queue_mapping) > 0xff);
+			off = offsetof(struct sk_buff, queue_mapping);
+			emit(ARM_LDRH_I(r_A, r_skb, off), ctx);
+			break;
+		case BPF_ANC | SKF_AD_PAY_OFFSET:
+			ctx->seen |= SEEN_SKB | SEEN_CALL;
+
+			emit(ARM_MOV_R(ARM_R0, r_skb), ctx);
+			emit_mov_i(ARM_R3, (unsigned int)skb_get_poff, ctx);
+			emit_blx_r(ARM_R3, ctx);
+			emit(ARM_MOV_R(r_A, ARM_R0), ctx);
+			break;
+		case BPF_LDX | BPF_W | BPF_ABS:
+			/*
+			 * load a 32bit word from struct seccomp_data.
+			 * seccomp_check_filter() will already have checked
+			 * that k is 32bit aligned and lies within the
+			 * struct seccomp_data.
+			 */
+			ctx->seen |= SEEN_SKB;
+			emit(ARM_LDR_I(r_A, r_skb, k), ctx);
+			break;
+		default:
+			return -1;
+		}
+
+		if (ctx->flags & FLAG_IMM_OVERFLOW)
+			/*
+			 * this instruction generated an overflow when
+			 * trying to access the literal pool, so
+			 * delegate this filter to the kernel interpreter.
+			 */
+			return -1;
+	}
+
+	/* compute offsets only during the first pass */
+	if (ctx->target == NULL)
+		ctx->offsets[i] = ctx->idx * 4;
+
+	return 0;
+}
+
+
+void bpf_jit_compile(struct bpf_prog *fp)
+{
+	struct bpf_binary_header *header;
+	struct jit_ctx ctx;
+	unsigned tmp_idx;
+	unsigned alloc_size;
+	u8 *target_ptr;
+
+	if (!bpf_jit_enable)
+		return;
+
+	memset(&ctx, 0, sizeof(ctx));
+	ctx.skf		= fp;
+	ctx.ret0_fp_idx = -1;
+
+	ctx.offsets = kzalloc(4 * (ctx.skf->len + 1), GFP_KERNEL);
+	if (ctx.offsets == NULL)
+		return;
+
+	/* fake pass to fill in the ctx->seen */
+	if (unlikely(build_body(&ctx)))
+		goto out;
+
+	tmp_idx = ctx.idx;
+	build_prologue(&ctx);
+	ctx.prologue_bytes = (ctx.idx - tmp_idx) * 4;
+
+#if __LINUX_ARM_ARCH__ < 7
+	tmp_idx = ctx.idx;
+	build_epilogue(&ctx);
+	ctx.epilogue_bytes = (ctx.idx - tmp_idx) * 4;
+
+	ctx.idx += ctx.imm_count;
+	if (ctx.imm_count) {
+		ctx.imms = kzalloc(4 * ctx.imm_count, GFP_KERNEL);
+		if (ctx.imms == NULL)
+			goto out;
+	}
+#else
+	/* there's nothing after the epilogue on ARMv7 */
+	build_epilogue(&ctx);
+#endif
+	alloc_size = 4 * ctx.idx;
+	header = bpf_jit_binary_alloc(alloc_size, &target_ptr,
+				      4, jit_fill_hole);
+	if (header == NULL)
+		goto out;
+
+	ctx.target = (u32 *) target_ptr;
+	ctx.idx = 0;
+
+	build_prologue(&ctx);
+	if (build_body(&ctx) < 0) {
+#if __LINUX_ARM_ARCH__ < 7
+		if (ctx.imm_count)
+			kfree(ctx.imms);
+#endif
+		bpf_jit_binary_free(header);
+		goto out;
+	}
+	build_epilogue(&ctx);
+
+	flush_icache_range((u32)header, (u32)(ctx.target + ctx.idx));
+
+#if __LINUX_ARM_ARCH__ < 7
+	if (ctx.imm_count)
+		kfree(ctx.imms);
+#endif
+
+	if (bpf_jit_enable > 1)
+		/* there are 2 passes here */
+		bpf_jit_dump(fp->len, alloc_size, 2, ctx.target);
+
+	set_memory_ro((unsigned long)header, header->pages);
+	fp->bpf_func = (void *)ctx.target;
+	fp->jited = 1;
+out:
+	kfree(ctx.offsets);
+	return;
+}
+
+void bpf_jit_free(struct bpf_prog *fp)
+{
+	unsigned long addr = (unsigned long)fp->bpf_func & PAGE_MASK;
+	struct bpf_binary_header *header = (void *)addr;
+
+	if (!fp->jited)
+		goto free_filter;
+
+	set_memory_rw(addr, header->pages);
+	bpf_jit_binary_free(header);
+
+free_filter:
+	bpf_prog_unlock_free(fp);
+}

arch/arm/net/bpf_jit_32.h

@@ -0,0 +1,218 @@
+/*
+ * Just-In-Time compiler for BPF filters on 32bit ARM
+ *
+ * Copyright (c) 2011 Mircea Gherzan <mgherzan@gmail.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; version 2 of the License.
+ */
+
+#ifndef PFILTER_OPCODES_ARM_H
+#define PFILTER_OPCODES_ARM_H
+
+#define ARM_R0	0
+#define ARM_R1	1
+#define ARM_R2	2
+#define ARM_R3	3
+#define ARM_R4	4
+#define ARM_R5	5
+#define ARM_R6	6
+#define ARM_R7	7
+#define ARM_R8	8
+#define ARM_R9	9
+#define ARM_R10	10
+#define ARM_FP	11
+#define ARM_IP	12
+#define ARM_SP	13
+#define ARM_LR	14
+#define ARM_PC	15
+
+#define ARM_COND_EQ		0x0
+#define ARM_COND_NE		0x1
+#define ARM_COND_CS		0x2
+#define ARM_COND_HS		ARM_COND_CS
+#define ARM_COND_CC		0x3
+#define ARM_COND_LO		ARM_COND_CC
+#define ARM_COND_MI		0x4
+#define ARM_COND_PL		0x5
+#define ARM_COND_VS		0x6
+#define ARM_COND_VC		0x7
+#define ARM_COND_HI		0x8
+#define ARM_COND_LS		0x9
+#define ARM_COND_GE		0xa
+#define ARM_COND_LT		0xb
+#define ARM_COND_GT		0xc
+#define ARM_COND_LE		0xd
+#define ARM_COND_AL		0xe
+
+/* register shift types */
+#define SRTYPE_LSL		0
+#define SRTYPE_LSR		1
+#define SRTYPE_ASR		2
+#define SRTYPE_ROR		3
+
+#define ARM_INST_ADD_R		0x00800000
+#define ARM_INST_ADD_I		0x02800000
+
+#define ARM_INST_AND_R		0x00000000
+#define ARM_INST_AND_I		0x02000000
+
+#define ARM_INST_BIC_R		0x01c00000
+#define ARM_INST_BIC_I		0x03c00000
+
+#define ARM_INST_B		0x0a000000
+#define ARM_INST_BX		0x012FFF10
+#define ARM_INST_BLX_R		0x012fff30
+
+#define ARM_INST_CMP_R		0x01500000
+#define ARM_INST_CMP_I		0x03500000
+
+#define ARM_INST_EOR_R		0x00200000
+#define ARM_INST_EOR_I		0x02200000
+
+#define ARM_INST_LDRB_I		0x05d00000
+#define ARM_INST_LDRB_R		0x07d00000
+#define ARM_INST_LDRH_I		0x01d000b0
+#define ARM_INST_LDRH_R		0x019000b0
+#define ARM_INST_LDR_I		0x05900000
+
+#define ARM_INST_LDM		0x08900000
+
+#define ARM_INST_LSL_I		0x01a00000
+#define ARM_INST_LSL_R		0x01a00010
+
+#define ARM_INST_LSR_I		0x01a00020
+#define ARM_INST_LSR_R		0x01a00030
+
+#define ARM_INST_MOV_R		0x01a00000
+#define ARM_INST_MOV_I		0x03a00000
+#define ARM_INST_MOVW		0x03000000
+#define ARM_INST_MOVT		0x03400000
+
+#define ARM_INST_MUL		0x00000090
+
+#define ARM_INST_POP		0x08bd0000
+#define ARM_INST_PUSH		0x092d0000
+
+#define ARM_INST_ORR_R		0x01800000
+#define ARM_INST_ORR_I		0x03800000
+
+#define ARM_INST_REV		0x06bf0f30
+#define ARM_INST_REV16		0x06bf0fb0
+
+#define ARM_INST_RSB_I		0x02600000
+
+#define ARM_INST_SUB_R		0x00400000
+#define ARM_INST_SUB_I		0x02400000
+
+#define ARM_INST_STR_I		0x05800000
+
+#define ARM_INST_TST_R		0x01100000
+#define ARM_INST_TST_I		0x03100000
+
+#define ARM_INST_UDIV		0x0730f010
+
+#define ARM_INST_UMULL		0x00800090
+
+#define ARM_INST_MLS		0x00600090
+
+/*
+ * Use a suitable undefined instruction to use for ARM/Thumb2 faulting.
+ * We need to be careful not to conflict with those used by other modules
+ * (BUG, kprobes, etc) and the register_undef_hook() system.
+ *
+ * The ARM architecture reference manual guarantees that the following
+ * instruction space will produce an undefined instruction exception on
+ * all CPUs:
+ *
+ * ARM:   xxxx 0111 1111 xxxx xxxx xxxx 1111 xxxx	ARMv7-AR, section A5.4
+ * Thumb: 1101 1110 xxxx xxxx				ARMv7-M, section A5.2.6
+ */
+#define ARM_INST_UDF		0xe7fddef1
+
+/* register */
+#define _AL3_R(op, rd, rn, rm)	((op ## _R) | (rd) << 12 | (rn) << 16 | (rm))
+/* immediate */
+#define _AL3_I(op, rd, rn, imm)	((op ## _I) | (rd) << 12 | (rn) << 16 | (imm))
+
+#define ARM_ADD_R(rd, rn, rm)	_AL3_R(ARM_INST_ADD, rd, rn, rm)
+#define ARM_ADD_I(rd, rn, imm)	_AL3_I(ARM_INST_ADD, rd, rn, imm)
+
+#define ARM_AND_R(rd, rn, rm)	_AL3_R(ARM_INST_AND, rd, rn, rm)
+#define ARM_AND_I(rd, rn, imm)	_AL3_I(ARM_INST_AND, rd, rn, imm)
+
+#define ARM_BIC_R(rd, rn, rm)	_AL3_R(ARM_INST_BIC, rd, rn, rm)
+#define ARM_BIC_I(rd, rn, imm)	_AL3_I(ARM_INST_BIC, rd, rn, imm)
+
+#define ARM_B(imm24)		(ARM_INST_B | ((imm24) & 0xffffff))
+#define ARM_BX(rm)		(ARM_INST_BX | (rm))
+#define ARM_BLX_R(rm)		(ARM_INST_BLX_R | (rm))
+
+#define ARM_CMP_R(rn, rm)	_AL3_R(ARM_INST_CMP, 0, rn, rm)
+#define ARM_CMP_I(rn, imm)	_AL3_I(ARM_INST_CMP, 0, rn, imm)
+
+#define ARM_EOR_R(rd, rn, rm)	_AL3_R(ARM_INST_EOR, rd, rn, rm)
+#define ARM_EOR_I(rd, rn, imm)	_AL3_I(ARM_INST_EOR, rd, rn, imm)
+
+#define ARM_LDR_I(rt, rn, off)	(ARM_INST_LDR_I | (rt) << 12 | (rn) << 16 \
+				 | (off))
+#define ARM_LDRB_I(rt, rn, off)	(ARM_INST_LDRB_I | (rt) << 12 | (rn) << 16 \
+				 | (off))
+#define ARM_LDRB_R(rt, rn, rm)	(ARM_INST_LDRB_R | (rt) << 12 | (rn) << 16 \
+				 | (rm))
+#define ARM_LDRH_I(rt, rn, off)	(ARM_INST_LDRH_I | (rt) << 12 | (rn) << 16 \
+				 | (((off) & 0xf0) << 4) | ((off) & 0xf))
+#define ARM_LDRH_R(rt, rn, rm)	(ARM_INST_LDRH_R | (rt) << 12 | (rn) << 16 \
+				 | (rm))
+
+#define ARM_LDM(rn, regs)	(ARM_INST_LDM | (rn) << 16 | (regs))
+
+#define ARM_LSL_R(rd, rn, rm)	(_AL3_R(ARM_INST_LSL, rd, 0, rn) | (rm) << 8)
+#define ARM_LSL_I(rd, rn, imm)	(_AL3_I(ARM_INST_LSL, rd, 0, rn) | (imm) << 7)
+
+#define ARM_LSR_R(rd, rn, rm)	(_AL3_R(ARM_INST_LSR, rd, 0, rn) | (rm) << 8)
+#define ARM_LSR_I(rd, rn, imm)	(_AL3_I(ARM_INST_LSR, rd, 0, rn) | (imm) << 7)
+
+#define ARM_MOV_R(rd, rm)	_AL3_R(ARM_INST_MOV, rd, 0, rm)
+#define ARM_MOV_I(rd, imm)	_AL3_I(ARM_INST_MOV, rd, 0, imm)
+
+#define ARM_MOVW(rd, imm)	\
+	(ARM_INST_MOVW | ((imm) >> 12) << 16 | (rd) << 12 | ((imm) & 0x0fff))
+
+#define ARM_MOVT(rd, imm)	\
+	(ARM_INST_MOVT | ((imm) >> 12) << 16 | (rd) << 12 | ((imm) & 0x0fff))
+
+#define ARM_MUL(rd, rm, rn)	(ARM_INST_MUL | (rd) << 16 | (rm) << 8 | (rn))
+
+#define ARM_POP(regs)		(ARM_INST_POP | (regs))
+#define ARM_PUSH(regs)		(ARM_INST_PUSH | (regs))
+
+#define ARM_ORR_R(rd, rn, rm)	_AL3_R(ARM_INST_ORR, rd, rn, rm)
+#define ARM_ORR_I(rd, rn, imm)	_AL3_I(ARM_INST_ORR, rd, rn, imm)
+#define ARM_ORR_S(rd, rn, rm, type, rs)	\
+	(ARM_ORR_R(rd, rn, rm) | (type) << 5 | (rs) << 7)
+
+#define ARM_REV(rd, rm)		(ARM_INST_REV | (rd) << 12 | (rm))
+#define ARM_REV16(rd, rm)	(ARM_INST_REV16 | (rd) << 12 | (rm))
+
+#define ARM_RSB_I(rd, rn, imm)	_AL3_I(ARM_INST_RSB, rd, rn, imm)
+
+#define ARM_SUB_R(rd, rn, rm)	_AL3_R(ARM_INST_SUB, rd, rn, rm)
+#define ARM_SUB_I(rd, rn, imm)	_AL3_I(ARM_INST_SUB, rd, rn, imm)
+
+#define ARM_STR_I(rt, rn, off)	(ARM_INST_STR_I | (rt) << 12 | (rn) << 16 \
+				 | (off))
+
+#define ARM_TST_R(rn, rm)	_AL3_R(ARM_INST_TST, 0, rn, rm)
+#define ARM_TST_I(rn, imm)	_AL3_I(ARM_INST_TST, 0, rn, imm)
+
+#define ARM_UDIV(rd, rn, rm)	(ARM_INST_UDIV | (rd) << 16 | (rn) | (rm) << 8)
+
+#define ARM_UMULL(rd_lo, rd_hi, rn, rm)	(ARM_INST_UMULL | (rd_hi) << 16 \
+					 | (rd_lo) << 12 | (rm) << 8 | rn)
+
+#define ARM_MLS(rd, rn, rm, ra)	(ARM_INST_MLS | (rd) << 16 | (rn) | (rm) << 8 \
+				 | (ra) << 12)
+
+#endif /* PFILTER_OPCODES_ARM_H */

arch/arm/nwfpe/ARM-gcc.h

@@ -0,0 +1,120 @@
+/*
+-------------------------------------------------------------------------------
+The macro `BITS64' can be defined to indicate that 64-bit integer types are
+supported by the compiler.
+-------------------------------------------------------------------------------
+*/
+#define BITS64
+
+/*
+-------------------------------------------------------------------------------
+Each of the following `typedef's defines the most convenient type that holds
+integers of at least as many bits as specified.  For example, `uint8' should
+be the most convenient type that can hold unsigned integers of as many as
+8 bits.  The `flag' type must be able to hold either a 0 or 1.  For most
+implementations of C, `flag', `uint8', and `int8' should all be `typedef'ed
+to the same as `int'.
+-------------------------------------------------------------------------------
+*/
+typedef char flag;
+typedef unsigned char uint8;
+typedef signed char int8;
+typedef int uint16;
+typedef int int16;
+typedef unsigned int uint32;
+typedef signed int int32;
+#ifdef BITS64
+typedef unsigned long long int bits64;
+typedef signed long long int sbits64;
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Each of the following `typedef's defines a type that holds integers
+of _exactly_ the number of bits specified.  For instance, for most
+implementation of C, `bits16' and `sbits16' should be `typedef'ed to
+`unsigned short int' and `signed short int' (or `short int'), respectively.
+-------------------------------------------------------------------------------
+*/
+typedef unsigned char bits8;
+typedef signed char sbits8;
+typedef unsigned short int bits16;
+typedef signed short int sbits16;
+typedef unsigned int bits32;
+typedef signed int sbits32;
+#ifdef BITS64
+typedef unsigned long long int uint64;
+typedef signed long long int int64;
+#endif
+
+#ifdef BITS64
+/*
+-------------------------------------------------------------------------------
+The `LIT64' macro takes as its argument a textual integer literal and if
+necessary ``marks'' the literal as having a 64-bit integer type.  For
+example, the Gnu C Compiler (`gcc') requires that 64-bit literals be
+appended with the letters `LL' standing for `long long', which is `gcc's
+name for the 64-bit integer type.  Some compilers may allow `LIT64' to be
+defined as the identity macro:  `#define LIT64( a ) a'.
+-------------------------------------------------------------------------------
+*/
+#define LIT64( a ) a##LL
+#endif
+
+/*
+-------------------------------------------------------------------------------
+The macro `INLINE' can be used before functions that should be inlined.  If
+a compiler does not support explicit inlining, this macro should be defined
+to be `static'.
+-------------------------------------------------------------------------------
+*/
+#define INLINE static inline
+
+
+/* For use as a GCC soft-float library we need some special function names. */
+
+#ifdef __LIBFLOAT__
+
+/* Some 32-bit ops can be mapped straight across by just changing the name. */
+#define float32_add			__addsf3
+#define float32_sub			__subsf3
+#define float32_mul			__mulsf3
+#define float32_div			__divsf3
+#define int32_to_float32		__floatsisf
+#define float32_to_int32_round_to_zero	__fixsfsi
+#define float32_to_uint32_round_to_zero	__fixunssfsi
+
+/* These ones go through the glue code.  To avoid namespace pollution
+   we rename the internal functions too.  */
+#define float32_eq			___float32_eq
+#define float32_le			___float32_le
+#define float32_lt			___float32_lt
+
+/* All the 64-bit ops have to go through the glue, so we pull the same
+   trick.  */
+#define float64_add			___float64_add
+#define float64_sub			___float64_sub
+#define float64_mul			___float64_mul
+#define float64_div			___float64_div
+#define int32_to_float64		___int32_to_float64
+#define float64_to_int32_round_to_zero	___float64_to_int32_round_to_zero
+#define float64_to_uint32_round_to_zero	___float64_to_uint32_round_to_zero
+#define float64_to_float32		___float64_to_float32
+#define float32_to_float64		___float32_to_float64
+#define float64_eq			___float64_eq
+#define float64_le			___float64_le
+#define float64_lt			___float64_lt
+
+#if 0
+#define float64_add			__adddf3
+#define float64_sub			__subdf3
+#define float64_mul			__muldf3
+#define float64_div			__divdf3
+#define int32_to_float64		__floatsidf
+#define float64_to_int32_round_to_zero	__fixdfsi
+#define float64_to_uint32_round_to_zero	__fixunsdfsi
+#define float64_to_float32		__truncdfsf2
+#define float32_to_float64		__extendsfdf2
+#endif
+
+#endif

arch/arm/nwfpe/ChangeLog

@@ -0,0 +1,91 @@
+2003-03-22  Ralph Siemsen <ralphs@netwinder.org>
+	* Reformat all but softfloat files to get a consistent coding style.
+	  Used "indent -kr -i8 -ts8 -sob -l132 -ss" and a few manual fixups.
+	* Removed dead code and fixed function protypes to match definitions.
+	* Consolidated use of (opcode && MASK_ARITHMETIC_OPCODE) >> 20.
+	* Make 80-bit precision a compile-time option. (1%)
+	* Only initialize FPE state once in repeat-FP situations. (6%)
+
+2002-01-19  Russell King <rmk@arm.linux.org.uk>
+
+	* fpa11.h - Add documentation
+	          - remove userRegisters pointer from this structure.
+	          - add new method to obtain integer register values.
+	* softfloat.c - Remove float128
+	* softfloat.h - Remove float128
+	* softfloat-specialize - Remove float128
+
+	* The FPA11 structure is not a kernel-specific data structure.
+	  It is used by users of ptrace to examine the values of the
+	  floating point registers.  Therefore, any changes to the
+	  FPA11 structure (size or position of elements contained
+	  within) have to be well thought out.
+
+	* Since 128-bit float requires the FPA11 structure to change
+	  size, it has been removed.  128-bit float is currently unused,
+	  and needs various things to be re-worked so that we won't
+	  overflow the available space in the task structure.
+
+	* The changes are designed to break any patch that goes on top
+	  of this code, so that the authors properly review their changes.
+
+1999-08-19  Scott Bambrough  <scottb@netwinder.org>
+
+	* fpmodule.c - Changed version number to 0.95
+	* fpa11.h - modified FPA11, FPREG structures
+	* fpa11.c - Changes due to FPA11, FPREG structure alterations.
+	* fpa11_cpdo.c - Changes due to FPA11, FPREG structure alterations.
+	* fpa11_cpdt.c - Changes due to FPA11, FPREG structure alterations.
+	* fpa11_cprt.c - Changes due to FPA11, FPREG structure alterations.
+	* single_cpdo.c - Changes due to FPA11, FPREG structure alterations.
+	* double_cpdo.c - Changes due to FPA11, FPREG structure alterations.
+	* extended_cpdo.c - Changes due to FPA11, FPREG structure alterations.
+
+	* I discovered several bugs.  First and worst is that the kernel
+	  passes in a pointer to the FPE's state area.	This is defined
+	  as a struct user_fp (see user.h).  This pointer was cast to a
+	  FPA11*.  Unfortunately FPA11 and user_fp are of different sizes;
+	  user_fp is smaller.  This meant that the FPE scribbled on things
+	  below its area, which is bad, as the area is in the thread_struct
+	  embedded in the process task structure.  Thus we were scribbling
+	  over one of the most important structures in the entire OS.
+
+	* user_fp and FPA11 have now been harmonized.  Most of the changes
+	  in the above code were dereferencing problems due to moving the
+	  register type out of FPREG, and getting rid of the union variable
+	  fpvalue.
+
+	* Second I noticed resetFPA11 was not always being called for a
+	  task.  This should happen on the first floating point exception
+	  that occurs.	It is controlled by init_flag in FPA11.  The
+	  comment in the code beside init_flag state the kernel guarantees
+	  this to be zero.  Not so.  I found that the kernel recycles task
+	  structures, and that recycled ones may not have init_flag zeroed.
+	  I couldn't even find anything that guarantees it is zeroed when
+	  when the task structure is initially allocated.  In any case
+	  I now initialize the entire FPE state in the thread structure to
+	  zero when allocated and recycled.  See alloc_task_struct() and
+	  flush_thread() in arch/arm/process.c.  The change to
+	  alloc_task_struct() may not be necessary, but I left it in for
+	  completeness (better safe than sorry).
+
+1998-11-23  Scott Bambrough  <scottb@netwinder.org>
+
+	* README.FPE - fix typo in description of lfm/sfm instructions
+	* NOTES - Added file to describe known bugs/problems
+	* fpmodule.c - Changed version number to 0.94
+
+1998-11-20  Scott Bambrough  <scottb@netwinder.org>
+
+	* README.FPE - fix description of URD, NRM instructions
+	* TODO - remove URD, NRM instructions from TODO list
+	* single_cpdo.c - implement URD, NRM
+	* double_cpdo.c - implement URD, NRM
+	* extended_cpdo.c - implement URD, NRM
+
+1998-11-19  Scott Bambrough  <scottb@netwinder.org>
+
+	* ChangeLog - Added this file to track changes made.
+	* fpa11.c - added code to initialize register types to typeNone
+	* fpa11_cpdt.c - fixed bug in storeExtended (typeExtended changed to
+	  typeDouble in switch statement)

arch/arm/nwfpe/Makefile

@@ -0,0 +1,12 @@
+#
+# Copyright (C) 1998, 1999, 2001 Philip Blundell
+#
+
+obj-$(CONFIG_FPE_NWFPE)		+= nwfpe.o
+
+nwfpe-y				+= fpa11.o fpa11_cpdo.o fpa11_cpdt.o \
+				   fpa11_cprt.o fpmodule.o fpopcode.o \
+				   softfloat.o single_cpdo.o double_cpdo.o \
+				   entry.o
+
+nwfpe-$(CONFIG_FPE_NWFPE_XP)	+= extended_cpdo.o

arch/arm/nwfpe/double_cpdo.c

@@ -0,0 +1,167 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+#include "softfloat.h"
+#include "fpopcode.h"
+
+union float64_components {
+	float64 f64;
+	unsigned int i[2];
+};
+
+float64 float64_exp(float64 Fm);
+float64 float64_ln(float64 Fm);
+float64 float64_sin(float64 rFm);
+float64 float64_cos(float64 rFm);
+float64 float64_arcsin(float64 rFm);
+float64 float64_arctan(float64 rFm);
+float64 float64_log(float64 rFm);
+float64 float64_tan(float64 rFm);
+float64 float64_arccos(float64 rFm);
+float64 float64_pow(float64 rFn, float64 rFm);
+float64 float64_pol(float64 rFn, float64 rFm);
+
+static float64 float64_rsf(struct roundingData *roundData, float64 rFn, float64 rFm)
+{
+	return float64_sub(roundData, rFm, rFn);
+}
+
+static float64 float64_rdv(struct roundingData *roundData, float64 rFn, float64 rFm)
+{
+	return float64_div(roundData, rFm, rFn);
+}
+
+static float64 (*const dyadic_double[16])(struct roundingData*, float64 rFn, float64 rFm) = {
+	[ADF_CODE >> 20] = float64_add,
+	[MUF_CODE >> 20] = float64_mul,
+	[SUF_CODE >> 20] = float64_sub,
+	[RSF_CODE >> 20] = float64_rsf,
+	[DVF_CODE >> 20] = float64_div,
+	[RDF_CODE >> 20] = float64_rdv,
+	[RMF_CODE >> 20] = float64_rem,
+
+	/* strictly, these opcodes should not be implemented */
+	[FML_CODE >> 20] = float64_mul,
+	[FDV_CODE >> 20] = float64_div,
+	[FRD_CODE >> 20] = float64_rdv,
+};
+
+static float64 float64_mvf(struct roundingData *roundData,float64 rFm)
+{
+	return rFm;
+}
+
+static float64 float64_mnf(struct roundingData *roundData,float64 rFm)
+{
+	union float64_components u;
+
+	u.f64 = rFm;
+#ifdef __ARMEB__
+	u.i[0] ^= 0x80000000;
+#else
+	u.i[1] ^= 0x80000000;
+#endif
+
+	return u.f64;
+}
+
+static float64 float64_abs(struct roundingData *roundData,float64 rFm)
+{
+	union float64_components u;
+
+	u.f64 = rFm;
+#ifdef __ARMEB__
+	u.i[0] &= 0x7fffffff;
+#else
+	u.i[1] &= 0x7fffffff;
+#endif
+
+	return u.f64;
+}
+
+static float64 (*const monadic_double[16])(struct roundingData *, float64 rFm) = {
+	[MVF_CODE >> 20] = float64_mvf,
+	[MNF_CODE >> 20] = float64_mnf,
+	[ABS_CODE >> 20] = float64_abs,
+	[RND_CODE >> 20] = float64_round_to_int,
+	[URD_CODE >> 20] = float64_round_to_int,
+	[SQT_CODE >> 20] = float64_sqrt,
+	[NRM_CODE >> 20] = float64_mvf,
+};
+
+unsigned int DoubleCPDO(struct roundingData *roundData, const unsigned int opcode, FPREG * rFd)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	float64 rFm;
+	unsigned int Fm, opc_mask_shift;
+
+	Fm = getFm(opcode);
+	if (CONSTANT_FM(opcode)) {
+		rFm = getDoubleConstant(Fm);
+	} else {
+		switch (fpa11->fType[Fm]) {
+		case typeSingle:
+			rFm = float32_to_float64(fpa11->fpreg[Fm].fSingle);
+			break;
+
+		case typeDouble:
+			rFm = fpa11->fpreg[Fm].fDouble;
+			break;
+
+		default:
+			return 0;
+		}
+	}
+
+	opc_mask_shift = (opcode & MASK_ARITHMETIC_OPCODE) >> 20;
+	if (!MONADIC_INSTRUCTION(opcode)) {
+		unsigned int Fn = getFn(opcode);
+		float64 rFn;
+
+		switch (fpa11->fType[Fn]) {
+		case typeSingle:
+			rFn = float32_to_float64(fpa11->fpreg[Fn].fSingle);
+			break;
+
+		case typeDouble:
+			rFn = fpa11->fpreg[Fn].fDouble;
+			break;
+
+		default:
+			return 0;
+		}
+
+		if (dyadic_double[opc_mask_shift]) {
+			rFd->fDouble = dyadic_double[opc_mask_shift](roundData, rFn, rFm);
+		} else {
+			return 0;
+		}
+	} else {
+		if (monadic_double[opc_mask_shift]) {
+			rFd->fDouble = monadic_double[opc_mask_shift](roundData, rFm);
+		} else {
+			return 0;
+		}
+	}
+
+	return 1;
+}

arch/arm/nwfpe/entry.S

@@ -0,0 +1,125 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998
+    (c) 1998, 1999 Philip Blundell
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+#include <asm/assembler.h>
+#include <asm/opcodes.h>
+
+/* This is the kernel's entry point into the floating point emulator.
+It is called from the kernel with code similar to this:
+
+	sub	r4, r5, #4
+	ldrt	r0, [r4]			@ r0  = instruction
+	adrsvc	al, r9, ret_from_exception	@ r9  = normal FP return
+	adrsvc	al, lr, fpundefinstr		@ lr  = undefined instr return
+
+	get_current_task r10
+	mov	r8, #1
+	strb	r8, [r10, #TSK_USED_MATH]	@ set current->used_math
+	add	r10, r10, #TSS_FPESAVE		@ r10 = workspace
+	ldr	r4, .LC2
+	ldr	pc, [r4]			@ Call FP emulator entry point
+
+The kernel expects the emulator to return via one of two possible
+points of return it passes to the emulator.  The emulator, if
+successful in its emulation, jumps to ret_from_exception (passed in
+r9) and the kernel takes care of returning control from the trap to
+the user code.  If the emulator is unable to emulate the instruction,
+it returns via _fpundefinstr (passed via lr) and the kernel halts the
+user program with a core dump.
+
+On entry to the emulator r10 points to an area of private FP workspace
+reserved in the thread structure for this process.  This is where the
+emulator saves its registers across calls.  The first word of this area
+is used as a flag to detect the first time a process uses floating point,
+so that the emulator startup cost can be avoided for tasks that don't
+want it.
+
+This routine does three things:
+
+1) The kernel has created a struct pt_regs on the stack and saved the
+user registers into it.  See /usr/include/asm/proc/ptrace.h for details.
+
+2) It calls EmulateAll to emulate a floating point instruction.
+EmulateAll returns 1 if the emulation was successful, or 0 if not.
+
+3) If an instruction has been emulated successfully, it looks ahead at
+the next instruction.  If it is a floating point instruction, it
+executes the instruction, without returning to user space.  In this
+way it repeatedly looks ahead and executes floating point instructions
+until it encounters a non floating point instruction, at which time it
+returns via _fpreturn.
+
+This is done to reduce the effect of the trap overhead on each
+floating point instructions.  GCC attempts to group floating point
+instructions to allow the emulator to spread the cost of the trap over
+several floating point instructions.  */
+
+#include <asm/asm-offsets.h>
+
+	.globl	nwfpe_enter
+nwfpe_enter:
+	mov	r4, lr			@ save the failure-return addresses
+	mov	sl, sp			@ we access the registers via 'sl'
+
+	ldr	r5, [sp, #S_PC]		@ get contents of PC;
+	mov	r6, r0			@ save the opcode
+emulate:
+	ldr	r1, [sp, #S_PSR]	@ fetch the PSR
+	bl	arm_check_condition	@ check the condition
+	cmp	r0, #ARM_OPCODE_CONDTEST_PASS	@ condition passed?
+
+	@ if condition code failed to match, next insn
+	bne	next			@ get the next instruction;
+
+	mov	r0, r6			@ prepare for EmulateAll()
+	bl	EmulateAll		@ emulate the instruction
+	cmp	r0, #0			@ was emulation successful
+	reteq	r4			@ no, return failure
+
+next:
+	uaccess_enable r3
+.Lx1:	ldrt	r6, [r5], #4		@ get the next instruction and
+					@ increment PC
+	uaccess_disable r3
+	and	r2, r6, #0x0F000000	@ test for FP insns
+	teq	r2, #0x0C000000
+	teqne	r2, #0x0D000000
+	teqne	r2, #0x0E000000
+	retne	r9			@ return ok if not a fp insn
+
+	str	r5, [sp, #S_PC]		@ update PC copy in regs
+
+	mov	r0, r6			@ save a copy
+	b	emulate			@ check condition and emulate
+
+	@ We need to be prepared for the instructions at .Lx1 and .Lx2 
+	@ to fault.  Emit the appropriate exception gunk to fix things up.
+	@ ??? For some reason, faults can happen at .Lx2 even with a
+	@ plain LDR instruction.  Weird, but it seems harmless.
+	.pushsection .text.fixup,"ax"
+	.align	2
+.Lfix:	ret	r9			@ let the user eat segfaults
+	.popsection
+
+	.pushsection __ex_table,"a"
+	.align	3
+	.long	.Lx1, .Lfix
+	.popsection

arch/arm/nwfpe/extended_cpdo.c

@@ -0,0 +1,154 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+#include "softfloat.h"
+#include "fpopcode.h"
+
+floatx80 floatx80_exp(floatx80 Fm);
+floatx80 floatx80_ln(floatx80 Fm);
+floatx80 floatx80_sin(floatx80 rFm);
+floatx80 floatx80_cos(floatx80 rFm);
+floatx80 floatx80_arcsin(floatx80 rFm);
+floatx80 floatx80_arctan(floatx80 rFm);
+floatx80 floatx80_log(floatx80 rFm);
+floatx80 floatx80_tan(floatx80 rFm);
+floatx80 floatx80_arccos(floatx80 rFm);
+floatx80 floatx80_pow(floatx80 rFn, floatx80 rFm);
+floatx80 floatx80_pol(floatx80 rFn, floatx80 rFm);
+
+static floatx80 floatx80_rsf(struct roundingData *roundData, floatx80 rFn, floatx80 rFm)
+{
+	return floatx80_sub(roundData, rFm, rFn);
+}
+
+static floatx80 floatx80_rdv(struct roundingData *roundData, floatx80 rFn, floatx80 rFm)
+{
+	return floatx80_div(roundData, rFm, rFn);
+}
+
+static floatx80 (*const dyadic_extended[16])(struct roundingData*, floatx80 rFn, floatx80 rFm) = {
+	[ADF_CODE >> 20] = floatx80_add,
+	[MUF_CODE >> 20] = floatx80_mul,
+	[SUF_CODE >> 20] = floatx80_sub,
+	[RSF_CODE >> 20] = floatx80_rsf,
+	[DVF_CODE >> 20] = floatx80_div,
+	[RDF_CODE >> 20] = floatx80_rdv,
+	[RMF_CODE >> 20] = floatx80_rem,
+
+	/* strictly, these opcodes should not be implemented */
+	[FML_CODE >> 20] = floatx80_mul,
+	[FDV_CODE >> 20] = floatx80_div,
+	[FRD_CODE >> 20] = floatx80_rdv,
+};
+
+static floatx80 floatx80_mvf(struct roundingData *roundData, floatx80 rFm)
+{
+	return rFm;
+}
+
+static floatx80 floatx80_mnf(struct roundingData *roundData, floatx80 rFm)
+{
+	rFm.high ^= 0x8000;
+	return rFm;
+}
+
+static floatx80 floatx80_abs(struct roundingData *roundData, floatx80 rFm)
+{
+	rFm.high &= 0x7fff;
+	return rFm;
+}
+
+static floatx80 (*const monadic_extended[16])(struct roundingData*, floatx80 rFm) = {
+	[MVF_CODE >> 20] = floatx80_mvf,
+	[MNF_CODE >> 20] = floatx80_mnf,
+	[ABS_CODE >> 20] = floatx80_abs,
+	[RND_CODE >> 20] = floatx80_round_to_int,
+	[URD_CODE >> 20] = floatx80_round_to_int,
+	[SQT_CODE >> 20] = floatx80_sqrt,
+	[NRM_CODE >> 20] = floatx80_mvf,
+};
+
+unsigned int ExtendedCPDO(struct roundingData *roundData, const unsigned int opcode, FPREG * rFd)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	floatx80 rFm;
+	unsigned int Fm, opc_mask_shift;
+
+	Fm = getFm(opcode);
+	if (CONSTANT_FM(opcode)) {
+		rFm = getExtendedConstant(Fm);
+	} else {
+		switch (fpa11->fType[Fm]) {
+		case typeSingle:
+			rFm = float32_to_floatx80(fpa11->fpreg[Fm].fSingle);
+			break;
+
+		case typeDouble:
+			rFm = float64_to_floatx80(fpa11->fpreg[Fm].fDouble);
+			break;
+
+		case typeExtended:
+			rFm = fpa11->fpreg[Fm].fExtended;
+			break;
+
+		default:
+			return 0;
+		}
+	}
+
+	opc_mask_shift = (opcode & MASK_ARITHMETIC_OPCODE) >> 20;
+	if (!MONADIC_INSTRUCTION(opcode)) {
+		unsigned int Fn = getFn(opcode);
+		floatx80 rFn;
+
+		switch (fpa11->fType[Fn]) {
+		case typeSingle:
+			rFn = float32_to_floatx80(fpa11->fpreg[Fn].fSingle);
+			break;
+
+		case typeDouble:
+			rFn = float64_to_floatx80(fpa11->fpreg[Fn].fDouble);
+			break;
+
+		case typeExtended:
+			rFn = fpa11->fpreg[Fn].fExtended;
+			break;
+
+		default:
+			return 0;
+		}
+
+		if (dyadic_extended[opc_mask_shift]) {
+			rFd->fExtended = dyadic_extended[opc_mask_shift](roundData, rFn, rFm);
+		} else {
+			return 0;
+		}
+	} else {
+		if (monadic_extended[opc_mask_shift]) {
+			rFd->fExtended = monadic_extended[opc_mask_shift](roundData, rFm);
+		} else {
+			return 0;
+		}
+	}
+
+	return 1;
+}

arch/arm/nwfpe/fpa11.c

@@ -0,0 +1,128 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+    (c) Philip Blundell, 2001
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+#include "fpopcode.h"
+
+#include "fpmodule.h"
+#include "fpmodule.inl"
+
+#include <linux/compiler.h>
+#include <linux/string.h>
+
+/* Reset the FPA11 chip.  Called to initialize and reset the emulator. */
+static void resetFPA11(void)
+{
+	int i;
+	FPA11 *fpa11 = GET_FPA11();
+
+	/* initialize the register type array */
+	for (i = 0; i <= 7; i++) {
+		fpa11->fType[i] = typeNone;
+	}
+
+	/* FPSR: set system id to FP_EMULATOR, set AC, clear all other bits */
+	fpa11->fpsr = FP_EMULATOR | BIT_AC;
+}
+
+int8 SetRoundingMode(const unsigned int opcode)
+{
+	switch (opcode & MASK_ROUNDING_MODE) {
+	default:
+	case ROUND_TO_NEAREST:
+		return float_round_nearest_even;
+
+	case ROUND_TO_PLUS_INFINITY:
+		return float_round_up;
+
+	case ROUND_TO_MINUS_INFINITY:
+		return float_round_down;
+
+	case ROUND_TO_ZERO:
+		return float_round_to_zero;
+	}
+}
+
+int8 SetRoundingPrecision(const unsigned int opcode)
+{
+#ifdef CONFIG_FPE_NWFPE_XP
+	switch (opcode & MASK_ROUNDING_PRECISION) {
+	case ROUND_SINGLE:
+		return 32;
+
+	case ROUND_DOUBLE:
+		return 64;
+
+	case ROUND_EXTENDED:
+		return 80;
+
+	default:
+		return 80;
+	}
+#endif
+	return 80;
+}
+
+void nwfpe_init_fpa(union fp_state *fp)
+{
+	FPA11 *fpa11 = (FPA11 *)fp;
+#ifdef NWFPE_DEBUG
+	printk("NWFPE: setting up state.\n");
+#endif
+ 	memset(fpa11, 0, sizeof(FPA11));
+	resetFPA11();
+	fpa11->initflag = 1;
+}
+
+/* Emulate the instruction in the opcode. */
+unsigned int EmulateAll(unsigned int opcode)
+{
+	unsigned int code;
+
+#ifdef NWFPE_DEBUG
+	printk("NWFPE: emulating opcode %08x\n", opcode);
+#endif
+	code = opcode & 0x00000f00;
+	if (code == 0x00000100 || code == 0x00000200) {
+		/* For coprocessor 1 or 2 (FPA11) */
+		code = opcode & 0x0e000000;
+		if (code == 0x0e000000) {
+			if (opcode & 0x00000010) {
+				/* Emulate conversion opcodes. */
+				/* Emulate register transfer opcodes. */
+				/* Emulate comparison opcodes. */
+				return EmulateCPRT(opcode);
+			} else {
+				/* Emulate monadic arithmetic opcodes. */
+				/* Emulate dyadic arithmetic opcodes. */
+				return EmulateCPDO(opcode);
+			}
+		} else if (code == 0x0c000000) {
+			/* Emulate load/store opcodes. */
+			/* Emulate load/store multiple opcodes. */
+			return EmulateCPDT(opcode);
+		}
+	}
+
+	/* Invalid instruction detected.  Return FALSE. */
+	return 0;
+}

arch/arm/nwfpe/fpa11.h

@@ -0,0 +1,121 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.com, 1998-1999
+    
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#ifndef __FPA11_H__
+#define __FPA11_H__
+
+#define GET_FPA11() ((FPA11 *)(&current_thread_info()->fpstate))
+
+/*
+ * The processes registers are always at the very top of the 8K
+ * stack+task struct.  Use the same method as 'current' uses to
+ * reach them.
+ */
+#define GET_USERREG() ((struct pt_regs *)(THREAD_START_SP + (unsigned long)current_thread_info()) - 1)
+
+#include <linux/thread_info.h>
+
+/* includes */
+#include "fpsr.h"		/* FP control and status register definitions */
+#include "milieu.h"
+
+struct roundingData {
+    int8 mode;
+    int8 precision;
+    signed char exception;
+};
+
+#include "softfloat.h"
+
+#define		typeNone		0x00
+#define		typeSingle		0x01
+#define		typeDouble		0x02
+#define		typeExtended		0x03
+
+/*
+ * This must be no more and no less than 12 bytes.
+ */
+typedef union tagFPREG {
+	float32 fSingle;
+	float64 fDouble;
+#ifdef CONFIG_FPE_NWFPE_XP
+	floatx80 fExtended;
+#else
+	u32 padding[3];
+#endif
+} __attribute__ ((packed,aligned(4))) FPREG;
+
+/*
+ * FPA11 device model.
+ *
+ * This structure is exported to user space.  Do not re-order.
+ * Only add new stuff to the end, and do not change the size of
+ * any element.  Elements of this structure are used by user
+ * space, and must match struct user_fp in <asm/user.h>.
+ * We include the byte offsets below for documentation purposes.
+ *
+ * The size of this structure and FPREG are checked by fpmodule.c
+ * on initialisation.  If the rules have been broken, NWFPE will
+ * not initialise.
+ */
+typedef struct tagFPA11 {
+/*   0 */ FPREG fpreg[8];	/* 8 floating point registers */
+/*  96 */ FPSR fpsr;		/* floating point status register */
+/* 100 */ FPCR fpcr;		/* floating point control register */
+/* 104 */ unsigned char fType[8];	/* type of floating point value held in
+					   floating point registers.  One of
+					   none, single, double or extended. */
+/* 112 */ int initflag;		/* this is special.  The kernel guarantees
+				   to set it to 0 when a thread is launched,
+				   so we can use it to detect whether this
+				   instance of the emulator needs to be
+				   initialised. */
+} __attribute__ ((packed,aligned(4))) FPA11;
+
+extern int8 SetRoundingMode(const unsigned int);
+extern int8 SetRoundingPrecision(const unsigned int);
+extern void nwfpe_init_fpa(union fp_state *fp);
+
+extern unsigned int EmulateAll(unsigned int opcode);
+
+extern unsigned int EmulateCPDT(const unsigned int opcode);
+extern unsigned int EmulateCPDO(const unsigned int opcode);
+extern unsigned int EmulateCPRT(const unsigned int opcode);
+
+/* fpa11_cpdt.c */
+extern unsigned int PerformLDF(const unsigned int opcode);
+extern unsigned int PerformSTF(const unsigned int opcode);
+extern unsigned int PerformLFM(const unsigned int opcode);
+extern unsigned int PerformSFM(const unsigned int opcode);
+
+/* single_cpdo.c */
+
+extern unsigned int SingleCPDO(struct roundingData *roundData,
+			       const unsigned int opcode, FPREG * rFd);
+/* double_cpdo.c */
+extern unsigned int DoubleCPDO(struct roundingData *roundData,
+			       const unsigned int opcode, FPREG * rFd);
+
+/* extneded_cpdo.c */
+extern unsigned int ExtendedCPDO(struct roundingData *roundData,
+				 const unsigned int opcode, FPREG * rFd);
+
+#endif

arch/arm/nwfpe/fpa11.inl

@@ -0,0 +1,51 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+
+/* Read and write floating point status register */
+static inline unsigned int readFPSR(void)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	return (fpa11->fpsr);
+}
+
+static inline void writeFPSR(FPSR reg)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	/* the sysid byte in the status register is readonly */
+	fpa11->fpsr = (fpa11->fpsr & MASK_SYSID) | (reg & ~MASK_SYSID);
+}
+
+/* Read and write floating point control register */
+static inline FPCR readFPCR(void)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	/* clear SB, AB and DA bits before returning FPCR */
+	return (fpa11->fpcr & ~MASK_RFC);
+}
+
+static inline void writeFPCR(FPCR reg)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	fpa11->fpcr &= ~MASK_WFC;		/* clear SB, AB and DA bits */
+	fpa11->fpcr |= (reg & MASK_WFC);	/* write SB, AB and DA bits */
+}

arch/arm/nwfpe/fpa11_cpdo.c

@@ -0,0 +1,137 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+    (c) Philip Blundell, 2001
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+#include "fpopcode.h"
+
+unsigned int SingleCPDO(struct roundingData *roundData, const unsigned int opcode, FPREG * rFd);
+unsigned int DoubleCPDO(struct roundingData *roundData, const unsigned int opcode, FPREG * rFd);
+unsigned int ExtendedCPDO(struct roundingData *roundData, const unsigned int opcode, FPREG * rFd);
+
+unsigned int EmulateCPDO(const unsigned int opcode)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	FPREG *rFd;
+	unsigned int nType, nDest, nRc;
+	struct roundingData roundData;
+
+	/* Get the destination size.  If not valid let Linux perform
+	   an invalid instruction trap. */
+	nDest = getDestinationSize(opcode);
+	if (typeNone == nDest)
+		return 0;
+
+	roundData.mode = SetRoundingMode(opcode);
+	roundData.precision = SetRoundingPrecision(opcode);
+	roundData.exception = 0;
+
+	/* Compare the size of the operands in Fn and Fm.
+	   Choose the largest size and perform operations in that size,
+	   in order to make use of all the precision of the operands.
+	   If Fm is a constant, we just grab a constant of a size
+	   matching the size of the operand in Fn. */
+	if (MONADIC_INSTRUCTION(opcode))
+		nType = nDest;
+	else
+		nType = fpa11->fType[getFn(opcode)];
+
+	if (!CONSTANT_FM(opcode)) {
+		register unsigned int Fm = getFm(opcode);
+		if (nType < fpa11->fType[Fm]) {
+			nType = fpa11->fType[Fm];
+		}
+	}
+
+	rFd = &fpa11->fpreg[getFd(opcode)];
+
+	switch (nType) {
+	case typeSingle:
+		nRc = SingleCPDO(&roundData, opcode, rFd);
+		break;
+	case typeDouble:
+		nRc = DoubleCPDO(&roundData, opcode, rFd);
+		break;
+#ifdef CONFIG_FPE_NWFPE_XP
+	case typeExtended:
+		nRc = ExtendedCPDO(&roundData, opcode, rFd);
+		break;
+#endif
+	default:
+		nRc = 0;
+	}
+
+	/* The CPDO functions used to always set the destination type
+	   to be the same as their working size. */
+
+	if (nRc != 0) {
+		/* If the operation succeeded, check to see if the result in the
+		   destination register is the correct size.  If not force it
+		   to be. */
+
+		fpa11->fType[getFd(opcode)] = nDest;
+
+#ifdef CONFIG_FPE_NWFPE_XP
+		if (nDest != nType) {
+			switch (nDest) {
+			case typeSingle:
+				{
+					if (typeDouble == nType)
+						rFd->fSingle = float64_to_float32(&roundData, rFd->fDouble);
+					else
+						rFd->fSingle = floatx80_to_float32(&roundData, rFd->fExtended);
+				}
+				break;
+
+			case typeDouble:
+				{
+					if (typeSingle == nType)
+						rFd->fDouble = float32_to_float64(rFd->fSingle);
+					else
+						rFd->fDouble = floatx80_to_float64(&roundData, rFd->fExtended);
+				}
+				break;
+
+			case typeExtended:
+				{
+					if (typeSingle == nType)
+						rFd->fExtended = float32_to_floatx80(rFd->fSingle);
+					else
+						rFd->fExtended = float64_to_floatx80(rFd->fDouble);
+				}
+				break;
+			}
+		}
+#else
+		if (nDest != nType) {
+			if (nDest == typeSingle)
+				rFd->fSingle = float64_to_float32(&roundData, rFd->fDouble);
+			else
+				rFd->fDouble = float32_to_float64(rFd->fSingle);
+		}
+#endif
+	}
+
+	if (roundData.exception)
+		float_raise(roundData.exception);
+
+	return nRc;
+}

arch/arm/nwfpe/fpa11_cpdt.c

@@ -0,0 +1,407 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.com, 1998-1999
+    (c) Philip Blundell, 1998, 2001
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+#include "softfloat.h"
+#include "fpopcode.h"
+#include "fpmodule.h"
+#include "fpmodule.inl"
+
+#include <linux/uaccess.h>
+
+static inline void loadSingle(const unsigned int Fn, const unsigned int __user *pMem)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	fpa11->fType[Fn] = typeSingle;
+	get_user(fpa11->fpreg[Fn].fSingle, pMem);
+}
+
+static inline void loadDouble(const unsigned int Fn, const unsigned int __user *pMem)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	unsigned int *p;
+	p = (unsigned int *) &fpa11->fpreg[Fn].fDouble;
+	fpa11->fType[Fn] = typeDouble;
+#ifdef __ARMEB__
+	get_user(p[0], &pMem[0]);	/* sign & exponent */
+	get_user(p[1], &pMem[1]);
+#else
+	get_user(p[0], &pMem[1]);
+	get_user(p[1], &pMem[0]);	/* sign & exponent */
+#endif
+}
+
+#ifdef CONFIG_FPE_NWFPE_XP
+static inline void loadExtended(const unsigned int Fn, const unsigned int __user *pMem)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	unsigned int *p;
+	p = (unsigned int *) &fpa11->fpreg[Fn].fExtended;
+	fpa11->fType[Fn] = typeExtended;
+	get_user(p[0], &pMem[0]);	/* sign & exponent */
+#ifdef __ARMEB__
+	get_user(p[1], &pMem[1]);	/* ms bits */
+	get_user(p[2], &pMem[2]);	/* ls bits */
+#else
+	get_user(p[1], &pMem[2]);	/* ls bits */
+	get_user(p[2], &pMem[1]);	/* ms bits */
+#endif
+}
+#endif
+
+static inline void loadMultiple(const unsigned int Fn, const unsigned int __user *pMem)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	register unsigned int *p;
+	unsigned long x;
+
+	p = (unsigned int *) &(fpa11->fpreg[Fn]);
+	get_user(x, &pMem[0]);
+	fpa11->fType[Fn] = (x >> 14) & 0x00000003;
+
+	switch (fpa11->fType[Fn]) {
+	case typeSingle:
+	case typeDouble:
+		{
+			get_user(p[0], &pMem[2]);	/* Single */
+			get_user(p[1], &pMem[1]);	/* double msw */
+			p[2] = 0;			/* empty */
+		}
+		break;
+
+#ifdef CONFIG_FPE_NWFPE_XP
+	case typeExtended:
+		{
+			get_user(p[1], &pMem[2]);
+			get_user(p[2], &pMem[1]);	/* msw */
+			p[0] = (x & 0x80003fff);
+		}
+		break;
+#endif
+	}
+}
+
+static inline void storeSingle(struct roundingData *roundData, const unsigned int Fn, unsigned int __user *pMem)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	union {
+		float32 f;
+		unsigned int i[1];
+	} val;
+
+	switch (fpa11->fType[Fn]) {
+	case typeDouble:
+		val.f = float64_to_float32(roundData, fpa11->fpreg[Fn].fDouble);
+		break;
+
+#ifdef CONFIG_FPE_NWFPE_XP
+	case typeExtended:
+		val.f = floatx80_to_float32(roundData, fpa11->fpreg[Fn].fExtended);
+		break;
+#endif
+
+	default:
+		val.f = fpa11->fpreg[Fn].fSingle;
+	}
+
+	put_user(val.i[0], pMem);
+}
+
+static inline void storeDouble(struct roundingData *roundData, const unsigned int Fn, unsigned int __user *pMem)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	union {
+		float64 f;
+		unsigned int i[2];
+	} val;
+
+	switch (fpa11->fType[Fn]) {
+	case typeSingle:
+		val.f = float32_to_float64(fpa11->fpreg[Fn].fSingle);
+		break;
+
+#ifdef CONFIG_FPE_NWFPE_XP
+	case typeExtended:
+		val.f = floatx80_to_float64(roundData, fpa11->fpreg[Fn].fExtended);
+		break;
+#endif
+
+	default:
+		val.f = fpa11->fpreg[Fn].fDouble;
+	}
+
+#ifdef __ARMEB__
+	put_user(val.i[0], &pMem[0]);	/* msw */
+	put_user(val.i[1], &pMem[1]);	/* lsw */
+#else
+	put_user(val.i[1], &pMem[0]);	/* msw */
+	put_user(val.i[0], &pMem[1]);	/* lsw */
+#endif
+}
+
+#ifdef CONFIG_FPE_NWFPE_XP
+static inline void storeExtended(const unsigned int Fn, unsigned int __user *pMem)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	union {
+		floatx80 f;
+		unsigned int i[3];
+	} val;
+
+	switch (fpa11->fType[Fn]) {
+	case typeSingle:
+		val.f = float32_to_floatx80(fpa11->fpreg[Fn].fSingle);
+		break;
+
+	case typeDouble:
+		val.f = float64_to_floatx80(fpa11->fpreg[Fn].fDouble);
+		break;
+
+	default:
+		val.f = fpa11->fpreg[Fn].fExtended;
+	}
+
+	put_user(val.i[0], &pMem[0]);	/* sign & exp */
+#ifdef __ARMEB__
+	put_user(val.i[1], &pMem[1]);	/* msw */
+	put_user(val.i[2], &pMem[2]);
+#else
+	put_user(val.i[1], &pMem[2]);
+	put_user(val.i[2], &pMem[1]);	/* msw */
+#endif
+}
+#endif
+
+static inline void storeMultiple(const unsigned int Fn, unsigned int __user *pMem)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	register unsigned int nType, *p;
+
+	p = (unsigned int *) &(fpa11->fpreg[Fn]);
+	nType = fpa11->fType[Fn];
+
+	switch (nType) {
+	case typeSingle:
+	case typeDouble:
+		{
+			put_user(p[0], &pMem[2]);	/* single */
+			put_user(p[1], &pMem[1]);	/* double msw */
+			put_user(nType << 14, &pMem[0]);
+		}
+		break;
+
+#ifdef CONFIG_FPE_NWFPE_XP
+	case typeExtended:
+		{
+			put_user(p[2], &pMem[1]);	/* msw */
+			put_user(p[1], &pMem[2]);
+			put_user((p[0] & 0x80003fff) | (nType << 14), &pMem[0]);
+		}
+		break;
+#endif
+	}
+}
+
+unsigned int PerformLDF(const unsigned int opcode)
+{
+	unsigned int __user *pBase, *pAddress, *pFinal;
+	unsigned int nRc = 1, write_back = WRITE_BACK(opcode);
+
+	pBase = (unsigned int __user *) readRegister(getRn(opcode));
+	if (REG_PC == getRn(opcode)) {
+		pBase += 2;
+		write_back = 0;
+	}
+
+	pFinal = pBase;
+	if (BIT_UP_SET(opcode))
+		pFinal += getOffset(opcode);
+	else
+		pFinal -= getOffset(opcode);
+
+	if (PREINDEXED(opcode))
+		pAddress = pFinal;
+	else
+		pAddress = pBase;
+
+	switch (opcode & MASK_TRANSFER_LENGTH) {
+	case TRANSFER_SINGLE:
+		loadSingle(getFd(opcode), pAddress);
+		break;
+	case TRANSFER_DOUBLE:
+		loadDouble(getFd(opcode), pAddress);
+		break;
+#ifdef CONFIG_FPE_NWFPE_XP
+	case TRANSFER_EXTENDED:
+		loadExtended(getFd(opcode), pAddress);
+		break;
+#endif
+	default:
+		nRc = 0;
+	}
+
+	if (write_back)
+		writeRegister(getRn(opcode), (unsigned long) pFinal);
+	return nRc;
+}
+
+unsigned int PerformSTF(const unsigned int opcode)
+{
+	unsigned int __user *pBase, *pAddress, *pFinal;
+	unsigned int nRc = 1, write_back = WRITE_BACK(opcode);
+	struct roundingData roundData;
+
+	roundData.mode = SetRoundingMode(opcode);
+	roundData.precision = SetRoundingPrecision(opcode);
+	roundData.exception = 0;
+
+	pBase = (unsigned int __user *) readRegister(getRn(opcode));
+	if (REG_PC == getRn(opcode)) {
+		pBase += 2;
+		write_back = 0;
+	}
+
+	pFinal = pBase;
+	if (BIT_UP_SET(opcode))
+		pFinal += getOffset(opcode);
+	else
+		pFinal -= getOffset(opcode);
+
+	if (PREINDEXED(opcode))
+		pAddress = pFinal;
+	else
+		pAddress = pBase;
+
+	switch (opcode & MASK_TRANSFER_LENGTH) {
+	case TRANSFER_SINGLE:
+		storeSingle(&roundData, getFd(opcode), pAddress);
+		break;
+	case TRANSFER_DOUBLE:
+		storeDouble(&roundData, getFd(opcode), pAddress);
+		break;
+#ifdef CONFIG_FPE_NWFPE_XP
+	case TRANSFER_EXTENDED:
+		storeExtended(getFd(opcode), pAddress);
+		break;
+#endif
+	default:
+		nRc = 0;
+	}
+
+	if (roundData.exception)
+		float_raise(roundData.exception);
+
+	if (write_back)
+		writeRegister(getRn(opcode), (unsigned long) pFinal);
+	return nRc;
+}
+
+unsigned int PerformLFM(const unsigned int opcode)
+{
+	unsigned int __user *pBase, *pAddress, *pFinal;
+	unsigned int i, Fd, write_back = WRITE_BACK(opcode);
+
+	pBase = (unsigned int __user *) readRegister(getRn(opcode));
+	if (REG_PC == getRn(opcode)) {
+		pBase += 2;
+		write_back = 0;
+	}
+
+	pFinal = pBase;
+	if (BIT_UP_SET(opcode))
+		pFinal += getOffset(opcode);
+	else
+		pFinal -= getOffset(opcode);
+
+	if (PREINDEXED(opcode))
+		pAddress = pFinal;
+	else
+		pAddress = pBase;
+
+	Fd = getFd(opcode);
+	for (i = getRegisterCount(opcode); i > 0; i--) {
+		loadMultiple(Fd, pAddress);
+		pAddress += 3;
+		Fd++;
+		if (Fd == 8)
+			Fd = 0;
+	}
+
+	if (write_back)
+		writeRegister(getRn(opcode), (unsigned long) pFinal);
+	return 1;
+}
+
+unsigned int PerformSFM(const unsigned int opcode)
+{
+	unsigned int __user *pBase, *pAddress, *pFinal;
+	unsigned int i, Fd, write_back = WRITE_BACK(opcode);
+
+	pBase = (unsigned int __user *) readRegister(getRn(opcode));
+	if (REG_PC == getRn(opcode)) {
+		pBase += 2;
+		write_back = 0;
+	}
+
+	pFinal = pBase;
+	if (BIT_UP_SET(opcode))
+		pFinal += getOffset(opcode);
+	else
+		pFinal -= getOffset(opcode);
+
+	if (PREINDEXED(opcode))
+		pAddress = pFinal;
+	else
+		pAddress = pBase;
+
+	Fd = getFd(opcode);
+	for (i = getRegisterCount(opcode); i > 0; i--) {
+		storeMultiple(Fd, pAddress);
+		pAddress += 3;
+		Fd++;
+		if (Fd == 8)
+			Fd = 0;
+	}
+
+	if (write_back)
+		writeRegister(getRn(opcode), (unsigned long) pFinal);
+	return 1;
+}
+
+unsigned int EmulateCPDT(const unsigned int opcode)
+{
+	unsigned int nRc = 0;
+
+	if (LDF_OP(opcode)) {
+		nRc = PerformLDF(opcode);
+	} else if (LFM_OP(opcode)) {
+		nRc = PerformLFM(opcode);
+	} else if (STF_OP(opcode)) {
+		nRc = PerformSTF(opcode);
+	} else if (SFM_OP(opcode)) {
+		nRc = PerformSFM(opcode);
+	} else {
+		nRc = 0;
+	}
+
+	return nRc;
+}

arch/arm/nwfpe/fpa11_cprt.c

@@ -0,0 +1,373 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+    (c) Philip Blundell, 1999, 2001
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+#include "fpopcode.h"
+#include "fpa11.inl"
+#include "fpmodule.h"
+#include "fpmodule.inl"
+#include "softfloat.h"
+
+unsigned int PerformFLT(const unsigned int opcode);
+unsigned int PerformFIX(const unsigned int opcode);
+
+static unsigned int PerformComparison(const unsigned int opcode);
+
+unsigned int EmulateCPRT(const unsigned int opcode)
+{
+
+	if (opcode & 0x800000) {
+		/* This is some variant of a comparison (PerformComparison
+		   will sort out which one).  Since most of the other CPRT
+		   instructions are oddball cases of some sort or other it
+		   makes sense to pull this out into a fast path.  */
+		return PerformComparison(opcode);
+	}
+
+	/* Hint to GCC that we'd like a jump table rather than a load of CMPs */
+	switch ((opcode & 0x700000) >> 20) {
+	case FLT_CODE >> 20:
+		return PerformFLT(opcode);
+		break;
+	case FIX_CODE >> 20:
+		return PerformFIX(opcode);
+		break;
+
+	case WFS_CODE >> 20:
+		writeFPSR(readRegister(getRd(opcode)));
+		break;
+	case RFS_CODE >> 20:
+		writeRegister(getRd(opcode), readFPSR());
+		break;
+
+	default:
+		return 0;
+	}
+
+	return 1;
+}
+
+unsigned int PerformFLT(const unsigned int opcode)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	struct roundingData roundData;
+
+	roundData.mode = SetRoundingMode(opcode);
+	roundData.precision = SetRoundingPrecision(opcode);
+	roundData.exception = 0;
+
+	switch (opcode & MASK_ROUNDING_PRECISION) {
+	case ROUND_SINGLE:
+		{
+			fpa11->fType[getFn(opcode)] = typeSingle;
+			fpa11->fpreg[getFn(opcode)].fSingle = int32_to_float32(&roundData, readRegister(getRd(opcode)));
+		}
+		break;
+
+	case ROUND_DOUBLE:
+		{
+			fpa11->fType[getFn(opcode)] = typeDouble;
+			fpa11->fpreg[getFn(opcode)].fDouble = int32_to_float64(readRegister(getRd(opcode)));
+		}
+		break;
+
+#ifdef CONFIG_FPE_NWFPE_XP
+	case ROUND_EXTENDED:
+		{
+			fpa11->fType[getFn(opcode)] = typeExtended;
+			fpa11->fpreg[getFn(opcode)].fExtended = int32_to_floatx80(readRegister(getRd(opcode)));
+		}
+		break;
+#endif
+
+	default:
+		return 0;
+	}
+
+	if (roundData.exception)
+		float_raise(roundData.exception);
+
+	return 1;
+}
+
+unsigned int PerformFIX(const unsigned int opcode)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	unsigned int Fn = getFm(opcode);
+	struct roundingData roundData;
+
+	roundData.mode = SetRoundingMode(opcode);
+	roundData.precision = SetRoundingPrecision(opcode);
+	roundData.exception = 0;
+
+	switch (fpa11->fType[Fn]) {
+	case typeSingle:
+		{
+			writeRegister(getRd(opcode), float32_to_int32(&roundData, fpa11->fpreg[Fn].fSingle));
+		}
+		break;
+
+	case typeDouble:
+		{
+			writeRegister(getRd(opcode), float64_to_int32(&roundData, fpa11->fpreg[Fn].fDouble));
+		}
+		break;
+
+#ifdef CONFIG_FPE_NWFPE_XP
+	case typeExtended:
+		{
+			writeRegister(getRd(opcode), floatx80_to_int32(&roundData, fpa11->fpreg[Fn].fExtended));
+		}
+		break;
+#endif
+
+	default:
+		return 0;
+	}
+
+	if (roundData.exception)
+		float_raise(roundData.exception);
+
+	return 1;
+}
+
+/* This instruction sets the flags N, Z, C, V in the FPSR. */
+static unsigned int PerformComparison(const unsigned int opcode)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	unsigned int Fn = getFn(opcode), Fm = getFm(opcode);
+	int e_flag = opcode & 0x400000;	/* 1 if CxFE */
+	int n_flag = opcode & 0x200000;	/* 1 if CNxx */
+	unsigned int flags = 0;
+
+#ifdef CONFIG_FPE_NWFPE_XP
+	floatx80 rFn, rFm;
+
+	/* Check for unordered condition and convert all operands to 80-bit
+	   format.
+	   ?? Might be some mileage in avoiding this conversion if possible.
+	   Eg, if both operands are 32-bit, detect this and do a 32-bit
+	   comparison (cheaper than an 80-bit one).  */
+	switch (fpa11->fType[Fn]) {
+	case typeSingle:
+		//printk("single.\n");
+		if (float32_is_nan(fpa11->fpreg[Fn].fSingle))
+			goto unordered;
+		rFn = float32_to_floatx80(fpa11->fpreg[Fn].fSingle);
+		break;
+
+	case typeDouble:
+		//printk("double.\n");
+		if (float64_is_nan(fpa11->fpreg[Fn].fDouble))
+			goto unordered;
+		rFn = float64_to_floatx80(fpa11->fpreg[Fn].fDouble);
+		break;
+
+	case typeExtended:
+		//printk("extended.\n");
+		if (floatx80_is_nan(fpa11->fpreg[Fn].fExtended))
+			goto unordered;
+		rFn = fpa11->fpreg[Fn].fExtended;
+		break;
+
+	default:
+		return 0;
+	}
+
+	if (CONSTANT_FM(opcode)) {
+		//printk("Fm is a constant: #%d.\n",Fm);
+		rFm = getExtendedConstant(Fm);
+		if (floatx80_is_nan(rFm))
+			goto unordered;
+	} else {
+		//printk("Fm = r%d which contains a ",Fm);
+		switch (fpa11->fType[Fm]) {
+		case typeSingle:
+			//printk("single.\n");
+			if (float32_is_nan(fpa11->fpreg[Fm].fSingle))
+				goto unordered;
+			rFm = float32_to_floatx80(fpa11->fpreg[Fm].fSingle);
+			break;
+
+		case typeDouble:
+			//printk("double.\n");
+			if (float64_is_nan(fpa11->fpreg[Fm].fDouble))
+				goto unordered;
+			rFm = float64_to_floatx80(fpa11->fpreg[Fm].fDouble);
+			break;
+
+		case typeExtended:
+			//printk("extended.\n");
+			if (floatx80_is_nan(fpa11->fpreg[Fm].fExtended))
+				goto unordered;
+			rFm = fpa11->fpreg[Fm].fExtended;
+			break;
+
+		default:
+			return 0;
+		}
+	}
+
+	if (n_flag)
+		rFm.high ^= 0x8000;
+
+	/* test for less than condition */
+	if (floatx80_lt(rFn, rFm))
+		flags |= CC_NEGATIVE;
+
+	/* test for equal condition */
+	if (floatx80_eq(rFn, rFm))
+		flags |= CC_ZERO;
+
+	/* test for greater than or equal condition */
+	if (floatx80_lt(rFm, rFn))
+		flags |= CC_CARRY;
+
+#else
+	if (CONSTANT_FM(opcode)) {
+		/* Fm is a constant.  Do the comparison in whatever precision
+		   Fn happens to be stored in.  */
+		if (fpa11->fType[Fn] == typeSingle) {
+			float32 rFm = getSingleConstant(Fm);
+			float32 rFn = fpa11->fpreg[Fn].fSingle;
+
+			if (float32_is_nan(rFn))
+				goto unordered;
+
+			if (n_flag)
+				rFm ^= 0x80000000;
+
+			/* test for less than condition */
+			if (float32_lt_nocheck(rFn, rFm))
+				flags |= CC_NEGATIVE;
+
+			/* test for equal condition */
+			if (float32_eq_nocheck(rFn, rFm))
+				flags |= CC_ZERO;
+
+			/* test for greater than or equal condition */
+			if (float32_lt_nocheck(rFm, rFn))
+				flags |= CC_CARRY;
+		} else {
+			float64 rFm = getDoubleConstant(Fm);
+			float64 rFn = fpa11->fpreg[Fn].fDouble;
+
+			if (float64_is_nan(rFn))
+				goto unordered;
+
+			if (n_flag)
+				rFm ^= 0x8000000000000000ULL;
+
+			/* test for less than condition */
+			if (float64_lt_nocheck(rFn, rFm))
+				flags |= CC_NEGATIVE;
+
+			/* test for equal condition */
+			if (float64_eq_nocheck(rFn, rFm))
+				flags |= CC_ZERO;
+
+			/* test for greater than or equal condition */
+			if (float64_lt_nocheck(rFm, rFn))
+				flags |= CC_CARRY;
+		}
+	} else {
+		/* Both operands are in registers.  */
+		if (fpa11->fType[Fn] == typeSingle
+		    && fpa11->fType[Fm] == typeSingle) {
+			float32 rFm = fpa11->fpreg[Fm].fSingle;
+			float32 rFn = fpa11->fpreg[Fn].fSingle;
+
+			if (float32_is_nan(rFn)
+			    || float32_is_nan(rFm))
+				goto unordered;
+
+			if (n_flag)
+				rFm ^= 0x80000000;
+
+			/* test for less than condition */
+			if (float32_lt_nocheck(rFn, rFm))
+				flags |= CC_NEGATIVE;
+
+			/* test for equal condition */
+			if (float32_eq_nocheck(rFn, rFm))
+				flags |= CC_ZERO;
+
+			/* test for greater than or equal condition */
+			if (float32_lt_nocheck(rFm, rFn))
+				flags |= CC_CARRY;
+		} else {
+			/* Promote 32-bit operand to 64 bits.  */
+			float64 rFm, rFn;
+
+			rFm = (fpa11->fType[Fm] == typeSingle) ?
+			    float32_to_float64(fpa11->fpreg[Fm].fSingle)
+			    : fpa11->fpreg[Fm].fDouble;
+
+			rFn = (fpa11->fType[Fn] == typeSingle) ?
+			    float32_to_float64(fpa11->fpreg[Fn].fSingle)
+			    : fpa11->fpreg[Fn].fDouble;
+
+			if (float64_is_nan(rFn)
+			    || float64_is_nan(rFm))
+				goto unordered;
+
+			if (n_flag)
+				rFm ^= 0x8000000000000000ULL;
+
+			/* test for less than condition */
+			if (float64_lt_nocheck(rFn, rFm))
+				flags |= CC_NEGATIVE;
+
+			/* test for equal condition */
+			if (float64_eq_nocheck(rFn, rFm))
+				flags |= CC_ZERO;
+
+			/* test for greater than or equal condition */
+			if (float64_lt_nocheck(rFm, rFn))
+				flags |= CC_CARRY;
+		}
+	}
+
+#endif
+
+	writeConditionCodes(flags);
+
+	return 1;
+
+      unordered:
+	/* ?? The FPA data sheet is pretty vague about this, in particular
+	   about whether the non-E comparisons can ever raise exceptions.
+	   This implementation is based on a combination of what it says in
+	   the data sheet, observation of how the Acorn emulator actually
+	   behaves (and how programs expect it to) and guesswork.  */
+	flags |= CC_OVERFLOW;
+	flags &= ~(CC_ZERO | CC_NEGATIVE);
+
+	if (BIT_AC & readFPSR())
+		flags |= CC_CARRY;
+
+	if (e_flag)
+		float_raise(float_flag_invalid);
+
+	writeConditionCodes(flags);
+	return 1;
+}

arch/arm/nwfpe/fpmodule.c

@@ -0,0 +1,190 @@
+
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.com, 1998-1999
+    (c) Philip Blundell, 1998-1999
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+
+/* XXX */
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/signal.h>
+#include <linux/sched.h>
+#include <linux/init.h>
+
+#include <asm/thread_notify.h>
+
+#include "softfloat.h"
+#include "fpopcode.h"
+#include "fpmodule.h"
+#include "fpa11.inl"
+
+/* kernel symbols required for signal handling */
+#ifdef CONFIG_FPE_NWFPE_XP
+#define NWFPE_BITS "extended"
+#else
+#define NWFPE_BITS "double"
+#endif
+
+#ifdef MODULE
+void fp_send_sig(unsigned long sig, struct task_struct *p, int priv);
+#else
+#define fp_send_sig	send_sig
+#define kern_fp_enter	fp_enter
+
+extern char fpe_type[];
+#endif
+
+static int nwfpe_notify(struct notifier_block *self, unsigned long cmd, void *v)
+{
+	struct thread_info *thread = v;
+
+	if (cmd == THREAD_NOTIFY_FLUSH)
+		nwfpe_init_fpa(&thread->fpstate);
+
+	return NOTIFY_DONE;
+}
+
+static struct notifier_block nwfpe_notifier_block = {
+	.notifier_call = nwfpe_notify,
+};
+
+/* kernel function prototypes required */
+void fp_setup(void);
+
+/* external declarations for saved kernel symbols */
+extern void (*kern_fp_enter)(void);
+
+/* Original value of fp_enter from kernel before patched by fpe_init. */
+static void (*orig_fp_enter)(void);
+
+/* forward declarations */
+extern void nwfpe_enter(void);
+
+static int __init fpe_init(void)
+{
+	if (sizeof(FPA11) > sizeof(union fp_state)) {
+		pr_err("nwfpe: bad structure size\n");
+		return -EINVAL;
+	}
+
+	if (sizeof(FPREG) != 12) {
+		pr_err("nwfpe: bad register size\n");
+		return -EINVAL;
+	}
+	if (fpe_type[0] && strcmp(fpe_type, "nwfpe"))
+		return 0;
+
+	/* Display title, version and copyright information. */
+	pr_info("NetWinder Floating Point Emulator V0.97 ("
+	        NWFPE_BITS " precision)\n");
+
+	thread_register_notifier(&nwfpe_notifier_block);
+
+	/* Save pointer to the old FP handler and then patch ourselves in */
+	orig_fp_enter = kern_fp_enter;
+	kern_fp_enter = nwfpe_enter;
+
+	return 0;
+}
+
+static void __exit fpe_exit(void)
+{
+	thread_unregister_notifier(&nwfpe_notifier_block);
+	/* Restore the values we saved earlier. */
+	kern_fp_enter = orig_fp_enter;
+}
+
+/*
+ScottB:  November 4, 1998
+
+Moved this function out of softfloat-specialize into fpmodule.c.
+This effectively isolates all the changes required for integrating with the
+Linux kernel into fpmodule.c.  Porting to NetBSD should only require modifying
+fpmodule.c to integrate with the NetBSD kernel (I hope!).
+
+[1/1/99: Not quite true any more unfortunately.  There is Linux-specific
+code to access data in user space in some other source files at the 
+moment (grep for get_user / put_user calls).  --philb]
+
+This function is called by the SoftFloat routines to raise a floating
+point exception.  We check the trap enable byte in the FPSR, and raise
+a SIGFPE exception if necessary.  If not the relevant bits in the 
+cumulative exceptions flag byte are set and we return.
+*/
+
+#ifdef CONFIG_DEBUG_USER
+/* By default, ignore inexact errors as there are far too many of them to log */
+static int debug = ~BIT_IXC;
+#endif
+
+void float_raise(signed char flags)
+{
+	register unsigned int fpsr, cumulativeTraps;
+
+#ifdef CONFIG_DEBUG_USER
+	if (flags & debug)
+ 		printk(KERN_DEBUG
+		       "NWFPE: %s[%d] takes exception %08x at %pf from %08lx\n",
+		       current->comm, current->pid, flags,
+		       __builtin_return_address(0), GET_USERREG()->ARM_pc);
+#endif
+
+	/* Read fpsr and initialize the cumulativeTraps.  */
+	fpsr = readFPSR();
+	cumulativeTraps = 0;
+
+	/* For each type of exception, the cumulative trap exception bit is only
+	   set if the corresponding trap enable bit is not set.  */
+	if ((!(fpsr & BIT_IXE)) && (flags & BIT_IXC))
+		cumulativeTraps |= BIT_IXC;
+	if ((!(fpsr & BIT_UFE)) && (flags & BIT_UFC))
+		cumulativeTraps |= BIT_UFC;
+	if ((!(fpsr & BIT_OFE)) && (flags & BIT_OFC))
+		cumulativeTraps |= BIT_OFC;
+	if ((!(fpsr & BIT_DZE)) && (flags & BIT_DZC))
+		cumulativeTraps |= BIT_DZC;
+	if ((!(fpsr & BIT_IOE)) && (flags & BIT_IOC))
+		cumulativeTraps |= BIT_IOC;
+
+	/* Set the cumulative exceptions flags.  */
+	if (cumulativeTraps)
+		writeFPSR(fpsr | cumulativeTraps);
+
+	/* Raise an exception if necessary.  */
+	if (fpsr & (flags << 16))
+		fp_send_sig(SIGFPE, current, 1);
+}
+
+module_init(fpe_init);
+module_exit(fpe_exit);
+
+MODULE_AUTHOR("Scott Bambrough <scottb@rebel.com>");
+MODULE_DESCRIPTION("NWFPE floating point emulator (" NWFPE_BITS " precision)");
+MODULE_LICENSE("GPL");
+
+#ifdef CONFIG_DEBUG_USER
+module_param(debug, int, 0644);
+#endif

arch/arm/nwfpe/fpmodule.h

@@ -0,0 +1,47 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.com, 1998-1999
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+    27/03/03 Ian Molton Clean up CONFIG_CPU
+*/
+
+#ifndef __FPMODULE_H__
+#define __FPMODULE_H__
+
+#define REG_ORIG_R0	17
+#define REG_CPSR	16
+#define REG_PC		15
+#define REG_LR		14
+#define REG_SP		13
+#define REG_IP		12
+#define REG_FP		11
+#define REG_R10		10
+#define REG_R9		9
+#define REG_R9		9
+#define REG_R8		8
+#define REG_R7		7
+#define REG_R6		6
+#define REG_R5		5
+#define REG_R4		4
+#define REG_R3		3
+#define REG_R2		2
+#define REG_R1		1
+#define REG_R0		0
+
+#endif

arch/arm/nwfpe/fpmodule.inl

@@ -0,0 +1,74 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+static inline unsigned long readRegister(const unsigned int nReg)
+{
+	/* Note: The CPU thinks it has dealt with the current instruction.
+	   As a result the program counter has been advanced to the next
+	   instruction, and points 4 bytes beyond the actual instruction
+	   that caused the invalid instruction trap to occur.  We adjust
+	   for this in this routine.  LDF/STF instructions with Rn = PC
+	   depend on the PC being correct, as they use PC+8 in their
+	   address calculations. */
+	struct pt_regs *regs = GET_USERREG();
+	unsigned int val = regs->uregs[nReg];
+	if (REG_PC == nReg)
+		val -= 4;
+	return val;
+}
+
+static inline void
+writeRegister(const unsigned int nReg, const unsigned long val)
+{
+	struct pt_regs *regs = GET_USERREG();
+	regs->uregs[nReg] = val;
+}
+
+static inline unsigned long readCPSR(void)
+{
+	return (readRegister(REG_CPSR));
+}
+
+static inline void writeCPSR(const unsigned long val)
+{
+	writeRegister(REG_CPSR, val);
+}
+
+static inline unsigned long readConditionCodes(void)
+{
+#ifdef __FPEM_TEST__
+	return (0);
+#else
+	return (readCPSR() & CC_MASK);
+#endif
+}
+
+static inline void writeConditionCodes(const unsigned long val)
+{
+	struct pt_regs *regs = GET_USERREG();
+	unsigned long rval;
+	/*
+	 * Operate directly on userRegisters since
+	 * the CPSR may be the PC register itself.
+	 */
+	rval = regs->ARM_cpsr & ~CC_MASK;
+	regs->ARM_cpsr = rval | (val & CC_MASK);
+}

arch/arm/nwfpe/fpopcode.c

@@ -0,0 +1,63 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+#include "softfloat.h"
+#include "fpopcode.h"
+#include "fpsr.h"
+#include "fpmodule.h"
+#include "fpmodule.inl"
+
+#ifdef CONFIG_FPE_NWFPE_XP
+const floatx80 floatx80Constant[] = {
+	{ .high = 0x0000, .low = 0x0000000000000000ULL},/* extended 0.0 */
+	{ .high = 0x3fff, .low = 0x8000000000000000ULL},/* extended 1.0 */
+	{ .high = 0x4000, .low = 0x8000000000000000ULL},/* extended 2.0 */
+	{ .high = 0x4000, .low = 0xc000000000000000ULL},/* extended 3.0 */
+	{ .high = 0x4001, .low = 0x8000000000000000ULL},/* extended 4.0 */
+	{ .high = 0x4001, .low = 0xa000000000000000ULL},/* extended 5.0 */
+	{ .high = 0x3ffe, .low = 0x8000000000000000ULL},/* extended 0.5 */
+	{ .high = 0x4002, .low = 0xa000000000000000ULL},/* extended 10.0 */
+};
+#endif
+
+const float64 float64Constant[] = {
+	0x0000000000000000ULL,	/* double 0.0 */
+	0x3ff0000000000000ULL,	/* double 1.0 */
+	0x4000000000000000ULL,	/* double 2.0 */
+	0x4008000000000000ULL,	/* double 3.0 */
+	0x4010000000000000ULL,	/* double 4.0 */
+	0x4014000000000000ULL,	/* double 5.0 */
+	0x3fe0000000000000ULL,	/* double 0.5 */
+	0x4024000000000000ULL	/* double 10.0 */
+};
+
+const float32 float32Constant[] = {
+	0x00000000,		/* single 0.0 */
+	0x3f800000,		/* single 1.0 */
+	0x40000000,		/* single 2.0 */
+	0x40400000,		/* single 3.0 */
+	0x40800000,		/* single 4.0 */
+	0x40a00000,		/* single 5.0 */
+	0x3f000000,		/* single 0.5 */
+	0x41200000		/* single 10.0 */
+};
+

arch/arm/nwfpe/fpopcode.h

@@ -0,0 +1,481 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+    (c) Philip Blundell, 2001
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#ifndef __FPOPCODE_H__
+#define __FPOPCODE_H__
+
+
+/*
+ARM Floating Point Instruction Classes
+| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 
+|c o n d|1 1 0 P|U|u|W|L|   Rn  |v|  Fd |0|0|0|1|  o f f s e t  | CPDT
+|c o n d|1 1 0 P|U|w|W|L|   Rn  |x|  Fd |0|0|1|0|  o f f s e t  | CPDT (copro 2)
+| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 
+|c o n d|1 1 1 0|a|b|c|d|e|  Fn |j|  Fd |0|0|0|1|f|g|h|0|i|  Fm | CPDO
+|c o n d|1 1 1 0|a|b|c|L|e|  Fn |   Rd  |0|0|0|1|f|g|h|1|i|  Fm | CPRT
+|c o n d|1 1 1 0|a|b|c|1|e|  Fn |1|1|1|1|0|0|0|1|f|g|h|1|i|  Fm | comparisons
+| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 
+
+CPDT		data transfer instructions
+		LDF, STF, LFM (copro 2), SFM (copro 2)
+		
+CPDO		dyadic arithmetic instructions
+		ADF, MUF, SUF, RSF, DVF, RDF,
+		POW, RPW, RMF, FML, FDV, FRD, POL
+
+CPDO		monadic arithmetic instructions
+		MVF, MNF, ABS, RND, SQT, LOG, LGN, EXP,
+		SIN, COS, TAN, ASN, ACS, ATN, URD, NRM
+		
+CPRT		joint arithmetic/data transfer instructions
+		FIX (arithmetic followed by load/store)
+		FLT (load/store followed by arithmetic)
+		CMF, CNF CMFE, CNFE (comparisons)
+		WFS, RFS (write/read floating point status register)
+		WFC, RFC (write/read floating point control register)
+
+cond		condition codes
+P		pre/post index bit: 0 = postindex, 1 = preindex
+U		up/down bit: 0 = stack grows down, 1 = stack grows up
+W		write back bit: 1 = update base register (Rn)
+L		load/store bit: 0 = store, 1 = load
+Rn		base register
+Rd		destination/source register		
+Fd		floating point destination register
+Fn		floating point source register
+Fm		floating point source register or floating point constant
+
+uv		transfer length (TABLE 1)
+wx		register count (TABLE 2)
+abcd		arithmetic opcode (TABLES 3 & 4)
+ef		destination size (rounding precision) (TABLE 5)
+gh		rounding mode (TABLE 6)
+j		dyadic/monadic bit: 0 = dyadic, 1 = monadic
+i 		constant bit: 1 = constant (TABLE 6)
+*/
+
+/*
+TABLE 1
++-------------------------+---+---+---------+---------+
+|  Precision              | u | v | FPSR.EP | length  |
++-------------------------+---+---+---------+---------+
+| Single                  | 0 | 0 |    x    | 1 words |
+| Double                  | 1 | 1 |    x    | 2 words |
+| Extended                | 1 | 1 |    x    | 3 words |
+| Packed decimal          | 1 | 1 |    0    | 3 words |
+| Expanded packed decimal | 1 | 1 |    1    | 4 words |
++-------------------------+---+---+---------+---------+
+Note: x = don't care
+*/
+
+/*
+TABLE 2
++---+---+---------------------------------+
+| w | x | Number of registers to transfer |
++---+---+---------------------------------+
+| 0 | 1 |  1                              |
+| 1 | 0 |  2                              |
+| 1 | 1 |  3                              |
+| 0 | 0 |  4                              |
++---+---+---------------------------------+
+*/
+
+/*
+TABLE 3: Dyadic Floating Point Opcodes
++---+---+---+---+----------+-----------------------+-----------------------+
+| a | b | c | d | Mnemonic | Description           | Operation             |
++---+---+---+---+----------+-----------------------+-----------------------+
+| 0 | 0 | 0 | 0 | ADF      | Add                   | Fd := Fn + Fm         |
+| 0 | 0 | 0 | 1 | MUF      | Multiply              | Fd := Fn * Fm         |
+| 0 | 0 | 1 | 0 | SUF      | Subtract              | Fd := Fn - Fm         |
+| 0 | 0 | 1 | 1 | RSF      | Reverse subtract      | Fd := Fm - Fn         |
+| 0 | 1 | 0 | 0 | DVF      | Divide                | Fd := Fn / Fm         |
+| 0 | 1 | 0 | 1 | RDF      | Reverse divide        | Fd := Fm / Fn         |
+| 0 | 1 | 1 | 0 | POW      | Power                 | Fd := Fn ^ Fm         |
+| 0 | 1 | 1 | 1 | RPW      | Reverse power         | Fd := Fm ^ Fn         |
+| 1 | 0 | 0 | 0 | RMF      | Remainder             | Fd := IEEE rem(Fn/Fm) |
+| 1 | 0 | 0 | 1 | FML      | Fast Multiply         | Fd := Fn * Fm         |
+| 1 | 0 | 1 | 0 | FDV      | Fast Divide           | Fd := Fn / Fm         |
+| 1 | 0 | 1 | 1 | FRD      | Fast reverse divide   | Fd := Fm / Fn         |
+| 1 | 1 | 0 | 0 | POL      | Polar angle (ArcTan2) | Fd := arctan2(Fn,Fm)  |
+| 1 | 1 | 0 | 1 |          | undefined instruction | trap                  |
+| 1 | 1 | 1 | 0 |          | undefined instruction | trap                  |
+| 1 | 1 | 1 | 1 |          | undefined instruction | trap                  |
++---+---+---+---+----------+-----------------------+-----------------------+
+Note: POW, RPW, POL are deprecated, and are available for backwards
+      compatibility only.
+*/
+
+/*
+TABLE 4: Monadic Floating Point Opcodes
++---+---+---+---+----------+-----------------------+-----------------------+
+| a | b | c | d | Mnemonic | Description           | Operation             |
++---+---+---+---+----------+-----------------------+-----------------------+
+| 0 | 0 | 0 | 0 | MVF      | Move                  | Fd := Fm              |
+| 0 | 0 | 0 | 1 | MNF      | Move negated          | Fd := - Fm            |
+| 0 | 0 | 1 | 0 | ABS      | Absolute value        | Fd := abs(Fm)         |
+| 0 | 0 | 1 | 1 | RND      | Round to integer      | Fd := int(Fm)         |
+| 0 | 1 | 0 | 0 | SQT      | Square root           | Fd := sqrt(Fm)        |
+| 0 | 1 | 0 | 1 | LOG      | Log base 10           | Fd := log10(Fm)       |
+| 0 | 1 | 1 | 0 | LGN      | Log base e            | Fd := ln(Fm)          |
+| 0 | 1 | 1 | 1 | EXP      | Exponent              | Fd := e ^ Fm          |
+| 1 | 0 | 0 | 0 | SIN      | Sine                  | Fd := sin(Fm)         |
+| 1 | 0 | 0 | 1 | COS      | Cosine                | Fd := cos(Fm)         |
+| 1 | 0 | 1 | 0 | TAN      | Tangent               | Fd := tan(Fm)         |
+| 1 | 0 | 1 | 1 | ASN      | Arc Sine              | Fd := arcsin(Fm)      |
+| 1 | 1 | 0 | 0 | ACS      | Arc Cosine            | Fd := arccos(Fm)      |
+| 1 | 1 | 0 | 1 | ATN      | Arc Tangent           | Fd := arctan(Fm)      |
+| 1 | 1 | 1 | 0 | URD      | Unnormalized round    | Fd := int(Fm)         |
+| 1 | 1 | 1 | 1 | NRM      | Normalize             | Fd := norm(Fm)        |
++---+---+---+---+----------+-----------------------+-----------------------+
+Note: LOG, LGN, EXP, SIN, COS, TAN, ASN, ACS, ATN are deprecated, and are
+      available for backwards compatibility only.
+*/
+
+/*
+TABLE 5
++-------------------------+---+---+
+|  Rounding Precision     | e | f |
++-------------------------+---+---+
+| IEEE Single precision   | 0 | 0 |
+| IEEE Double precision   | 0 | 1 |
+| IEEE Extended precision | 1 | 0 |
+| undefined (trap)        | 1 | 1 |
++-------------------------+---+---+
+*/
+
+/*
+TABLE 5
++---------------------------------+---+---+
+|  Rounding Mode                  | g | h |
++---------------------------------+---+---+
+| Round to nearest (default)      | 0 | 0 |
+| Round toward plus infinity      | 0 | 1 |
+| Round toward negative infinity  | 1 | 0 |
+| Round toward zero               | 1 | 1 |
++---------------------------------+---+---+
+*/
+
+/*
+===
+=== Definitions for load and store instructions
+===
+*/
+
+/* bit masks */
+#define BIT_PREINDEX	0x01000000
+#define BIT_UP		0x00800000
+#define BIT_WRITE_BACK	0x00200000
+#define BIT_LOAD	0x00100000
+
+/* masks for load/store */
+#define MASK_CPDT		0x0c000000	/* data processing opcode */
+#define MASK_OFFSET		0x000000ff
+#define MASK_TRANSFER_LENGTH	0x00408000
+#define MASK_REGISTER_COUNT	MASK_TRANSFER_LENGTH
+#define MASK_COPROCESSOR	0x00000f00
+
+/* Tests for transfer length */
+#define TRANSFER_SINGLE		0x00000000
+#define TRANSFER_DOUBLE		0x00008000
+#define TRANSFER_EXTENDED	0x00400000
+#define TRANSFER_PACKED		MASK_TRANSFER_LENGTH
+
+/* Get the coprocessor number from the opcode. */
+#define getCoprocessorNumber(opcode)	((opcode & MASK_COPROCESSOR) >> 8)
+
+/* Get the offset from the opcode. */
+#define getOffset(opcode)		(opcode & MASK_OFFSET)
+
+/* Tests for specific data transfer load/store opcodes. */
+#define TEST_OPCODE(opcode,mask)	(((opcode) & (mask)) == (mask))
+
+#define LOAD_OP(opcode)   TEST_OPCODE((opcode),MASK_CPDT | BIT_LOAD)
+#define STORE_OP(opcode)  ((opcode & (MASK_CPDT | BIT_LOAD)) == MASK_CPDT)
+
+#define LDF_OP(opcode)	(LOAD_OP(opcode) && (getCoprocessorNumber(opcode) == 1))
+#define LFM_OP(opcode)	(LOAD_OP(opcode) && (getCoprocessorNumber(opcode) == 2))
+#define STF_OP(opcode)	(STORE_OP(opcode) && (getCoprocessorNumber(opcode) == 1))
+#define SFM_OP(opcode)	(STORE_OP(opcode) && (getCoprocessorNumber(opcode) == 2))
+
+#define PREINDEXED(opcode)		((opcode & BIT_PREINDEX) != 0)
+#define POSTINDEXED(opcode)		((opcode & BIT_PREINDEX) == 0)
+#define BIT_UP_SET(opcode)		((opcode & BIT_UP) != 0)
+#define BIT_UP_CLEAR(opcode)		((opcode & BIT_DOWN) == 0)
+#define WRITE_BACK(opcode)		((opcode & BIT_WRITE_BACK) != 0)
+#define LOAD(opcode)			((opcode & BIT_LOAD) != 0)
+#define STORE(opcode)			((opcode & BIT_LOAD) == 0)
+
+/*
+===
+=== Definitions for arithmetic instructions
+===
+*/
+/* bit masks */
+#define BIT_MONADIC	0x00008000
+#define BIT_CONSTANT	0x00000008
+
+#define CONSTANT_FM(opcode)		((opcode & BIT_CONSTANT) != 0)
+#define MONADIC_INSTRUCTION(opcode)	((opcode & BIT_MONADIC) != 0)
+
+/* instruction identification masks */
+#define MASK_CPDO		0x0e000000	/* arithmetic opcode */
+#define MASK_ARITHMETIC_OPCODE	0x00f08000
+#define MASK_DESTINATION_SIZE	0x00080080
+
+/* dyadic arithmetic opcodes. */
+#define ADF_CODE	0x00000000
+#define MUF_CODE	0x00100000
+#define SUF_CODE	0x00200000
+#define RSF_CODE	0x00300000
+#define DVF_CODE	0x00400000
+#define RDF_CODE	0x00500000
+#define POW_CODE	0x00600000
+#define RPW_CODE	0x00700000
+#define RMF_CODE	0x00800000
+#define FML_CODE	0x00900000
+#define FDV_CODE	0x00a00000
+#define FRD_CODE	0x00b00000
+#define POL_CODE	0x00c00000
+/* 0x00d00000 is an invalid dyadic arithmetic opcode */
+/* 0x00e00000 is an invalid dyadic arithmetic opcode */
+/* 0x00f00000 is an invalid dyadic arithmetic opcode */
+
+/* monadic arithmetic opcodes. */
+#define MVF_CODE	0x00008000
+#define MNF_CODE	0x00108000
+#define ABS_CODE	0x00208000
+#define RND_CODE	0x00308000
+#define SQT_CODE	0x00408000
+#define LOG_CODE	0x00508000
+#define LGN_CODE	0x00608000
+#define EXP_CODE	0x00708000
+#define SIN_CODE	0x00808000
+#define COS_CODE	0x00908000
+#define TAN_CODE	0x00a08000
+#define ASN_CODE	0x00b08000
+#define ACS_CODE	0x00c08000
+#define ATN_CODE	0x00d08000
+#define URD_CODE	0x00e08000
+#define NRM_CODE	0x00f08000
+
+/*
+===
+=== Definitions for register transfer and comparison instructions
+===
+*/
+
+#define MASK_CPRT		0x0e000010	/* register transfer opcode */
+#define MASK_CPRT_CODE		0x00f00000
+#define FLT_CODE		0x00000000
+#define FIX_CODE		0x00100000
+#define WFS_CODE		0x00200000
+#define RFS_CODE		0x00300000
+#define WFC_CODE		0x00400000
+#define RFC_CODE		0x00500000
+#define CMF_CODE		0x00900000
+#define CNF_CODE		0x00b00000
+#define CMFE_CODE		0x00d00000
+#define CNFE_CODE		0x00f00000
+
+/*
+===
+=== Common definitions
+===
+*/
+
+/* register masks */
+#define MASK_Rd		0x0000f000
+#define MASK_Rn		0x000f0000
+#define MASK_Fd		0x00007000
+#define MASK_Fm		0x00000007
+#define MASK_Fn		0x00070000
+
+/* condition code masks */
+#define CC_MASK		0xf0000000
+#define CC_NEGATIVE	0x80000000
+#define CC_ZERO		0x40000000
+#define CC_CARRY	0x20000000
+#define CC_OVERFLOW	0x10000000
+#define CC_EQ		0x00000000
+#define CC_NE		0x10000000
+#define CC_CS		0x20000000
+#define CC_HS		CC_CS
+#define CC_CC		0x30000000
+#define CC_LO		CC_CC
+#define CC_MI		0x40000000
+#define CC_PL		0x50000000
+#define CC_VS		0x60000000
+#define CC_VC		0x70000000
+#define CC_HI		0x80000000
+#define CC_LS		0x90000000
+#define CC_GE		0xa0000000
+#define CC_LT		0xb0000000
+#define CC_GT		0xc0000000
+#define CC_LE		0xd0000000
+#define CC_AL		0xe0000000
+#define CC_NV		0xf0000000
+
+/* rounding masks/values */
+#define MASK_ROUNDING_MODE	0x00000060
+#define ROUND_TO_NEAREST	0x00000000
+#define ROUND_TO_PLUS_INFINITY	0x00000020
+#define ROUND_TO_MINUS_INFINITY	0x00000040
+#define ROUND_TO_ZERO		0x00000060
+
+#define MASK_ROUNDING_PRECISION	0x00080080
+#define ROUND_SINGLE		0x00000000
+#define ROUND_DOUBLE		0x00000080
+#define ROUND_EXTENDED		0x00080000
+
+/* Get the condition code from the opcode. */
+#define getCondition(opcode)		(opcode >> 28)
+
+/* Get the source register from the opcode. */
+#define getRn(opcode)			((opcode & MASK_Rn) >> 16)
+
+/* Get the destination floating point register from the opcode. */
+#define getFd(opcode)			((opcode & MASK_Fd) >> 12)
+
+/* Get the first source floating point register from the opcode. */
+#define getFn(opcode)		((opcode & MASK_Fn) >> 16)
+
+/* Get the second source floating point register from the opcode. */
+#define getFm(opcode)		(opcode & MASK_Fm)
+
+/* Get the destination register from the opcode. */
+#define getRd(opcode)		((opcode & MASK_Rd) >> 12)
+
+/* Get the rounding mode from the opcode. */
+#define getRoundingMode(opcode)		((opcode & MASK_ROUNDING_MODE) >> 5)
+
+#ifdef CONFIG_FPE_NWFPE_XP
+static inline floatx80 __pure getExtendedConstant(const unsigned int nIndex)
+{
+	extern const floatx80 floatx80Constant[];
+	return floatx80Constant[nIndex];
+}
+#endif
+
+static inline float64 __pure getDoubleConstant(const unsigned int nIndex)
+{
+	extern const float64 float64Constant[];
+	return float64Constant[nIndex];
+}
+
+static inline float32 __pure getSingleConstant(const unsigned int nIndex)
+{
+	extern const float32 float32Constant[];
+	return float32Constant[nIndex];
+}
+
+static inline unsigned int getTransferLength(const unsigned int opcode)
+{
+	unsigned int nRc;
+
+	switch (opcode & MASK_TRANSFER_LENGTH) {
+	case 0x00000000:
+		nRc = 1;
+		break;		/* single precision */
+	case 0x00008000:
+		nRc = 2;
+		break;		/* double precision */
+	case 0x00400000:
+		nRc = 3;
+		break;		/* extended precision */
+	default:
+		nRc = 0;
+	}
+
+	return (nRc);
+}
+
+static inline unsigned int getRegisterCount(const unsigned int opcode)
+{
+	unsigned int nRc;
+
+	switch (opcode & MASK_REGISTER_COUNT) {
+	case 0x00000000:
+		nRc = 4;
+		break;
+	case 0x00008000:
+		nRc = 1;
+		break;
+	case 0x00400000:
+		nRc = 2;
+		break;
+	case 0x00408000:
+		nRc = 3;
+		break;
+	default:
+		nRc = 0;
+	}
+
+	return (nRc);
+}
+
+static inline unsigned int getRoundingPrecision(const unsigned int opcode)
+{
+	unsigned int nRc;
+
+	switch (opcode & MASK_ROUNDING_PRECISION) {
+	case 0x00000000:
+		nRc = 1;
+		break;
+	case 0x00000080:
+		nRc = 2;
+		break;
+	case 0x00080000:
+		nRc = 3;
+		break;
+	default:
+		nRc = 0;
+	}
+
+	return (nRc);
+}
+
+static inline unsigned int getDestinationSize(const unsigned int opcode)
+{
+	unsigned int nRc;
+
+	switch (opcode & MASK_DESTINATION_SIZE) {
+	case 0x00000000:
+		nRc = typeSingle;
+		break;
+	case 0x00000080:
+		nRc = typeDouble;
+		break;
+	case 0x00080000:
+		nRc = typeExtended;
+		break;
+	default:
+		nRc = typeNone;
+	}
+
+	return (nRc);
+}
+
+extern const float64 float64Constant[];
+extern const float32 float32Constant[];
+
+#endif

arch/arm/nwfpe/fpsr.h

@@ -0,0 +1,108 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.com, 1998-1999
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#ifndef __FPSR_H__
+#define __FPSR_H__
+
+/*
+The FPSR is a 32 bit register consisting of 4 parts, each exactly
+one byte.
+
+	SYSTEM ID
+	EXCEPTION TRAP ENABLE BYTE
+	SYSTEM CONTROL BYTE
+	CUMULATIVE EXCEPTION FLAGS BYTE
+
+The FPCR is a 32 bit register consisting of bit flags.
+*/
+
+/* SYSTEM ID
+------------
+Note: the system id byte is read only  */
+
+typedef unsigned int FPSR;	/* type for floating point status register */
+typedef unsigned int FPCR;	/* type for floating point control register */
+
+#define MASK_SYSID		0xff000000
+#define BIT_HARDWARE		0x80000000
+#define FP_EMULATOR		0x01000000	/* System ID for emulator */
+#define FP_ACCELERATOR		0x81000000	/* System ID for FPA11 */
+
+/* EXCEPTION TRAP ENABLE BYTE
+----------------------------- */
+
+#define MASK_TRAP_ENABLE	0x00ff0000
+#define MASK_TRAP_ENABLE_STRICT	0x001f0000
+#define BIT_IXE		0x00100000	/* inexact exception enable */
+#define BIT_UFE		0x00080000	/* underflow exception enable */
+#define BIT_OFE		0x00040000	/* overflow exception enable */
+#define BIT_DZE		0x00020000	/* divide by zero exception enable */
+#define BIT_IOE		0x00010000	/* invalid operation exception enable */
+
+/* SYSTEM CONTROL BYTE
+---------------------- */
+
+#define MASK_SYSTEM_CONTROL	0x0000ff00
+#define MASK_TRAP_STRICT	0x00001f00
+
+#define BIT_AC	0x00001000	/* use alternative C-flag definition
+				   for compares */
+#define BIT_EP	0x00000800	/* use expanded packed decimal format */
+#define BIT_SO	0x00000400	/* select synchronous operation of FPA */
+#define BIT_NE	0x00000200	/* NaN exception bit */
+#define BIT_ND	0x00000100	/* no denormalized numbers bit */
+
+/* CUMULATIVE EXCEPTION FLAGS BYTE
+---------------------------------- */
+
+#define MASK_EXCEPTION_FLAGS		0x000000ff
+#define MASK_EXCEPTION_FLAGS_STRICT	0x0000001f
+
+#define BIT_IXC		0x00000010	/* inexact exception flag */
+#define BIT_UFC		0x00000008	/* underflow exception flag */
+#define BIT_OFC		0x00000004	/* overfloat exception flag */
+#define BIT_DZC		0x00000002	/* divide by zero exception flag */
+#define BIT_IOC		0x00000001	/* invalid operation exception flag */
+
+/* Floating Point Control Register
+----------------------------------*/
+
+#define BIT_RU		0x80000000	/* rounded up bit */
+#define BIT_IE		0x10000000	/* inexact bit */
+#define BIT_MO		0x08000000	/* mantissa overflow bit */
+#define BIT_EO		0x04000000	/* exponent overflow bit */
+#define BIT_SB		0x00000800	/* store bounce */
+#define BIT_AB		0x00000400	/* arithmetic bounce */
+#define BIT_RE		0x00000200	/* rounding exception */
+#define BIT_DA		0x00000100	/* disable FPA */
+
+#define MASK_OP		0x00f08010	/* AU operation code */
+#define MASK_PR		0x00080080	/* AU precision */
+#define MASK_S1		0x00070000	/* AU source register 1 */
+#define MASK_S2		0x00000007	/* AU source register 2 */
+#define MASK_DS		0x00007000	/* AU destination register */
+#define MASK_RM		0x00000060	/* AU rounding mode */
+#define MASK_ALU	0x9cfff2ff	/* only ALU can write these bits */
+#define MASK_RESET	0x00000d00	/* bits set on reset, all others cleared */
+#define MASK_WFC	MASK_RESET
+#define MASK_RFC	~MASK_RESET
+
+#endif

arch/arm/nwfpe/milieu.h

@@ -0,0 +1,48 @@
+
+/*
+===============================================================================
+
+This C header file is part of the SoftFloat IEC/IEEE Floating-point
+Arithmetic Package, Release 2.
+
+Written by John R. Hauser.  This work was made possible in part by the
+International Computer Science Institute, located at Suite 600, 1947 Center
+Street, Berkeley, California 94704.  Funding was partially provided by the
+National Science Foundation under grant MIP-9311980.  The original version
+of this code was written as part of a project to build a fixed-point vector
+processor in collaboration with the University of California at Berkeley,
+overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
+is available through the Web page
+http://www.jhauser.us/arithmetic/SoftFloat-2b/SoftFloat-source.txt
+
+THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort
+has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
+TIMES RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO
+PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
+AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
+
+Derivative works are acceptable, even for commercial purposes, so long as
+(1) they include prominent notice that the work is derivative, and (2) they
+include prominent notice akin to these three paragraphs for those parts of
+this code that are retained.
+
+===============================================================================
+*/
+
+/*
+-------------------------------------------------------------------------------
+Include common integer types and flags.
+-------------------------------------------------------------------------------
+*/
+#include "ARM-gcc.h"
+
+/*
+-------------------------------------------------------------------------------
+Symbolic Boolean literals.
+-------------------------------------------------------------------------------
+*/
+enum {
+    FALSE = 0,
+    TRUE  = 1
+};
+

arch/arm/nwfpe/single_cpdo.c

@@ -0,0 +1,124 @@
+/*
+    NetWinder Floating Point Emulator
+    (c) Rebel.COM, 1998,1999
+    (c) Philip Blundell, 2001
+
+    Direct questions, comments to Scott Bambrough <scottb@netwinder.org>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+*/
+
+#include "fpa11.h"
+#include "softfloat.h"
+#include "fpopcode.h"
+
+float32 float32_exp(float32 Fm);
+float32 float32_ln(float32 Fm);
+float32 float32_sin(float32 rFm);
+float32 float32_cos(float32 rFm);
+float32 float32_arcsin(float32 rFm);
+float32 float32_arctan(float32 rFm);
+float32 float32_log(float32 rFm);
+float32 float32_tan(float32 rFm);
+float32 float32_arccos(float32 rFm);
+float32 float32_pow(float32 rFn, float32 rFm);
+float32 float32_pol(float32 rFn, float32 rFm);
+
+static float32 float32_rsf(struct roundingData *roundData, float32 rFn, float32 rFm)
+{
+	return float32_sub(roundData, rFm, rFn);
+}
+
+static float32 float32_rdv(struct roundingData *roundData, float32 rFn, float32 rFm)
+{
+	return float32_div(roundData, rFm, rFn);
+}
+
+static float32 (*const dyadic_single[16])(struct roundingData *, float32 rFn, float32 rFm) = {
+	[ADF_CODE >> 20] = float32_add,
+	[MUF_CODE >> 20] = float32_mul,
+	[SUF_CODE >> 20] = float32_sub,
+	[RSF_CODE >> 20] = float32_rsf,
+	[DVF_CODE >> 20] = float32_div,
+	[RDF_CODE >> 20] = float32_rdv,
+	[RMF_CODE >> 20] = float32_rem,
+
+	[FML_CODE >> 20] = float32_mul,
+	[FDV_CODE >> 20] = float32_div,
+	[FRD_CODE >> 20] = float32_rdv,
+};
+
+static float32 float32_mvf(struct roundingData *roundData, float32 rFm)
+{
+	return rFm;
+}
+
+static float32 float32_mnf(struct roundingData *roundData, float32 rFm)
+{
+	return rFm ^ 0x80000000;
+}
+
+static float32 float32_abs(struct roundingData *roundData, float32 rFm)
+{
+	return rFm & 0x7fffffff;
+}
+
+static float32 (*const monadic_single[16])(struct roundingData*, float32 rFm) = {
+	[MVF_CODE >> 20] = float32_mvf,
+	[MNF_CODE >> 20] = float32_mnf,
+	[ABS_CODE >> 20] = float32_abs,
+	[RND_CODE >> 20] = float32_round_to_int,
+	[URD_CODE >> 20] = float32_round_to_int,
+	[SQT_CODE >> 20] = float32_sqrt,
+	[NRM_CODE >> 20] = float32_mvf,
+};
+
+unsigned int SingleCPDO(struct roundingData *roundData, const unsigned int opcode, FPREG * rFd)
+{
+	FPA11 *fpa11 = GET_FPA11();
+	float32 rFm;
+	unsigned int Fm, opc_mask_shift;
+
+	Fm = getFm(opcode);
+	if (CONSTANT_FM(opcode)) {
+		rFm = getSingleConstant(Fm);
+	} else if (fpa11->fType[Fm] == typeSingle) {
+		rFm = fpa11->fpreg[Fm].fSingle;
+	} else {
+		return 0;
+	}
+
+	opc_mask_shift = (opcode & MASK_ARITHMETIC_OPCODE) >> 20;
+	if (!MONADIC_INSTRUCTION(opcode)) {
+		unsigned int Fn = getFn(opcode);
+		float32 rFn;
+
+		if (fpa11->fType[Fn] == typeSingle &&
+		    dyadic_single[opc_mask_shift]) {
+			rFn = fpa11->fpreg[Fn].fSingle;
+			rFd->fSingle = dyadic_single[opc_mask_shift](roundData, rFn, rFm);
+		} else {
+			return 0;
+		}
+	} else {
+		if (monadic_single[opc_mask_shift]) {
+			rFd->fSingle = monadic_single[opc_mask_shift](roundData, rFm);
+		} else {
+			return 0;
+		}
+	}
+
+	return 1;
+}

arch/arm/nwfpe/softfloat-macros

@@ -0,0 +1,754 @@
+
+/*
+===============================================================================
+
+This C source fragment is part of the SoftFloat IEC/IEEE Floating-point
+Arithmetic Package, Release 2.
+
+Written by John R. Hauser.  This work was made possible in part by the
+International Computer Science Institute, located at Suite 600, 1947 Center
+Street, Berkeley, California 94704.  Funding was partially provided by the
+National Science Foundation under grant MIP-9311980.  The original version
+of this code was written as part of a project to build a fixed-point vector
+processor in collaboration with the University of California at Berkeley,
+overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
+is available through the web page
+http://www.jhauser.us/arithmetic/SoftFloat-2b/SoftFloat-source.txt
+
+THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort
+has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
+TIMES RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO
+PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
+AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
+
+Derivative works are acceptable, even for commercial purposes, so long as
+(1) they include prominent notice that the work is derivative, and (2) they
+include prominent notice akin to these three paragraphs for those parts of
+this code that are retained.
+
+===============================================================================
+*/
+
+/*
+-------------------------------------------------------------------------------
+Shifts `a' right by the number of bits given in `count'.  If any nonzero
+bits are shifted off, they are ``jammed'' into the least significant bit of
+the result by setting the least significant bit to 1.  The value of `count'
+can be arbitrarily large; in particular, if `count' is greater than 32, the
+result will be either 0 or 1, depending on whether `a' is zero or nonzero.
+The result is stored in the location pointed to by `zPtr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void shift32RightJamming( bits32 a, int16 count, bits32 *zPtr )
+{
+    bits32 z;
+    if ( count == 0 ) {
+        z = a;
+    }
+    else if ( count < 32 ) {
+        z = ( a>>count ) | ( ( a<<( ( - count ) & 31 ) ) != 0 );
+    }
+    else {
+        z = ( a != 0 );
+    }
+    *zPtr = z;
+}
+
+/*
+-------------------------------------------------------------------------------
+Shifts `a' right by the number of bits given in `count'.  If any nonzero
+bits are shifted off, they are ``jammed'' into the least significant bit of
+the result by setting the least significant bit to 1.  The value of `count'
+can be arbitrarily large; in particular, if `count' is greater than 64, the
+result will be either 0 or 1, depending on whether `a' is zero or nonzero.
+The result is stored in the location pointed to by `zPtr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void shift64RightJamming( bits64 a, int16 count, bits64 *zPtr )
+{
+    bits64 z;
+
+ __asm__("@shift64RightJamming -- start");   
+    if ( count == 0 ) {
+        z = a;
+    }
+    else if ( count < 64 ) {
+        z = ( a>>count ) | ( ( a<<( ( - count ) & 63 ) ) != 0 );
+    }
+    else {
+        z = ( a != 0 );
+    }
+ __asm__("@shift64RightJamming -- end");   
+    *zPtr = z;
+}
+
+/*
+-------------------------------------------------------------------------------
+Shifts the 128-bit value formed by concatenating `a0' and `a1' right by 64
+_plus_ the number of bits given in `count'.  The shifted result is at most
+64 nonzero bits; this is stored at the location pointed to by `z0Ptr'.  The
+bits shifted off form a second 64-bit result as follows:  The _last_ bit
+shifted off is the most-significant bit of the extra result, and the other
+63 bits of the extra result are all zero if and only if _all_but_the_last_
+bits shifted off were all zero.  This extra result is stored in the location
+pointed to by `z1Ptr'.  The value of `count' can be arbitrarily large.
+    (This routine makes more sense if `a0' and `a1' are considered to form a
+fixed-point value with binary point between `a0' and `a1'.  This fixed-point
+value is shifted right by the number of bits given in `count', and the
+integer part of the result is returned at the location pointed to by
+`z0Ptr'.  The fractional part of the result may be slightly corrupted as
+described above, and is returned at the location pointed to by `z1Ptr'.)
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ shift64ExtraRightJamming(
+     bits64 a0, bits64 a1, int16 count, bits64 *z0Ptr, bits64 *z1Ptr )
+{
+    bits64 z0, z1;
+    int8 negCount = ( - count ) & 63;
+
+    if ( count == 0 ) {
+        z1 = a1;
+        z0 = a0;
+    }
+    else if ( count < 64 ) {
+        z1 = ( a0<<negCount ) | ( a1 != 0 );
+        z0 = a0>>count;
+    }
+    else {
+        if ( count == 64 ) {
+            z1 = a0 | ( a1 != 0 );
+        }
+        else {
+            z1 = ( ( a0 | a1 ) != 0 );
+        }
+        z0 = 0;
+    }
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Shifts the 128-bit value formed by concatenating `a0' and `a1' right by the
+number of bits given in `count'.  Any bits shifted off are lost.  The value
+of `count' can be arbitrarily large; in particular, if `count' is greater
+than 128, the result will be 0.  The result is broken into two 64-bit pieces
+which are stored at the locations pointed to by `z0Ptr' and `z1Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ shift128Right(
+     bits64 a0, bits64 a1, int16 count, bits64 *z0Ptr, bits64 *z1Ptr )
+{
+    bits64 z0, z1;
+    int8 negCount = ( - count ) & 63;
+
+    if ( count == 0 ) {
+        z1 = a1;
+        z0 = a0;
+    }
+    else if ( count < 64 ) {
+        z1 = ( a0<<negCount ) | ( a1>>count );
+        z0 = a0>>count;
+    }
+    else {
+        z1 = ( count < 64 ) ? ( a0>>( count & 63 ) ) : 0;
+        z0 = 0;
+    }
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Shifts the 128-bit value formed by concatenating `a0' and `a1' right by the
+number of bits given in `count'.  If any nonzero bits are shifted off, they
+are ``jammed'' into the least significant bit of the result by setting the
+least significant bit to 1.  The value of `count' can be arbitrarily large;
+in particular, if `count' is greater than 128, the result will be either 0
+or 1, depending on whether the concatenation of `a0' and `a1' is zero or
+nonzero.  The result is broken into two 64-bit pieces which are stored at
+the locations pointed to by `z0Ptr' and `z1Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ shift128RightJamming(
+     bits64 a0, bits64 a1, int16 count, bits64 *z0Ptr, bits64 *z1Ptr )
+{
+    bits64 z0, z1;
+    int8 negCount = ( - count ) & 63;
+
+    if ( count == 0 ) {
+        z1 = a1;
+        z0 = a0;
+    }
+    else if ( count < 64 ) {
+        z1 = ( a0<<negCount ) | ( a1>>count ) | ( ( a1<<negCount ) != 0 );
+        z0 = a0>>count;
+    }
+    else {
+        if ( count == 64 ) {
+            z1 = a0 | ( a1 != 0 );
+        }
+        else if ( count < 128 ) {
+            z1 = ( a0>>( count & 63 ) ) | ( ( ( a0<<negCount ) | a1 ) != 0 );
+        }
+        else {
+            z1 = ( ( a0 | a1 ) != 0 );
+        }
+        z0 = 0;
+    }
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Shifts the 192-bit value formed by concatenating `a0', `a1', and `a2' right
+by 64 _plus_ the number of bits given in `count'.  The shifted result is
+at most 128 nonzero bits; these are broken into two 64-bit pieces which are
+stored at the locations pointed to by `z0Ptr' and `z1Ptr'.  The bits shifted
+off form a third 64-bit result as follows:  The _last_ bit shifted off is
+the most-significant bit of the extra result, and the other 63 bits of the
+extra result are all zero if and only if _all_but_the_last_ bits shifted off
+were all zero.  This extra result is stored in the location pointed to by
+`z2Ptr'.  The value of `count' can be arbitrarily large.
+    (This routine makes more sense if `a0', `a1', and `a2' are considered
+to form a fixed-point value with binary point between `a1' and `a2'.  This
+fixed-point value is shifted right by the number of bits given in `count',
+and the integer part of the result is returned at the locations pointed to
+by `z0Ptr' and `z1Ptr'.  The fractional part of the result may be slightly
+corrupted as described above, and is returned at the location pointed to by
+`z2Ptr'.)
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ shift128ExtraRightJamming(
+     bits64 a0,
+     bits64 a1,
+     bits64 a2,
+     int16 count,
+     bits64 *z0Ptr,
+     bits64 *z1Ptr,
+     bits64 *z2Ptr
+ )
+{
+    bits64 z0, z1, z2;
+    int8 negCount = ( - count ) & 63;
+
+    if ( count == 0 ) {
+        z2 = a2;
+        z1 = a1;
+        z0 = a0;
+    }
+    else {
+        if ( count < 64 ) {
+            z2 = a1<<negCount;
+            z1 = ( a0<<negCount ) | ( a1>>count );
+            z0 = a0>>count;
+        }
+        else {
+            if ( count == 64 ) {
+                z2 = a1;
+                z1 = a0;
+            }
+            else {
+                a2 |= a1;
+                if ( count < 128 ) {
+                    z2 = a0<<negCount;
+                    z1 = a0>>( count & 63 );
+                }
+                else {
+                    z2 = ( count == 128 ) ? a0 : ( a0 != 0 );
+                    z1 = 0;
+                }
+            }
+            z0 = 0;
+        }
+        z2 |= ( a2 != 0 );
+    }
+    *z2Ptr = z2;
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Shifts the 128-bit value formed by concatenating `a0' and `a1' left by the
+number of bits given in `count'.  Any bits shifted off are lost.  The value
+of `count' must be less than 64.  The result is broken into two 64-bit
+pieces which are stored at the locations pointed to by `z0Ptr' and `z1Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ shortShift128Left(
+     bits64 a0, bits64 a1, int16 count, bits64 *z0Ptr, bits64 *z1Ptr )
+{
+
+    *z1Ptr = a1<<count;
+    *z0Ptr =
+        ( count == 0 ) ? a0 : ( a0<<count ) | ( a1>>( ( - count ) & 63 ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Shifts the 192-bit value formed by concatenating `a0', `a1', and `a2' left
+by the number of bits given in `count'.  Any bits shifted off are lost.
+The value of `count' must be less than 64.  The result is broken into three
+64-bit pieces which are stored at the locations pointed to by `z0Ptr',
+`z1Ptr', and `z2Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ shortShift192Left(
+     bits64 a0,
+     bits64 a1,
+     bits64 a2,
+     int16 count,
+     bits64 *z0Ptr,
+     bits64 *z1Ptr,
+     bits64 *z2Ptr
+ )
+{
+    bits64 z0, z1, z2;
+    int8 negCount;
+
+    z2 = a2<<count;
+    z1 = a1<<count;
+    z0 = a0<<count;
+    if ( 0 < count ) {
+        negCount = ( ( - count ) & 63 );
+        z1 |= a2>>negCount;
+        z0 |= a1>>negCount;
+    }
+    *z2Ptr = z2;
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Adds the 128-bit value formed by concatenating `a0' and `a1' to the 128-bit
+value formed by concatenating `b0' and `b1'.  Addition is modulo 2^128, so
+any carry out is lost.  The result is broken into two 64-bit pieces which
+are stored at the locations pointed to by `z0Ptr' and `z1Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ add128(
+     bits64 a0, bits64 a1, bits64 b0, bits64 b1, bits64 *z0Ptr, bits64 *z1Ptr )
+{
+    bits64 z1;
+
+    z1 = a1 + b1;
+    *z1Ptr = z1;
+    *z0Ptr = a0 + b0 + ( z1 < a1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Adds the 192-bit value formed by concatenating `a0', `a1', and `a2' to the
+192-bit value formed by concatenating `b0', `b1', and `b2'.  Addition is
+modulo 2^192, so any carry out is lost.  The result is broken into three
+64-bit pieces which are stored at the locations pointed to by `z0Ptr',
+`z1Ptr', and `z2Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ add192(
+     bits64 a0,
+     bits64 a1,
+     bits64 a2,
+     bits64 b0,
+     bits64 b1,
+     bits64 b2,
+     bits64 *z0Ptr,
+     bits64 *z1Ptr,
+     bits64 *z2Ptr
+ )
+{
+    bits64 z0, z1, z2;
+    int8 carry0, carry1;
+
+    z2 = a2 + b2;
+    carry1 = ( z2 < a2 );
+    z1 = a1 + b1;
+    carry0 = ( z1 < a1 );
+    z0 = a0 + b0;
+    z1 += carry1;
+    z0 += ( z1 < carry1 );
+    z0 += carry0;
+    *z2Ptr = z2;
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Subtracts the 128-bit value formed by concatenating `b0' and `b1' from the
+128-bit value formed by concatenating `a0' and `a1'.  Subtraction is modulo
+2^128, so any borrow out (carry out) is lost.  The result is broken into two
+64-bit pieces which are stored at the locations pointed to by `z0Ptr' and
+`z1Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ sub128(
+     bits64 a0, bits64 a1, bits64 b0, bits64 b1, bits64 *z0Ptr, bits64 *z1Ptr )
+{
+
+    *z1Ptr = a1 - b1;
+    *z0Ptr = a0 - b0 - ( a1 < b1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Subtracts the 192-bit value formed by concatenating `b0', `b1', and `b2'
+from the 192-bit value formed by concatenating `a0', `a1', and `a2'.
+Subtraction is modulo 2^192, so any borrow out (carry out) is lost.  The
+result is broken into three 64-bit pieces which are stored at the locations
+pointed to by `z0Ptr', `z1Ptr', and `z2Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ sub192(
+     bits64 a0,
+     bits64 a1,
+     bits64 a2,
+     bits64 b0,
+     bits64 b1,
+     bits64 b2,
+     bits64 *z0Ptr,
+     bits64 *z1Ptr,
+     bits64 *z2Ptr
+ )
+{
+    bits64 z0, z1, z2;
+    int8 borrow0, borrow1;
+
+    z2 = a2 - b2;
+    borrow1 = ( a2 < b2 );
+    z1 = a1 - b1;
+    borrow0 = ( a1 < b1 );
+    z0 = a0 - b0;
+    z0 -= ( z1 < borrow1 );
+    z1 -= borrow1;
+    z0 -= borrow0;
+    *z2Ptr = z2;
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Multiplies `a' by `b' to obtain a 128-bit product.  The product is broken
+into two 64-bit pieces which are stored at the locations pointed to by
+`z0Ptr' and `z1Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void mul64To128( bits64 a, bits64 b, bits64 *z0Ptr, bits64 *z1Ptr )
+{
+    bits32 aHigh, aLow, bHigh, bLow;
+    bits64 z0, zMiddleA, zMiddleB, z1;
+
+    aLow = a;
+    aHigh = a>>32;
+    bLow = b;
+    bHigh = b>>32;
+    z1 = ( (bits64) aLow ) * bLow;
+    zMiddleA = ( (bits64) aLow ) * bHigh;
+    zMiddleB = ( (bits64) aHigh ) * bLow;
+    z0 = ( (bits64) aHigh ) * bHigh;
+    zMiddleA += zMiddleB;
+    z0 += ( ( (bits64) ( zMiddleA < zMiddleB ) )<<32 ) + ( zMiddleA>>32 );
+    zMiddleA <<= 32;
+    z1 += zMiddleA;
+    z0 += ( z1 < zMiddleA );
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Multiplies the 128-bit value formed by concatenating `a0' and `a1' by `b' to
+obtain a 192-bit product.  The product is broken into three 64-bit pieces
+which are stored at the locations pointed to by `z0Ptr', `z1Ptr', and
+`z2Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ mul128By64To192(
+     bits64 a0,
+     bits64 a1,
+     bits64 b,
+     bits64 *z0Ptr,
+     bits64 *z1Ptr,
+     bits64 *z2Ptr
+ )
+{
+    bits64 z0, z1, z2, more1;
+
+    mul64To128( a1, b, &z1, &z2 );
+    mul64To128( a0, b, &z0, &more1 );
+    add128( z0, more1, 0, z1, &z0, &z1 );
+    *z2Ptr = z2;
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Multiplies the 128-bit value formed by concatenating `a0' and `a1' to the
+128-bit value formed by concatenating `b0' and `b1' to obtain a 256-bit
+product.  The product is broken into four 64-bit pieces which are stored at
+the locations pointed to by `z0Ptr', `z1Ptr', `z2Ptr', and `z3Ptr'.
+-------------------------------------------------------------------------------
+*/
+INLINE void
+ mul128To256(
+     bits64 a0,
+     bits64 a1,
+     bits64 b0,
+     bits64 b1,
+     bits64 *z0Ptr,
+     bits64 *z1Ptr,
+     bits64 *z2Ptr,
+     bits64 *z3Ptr
+ )
+{
+    bits64 z0, z1, z2, z3;
+    bits64 more1, more2;
+
+    mul64To128( a1, b1, &z2, &z3 );
+    mul64To128( a1, b0, &z1, &more2 );
+    add128( z1, more2, 0, z2, &z1, &z2 );
+    mul64To128( a0, b0, &z0, &more1 );
+    add128( z0, more1, 0, z1, &z0, &z1 );
+    mul64To128( a0, b1, &more1, &more2 );
+    add128( more1, more2, 0, z2, &more1, &z2 );
+    add128( z0, z1, 0, more1, &z0, &z1 );
+    *z3Ptr = z3;
+    *z2Ptr = z2;
+    *z1Ptr = z1;
+    *z0Ptr = z0;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns an approximation to the 64-bit integer quotient obtained by dividing
+`b' into the 128-bit value formed by concatenating `a0' and `a1'.  The
+divisor `b' must be at least 2^63.  If q is the exact quotient truncated
+toward zero, the approximation returned lies between q and q + 2 inclusive.
+If the exact quotient q is larger than 64 bits, the maximum positive 64-bit
+unsigned integer is returned.
+-------------------------------------------------------------------------------
+*/
+static bits64 estimateDiv128To64( bits64 a0, bits64 a1, bits64 b )
+{
+    bits64 b0, b1;
+    bits64 rem0, rem1, term0, term1;
+    bits64 z;
+    if ( b <= a0 ) return LIT64( 0xFFFFFFFFFFFFFFFF );
+    b0 = b>>32;  /* hence b0 is 32 bits wide now */
+    if ( b0<<32 <= a0 ) {
+        z = LIT64( 0xFFFFFFFF00000000 );
+    }  else {
+        z = a0;
+        do_div( z, b0 );
+        z <<= 32;
+    }
+    mul64To128( b, z, &term0, &term1 );
+    sub128( a0, a1, term0, term1, &rem0, &rem1 );
+    while ( ( (sbits64) rem0 ) < 0 ) {
+        z -= LIT64( 0x100000000 );
+        b1 = b<<32;
+        add128( rem0, rem1, b0, b1, &rem0, &rem1 );
+    }
+    rem0 = ( rem0<<32 ) | ( rem1>>32 );
+    if ( b0<<32 <= rem0 ) {
+        z |= 0xFFFFFFFF;
+    } else {
+        do_div( rem0, b0 );
+        z |= rem0;
+    }
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns an approximation to the square root of the 32-bit significand given
+by `a'.  Considered as an integer, `a' must be at least 2^31.  If bit 0 of
+`aExp' (the least significant bit) is 1, the integer returned approximates
+2^31*sqrt(`a'/2^31), where `a' is considered an integer.  If bit 0 of `aExp'
+is 0, the integer returned approximates 2^31*sqrt(`a'/2^30).  In either
+case, the approximation returned lies strictly within +/-2 of the exact
+value.
+-------------------------------------------------------------------------------
+*/
+static bits32 estimateSqrt32( int16 aExp, bits32 a )
+{
+    static const bits16 sqrtOddAdjustments[] = {
+        0x0004, 0x0022, 0x005D, 0x00B1, 0x011D, 0x019F, 0x0236, 0x02E0,
+        0x039C, 0x0468, 0x0545, 0x0631, 0x072B, 0x0832, 0x0946, 0x0A67
+    };
+    static const bits16 sqrtEvenAdjustments[] = {
+        0x0A2D, 0x08AF, 0x075A, 0x0629, 0x051A, 0x0429, 0x0356, 0x029E,
+        0x0200, 0x0179, 0x0109, 0x00AF, 0x0068, 0x0034, 0x0012, 0x0002
+    };
+    int8 index;
+    bits32 z;
+    bits64 A;
+
+    index = ( a>>27 ) & 15;
+    if ( aExp & 1 ) {
+        z = 0x4000 + ( a>>17 ) - sqrtOddAdjustments[ index ];
+        z = ( ( a / z )<<14 ) + ( z<<15 );
+        a >>= 1;
+    }
+    else {
+        z = 0x8000 + ( a>>17 ) - sqrtEvenAdjustments[ index ];
+        z = a / z + z;
+        z = ( 0x20000 <= z ) ? 0xFFFF8000 : ( z<<15 );
+        if ( z <= a ) return (bits32) ( ( (sbits32) a )>>1 );
+    }
+    A = ( (bits64) a )<<31;
+    do_div( A, z );
+    return ( (bits32) A ) + ( z>>1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the number of leading 0 bits before the most-significant 1 bit
+of `a'.  If `a' is zero, 32 is returned.
+-------------------------------------------------------------------------------
+*/
+static int8 countLeadingZeros32( bits32 a )
+{
+    static const int8 countLeadingZerosHigh[] = {
+        8, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4,
+        3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
+        2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+        2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
+    };
+    int8 shiftCount;
+
+    shiftCount = 0;
+    if ( a < 0x10000 ) {
+        shiftCount += 16;
+        a <<= 16;
+    }
+    if ( a < 0x1000000 ) {
+        shiftCount += 8;
+        a <<= 8;
+    }
+    shiftCount += countLeadingZerosHigh[ a>>24 ];
+    return shiftCount;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the number of leading 0 bits before the most-significant 1 bit
+of `a'.  If `a' is zero, 64 is returned.
+-------------------------------------------------------------------------------
+*/
+static int8 countLeadingZeros64( bits64 a )
+{
+    int8 shiftCount;
+
+    shiftCount = 0;
+    if ( a < ( (bits64) 1 )<<32 ) {
+        shiftCount += 32;
+    }
+    else {
+        a >>= 32;
+    }
+    shiftCount += countLeadingZeros32( a );
+    return shiftCount;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the 128-bit value formed by concatenating `a0' and `a1'
+is equal to the 128-bit value formed by concatenating `b0' and `b1'.
+Otherwise, returns 0.
+-------------------------------------------------------------------------------
+*/
+INLINE flag eq128( bits64 a0, bits64 a1, bits64 b0, bits64 b1 )
+{
+
+    return ( a0 == b0 ) && ( a1 == b1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the 128-bit value formed by concatenating `a0' and `a1' is less
+than or equal to the 128-bit value formed by concatenating `b0' and `b1'.
+Otherwise, returns 0.
+-------------------------------------------------------------------------------
+*/
+INLINE flag le128( bits64 a0, bits64 a1, bits64 b0, bits64 b1 )
+{
+
+    return ( a0 < b0 ) || ( ( a0 == b0 ) && ( a1 <= b1 ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the 128-bit value formed by concatenating `a0' and `a1' is less
+than the 128-bit value formed by concatenating `b0' and `b1'.  Otherwise,
+returns 0.
+-------------------------------------------------------------------------------
+*/
+INLINE flag lt128( bits64 a0, bits64 a1, bits64 b0, bits64 b1 )
+{
+
+    return ( a0 < b0 ) || ( ( a0 == b0 ) && ( a1 < b1 ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the 128-bit value formed by concatenating `a0' and `a1' is
+not equal to the 128-bit value formed by concatenating `b0' and `b1'.
+Otherwise, returns 0.
+-------------------------------------------------------------------------------
+*/
+INLINE flag ne128( bits64 a0, bits64 a1, bits64 b0, bits64 b1 )
+{
+
+    return ( a0 != b0 ) || ( a1 != b1 );
+
+}
+

arch/arm/nwfpe/softfloat-specialize

@@ -0,0 +1,367 @@
+
+/*
+===============================================================================
+
+This C source fragment is part of the SoftFloat IEC/IEEE Floating-point
+Arithmetic Package, Release 2.
+
+Written by John R. Hauser.  This work was made possible in part by the
+International Computer Science Institute, located at Suite 600, 1947 Center
+Street, Berkeley, California 94704.  Funding was partially provided by the
+National Science Foundation under grant MIP-9311980.  The original version
+of this code was written as part of a project to build a fixed-point vector
+processor in collaboration with the University of California at Berkeley,
+overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
+is available through the Web page
+http://www.jhauser.us/arithmetic/SoftFloat-2b/SoftFloat-source.txt
+
+THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort
+has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
+TIMES RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO
+PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
+AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
+
+Derivative works are acceptable, even for commercial purposes, so long as
+(1) they include prominent notice that the work is derivative, and (2) they
+include prominent notice akin to these three paragraphs for those parts of
+this code that are retained.
+
+===============================================================================
+*/
+
+/*
+-------------------------------------------------------------------------------
+Underflow tininess-detection mode, statically initialized to default value.
+(The declaration in `softfloat.h' must match the `int8' type here.)
+-------------------------------------------------------------------------------
+*/
+int8 float_detect_tininess = float_tininess_after_rounding;
+
+/*
+-------------------------------------------------------------------------------
+Raises the exceptions specified by `flags'.  Floating-point traps can be
+defined here if desired.  It is currently not possible for such a trap to
+substitute a result value.  If traps are not implemented, this routine
+should be simply `float_exception_flags |= flags;'.
+
+ScottB:  November 4, 1998
+Moved this function out of softfloat-specialize into fpmodule.c.
+This effectively isolates all the changes required for integrating with the
+Linux kernel into fpmodule.c.  Porting to NetBSD should only require modifying
+fpmodule.c to integrate with the NetBSD kernel (I hope!).
+-------------------------------------------------------------------------------
+void float_raise( int8 flags )
+{
+    float_exception_flags |= flags;
+}
+*/
+
+/*
+-------------------------------------------------------------------------------
+Internal canonical NaN format.
+-------------------------------------------------------------------------------
+*/
+typedef struct {
+    flag sign;
+    bits64 high, low;
+} commonNaNT;
+
+/*
+-------------------------------------------------------------------------------
+The pattern for a default generated single-precision NaN.
+-------------------------------------------------------------------------------
+*/
+#define float32_default_nan 0xFFFFFFFF
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the single-precision floating-point value `a' is a NaN;
+otherwise returns 0.
+-------------------------------------------------------------------------------
+*/
+flag float32_is_nan( float32 a )
+{
+
+    return ( 0xFF000000 < (bits32) ( a<<1 ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the single-precision floating-point value `a' is a signaling
+NaN; otherwise returns 0.
+-------------------------------------------------------------------------------
+*/
+flag float32_is_signaling_nan( float32 a )
+{
+
+    return ( ( ( a>>22 ) & 0x1FF ) == 0x1FE ) && ( a & 0x003FFFFF );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the single-precision floating-point NaN
+`a' to the canonical NaN format.  If `a' is a signaling NaN, the invalid
+exception is raised.
+-------------------------------------------------------------------------------
+*/
+static commonNaNT float32ToCommonNaN( float32 a )
+{
+    commonNaNT z;
+
+    if ( float32_is_signaling_nan( a ) ) float_raise( float_flag_invalid );
+    z.sign = a>>31;
+    z.low = 0;
+    z.high = ( (bits64) a )<<41;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the canonical NaN `a' to the single-
+precision floating-point format.
+-------------------------------------------------------------------------------
+*/
+static float32 commonNaNToFloat32( commonNaNT a )
+{
+
+    return ( ( (bits32) a.sign )<<31 ) | 0x7FC00000 | ( a.high>>41 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes two single-precision floating-point values `a' and `b', one of which
+is a NaN, and returns the appropriate NaN result.  If either `a' or `b' is a
+signaling NaN, the invalid exception is raised.
+-------------------------------------------------------------------------------
+*/
+static float32 propagateFloat32NaN( float32 a, float32 b )
+{
+    flag aIsNaN, aIsSignalingNaN, bIsNaN, bIsSignalingNaN;
+
+    aIsNaN = float32_is_nan( a );
+    aIsSignalingNaN = float32_is_signaling_nan( a );
+    bIsNaN = float32_is_nan( b );
+    bIsSignalingNaN = float32_is_signaling_nan( b );
+    a |= 0x00400000;
+    b |= 0x00400000;
+    if ( aIsSignalingNaN | bIsSignalingNaN ) float_raise( float_flag_invalid );
+    if ( aIsNaN ) {
+        return ( aIsSignalingNaN & bIsNaN ) ? b : a;
+    }
+    else {
+        return b;
+    }
+
+}
+
+/*
+-------------------------------------------------------------------------------
+The pattern for a default generated double-precision NaN.
+-------------------------------------------------------------------------------
+*/
+#define float64_default_nan LIT64( 0xFFFFFFFFFFFFFFFF )
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the double-precision floating-point value `a' is a NaN;
+otherwise returns 0.
+-------------------------------------------------------------------------------
+*/
+flag float64_is_nan( float64 a )
+{
+
+    return ( LIT64( 0xFFE0000000000000 ) < (bits64) ( a<<1 ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the double-precision floating-point value `a' is a signaling
+NaN; otherwise returns 0.
+-------------------------------------------------------------------------------
+*/
+flag float64_is_signaling_nan( float64 a )
+{
+
+    return
+           ( ( ( a>>51 ) & 0xFFF ) == 0xFFE )
+        && ( a & LIT64( 0x0007FFFFFFFFFFFF ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the double-precision floating-point NaN
+`a' to the canonical NaN format.  If `a' is a signaling NaN, the invalid
+exception is raised.
+-------------------------------------------------------------------------------
+*/
+static commonNaNT float64ToCommonNaN( float64 a )
+{
+    commonNaNT z;
+
+    if ( float64_is_signaling_nan( a ) ) float_raise( float_flag_invalid );
+    z.sign = a>>63;
+    z.low = 0;
+    z.high = a<<12;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the canonical NaN `a' to the double-
+precision floating-point format.
+-------------------------------------------------------------------------------
+*/
+static float64 commonNaNToFloat64( commonNaNT a )
+{
+
+    return
+          ( ( (bits64) a.sign )<<63 )
+        | LIT64( 0x7FF8000000000000 )
+        | ( a.high>>12 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes two double-precision floating-point values `a' and `b', one of which
+is a NaN, and returns the appropriate NaN result.  If either `a' or `b' is a
+signaling NaN, the invalid exception is raised.
+-------------------------------------------------------------------------------
+*/
+static float64 propagateFloat64NaN( float64 a, float64 b )
+{
+    flag aIsNaN, aIsSignalingNaN, bIsNaN, bIsSignalingNaN;
+
+    aIsNaN = float64_is_nan( a );
+    aIsSignalingNaN = float64_is_signaling_nan( a );
+    bIsNaN = float64_is_nan( b );
+    bIsSignalingNaN = float64_is_signaling_nan( b );
+    a |= LIT64( 0x0008000000000000 );
+    b |= LIT64( 0x0008000000000000 );
+    if ( aIsSignalingNaN | bIsSignalingNaN ) float_raise( float_flag_invalid );
+    if ( aIsNaN ) {
+        return ( aIsSignalingNaN & bIsNaN ) ? b : a;
+    }
+    else {
+        return b;
+    }
+
+}
+
+#ifdef FLOATX80
+
+/*
+-------------------------------------------------------------------------------
+The pattern for a default generated extended double-precision NaN.  The
+`high' and `low' values hold the most- and least-significant bits,
+respectively.
+-------------------------------------------------------------------------------
+*/
+#define floatx80_default_nan_high 0xFFFF
+#define floatx80_default_nan_low  LIT64( 0xFFFFFFFFFFFFFFFF )
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the extended double-precision floating-point value `a' is a
+NaN; otherwise returns 0.
+-------------------------------------------------------------------------------
+*/
+flag floatx80_is_nan( floatx80 a )
+{
+
+    return ( ( a.high & 0x7FFF ) == 0x7FFF ) && (bits64) ( a.low<<1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the extended double-precision floating-point value `a' is a
+signaling NaN; otherwise returns 0.
+-------------------------------------------------------------------------------
+*/
+flag floatx80_is_signaling_nan( floatx80 a )
+{
+    //register int lr;
+    bits64 aLow;
+
+    //__asm__("mov %0, lr" : : "g" (lr));
+    //fp_printk("floatx80_is_signalling_nan() called from 0x%08x\n",lr);
+    aLow = a.low & ~ LIT64( 0x4000000000000000 );
+    return
+           ( ( a.high & 0x7FFF ) == 0x7FFF )
+        && (bits64) ( aLow<<1 )
+        && ( a.low == aLow );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the extended double-precision floating-
+point NaN `a' to the canonical NaN format.  If `a' is a signaling NaN, the
+invalid exception is raised.
+-------------------------------------------------------------------------------
+*/
+static commonNaNT floatx80ToCommonNaN( floatx80 a )
+{
+    commonNaNT z;
+
+    if ( floatx80_is_signaling_nan( a ) ) float_raise( float_flag_invalid );
+    z.sign = a.high>>15;
+    z.low = 0;
+    z.high = a.low<<1;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the canonical NaN `a' to the extended
+double-precision floating-point format.
+-------------------------------------------------------------------------------
+*/
+static floatx80 commonNaNToFloatx80( commonNaNT a )
+{
+    floatx80 z;
+
+    z.low = LIT64( 0xC000000000000000 ) | ( a.high>>1 );
+    z.high = ( ( (bits16) a.sign )<<15 ) | 0x7FFF;
+    z.__padding = 0;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes two extended double-precision floating-point values `a' and `b', one
+of which is a NaN, and returns the appropriate NaN result.  If either `a' or
+`b' is a signaling NaN, the invalid exception is raised.
+-------------------------------------------------------------------------------
+*/
+static floatx80 propagateFloatx80NaN( floatx80 a, floatx80 b )
+{
+    flag aIsNaN, aIsSignalingNaN, bIsNaN, bIsSignalingNaN;
+
+    aIsNaN = floatx80_is_nan( a );
+    aIsSignalingNaN = floatx80_is_signaling_nan( a );
+    bIsNaN = floatx80_is_nan( b );
+    bIsSignalingNaN = floatx80_is_signaling_nan( b );
+    a.low |= LIT64( 0xC000000000000000 );
+    b.low |= LIT64( 0xC000000000000000 );
+    if ( aIsSignalingNaN | bIsSignalingNaN ) float_raise( float_flag_invalid );
+    if ( aIsNaN ) {
+        return ( aIsSignalingNaN & bIsNaN ) ? b : a;
+    }
+    else {
+        return b;
+    }
+
+}
+
+#endif

arch/arm/nwfpe/softfloat.c

@@ -0,0 +1,3435 @@
+/*
+===============================================================================
+
+This C source file is part of the SoftFloat IEC/IEEE Floating-point
+Arithmetic Package, Release 2.
+
+Written by John R. Hauser.  This work was made possible in part by the
+International Computer Science Institute, located at Suite 600, 1947 Center
+Street, Berkeley, California 94704.  Funding was partially provided by the
+National Science Foundation under grant MIP-9311980.  The original version
+of this code was written as part of a project to build a fixed-point vector
+processor in collaboration with the University of California at Berkeley,
+overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
+is available through the web page
+http://www.jhauser.us/arithmetic/SoftFloat-2b/SoftFloat-source.txt
+
+THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort
+has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
+TIMES RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO
+PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
+AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
+
+Derivative works are acceptable, even for commercial purposes, so long as
+(1) they include prominent notice that the work is derivative, and (2) they
+include prominent notice akin to these three paragraphs for those parts of
+this code that are retained.
+
+===============================================================================
+*/
+
+#include <asm/div64.h>
+
+#include "fpa11.h"
+//#include "milieu.h"
+//#include "softfloat.h"
+
+/*
+-------------------------------------------------------------------------------
+Primitive arithmetic functions, including multi-word arithmetic, and
+division and square root approximations.  (Can be specialized to target if
+desired.)
+-------------------------------------------------------------------------------
+*/
+#include "softfloat-macros"
+
+/*
+-------------------------------------------------------------------------------
+Functions and definitions to determine:  (1) whether tininess for underflow
+is detected before or after rounding by default, (2) what (if anything)
+happens when exceptions are raised, (3) how signaling NaNs are distinguished
+from quiet NaNs, (4) the default generated quiet NaNs, and (5) how NaNs
+are propagated from function inputs to output.  These details are target-
+specific.
+-------------------------------------------------------------------------------
+*/
+#include "softfloat-specialize"
+
+/*
+-------------------------------------------------------------------------------
+Takes a 64-bit fixed-point value `absZ' with binary point between bits 6
+and 7, and returns the properly rounded 32-bit integer corresponding to the
+input.  If `zSign' is nonzero, the input is negated before being converted
+to an integer.  Bit 63 of `absZ' must be zero.  Ordinarily, the fixed-point
+input is simply rounded to an integer, with the inexact exception raised if
+the input cannot be represented exactly as an integer.  If the fixed-point
+input is too large, however, the invalid exception is raised and the largest
+positive or negative integer is returned.
+-------------------------------------------------------------------------------
+*/
+static int32 roundAndPackInt32( struct roundingData *roundData, flag zSign, bits64 absZ )
+{
+    int8 roundingMode;
+    flag roundNearestEven;
+    int8 roundIncrement, roundBits;
+    int32 z;
+
+    roundingMode = roundData->mode;
+    roundNearestEven = ( roundingMode == float_round_nearest_even );
+    roundIncrement = 0x40;
+    if ( ! roundNearestEven ) {
+        if ( roundingMode == float_round_to_zero ) {
+            roundIncrement = 0;
+        }
+        else {
+            roundIncrement = 0x7F;
+            if ( zSign ) {
+                if ( roundingMode == float_round_up ) roundIncrement = 0;
+            }
+            else {
+                if ( roundingMode == float_round_down ) roundIncrement = 0;
+            }
+        }
+    }
+    roundBits = absZ & 0x7F;
+    absZ = ( absZ + roundIncrement )>>7;
+    absZ &= ~ ( ( ( roundBits ^ 0x40 ) == 0 ) & roundNearestEven );
+    z = absZ;
+    if ( zSign ) z = - z;
+    if ( ( absZ>>32 ) || ( z && ( ( z < 0 ) ^ zSign ) ) ) {
+        roundData->exception |= float_flag_invalid;
+        return zSign ? 0x80000000 : 0x7FFFFFFF;
+    }
+    if ( roundBits ) roundData->exception |= float_flag_inexact;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the fraction bits of the single-precision floating-point value `a'.
+-------------------------------------------------------------------------------
+*/
+INLINE bits32 extractFloat32Frac( float32 a )
+{
+
+    return a & 0x007FFFFF;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the exponent bits of the single-precision floating-point value `a'.
+-------------------------------------------------------------------------------
+*/
+INLINE int16 extractFloat32Exp( float32 a )
+{
+
+    return ( a>>23 ) & 0xFF;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the sign bit of the single-precision floating-point value `a'.
+-------------------------------------------------------------------------------
+*/
+#if 0	/* in softfloat.h */
+INLINE flag extractFloat32Sign( float32 a )
+{
+
+    return a>>31;
+
+}
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Normalizes the subnormal single-precision floating-point value represented
+by the denormalized significand `aSig'.  The normalized exponent and
+significand are stored at the locations pointed to by `zExpPtr' and
+`zSigPtr', respectively.
+-------------------------------------------------------------------------------
+*/
+static void
+ normalizeFloat32Subnormal( bits32 aSig, int16 *zExpPtr, bits32 *zSigPtr )
+{
+    int8 shiftCount;
+
+    shiftCount = countLeadingZeros32( aSig ) - 8;
+    *zSigPtr = aSig<<shiftCount;
+    *zExpPtr = 1 - shiftCount;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Packs the sign `zSign', exponent `zExp', and significand `zSig' into a
+single-precision floating-point value, returning the result.  After being
+shifted into the proper positions, the three fields are simply added
+together to form the result.  This means that any integer portion of `zSig'
+will be added into the exponent.  Since a properly normalized significand
+will have an integer portion equal to 1, the `zExp' input should be 1 less
+than the desired result exponent whenever `zSig' is a complete, normalized
+significand.
+-------------------------------------------------------------------------------
+*/
+INLINE float32 packFloat32( flag zSign, int16 zExp, bits32 zSig )
+{
+#if 0
+   float32 f;
+   __asm__("@ packFloat32				\n\
+   	    mov %0, %1, asl #31				\n\
+   	    orr %0, %2, asl #23				\n\
+   	    orr %0, %3"
+   	    : /* no outputs */
+   	    : "g" (f), "g" (zSign), "g" (zExp), "g" (zSig)
+   	    : "cc");
+   return f;
+#else
+    return ( ( (bits32) zSign )<<31 ) + ( ( (bits32) zExp )<<23 ) + zSig;
+#endif 
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes an abstract floating-point value having sign `zSign', exponent `zExp',
+and significand `zSig', and returns the proper single-precision floating-
+point value corresponding to the abstract input.  Ordinarily, the abstract
+value is simply rounded and packed into the single-precision format, with
+the inexact exception raised if the abstract input cannot be represented
+exactly.  If the abstract value is too large, however, the overflow and
+inexact exceptions are raised and an infinity or maximal finite value is
+returned.  If the abstract value is too small, the input value is rounded to
+a subnormal number, and the underflow and inexact exceptions are raised if
+the abstract input cannot be represented exactly as a subnormal single-
+precision floating-point number.
+    The input significand `zSig' has its binary point between bits 30
+and 29, which is 7 bits to the left of the usual location.  This shifted
+significand must be normalized or smaller.  If `zSig' is not normalized,
+`zExp' must be 0; in that case, the result returned is a subnormal number,
+and it must not require rounding.  In the usual case that `zSig' is
+normalized, `zExp' must be 1 less than the ``true'' floating-point exponent.
+The handling of underflow and overflow follows the IEC/IEEE Standard for
+Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static float32 roundAndPackFloat32( struct roundingData *roundData, flag zSign, int16 zExp, bits32 zSig )
+{
+    int8 roundingMode;
+    flag roundNearestEven;
+    int8 roundIncrement, roundBits;
+    flag isTiny;
+
+    roundingMode = roundData->mode;
+    roundNearestEven = ( roundingMode == float_round_nearest_even );
+    roundIncrement = 0x40;
+    if ( ! roundNearestEven ) {
+        if ( roundingMode == float_round_to_zero ) {
+            roundIncrement = 0;
+        }
+        else {
+            roundIncrement = 0x7F;
+            if ( zSign ) {
+                if ( roundingMode == float_round_up ) roundIncrement = 0;
+            }
+            else {
+                if ( roundingMode == float_round_down ) roundIncrement = 0;
+            }
+        }
+    }
+    roundBits = zSig & 0x7F;
+    if ( 0xFD <= (bits16) zExp ) {
+        if (    ( 0xFD < zExp )
+             || (    ( zExp == 0xFD )
+                  && ( (sbits32) ( zSig + roundIncrement ) < 0 ) )
+           ) {
+            roundData->exception |= float_flag_overflow | float_flag_inexact;
+            return packFloat32( zSign, 0xFF, 0 ) - ( roundIncrement == 0 );
+        }
+        if ( zExp < 0 ) {
+            isTiny =
+                   ( float_detect_tininess == float_tininess_before_rounding )
+                || ( zExp < -1 )
+                || ( zSig + roundIncrement < 0x80000000 );
+            shift32RightJamming( zSig, - zExp, &zSig );
+            zExp = 0;
+            roundBits = zSig & 0x7F;
+            if ( isTiny && roundBits ) roundData->exception |= float_flag_underflow;
+        }
+    }
+    if ( roundBits ) roundData->exception |= float_flag_inexact;
+    zSig = ( zSig + roundIncrement )>>7;
+    zSig &= ~ ( ( ( roundBits ^ 0x40 ) == 0 ) & roundNearestEven );
+    if ( zSig == 0 ) zExp = 0;
+    return packFloat32( zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes an abstract floating-point value having sign `zSign', exponent `zExp',
+and significand `zSig', and returns the proper single-precision floating-
+point value corresponding to the abstract input.  This routine is just like
+`roundAndPackFloat32' except that `zSig' does not have to be normalized in
+any way.  In all cases, `zExp' must be 1 less than the ``true'' floating-
+point exponent.
+-------------------------------------------------------------------------------
+*/
+static float32
+ normalizeRoundAndPackFloat32( struct roundingData *roundData, flag zSign, int16 zExp, bits32 zSig )
+{
+    int8 shiftCount;
+
+    shiftCount = countLeadingZeros32( zSig ) - 1;
+    return roundAndPackFloat32( roundData, zSign, zExp - shiftCount, zSig<<shiftCount );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the fraction bits of the double-precision floating-point value `a'.
+-------------------------------------------------------------------------------
+*/
+INLINE bits64 extractFloat64Frac( float64 a )
+{
+
+    return a & LIT64( 0x000FFFFFFFFFFFFF );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the exponent bits of the double-precision floating-point value `a'.
+-------------------------------------------------------------------------------
+*/
+INLINE int16 extractFloat64Exp( float64 a )
+{
+
+    return ( a>>52 ) & 0x7FF;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the sign bit of the double-precision floating-point value `a'.
+-------------------------------------------------------------------------------
+*/
+#if 0	/* in softfloat.h */
+INLINE flag extractFloat64Sign( float64 a )
+{
+
+    return a>>63;
+
+}
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Normalizes the subnormal double-precision floating-point value represented
+by the denormalized significand `aSig'.  The normalized exponent and
+significand are stored at the locations pointed to by `zExpPtr' and
+`zSigPtr', respectively.
+-------------------------------------------------------------------------------
+*/
+static void
+ normalizeFloat64Subnormal( bits64 aSig, int16 *zExpPtr, bits64 *zSigPtr )
+{
+    int8 shiftCount;
+
+    shiftCount = countLeadingZeros64( aSig ) - 11;
+    *zSigPtr = aSig<<shiftCount;
+    *zExpPtr = 1 - shiftCount;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Packs the sign `zSign', exponent `zExp', and significand `zSig' into a
+double-precision floating-point value, returning the result.  After being
+shifted into the proper positions, the three fields are simply added
+together to form the result.  This means that any integer portion of `zSig'
+will be added into the exponent.  Since a properly normalized significand
+will have an integer portion equal to 1, the `zExp' input should be 1 less
+than the desired result exponent whenever `zSig' is a complete, normalized
+significand.
+-------------------------------------------------------------------------------
+*/
+INLINE float64 packFloat64( flag zSign, int16 zExp, bits64 zSig )
+{
+
+    return ( ( (bits64) zSign )<<63 ) + ( ( (bits64) zExp )<<52 ) + zSig;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes an abstract floating-point value having sign `zSign', exponent `zExp',
+and significand `zSig', and returns the proper double-precision floating-
+point value corresponding to the abstract input.  Ordinarily, the abstract
+value is simply rounded and packed into the double-precision format, with
+the inexact exception raised if the abstract input cannot be represented
+exactly.  If the abstract value is too large, however, the overflow and
+inexact exceptions are raised and an infinity or maximal finite value is
+returned.  If the abstract value is too small, the input value is rounded to
+a subnormal number, and the underflow and inexact exceptions are raised if
+the abstract input cannot be represented exactly as a subnormal double-
+precision floating-point number.
+    The input significand `zSig' has its binary point between bits 62
+and 61, which is 10 bits to the left of the usual location.  This shifted
+significand must be normalized or smaller.  If `zSig' is not normalized,
+`zExp' must be 0; in that case, the result returned is a subnormal number,
+and it must not require rounding.  In the usual case that `zSig' is
+normalized, `zExp' must be 1 less than the ``true'' floating-point exponent.
+The handling of underflow and overflow follows the IEC/IEEE Standard for
+Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static float64 roundAndPackFloat64( struct roundingData *roundData, flag zSign, int16 zExp, bits64 zSig )
+{
+    int8 roundingMode;
+    flag roundNearestEven;
+    int16 roundIncrement, roundBits;
+    flag isTiny;
+
+    roundingMode = roundData->mode;
+    roundNearestEven = ( roundingMode == float_round_nearest_even );
+    roundIncrement = 0x200;
+    if ( ! roundNearestEven ) {
+        if ( roundingMode == float_round_to_zero ) {
+            roundIncrement = 0;
+        }
+        else {
+            roundIncrement = 0x3FF;
+            if ( zSign ) {
+                if ( roundingMode == float_round_up ) roundIncrement = 0;
+            }
+            else {
+                if ( roundingMode == float_round_down ) roundIncrement = 0;
+            }
+        }
+    }
+    roundBits = zSig & 0x3FF;
+    if ( 0x7FD <= (bits16) zExp ) {
+        if (    ( 0x7FD < zExp )
+             || (    ( zExp == 0x7FD )
+                  && ( (sbits64) ( zSig + roundIncrement ) < 0 ) )
+           ) {
+            //register int lr = __builtin_return_address(0);
+            //printk("roundAndPackFloat64 called from 0x%08x\n",lr);
+            roundData->exception |= float_flag_overflow | float_flag_inexact;
+            return packFloat64( zSign, 0x7FF, 0 ) - ( roundIncrement == 0 );
+        }
+        if ( zExp < 0 ) {
+            isTiny =
+                   ( float_detect_tininess == float_tininess_before_rounding )
+                || ( zExp < -1 )
+                || ( zSig + roundIncrement < LIT64( 0x8000000000000000 ) );
+            shift64RightJamming( zSig, - zExp, &zSig );
+            zExp = 0;
+            roundBits = zSig & 0x3FF;
+            if ( isTiny && roundBits ) roundData->exception |= float_flag_underflow;
+        }
+    }
+    if ( roundBits ) roundData->exception |= float_flag_inexact;
+    zSig = ( zSig + roundIncrement )>>10;
+    zSig &= ~ ( ( ( roundBits ^ 0x200 ) == 0 ) & roundNearestEven );
+    if ( zSig == 0 ) zExp = 0;
+    return packFloat64( zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes an abstract floating-point value having sign `zSign', exponent `zExp',
+and significand `zSig', and returns the proper double-precision floating-
+point value corresponding to the abstract input.  This routine is just like
+`roundAndPackFloat64' except that `zSig' does not have to be normalized in
+any way.  In all cases, `zExp' must be 1 less than the ``true'' floating-
+point exponent.
+-------------------------------------------------------------------------------
+*/
+static float64
+ normalizeRoundAndPackFloat64( struct roundingData *roundData, flag zSign, int16 zExp, bits64 zSig )
+{
+    int8 shiftCount;
+
+    shiftCount = countLeadingZeros64( zSig ) - 1;
+    return roundAndPackFloat64( roundData, zSign, zExp - shiftCount, zSig<<shiftCount );
+
+}
+
+#ifdef FLOATX80
+
+/*
+-------------------------------------------------------------------------------
+Returns the fraction bits of the extended double-precision floating-point
+value `a'.
+-------------------------------------------------------------------------------
+*/
+INLINE bits64 extractFloatx80Frac( floatx80 a )
+{
+
+    return a.low;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the exponent bits of the extended double-precision floating-point
+value `a'.
+-------------------------------------------------------------------------------
+*/
+INLINE int32 extractFloatx80Exp( floatx80 a )
+{
+
+    return a.high & 0x7FFF;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the sign bit of the extended double-precision floating-point value
+`a'.
+-------------------------------------------------------------------------------
+*/
+INLINE flag extractFloatx80Sign( floatx80 a )
+{
+
+    return a.high>>15;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Normalizes the subnormal extended double-precision floating-point value
+represented by the denormalized significand `aSig'.  The normalized exponent
+and significand are stored at the locations pointed to by `zExpPtr' and
+`zSigPtr', respectively.
+-------------------------------------------------------------------------------
+*/
+static void
+ normalizeFloatx80Subnormal( bits64 aSig, int32 *zExpPtr, bits64 *zSigPtr )
+{
+    int8 shiftCount;
+
+    shiftCount = countLeadingZeros64( aSig );
+    *zSigPtr = aSig<<shiftCount;
+    *zExpPtr = 1 - shiftCount;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Packs the sign `zSign', exponent `zExp', and significand `zSig' into an
+extended double-precision floating-point value, returning the result.
+-------------------------------------------------------------------------------
+*/
+INLINE floatx80 packFloatx80( flag zSign, int32 zExp, bits64 zSig )
+{
+    floatx80 z;
+
+    z.low = zSig;
+    z.high = ( ( (bits16) zSign )<<15 ) + zExp;
+    z.__padding = 0;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes an abstract floating-point value having sign `zSign', exponent `zExp',
+and extended significand formed by the concatenation of `zSig0' and `zSig1',
+and returns the proper extended double-precision floating-point value
+corresponding to the abstract input.  Ordinarily, the abstract value is
+rounded and packed into the extended double-precision format, with the
+inexact exception raised if the abstract input cannot be represented
+exactly.  If the abstract value is too large, however, the overflow and
+inexact exceptions are raised and an infinity or maximal finite value is
+returned.  If the abstract value is too small, the input value is rounded to
+a subnormal number, and the underflow and inexact exceptions are raised if
+the abstract input cannot be represented exactly as a subnormal extended
+double-precision floating-point number.
+    If `roundingPrecision' is 32 or 64, the result is rounded to the same
+number of bits as single or double precision, respectively.  Otherwise, the
+result is rounded to the full precision of the extended double-precision
+format.
+    The input significand must be normalized or smaller.  If the input
+significand is not normalized, `zExp' must be 0; in that case, the result
+returned is a subnormal number, and it must not require rounding.  The
+handling of underflow and overflow follows the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static floatx80
+ roundAndPackFloatx80(
+     struct roundingData *roundData, flag zSign, int32 zExp, bits64 zSig0, bits64 zSig1
+ )
+{
+    int8 roundingMode, roundingPrecision;
+    flag roundNearestEven, increment, isTiny;
+    int64 roundIncrement, roundMask, roundBits;
+
+    roundingMode = roundData->mode;
+    roundingPrecision = roundData->precision;
+    roundNearestEven = ( roundingMode == float_round_nearest_even );
+    if ( roundingPrecision == 80 ) goto precision80;
+    if ( roundingPrecision == 64 ) {
+        roundIncrement = LIT64( 0x0000000000000400 );
+        roundMask = LIT64( 0x00000000000007FF );
+    }
+    else if ( roundingPrecision == 32 ) {
+        roundIncrement = LIT64( 0x0000008000000000 );
+        roundMask = LIT64( 0x000000FFFFFFFFFF );
+    }
+    else {
+        goto precision80;
+    }
+    zSig0 |= ( zSig1 != 0 );
+    if ( ! roundNearestEven ) {
+        if ( roundingMode == float_round_to_zero ) {
+            roundIncrement = 0;
+        }
+        else {
+            roundIncrement = roundMask;
+            if ( zSign ) {
+                if ( roundingMode == float_round_up ) roundIncrement = 0;
+            }
+            else {
+                if ( roundingMode == float_round_down ) roundIncrement = 0;
+            }
+        }
+    }
+    roundBits = zSig0 & roundMask;
+    if ( 0x7FFD <= (bits32) ( zExp - 1 ) ) {
+        if (    ( 0x7FFE < zExp )
+             || ( ( zExp == 0x7FFE ) && ( zSig0 + roundIncrement < zSig0 ) )
+           ) {
+            goto overflow;
+        }
+        if ( zExp <= 0 ) {
+            isTiny =
+                   ( float_detect_tininess == float_tininess_before_rounding )
+                || ( zExp < 0 )
+                || ( zSig0 <= zSig0 + roundIncrement );
+            shift64RightJamming( zSig0, 1 - zExp, &zSig0 );
+            zExp = 0;
+            roundBits = zSig0 & roundMask;
+            if ( isTiny && roundBits ) roundData->exception |= float_flag_underflow;
+            if ( roundBits ) roundData->exception |= float_flag_inexact;
+            zSig0 += roundIncrement;
+            if ( (sbits64) zSig0 < 0 ) zExp = 1;
+            roundIncrement = roundMask + 1;
+            if ( roundNearestEven && ( roundBits<<1 == roundIncrement ) ) {
+                roundMask |= roundIncrement;
+            }
+            zSig0 &= ~ roundMask;
+            return packFloatx80( zSign, zExp, zSig0 );
+        }
+    }
+    if ( roundBits ) roundData->exception |= float_flag_inexact;
+    zSig0 += roundIncrement;
+    if ( zSig0 < roundIncrement ) {
+        ++zExp;
+        zSig0 = LIT64( 0x8000000000000000 );
+    }
+    roundIncrement = roundMask + 1;
+    if ( roundNearestEven && ( roundBits<<1 == roundIncrement ) ) {
+        roundMask |= roundIncrement;
+    }
+    zSig0 &= ~ roundMask;
+    if ( zSig0 == 0 ) zExp = 0;
+    return packFloatx80( zSign, zExp, zSig0 );
+ precision80:
+    increment = ( (sbits64) zSig1 < 0 );
+    if ( ! roundNearestEven ) {
+        if ( roundingMode == float_round_to_zero ) {
+            increment = 0;
+        }
+        else {
+            if ( zSign ) {
+                increment = ( roundingMode == float_round_down ) && zSig1;
+            }
+            else {
+                increment = ( roundingMode == float_round_up ) && zSig1;
+            }
+        }
+    }
+    if ( 0x7FFD <= (bits32) ( zExp - 1 ) ) {
+        if (    ( 0x7FFE < zExp )
+             || (    ( zExp == 0x7FFE )
+                  && ( zSig0 == LIT64( 0xFFFFFFFFFFFFFFFF ) )
+                  && increment
+                )
+           ) {
+            roundMask = 0;
+ overflow:
+            roundData->exception |= float_flag_overflow | float_flag_inexact;
+            if (    ( roundingMode == float_round_to_zero )
+                 || ( zSign && ( roundingMode == float_round_up ) )
+                 || ( ! zSign && ( roundingMode == float_round_down ) )
+               ) {
+                return packFloatx80( zSign, 0x7FFE, ~ roundMask );
+            }
+            return packFloatx80( zSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
+        }
+        if ( zExp <= 0 ) {
+            isTiny =
+                   ( float_detect_tininess == float_tininess_before_rounding )
+                || ( zExp < 0 )
+                || ! increment
+                || ( zSig0 < LIT64( 0xFFFFFFFFFFFFFFFF ) );
+            shift64ExtraRightJamming( zSig0, zSig1, 1 - zExp, &zSig0, &zSig1 );
+            zExp = 0;
+            if ( isTiny && zSig1 ) roundData->exception |= float_flag_underflow;
+            if ( zSig1 ) roundData->exception |= float_flag_inexact;
+            if ( roundNearestEven ) {
+                increment = ( (sbits64) zSig1 < 0 );
+            }
+            else {
+                if ( zSign ) {
+                    increment = ( roundingMode == float_round_down ) && zSig1;
+                }
+                else {
+                    increment = ( roundingMode == float_round_up ) && zSig1;
+                }
+            }
+            if ( increment ) {
+                ++zSig0;
+                zSig0 &= ~ ( ( zSig1 + zSig1 == 0 ) & roundNearestEven );
+                if ( (sbits64) zSig0 < 0 ) zExp = 1;
+            }
+            return packFloatx80( zSign, zExp, zSig0 );
+        }
+    }
+    if ( zSig1 ) roundData->exception |= float_flag_inexact;
+    if ( increment ) {
+        ++zSig0;
+        if ( zSig0 == 0 ) {
+            ++zExp;
+            zSig0 = LIT64( 0x8000000000000000 );
+        }
+        else {
+            zSig0 &= ~ ( ( zSig1 + zSig1 == 0 ) & roundNearestEven );
+        }
+    }
+    else {
+        if ( zSig0 == 0 ) zExp = 0;
+    }
+    
+    return packFloatx80( zSign, zExp, zSig0 );
+}
+
+/*
+-------------------------------------------------------------------------------
+Takes an abstract floating-point value having sign `zSign', exponent
+`zExp', and significand formed by the concatenation of `zSig0' and `zSig1',
+and returns the proper extended double-precision floating-point value
+corresponding to the abstract input.  This routine is just like
+`roundAndPackFloatx80' except that the input significand does not have to be
+normalized.
+-------------------------------------------------------------------------------
+*/
+static floatx80
+ normalizeRoundAndPackFloatx80(
+     struct roundingData *roundData, flag zSign, int32 zExp, bits64 zSig0, bits64 zSig1
+ )
+{
+    int8 shiftCount;
+
+    if ( zSig0 == 0 ) {
+        zSig0 = zSig1;
+        zSig1 = 0;
+        zExp -= 64;
+    }
+    shiftCount = countLeadingZeros64( zSig0 );
+    shortShift128Left( zSig0, zSig1, shiftCount, &zSig0, &zSig1 );
+    zExp -= shiftCount;
+    return
+        roundAndPackFloatx80( roundData, zSign, zExp, zSig0, zSig1 );
+
+}
+
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the 32-bit two's complement integer `a' to
+the single-precision floating-point format.  The conversion is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 int32_to_float32(struct roundingData *roundData, int32 a)
+{
+    flag zSign;
+
+    if ( a == 0 ) return 0;
+    if ( a == 0x80000000 ) return packFloat32( 1, 0x9E, 0 );
+    zSign = ( a < 0 );
+    return normalizeRoundAndPackFloat32( roundData, zSign, 0x9C, zSign ? - a : a );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the 32-bit two's complement integer `a' to
+the double-precision floating-point format.  The conversion is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 int32_to_float64( int32 a )
+{
+    flag aSign;
+    uint32 absA;
+    int8 shiftCount;
+    bits64 zSig;
+
+    if ( a == 0 ) return 0;
+    aSign = ( a < 0 );
+    absA = aSign ? - a : a;
+    shiftCount = countLeadingZeros32( absA ) + 21;
+    zSig = absA;
+    return packFloat64( aSign, 0x432 - shiftCount, zSig<<shiftCount );
+
+}
+
+#ifdef FLOATX80
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the 32-bit two's complement integer `a'
+to the extended double-precision floating-point format.  The conversion
+is performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 int32_to_floatx80( int32 a )
+{
+    flag zSign;
+    uint32 absA;
+    int8 shiftCount;
+    bits64 zSig;
+
+    if ( a == 0 ) return packFloatx80( 0, 0, 0 );
+    zSign = ( a < 0 );
+    absA = zSign ? - a : a;
+    shiftCount = countLeadingZeros32( absA ) + 32;
+    zSig = absA;
+    return packFloatx80( zSign, 0x403E - shiftCount, zSig<<shiftCount );
+
+}
+
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the single-precision floating-point value
+`a' to the 32-bit two's complement integer format.  The conversion is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic---which means in particular that the conversion is rounded
+according to the current rounding mode.  If `a' is a NaN, the largest
+positive integer is returned.  Otherwise, if the conversion overflows, the
+largest integer with the same sign as `a' is returned.
+-------------------------------------------------------------------------------
+*/
+int32 float32_to_int32( struct roundingData *roundData, float32 a )
+{
+    flag aSign;
+    int16 aExp, shiftCount;
+    bits32 aSig;
+    bits64 zSig;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    aSign = extractFloat32Sign( a );
+    if ( ( aExp == 0x7FF ) && aSig ) aSign = 0;
+    if ( aExp ) aSig |= 0x00800000;
+    shiftCount = 0xAF - aExp;
+    zSig = aSig;
+    zSig <<= 32;
+    if ( 0 < shiftCount ) shift64RightJamming( zSig, shiftCount, &zSig );
+    return roundAndPackInt32( roundData, aSign, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the single-precision floating-point value
+`a' to the 32-bit two's complement integer format.  The conversion is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic, except that the conversion is always rounded toward zero.  If
+`a' is a NaN, the largest positive integer is returned.  Otherwise, if the
+conversion overflows, the largest integer with the same sign as `a' is
+returned.
+-------------------------------------------------------------------------------
+*/
+int32 float32_to_int32_round_to_zero( float32 a )
+{
+    flag aSign;
+    int16 aExp, shiftCount;
+    bits32 aSig;
+    int32 z;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    aSign = extractFloat32Sign( a );
+    shiftCount = aExp - 0x9E;
+    if ( 0 <= shiftCount ) {
+        if ( a == 0xCF000000 ) return 0x80000000;
+        float_raise( float_flag_invalid );
+        if ( ! aSign || ( ( aExp == 0xFF ) && aSig ) ) return 0x7FFFFFFF;
+        return 0x80000000;
+    }
+    else if ( aExp <= 0x7E ) {
+        if ( aExp | aSig ) float_raise( float_flag_inexact );
+        return 0;
+    }
+    aSig = ( aSig | 0x00800000 )<<8;
+    z = aSig>>( - shiftCount );
+    if ( (bits32) ( aSig<<( shiftCount & 31 ) ) ) {
+        float_raise( float_flag_inexact );
+    }
+    return aSign ? - z : z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the single-precision floating-point value
+`a' to the double-precision floating-point format.  The conversion is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 float32_to_float64( float32 a )
+{
+    flag aSign;
+    int16 aExp;
+    bits32 aSig;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    aSign = extractFloat32Sign( a );
+    if ( aExp == 0xFF ) {
+        if ( aSig ) return commonNaNToFloat64( float32ToCommonNaN( a ) );
+        return packFloat64( aSign, 0x7FF, 0 );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloat64( aSign, 0, 0 );
+        normalizeFloat32Subnormal( aSig, &aExp, &aSig );
+        --aExp;
+    }
+    return packFloat64( aSign, aExp + 0x380, ( (bits64) aSig )<<29 );
+
+}
+
+#ifdef FLOATX80
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the single-precision floating-point value
+`a' to the extended double-precision floating-point format.  The conversion
+is performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 float32_to_floatx80( float32 a )
+{
+    flag aSign;
+    int16 aExp;
+    bits32 aSig;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    aSign = extractFloat32Sign( a );
+    if ( aExp == 0xFF ) {
+        if ( aSig ) return commonNaNToFloatx80( float32ToCommonNaN( a ) );
+        return packFloatx80( aSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloatx80( aSign, 0, 0 );
+        normalizeFloat32Subnormal( aSig, &aExp, &aSig );
+    }
+    aSig |= 0x00800000;
+    return packFloatx80( aSign, aExp + 0x3F80, ( (bits64) aSig )<<40 );
+
+}
+
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Rounds the single-precision floating-point value `a' to an integer, and
+returns the result as a single-precision floating-point value.  The
+operation is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 float32_round_to_int( struct roundingData *roundData, float32 a )
+{
+    flag aSign;
+    int16 aExp;
+    bits32 lastBitMask, roundBitsMask;
+    int8 roundingMode;
+    float32 z;
+
+    aExp = extractFloat32Exp( a );
+    if ( 0x96 <= aExp ) {
+        if ( ( aExp == 0xFF ) && extractFloat32Frac( a ) ) {
+            return propagateFloat32NaN( a, a );
+        }
+        return a;
+    }
+    roundingMode = roundData->mode;
+    if ( aExp <= 0x7E ) {
+        if ( (bits32) ( a<<1 ) == 0 ) return a;
+        roundData->exception |= float_flag_inexact;
+        aSign = extractFloat32Sign( a );
+        switch ( roundingMode ) {
+         case float_round_nearest_even:
+            if ( ( aExp == 0x7E ) && extractFloat32Frac( a ) ) {
+                return packFloat32( aSign, 0x7F, 0 );
+            }
+            break;
+         case float_round_down:
+            return aSign ? 0xBF800000 : 0;
+         case float_round_up:
+            return aSign ? 0x80000000 : 0x3F800000;
+        }
+        return packFloat32( aSign, 0, 0 );
+    }
+    lastBitMask = 1;
+    lastBitMask <<= 0x96 - aExp;
+    roundBitsMask = lastBitMask - 1;
+    z = a;
+    if ( roundingMode == float_round_nearest_even ) {
+        z += lastBitMask>>1;
+        if ( ( z & roundBitsMask ) == 0 ) z &= ~ lastBitMask;
+    }
+    else if ( roundingMode != float_round_to_zero ) {
+        if ( extractFloat32Sign( z ) ^ ( roundingMode == float_round_up ) ) {
+            z += roundBitsMask;
+        }
+    }
+    z &= ~ roundBitsMask;
+    if ( z != a ) roundData->exception |= float_flag_inexact;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of adding the absolute values of the single-precision
+floating-point values `a' and `b'.  If `zSign' is true, the sum is negated
+before being returned.  `zSign' is ignored if the result is a NaN.  The
+addition is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static float32 addFloat32Sigs( struct roundingData *roundData, float32 a, float32 b, flag zSign )
+{
+    int16 aExp, bExp, zExp;
+    bits32 aSig, bSig, zSig;
+    int16 expDiff;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    bSig = extractFloat32Frac( b );
+    bExp = extractFloat32Exp( b );
+    expDiff = aExp - bExp;
+    aSig <<= 6;
+    bSig <<= 6;
+    if ( 0 < expDiff ) {
+        if ( aExp == 0xFF ) {
+            if ( aSig ) return propagateFloat32NaN( a, b );
+            return a;
+        }
+        if ( bExp == 0 ) {
+            --expDiff;
+        }
+        else {
+            bSig |= 0x20000000;
+        }
+        shift32RightJamming( bSig, expDiff, &bSig );
+        zExp = aExp;
+    }
+    else if ( expDiff < 0 ) {
+        if ( bExp == 0xFF ) {
+            if ( bSig ) return propagateFloat32NaN( a, b );
+            return packFloat32( zSign, 0xFF, 0 );
+        }
+        if ( aExp == 0 ) {
+            ++expDiff;
+        }
+        else {
+            aSig |= 0x20000000;
+        }
+        shift32RightJamming( aSig, - expDiff, &aSig );
+        zExp = bExp;
+    }
+    else {
+        if ( aExp == 0xFF ) {
+            if ( aSig | bSig ) return propagateFloat32NaN( a, b );
+            return a;
+        }
+        if ( aExp == 0 ) return packFloat32( zSign, 0, ( aSig + bSig )>>6 );
+        zSig = 0x40000000 + aSig + bSig;
+        zExp = aExp;
+        goto roundAndPack;
+    }
+    aSig |= 0x20000000;
+    zSig = ( aSig + bSig )<<1;
+    --zExp;
+    if ( (sbits32) zSig < 0 ) {
+        zSig = aSig + bSig;
+        ++zExp;
+    }
+ roundAndPack:
+    return roundAndPackFloat32( roundData, zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of subtracting the absolute values of the single-
+precision floating-point values `a' and `b'.  If `zSign' is true, the
+difference is negated before being returned.  `zSign' is ignored if the
+result is a NaN.  The subtraction is performed according to the IEC/IEEE
+Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static float32 subFloat32Sigs( struct roundingData *roundData, float32 a, float32 b, flag zSign )
+{
+    int16 aExp, bExp, zExp;
+    bits32 aSig, bSig, zSig;
+    int16 expDiff;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    bSig = extractFloat32Frac( b );
+    bExp = extractFloat32Exp( b );
+    expDiff = aExp - bExp;
+    aSig <<= 7;
+    bSig <<= 7;
+    if ( 0 < expDiff ) goto aExpBigger;
+    if ( expDiff < 0 ) goto bExpBigger;
+    if ( aExp == 0xFF ) {
+        if ( aSig | bSig ) return propagateFloat32NaN( a, b );
+        roundData->exception |= float_flag_invalid;
+        return float32_default_nan;
+    }
+    if ( aExp == 0 ) {
+        aExp = 1;
+        bExp = 1;
+    }
+    if ( bSig < aSig ) goto aBigger;
+    if ( aSig < bSig ) goto bBigger;
+    return packFloat32( roundData->mode == float_round_down, 0, 0 );
+ bExpBigger:
+    if ( bExp == 0xFF ) {
+        if ( bSig ) return propagateFloat32NaN( a, b );
+        return packFloat32( zSign ^ 1, 0xFF, 0 );
+    }
+    if ( aExp == 0 ) {
+        ++expDiff;
+    }
+    else {
+        aSig |= 0x40000000;
+    }
+    shift32RightJamming( aSig, - expDiff, &aSig );
+    bSig |= 0x40000000;
+ bBigger:
+    zSig = bSig - aSig;
+    zExp = bExp;
+    zSign ^= 1;
+    goto normalizeRoundAndPack;
+ aExpBigger:
+    if ( aExp == 0xFF ) {
+        if ( aSig ) return propagateFloat32NaN( a, b );
+        return a;
+    }
+    if ( bExp == 0 ) {
+        --expDiff;
+    }
+    else {
+        bSig |= 0x40000000;
+    }
+    shift32RightJamming( bSig, expDiff, &bSig );
+    aSig |= 0x40000000;
+ aBigger:
+    zSig = aSig - bSig;
+    zExp = aExp;
+ normalizeRoundAndPack:
+    --zExp;
+    return normalizeRoundAndPackFloat32( roundData, zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of adding the single-precision floating-point values `a'
+and `b'.  The operation is performed according to the IEC/IEEE Standard for
+Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 float32_add( struct roundingData *roundData, float32 a, float32 b )
+{
+    flag aSign, bSign;
+
+    aSign = extractFloat32Sign( a );
+    bSign = extractFloat32Sign( b );
+    if ( aSign == bSign ) {
+        return addFloat32Sigs( roundData, a, b, aSign );
+    }
+    else {
+        return subFloat32Sigs( roundData, a, b, aSign );
+    }
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of subtracting the single-precision floating-point values
+`a' and `b'.  The operation is performed according to the IEC/IEEE Standard
+for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 float32_sub( struct roundingData *roundData, float32 a, float32 b )
+{
+    flag aSign, bSign;
+
+    aSign = extractFloat32Sign( a );
+    bSign = extractFloat32Sign( b );
+    if ( aSign == bSign ) {
+        return subFloat32Sigs( roundData, a, b, aSign );
+    }
+    else {
+        return addFloat32Sigs( roundData, a, b, aSign );
+    }
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of multiplying the single-precision floating-point values
+`a' and `b'.  The operation is performed according to the IEC/IEEE Standard
+for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 float32_mul( struct roundingData *roundData, float32 a, float32 b )
+{
+    flag aSign, bSign, zSign;
+    int16 aExp, bExp, zExp;
+    bits32 aSig, bSig;
+    bits64 zSig64;
+    bits32 zSig;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    aSign = extractFloat32Sign( a );
+    bSig = extractFloat32Frac( b );
+    bExp = extractFloat32Exp( b );
+    bSign = extractFloat32Sign( b );
+    zSign = aSign ^ bSign;
+    if ( aExp == 0xFF ) {
+        if ( aSig || ( ( bExp == 0xFF ) && bSig ) ) {
+            return propagateFloat32NaN( a, b );
+        }
+        if ( ( bExp | bSig ) == 0 ) {
+            roundData->exception |= float_flag_invalid;
+            return float32_default_nan;
+        }
+        return packFloat32( zSign, 0xFF, 0 );
+    }
+    if ( bExp == 0xFF ) {
+        if ( bSig ) return propagateFloat32NaN( a, b );
+        if ( ( aExp | aSig ) == 0 ) {
+            roundData->exception |= float_flag_invalid;
+            return float32_default_nan;
+        }
+        return packFloat32( zSign, 0xFF, 0 );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloat32( zSign, 0, 0 );
+        normalizeFloat32Subnormal( aSig, &aExp, &aSig );
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) return packFloat32( zSign, 0, 0 );
+        normalizeFloat32Subnormal( bSig, &bExp, &bSig );
+    }
+    zExp = aExp + bExp - 0x7F;
+    aSig = ( aSig | 0x00800000 )<<7;
+    bSig = ( bSig | 0x00800000 )<<8;
+    shift64RightJamming( ( (bits64) aSig ) * bSig, 32, &zSig64 );
+    zSig = zSig64;
+    if ( 0 <= (sbits32) ( zSig<<1 ) ) {
+        zSig <<= 1;
+        --zExp;
+    }
+    return roundAndPackFloat32( roundData, zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of dividing the single-precision floating-point value `a'
+by the corresponding value `b'.  The operation is performed according to the
+IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 float32_div( struct roundingData *roundData, float32 a, float32 b )
+{
+    flag aSign, bSign, zSign;
+    int16 aExp, bExp, zExp;
+    bits32 aSig, bSig, zSig;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    aSign = extractFloat32Sign( a );
+    bSig = extractFloat32Frac( b );
+    bExp = extractFloat32Exp( b );
+    bSign = extractFloat32Sign( b );
+    zSign = aSign ^ bSign;
+    if ( aExp == 0xFF ) {
+        if ( aSig ) return propagateFloat32NaN( a, b );
+        if ( bExp == 0xFF ) {
+            if ( bSig ) return propagateFloat32NaN( a, b );
+            roundData->exception |= float_flag_invalid;
+            return float32_default_nan;
+        }
+        return packFloat32( zSign, 0xFF, 0 );
+    }
+    if ( bExp == 0xFF ) {
+        if ( bSig ) return propagateFloat32NaN( a, b );
+        return packFloat32( zSign, 0, 0 );
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) {
+            if ( ( aExp | aSig ) == 0 ) {
+                roundData->exception |= float_flag_invalid;
+                return float32_default_nan;
+            }
+            roundData->exception |= float_flag_divbyzero;
+            return packFloat32( zSign, 0xFF, 0 );
+        }
+        normalizeFloat32Subnormal( bSig, &bExp, &bSig );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloat32( zSign, 0, 0 );
+        normalizeFloat32Subnormal( aSig, &aExp, &aSig );
+    }
+    zExp = aExp - bExp + 0x7D;
+    aSig = ( aSig | 0x00800000 )<<7;
+    bSig = ( bSig | 0x00800000 )<<8;
+    if ( bSig <= ( aSig + aSig ) ) {
+        aSig >>= 1;
+        ++zExp;
+    }
+    {
+        bits64 tmp = ( (bits64) aSig )<<32;
+        do_div( tmp, bSig );
+        zSig = tmp;
+    }
+    if ( ( zSig & 0x3F ) == 0 ) {
+        zSig |= ( ( (bits64) bSig ) * zSig != ( (bits64) aSig )<<32 );
+    }
+    return roundAndPackFloat32( roundData, zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the remainder of the single-precision floating-point value `a'
+with respect to the corresponding value `b'.  The operation is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 float32_rem( struct roundingData *roundData, float32 a, float32 b )
+{
+    flag aSign, bSign, zSign;
+    int16 aExp, bExp, expDiff;
+    bits32 aSig, bSig;
+    bits32 q;
+    bits64 aSig64, bSig64, q64;
+    bits32 alternateASig;
+    sbits32 sigMean;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    aSign = extractFloat32Sign( a );
+    bSig = extractFloat32Frac( b );
+    bExp = extractFloat32Exp( b );
+    bSign = extractFloat32Sign( b );
+    if ( aExp == 0xFF ) {
+        if ( aSig || ( ( bExp == 0xFF ) && bSig ) ) {
+            return propagateFloat32NaN( a, b );
+        }
+        roundData->exception |= float_flag_invalid;
+        return float32_default_nan;
+    }
+    if ( bExp == 0xFF ) {
+        if ( bSig ) return propagateFloat32NaN( a, b );
+        return a;
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) {
+            roundData->exception |= float_flag_invalid;
+            return float32_default_nan;
+        }
+        normalizeFloat32Subnormal( bSig, &bExp, &bSig );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return a;
+        normalizeFloat32Subnormal( aSig, &aExp, &aSig );
+    }
+    expDiff = aExp - bExp;
+    aSig |= 0x00800000;
+    bSig |= 0x00800000;
+    if ( expDiff < 32 ) {
+        aSig <<= 8;
+        bSig <<= 8;
+        if ( expDiff < 0 ) {
+            if ( expDiff < -1 ) return a;
+            aSig >>= 1;
+        }
+        q = ( bSig <= aSig );
+        if ( q ) aSig -= bSig;
+        if ( 0 < expDiff ) {
+            bits64 tmp = ( (bits64) aSig )<<32;
+            do_div( tmp, bSig );
+            q = tmp;
+            q >>= 32 - expDiff;
+            bSig >>= 2;
+            aSig = ( ( aSig>>1 )<<( expDiff - 1 ) ) - bSig * q;
+        }
+        else {
+            aSig >>= 2;
+            bSig >>= 2;
+        }
+    }
+    else {
+        if ( bSig <= aSig ) aSig -= bSig;
+        aSig64 = ( (bits64) aSig )<<40;
+        bSig64 = ( (bits64) bSig )<<40;
+        expDiff -= 64;
+        while ( 0 < expDiff ) {
+            q64 = estimateDiv128To64( aSig64, 0, bSig64 );
+            q64 = ( 2 < q64 ) ? q64 - 2 : 0;
+            aSig64 = - ( ( bSig * q64 )<<38 );
+            expDiff -= 62;
+        }
+        expDiff += 64;
+        q64 = estimateDiv128To64( aSig64, 0, bSig64 );
+        q64 = ( 2 < q64 ) ? q64 - 2 : 0;
+        q = q64>>( 64 - expDiff );
+        bSig <<= 6;
+        aSig = ( ( aSig64>>33 )<<( expDiff - 1 ) ) - bSig * q;
+    }
+    do {
+        alternateASig = aSig;
+        ++q;
+        aSig -= bSig;
+    } while ( 0 <= (sbits32) aSig );
+    sigMean = aSig + alternateASig;
+    if ( ( sigMean < 0 ) || ( ( sigMean == 0 ) && ( q & 1 ) ) ) {
+        aSig = alternateASig;
+    }
+    zSign = ( (sbits32) aSig < 0 );
+    if ( zSign ) aSig = - aSig;
+    return normalizeRoundAndPackFloat32( roundData, aSign ^ zSign, bExp, aSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the square root of the single-precision floating-point value `a'.
+The operation is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 float32_sqrt( struct roundingData *roundData, float32 a )
+{
+    flag aSign;
+    int16 aExp, zExp;
+    bits32 aSig, zSig;
+    bits64 rem, term;
+
+    aSig = extractFloat32Frac( a );
+    aExp = extractFloat32Exp( a );
+    aSign = extractFloat32Sign( a );
+    if ( aExp == 0xFF ) {
+        if ( aSig ) return propagateFloat32NaN( a, 0 );
+        if ( ! aSign ) return a;
+        roundData->exception |= float_flag_invalid;
+        return float32_default_nan;
+    }
+    if ( aSign ) {
+        if ( ( aExp | aSig ) == 0 ) return a;
+        roundData->exception |= float_flag_invalid;
+        return float32_default_nan;
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return 0;
+        normalizeFloat32Subnormal( aSig, &aExp, &aSig );
+    }
+    zExp = ( ( aExp - 0x7F )>>1 ) + 0x7E;
+    aSig = ( aSig | 0x00800000 )<<8;
+    zSig = estimateSqrt32( aExp, aSig ) + 2;
+    if ( ( zSig & 0x7F ) <= 5 ) {
+        if ( zSig < 2 ) {
+            zSig = 0xFFFFFFFF;
+        }
+        else {
+            aSig >>= aExp & 1;
+            term = ( (bits64) zSig ) * zSig;
+            rem = ( ( (bits64) aSig )<<32 ) - term;
+            while ( (sbits64) rem < 0 ) {
+                --zSig;
+                rem += ( ( (bits64) zSig )<<1 ) | 1;
+            }
+            zSig |= ( rem != 0 );
+        }
+    }
+    shift32RightJamming( zSig, 1, &zSig );
+    return roundAndPackFloat32( roundData, 0, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the single-precision floating-point value `a' is equal to the
+corresponding value `b', and 0 otherwise.  The comparison is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float32_eq( float32 a, float32 b )
+{
+
+    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
+         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
+       ) {
+        if ( float32_is_signaling_nan( a ) || float32_is_signaling_nan( b ) ) {
+            float_raise( float_flag_invalid );
+        }
+        return 0;
+    }
+    return ( a == b ) || ( (bits32) ( ( a | b )<<1 ) == 0 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the single-precision floating-point value `a' is less than or
+equal to the corresponding value `b', and 0 otherwise.  The comparison is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float32_le( float32 a, float32 b )
+{
+    flag aSign, bSign;
+
+    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
+         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    aSign = extractFloat32Sign( a );
+    bSign = extractFloat32Sign( b );
+    if ( aSign != bSign ) return aSign || ( (bits32) ( ( a | b )<<1 ) == 0 );
+    return ( a == b ) || ( aSign ^ ( a < b ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the single-precision floating-point value `a' is less than
+the corresponding value `b', and 0 otherwise.  The comparison is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float32_lt( float32 a, float32 b )
+{
+    flag aSign, bSign;
+
+    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
+         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    aSign = extractFloat32Sign( a );
+    bSign = extractFloat32Sign( b );
+    if ( aSign != bSign ) return aSign && ( (bits32) ( ( a | b )<<1 ) != 0 );
+    return ( a != b ) && ( aSign ^ ( a < b ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the single-precision floating-point value `a' is equal to the
+corresponding value `b', and 0 otherwise.  The invalid exception is raised
+if either operand is a NaN.  Otherwise, the comparison is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float32_eq_signaling( float32 a, float32 b )
+{
+
+    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
+         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    return ( a == b ) || ( (bits32) ( ( a | b )<<1 ) == 0 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the single-precision floating-point value `a' is less than or
+equal to the corresponding value `b', and 0 otherwise.  Quiet NaNs do not
+cause an exception.  Otherwise, the comparison is performed according to the
+IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float32_le_quiet( float32 a, float32 b )
+{
+    flag aSign, bSign;
+    //int16 aExp, bExp;
+
+    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
+         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
+       ) {
+        /* Do nothing, even if NaN as we're quiet */
+        return 0;
+    }
+    aSign = extractFloat32Sign( a );
+    bSign = extractFloat32Sign( b );
+    if ( aSign != bSign ) return aSign || ( (bits32) ( ( a | b )<<1 ) == 0 );
+    return ( a == b ) || ( aSign ^ ( a < b ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the single-precision floating-point value `a' is less than
+the corresponding value `b', and 0 otherwise.  Quiet NaNs do not cause an
+exception.  Otherwise, the comparison is performed according to the IEC/IEEE
+Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float32_lt_quiet( float32 a, float32 b )
+{
+    flag aSign, bSign;
+
+    if (    ( ( extractFloat32Exp( a ) == 0xFF ) && extractFloat32Frac( a ) )
+         || ( ( extractFloat32Exp( b ) == 0xFF ) && extractFloat32Frac( b ) )
+       ) {
+        /* Do nothing, even if NaN as we're quiet */
+        return 0;
+    }
+    aSign = extractFloat32Sign( a );
+    bSign = extractFloat32Sign( b );
+    if ( aSign != bSign ) return aSign && ( (bits32) ( ( a | b )<<1 ) != 0 );
+    return ( a != b ) && ( aSign ^ ( a < b ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the double-precision floating-point value
+`a' to the 32-bit two's complement integer format.  The conversion is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic---which means in particular that the conversion is rounded
+according to the current rounding mode.  If `a' is a NaN, the largest
+positive integer is returned.  Otherwise, if the conversion overflows, the
+largest integer with the same sign as `a' is returned.
+-------------------------------------------------------------------------------
+*/
+int32 float64_to_int32( struct roundingData *roundData, float64 a )
+{
+    flag aSign;
+    int16 aExp, shiftCount;
+    bits64 aSig;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    if ( ( aExp == 0x7FF ) && aSig ) aSign = 0;
+    if ( aExp ) aSig |= LIT64( 0x0010000000000000 );
+    shiftCount = 0x42C - aExp;
+    if ( 0 < shiftCount ) shift64RightJamming( aSig, shiftCount, &aSig );
+    return roundAndPackInt32( roundData, aSign, aSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the double-precision floating-point value
+`a' to the 32-bit two's complement integer format.  The conversion is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic, except that the conversion is always rounded toward zero.  If
+`a' is a NaN, the largest positive integer is returned.  Otherwise, if the
+conversion overflows, the largest integer with the same sign as `a' is
+returned.
+-------------------------------------------------------------------------------
+*/
+int32 float64_to_int32_round_to_zero( float64 a )
+{
+    flag aSign;
+    int16 aExp, shiftCount;
+    bits64 aSig, savedASig;
+    int32 z;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    shiftCount = 0x433 - aExp;
+    if ( shiftCount < 21 ) {
+        if ( ( aExp == 0x7FF ) && aSig ) aSign = 0;
+        goto invalid;
+    }
+    else if ( 52 < shiftCount ) {
+        if ( aExp || aSig ) float_raise( float_flag_inexact );
+        return 0;
+    }
+    aSig |= LIT64( 0x0010000000000000 );
+    savedASig = aSig;
+    aSig >>= shiftCount;
+    z = aSig;
+    if ( aSign ) z = - z;
+    if ( ( z < 0 ) ^ aSign ) {
+ invalid:
+        float_raise( float_flag_invalid );
+        return aSign ? 0x80000000 : 0x7FFFFFFF;
+    }
+    if ( ( aSig<<shiftCount ) != savedASig ) {
+        float_raise( float_flag_inexact );
+    }
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the double-precision floating-point value
+`a' to the 32-bit two's complement unsigned integer format.  The conversion
+is performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic---which means in particular that the conversion is rounded
+according to the current rounding mode.  If `a' is a NaN, the largest
+positive integer is returned.  Otherwise, if the conversion overflows, the
+largest positive integer is returned.
+-------------------------------------------------------------------------------
+*/
+int32 float64_to_uint32( struct roundingData *roundData, float64 a )
+{
+    flag aSign;
+    int16 aExp, shiftCount;
+    bits64 aSig;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = 0; //extractFloat64Sign( a );
+    //if ( ( aExp == 0x7FF ) && aSig ) aSign = 0;
+    if ( aExp ) aSig |= LIT64( 0x0010000000000000 );
+    shiftCount = 0x42C - aExp;
+    if ( 0 < shiftCount ) shift64RightJamming( aSig, shiftCount, &aSig );
+    return roundAndPackInt32( roundData, aSign, aSig );
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the double-precision floating-point value
+`a' to the 32-bit two's complement integer format.  The conversion is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic, except that the conversion is always rounded toward zero.  If
+`a' is a NaN, the largest positive integer is returned.  Otherwise, if the
+conversion overflows, the largest positive integer is returned.
+-------------------------------------------------------------------------------
+*/
+int32 float64_to_uint32_round_to_zero( float64 a )
+{
+    flag aSign;
+    int16 aExp, shiftCount;
+    bits64 aSig, savedASig;
+    int32 z;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    shiftCount = 0x433 - aExp;
+    if ( shiftCount < 21 ) {
+        if ( ( aExp == 0x7FF ) && aSig ) aSign = 0;
+        goto invalid;
+    }
+    else if ( 52 < shiftCount ) {
+        if ( aExp || aSig ) float_raise( float_flag_inexact );
+        return 0;
+    }
+    aSig |= LIT64( 0x0010000000000000 );
+    savedASig = aSig;
+    aSig >>= shiftCount;
+    z = aSig;
+    if ( aSign ) z = - z;
+    if ( ( z < 0 ) ^ aSign ) {
+ invalid:
+        float_raise( float_flag_invalid );
+        return aSign ? 0x80000000 : 0x7FFFFFFF;
+    }
+    if ( ( aSig<<shiftCount ) != savedASig ) {
+        float_raise( float_flag_inexact );
+    }
+    return z;
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the double-precision floating-point value
+`a' to the single-precision floating-point format.  The conversion is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 float64_to_float32( struct roundingData *roundData, float64 a )
+{
+    flag aSign;
+    int16 aExp;
+    bits64 aSig;
+    bits32 zSig;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    if ( aExp == 0x7FF ) {
+        if ( aSig ) return commonNaNToFloat32( float64ToCommonNaN( a ) );
+        return packFloat32( aSign, 0xFF, 0 );
+    }
+    shift64RightJamming( aSig, 22, &aSig );
+    zSig = aSig;
+    if ( aExp || zSig ) {
+        zSig |= 0x40000000;
+        aExp -= 0x381;
+    }
+    return roundAndPackFloat32( roundData, aSign, aExp, zSig );
+
+}
+
+#ifdef FLOATX80
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the double-precision floating-point value
+`a' to the extended double-precision floating-point format.  The conversion
+is performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 float64_to_floatx80( float64 a )
+{
+    flag aSign;
+    int16 aExp;
+    bits64 aSig;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    if ( aExp == 0x7FF ) {
+        if ( aSig ) return commonNaNToFloatx80( float64ToCommonNaN( a ) );
+        return packFloatx80( aSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloatx80( aSign, 0, 0 );
+        normalizeFloat64Subnormal( aSig, &aExp, &aSig );
+    }
+    return
+        packFloatx80(
+            aSign, aExp + 0x3C00, ( aSig | LIT64( 0x0010000000000000 ) )<<11 );
+
+}
+
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Rounds the double-precision floating-point value `a' to an integer, and
+returns the result as a double-precision floating-point value.  The
+operation is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 float64_round_to_int( struct roundingData *roundData, float64 a )
+{
+    flag aSign;
+    int16 aExp;
+    bits64 lastBitMask, roundBitsMask;
+    int8 roundingMode;
+    float64 z;
+
+    aExp = extractFloat64Exp( a );
+    if ( 0x433 <= aExp ) {
+        if ( ( aExp == 0x7FF ) && extractFloat64Frac( a ) ) {
+            return propagateFloat64NaN( a, a );
+        }
+        return a;
+    }
+    if ( aExp <= 0x3FE ) {
+        if ( (bits64) ( a<<1 ) == 0 ) return a;
+        roundData->exception |= float_flag_inexact;
+        aSign = extractFloat64Sign( a );
+        switch ( roundData->mode ) {
+         case float_round_nearest_even:
+            if ( ( aExp == 0x3FE ) && extractFloat64Frac( a ) ) {
+                return packFloat64( aSign, 0x3FF, 0 );
+            }
+            break;
+         case float_round_down:
+            return aSign ? LIT64( 0xBFF0000000000000 ) : 0;
+         case float_round_up:
+            return
+            aSign ? LIT64( 0x8000000000000000 ) : LIT64( 0x3FF0000000000000 );
+        }
+        return packFloat64( aSign, 0, 0 );
+    }
+    lastBitMask = 1;
+    lastBitMask <<= 0x433 - aExp;
+    roundBitsMask = lastBitMask - 1;
+    z = a;
+    roundingMode = roundData->mode;
+    if ( roundingMode == float_round_nearest_even ) {
+        z += lastBitMask>>1;
+        if ( ( z & roundBitsMask ) == 0 ) z &= ~ lastBitMask;
+    }
+    else if ( roundingMode != float_round_to_zero ) {
+        if ( extractFloat64Sign( z ) ^ ( roundingMode == float_round_up ) ) {
+            z += roundBitsMask;
+        }
+    }
+    z &= ~ roundBitsMask;
+    if ( z != a ) roundData->exception |= float_flag_inexact;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of adding the absolute values of the double-precision
+floating-point values `a' and `b'.  If `zSign' is true, the sum is negated
+before being returned.  `zSign' is ignored if the result is a NaN.  The
+addition is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static float64 addFloat64Sigs( struct roundingData *roundData, float64 a, float64 b, flag zSign )
+{
+    int16 aExp, bExp, zExp;
+    bits64 aSig, bSig, zSig;
+    int16 expDiff;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    bSig = extractFloat64Frac( b );
+    bExp = extractFloat64Exp( b );
+    expDiff = aExp - bExp;
+    aSig <<= 9;
+    bSig <<= 9;
+    if ( 0 < expDiff ) {
+        if ( aExp == 0x7FF ) {
+            if ( aSig ) return propagateFloat64NaN( a, b );
+            return a;
+        }
+        if ( bExp == 0 ) {
+            --expDiff;
+        }
+        else {
+            bSig |= LIT64( 0x2000000000000000 );
+        }
+        shift64RightJamming( bSig, expDiff, &bSig );
+        zExp = aExp;
+    }
+    else if ( expDiff < 0 ) {
+        if ( bExp == 0x7FF ) {
+            if ( bSig ) return propagateFloat64NaN( a, b );
+            return packFloat64( zSign, 0x7FF, 0 );
+        }
+        if ( aExp == 0 ) {
+            ++expDiff;
+        }
+        else {
+            aSig |= LIT64( 0x2000000000000000 );
+        }
+        shift64RightJamming( aSig, - expDiff, &aSig );
+        zExp = bExp;
+    }
+    else {
+        if ( aExp == 0x7FF ) {
+            if ( aSig | bSig ) return propagateFloat64NaN( a, b );
+            return a;
+        }
+        if ( aExp == 0 ) return packFloat64( zSign, 0, ( aSig + bSig )>>9 );
+        zSig = LIT64( 0x4000000000000000 ) + aSig + bSig;
+        zExp = aExp;
+        goto roundAndPack;
+    }
+    aSig |= LIT64( 0x2000000000000000 );
+    zSig = ( aSig + bSig )<<1;
+    --zExp;
+    if ( (sbits64) zSig < 0 ) {
+        zSig = aSig + bSig;
+        ++zExp;
+    }
+ roundAndPack:
+    return roundAndPackFloat64( roundData, zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of subtracting the absolute values of the double-
+precision floating-point values `a' and `b'.  If `zSign' is true, the
+difference is negated before being returned.  `zSign' is ignored if the
+result is a NaN.  The subtraction is performed according to the IEC/IEEE
+Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static float64 subFloat64Sigs( struct roundingData *roundData, float64 a, float64 b, flag zSign )
+{
+    int16 aExp, bExp, zExp;
+    bits64 aSig, bSig, zSig;
+    int16 expDiff;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    bSig = extractFloat64Frac( b );
+    bExp = extractFloat64Exp( b );
+    expDiff = aExp - bExp;
+    aSig <<= 10;
+    bSig <<= 10;
+    if ( 0 < expDiff ) goto aExpBigger;
+    if ( expDiff < 0 ) goto bExpBigger;
+    if ( aExp == 0x7FF ) {
+        if ( aSig | bSig ) return propagateFloat64NaN( a, b );
+        roundData->exception |= float_flag_invalid;
+        return float64_default_nan;
+    }
+    if ( aExp == 0 ) {
+        aExp = 1;
+        bExp = 1;
+    }
+    if ( bSig < aSig ) goto aBigger;
+    if ( aSig < bSig ) goto bBigger;
+    return packFloat64( roundData->mode == float_round_down, 0, 0 );
+ bExpBigger:
+    if ( bExp == 0x7FF ) {
+        if ( bSig ) return propagateFloat64NaN( a, b );
+        return packFloat64( zSign ^ 1, 0x7FF, 0 );
+    }
+    if ( aExp == 0 ) {
+        ++expDiff;
+    }
+    else {
+        aSig |= LIT64( 0x4000000000000000 );
+    }
+    shift64RightJamming( aSig, - expDiff, &aSig );
+    bSig |= LIT64( 0x4000000000000000 );
+ bBigger:
+    zSig = bSig - aSig;
+    zExp = bExp;
+    zSign ^= 1;
+    goto normalizeRoundAndPack;
+ aExpBigger:
+    if ( aExp == 0x7FF ) {
+        if ( aSig ) return propagateFloat64NaN( a, b );
+        return a;
+    }
+    if ( bExp == 0 ) {
+        --expDiff;
+    }
+    else {
+        bSig |= LIT64( 0x4000000000000000 );
+    }
+    shift64RightJamming( bSig, expDiff, &bSig );
+    aSig |= LIT64( 0x4000000000000000 );
+ aBigger:
+    zSig = aSig - bSig;
+    zExp = aExp;
+ normalizeRoundAndPack:
+    --zExp;
+    return normalizeRoundAndPackFloat64( roundData, zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of adding the double-precision floating-point values `a'
+and `b'.  The operation is performed according to the IEC/IEEE Standard for
+Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 float64_add( struct roundingData *roundData, float64 a, float64 b )
+{
+    flag aSign, bSign;
+
+    aSign = extractFloat64Sign( a );
+    bSign = extractFloat64Sign( b );
+    if ( aSign == bSign ) {
+        return addFloat64Sigs( roundData, a, b, aSign );
+    }
+    else {
+        return subFloat64Sigs( roundData, a, b, aSign );
+    }
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of subtracting the double-precision floating-point values
+`a' and `b'.  The operation is performed according to the IEC/IEEE Standard
+for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 float64_sub( struct roundingData *roundData, float64 a, float64 b )
+{
+    flag aSign, bSign;
+
+    aSign = extractFloat64Sign( a );
+    bSign = extractFloat64Sign( b );
+    if ( aSign == bSign ) {
+        return subFloat64Sigs( roundData, a, b, aSign );
+    }
+    else {
+        return addFloat64Sigs( roundData, a, b, aSign );
+    }
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of multiplying the double-precision floating-point values
+`a' and `b'.  The operation is performed according to the IEC/IEEE Standard
+for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 float64_mul( struct roundingData *roundData, float64 a, float64 b )
+{
+    flag aSign, bSign, zSign;
+    int16 aExp, bExp, zExp;
+    bits64 aSig, bSig, zSig0, zSig1;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    bSig = extractFloat64Frac( b );
+    bExp = extractFloat64Exp( b );
+    bSign = extractFloat64Sign( b );
+    zSign = aSign ^ bSign;
+    if ( aExp == 0x7FF ) {
+        if ( aSig || ( ( bExp == 0x7FF ) && bSig ) ) {
+            return propagateFloat64NaN( a, b );
+        }
+        if ( ( bExp | bSig ) == 0 ) {
+            roundData->exception |= float_flag_invalid;
+            return float64_default_nan;
+        }
+        return packFloat64( zSign, 0x7FF, 0 );
+    }
+    if ( bExp == 0x7FF ) {
+        if ( bSig ) return propagateFloat64NaN( a, b );
+        if ( ( aExp | aSig ) == 0 ) {
+            roundData->exception |= float_flag_invalid;
+            return float64_default_nan;
+        }
+        return packFloat64( zSign, 0x7FF, 0 );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloat64( zSign, 0, 0 );
+        normalizeFloat64Subnormal( aSig, &aExp, &aSig );
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) return packFloat64( zSign, 0, 0 );
+        normalizeFloat64Subnormal( bSig, &bExp, &bSig );
+    }
+    zExp = aExp + bExp - 0x3FF;
+    aSig = ( aSig | LIT64( 0x0010000000000000 ) )<<10;
+    bSig = ( bSig | LIT64( 0x0010000000000000 ) )<<11;
+    mul64To128( aSig, bSig, &zSig0, &zSig1 );
+    zSig0 |= ( zSig1 != 0 );
+    if ( 0 <= (sbits64) ( zSig0<<1 ) ) {
+        zSig0 <<= 1;
+        --zExp;
+    }
+    return roundAndPackFloat64( roundData, zSign, zExp, zSig0 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of dividing the double-precision floating-point value `a'
+by the corresponding value `b'.  The operation is performed according to
+the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 float64_div( struct roundingData *roundData, float64 a, float64 b )
+{
+    flag aSign, bSign, zSign;
+    int16 aExp, bExp, zExp;
+    bits64 aSig, bSig, zSig;
+    bits64 rem0, rem1;
+    bits64 term0, term1;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    bSig = extractFloat64Frac( b );
+    bExp = extractFloat64Exp( b );
+    bSign = extractFloat64Sign( b );
+    zSign = aSign ^ bSign;
+    if ( aExp == 0x7FF ) {
+        if ( aSig ) return propagateFloat64NaN( a, b );
+        if ( bExp == 0x7FF ) {
+            if ( bSig ) return propagateFloat64NaN( a, b );
+            roundData->exception |= float_flag_invalid;
+            return float64_default_nan;
+        }
+        return packFloat64( zSign, 0x7FF, 0 );
+    }
+    if ( bExp == 0x7FF ) {
+        if ( bSig ) return propagateFloat64NaN( a, b );
+        return packFloat64( zSign, 0, 0 );
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) {
+            if ( ( aExp | aSig ) == 0 ) {
+                roundData->exception |= float_flag_invalid;
+                return float64_default_nan;
+            }
+            roundData->exception |= float_flag_divbyzero;
+            return packFloat64( zSign, 0x7FF, 0 );
+        }
+        normalizeFloat64Subnormal( bSig, &bExp, &bSig );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloat64( zSign, 0, 0 );
+        normalizeFloat64Subnormal( aSig, &aExp, &aSig );
+    }
+    zExp = aExp - bExp + 0x3FD;
+    aSig = ( aSig | LIT64( 0x0010000000000000 ) )<<10;
+    bSig = ( bSig | LIT64( 0x0010000000000000 ) )<<11;
+    if ( bSig <= ( aSig + aSig ) ) {
+        aSig >>= 1;
+        ++zExp;
+    }
+    zSig = estimateDiv128To64( aSig, 0, bSig );
+    if ( ( zSig & 0x1FF ) <= 2 ) {
+        mul64To128( bSig, zSig, &term0, &term1 );
+        sub128( aSig, 0, term0, term1, &rem0, &rem1 );
+        while ( (sbits64) rem0 < 0 ) {
+            --zSig;
+            add128( rem0, rem1, 0, bSig, &rem0, &rem1 );
+        }
+        zSig |= ( rem1 != 0 );
+    }
+    return roundAndPackFloat64( roundData, zSign, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the remainder of the double-precision floating-point value `a'
+with respect to the corresponding value `b'.  The operation is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 float64_rem( struct roundingData *roundData, float64 a, float64 b )
+{
+    flag aSign, bSign, zSign;
+    int16 aExp, bExp, expDiff;
+    bits64 aSig, bSig;
+    bits64 q, alternateASig;
+    sbits64 sigMean;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    bSig = extractFloat64Frac( b );
+    bExp = extractFloat64Exp( b );
+    bSign = extractFloat64Sign( b );
+    if ( aExp == 0x7FF ) {
+        if ( aSig || ( ( bExp == 0x7FF ) && bSig ) ) {
+            return propagateFloat64NaN( a, b );
+        }
+        roundData->exception |= float_flag_invalid;
+        return float64_default_nan;
+    }
+    if ( bExp == 0x7FF ) {
+        if ( bSig ) return propagateFloat64NaN( a, b );
+        return a;
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) {
+            roundData->exception |= float_flag_invalid;
+            return float64_default_nan;
+        }
+        normalizeFloat64Subnormal( bSig, &bExp, &bSig );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return a;
+        normalizeFloat64Subnormal( aSig, &aExp, &aSig );
+    }
+    expDiff = aExp - bExp;
+    aSig = ( aSig | LIT64( 0x0010000000000000 ) )<<11;
+    bSig = ( bSig | LIT64( 0x0010000000000000 ) )<<11;
+    if ( expDiff < 0 ) {
+        if ( expDiff < -1 ) return a;
+        aSig >>= 1;
+    }
+    q = ( bSig <= aSig );
+    if ( q ) aSig -= bSig;
+    expDiff -= 64;
+    while ( 0 < expDiff ) {
+        q = estimateDiv128To64( aSig, 0, bSig );
+        q = ( 2 < q ) ? q - 2 : 0;
+        aSig = - ( ( bSig>>2 ) * q );
+        expDiff -= 62;
+    }
+    expDiff += 64;
+    if ( 0 < expDiff ) {
+        q = estimateDiv128To64( aSig, 0, bSig );
+        q = ( 2 < q ) ? q - 2 : 0;
+        q >>= 64 - expDiff;
+        bSig >>= 2;
+        aSig = ( ( aSig>>1 )<<( expDiff - 1 ) ) - bSig * q;
+    }
+    else {
+        aSig >>= 2;
+        bSig >>= 2;
+    }
+    do {
+        alternateASig = aSig;
+        ++q;
+        aSig -= bSig;
+    } while ( 0 <= (sbits64) aSig );
+    sigMean = aSig + alternateASig;
+    if ( ( sigMean < 0 ) || ( ( sigMean == 0 ) && ( q & 1 ) ) ) {
+        aSig = alternateASig;
+    }
+    zSign = ( (sbits64) aSig < 0 );
+    if ( zSign ) aSig = - aSig;
+    return normalizeRoundAndPackFloat64( roundData, aSign ^ zSign, bExp, aSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the square root of the double-precision floating-point value `a'.
+The operation is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 float64_sqrt( struct roundingData *roundData, float64 a )
+{
+    flag aSign;
+    int16 aExp, zExp;
+    bits64 aSig, zSig;
+    bits64 rem0, rem1, term0, term1; //, shiftedRem;
+    //float64 z;
+
+    aSig = extractFloat64Frac( a );
+    aExp = extractFloat64Exp( a );
+    aSign = extractFloat64Sign( a );
+    if ( aExp == 0x7FF ) {
+        if ( aSig ) return propagateFloat64NaN( a, a );
+        if ( ! aSign ) return a;
+        roundData->exception |= float_flag_invalid;
+        return float64_default_nan;
+    }
+    if ( aSign ) {
+        if ( ( aExp | aSig ) == 0 ) return a;
+        roundData->exception |= float_flag_invalid;
+        return float64_default_nan;
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return 0;
+        normalizeFloat64Subnormal( aSig, &aExp, &aSig );
+    }
+    zExp = ( ( aExp - 0x3FF )>>1 ) + 0x3FE;
+    aSig |= LIT64( 0x0010000000000000 );
+    zSig = estimateSqrt32( aExp, aSig>>21 );
+    zSig <<= 31;
+    aSig <<= 9 - ( aExp & 1 );
+    zSig = estimateDiv128To64( aSig, 0, zSig ) + zSig + 2;
+    if ( ( zSig & 0x3FF ) <= 5 ) {
+        if ( zSig < 2 ) {
+            zSig = LIT64( 0xFFFFFFFFFFFFFFFF );
+        }
+        else {
+            aSig <<= 2;
+            mul64To128( zSig, zSig, &term0, &term1 );
+            sub128( aSig, 0, term0, term1, &rem0, &rem1 );
+            while ( (sbits64) rem0 < 0 ) {
+                --zSig;
+                shortShift128Left( 0, zSig, 1, &term0, &term1 );
+                term1 |= 1;
+                add128( rem0, rem1, term0, term1, &rem0, &rem1 );
+            }
+            zSig |= ( ( rem0 | rem1 ) != 0 );
+        }
+    }
+    shift64RightJamming( zSig, 1, &zSig );
+    return roundAndPackFloat64( roundData, 0, zExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the double-precision floating-point value `a' is equal to the
+corresponding value `b', and 0 otherwise.  The comparison is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float64_eq( float64 a, float64 b )
+{
+
+    if (    ( ( extractFloat64Exp( a ) == 0x7FF ) && extractFloat64Frac( a ) )
+         || ( ( extractFloat64Exp( b ) == 0x7FF ) && extractFloat64Frac( b ) )
+       ) {
+        if ( float64_is_signaling_nan( a ) || float64_is_signaling_nan( b ) ) {
+            float_raise( float_flag_invalid );
+        }
+        return 0;
+    }
+    return ( a == b ) || ( (bits64) ( ( a | b )<<1 ) == 0 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the double-precision floating-point value `a' is less than or
+equal to the corresponding value `b', and 0 otherwise.  The comparison is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float64_le( float64 a, float64 b )
+{
+    flag aSign, bSign;
+
+    if (    ( ( extractFloat64Exp( a ) == 0x7FF ) && extractFloat64Frac( a ) )
+         || ( ( extractFloat64Exp( b ) == 0x7FF ) && extractFloat64Frac( b ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    aSign = extractFloat64Sign( a );
+    bSign = extractFloat64Sign( b );
+    if ( aSign != bSign ) return aSign || ( (bits64) ( ( a | b )<<1 ) == 0 );
+    return ( a == b ) || ( aSign ^ ( a < b ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the double-precision floating-point value `a' is less than
+the corresponding value `b', and 0 otherwise.  The comparison is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float64_lt( float64 a, float64 b )
+{
+    flag aSign, bSign;
+
+    if (    ( ( extractFloat64Exp( a ) == 0x7FF ) && extractFloat64Frac( a ) )
+         || ( ( extractFloat64Exp( b ) == 0x7FF ) && extractFloat64Frac( b ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    aSign = extractFloat64Sign( a );
+    bSign = extractFloat64Sign( b );
+    if ( aSign != bSign ) return aSign && ( (bits64) ( ( a | b )<<1 ) != 0 );
+    return ( a != b ) && ( aSign ^ ( a < b ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the double-precision floating-point value `a' is equal to the
+corresponding value `b', and 0 otherwise.  The invalid exception is raised
+if either operand is a NaN.  Otherwise, the comparison is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float64_eq_signaling( float64 a, float64 b )
+{
+
+    if (    ( ( extractFloat64Exp( a ) == 0x7FF ) && extractFloat64Frac( a ) )
+         || ( ( extractFloat64Exp( b ) == 0x7FF ) && extractFloat64Frac( b ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    return ( a == b ) || ( (bits64) ( ( a | b )<<1 ) == 0 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the double-precision floating-point value `a' is less than or
+equal to the corresponding value `b', and 0 otherwise.  Quiet NaNs do not
+cause an exception.  Otherwise, the comparison is performed according to the
+IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float64_le_quiet( float64 a, float64 b )
+{
+    flag aSign, bSign;
+    //int16 aExp, bExp;
+
+    if (    ( ( extractFloat64Exp( a ) == 0x7FF ) && extractFloat64Frac( a ) )
+         || ( ( extractFloat64Exp( b ) == 0x7FF ) && extractFloat64Frac( b ) )
+       ) {
+        /* Do nothing, even if NaN as we're quiet */
+        return 0;
+    }
+    aSign = extractFloat64Sign( a );
+    bSign = extractFloat64Sign( b );
+    if ( aSign != bSign ) return aSign || ( (bits64) ( ( a | b )<<1 ) == 0 );
+    return ( a == b ) || ( aSign ^ ( a < b ) );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the double-precision floating-point value `a' is less than
+the corresponding value `b', and 0 otherwise.  Quiet NaNs do not cause an
+exception.  Otherwise, the comparison is performed according to the IEC/IEEE
+Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag float64_lt_quiet( float64 a, float64 b )
+{
+    flag aSign, bSign;
+
+    if (    ( ( extractFloat64Exp( a ) == 0x7FF ) && extractFloat64Frac( a ) )
+         || ( ( extractFloat64Exp( b ) == 0x7FF ) && extractFloat64Frac( b ) )
+       ) {
+        /* Do nothing, even if NaN as we're quiet */
+        return 0;
+    }
+    aSign = extractFloat64Sign( a );
+    bSign = extractFloat64Sign( b );
+    if ( aSign != bSign ) return aSign && ( (bits64) ( ( a | b )<<1 ) != 0 );
+    return ( a != b ) && ( aSign ^ ( a < b ) );
+
+}
+
+#ifdef FLOATX80
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the extended double-precision floating-
+point value `a' to the 32-bit two's complement integer format.  The
+conversion is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic---which means in particular that the conversion
+is rounded according to the current rounding mode.  If `a' is a NaN, the
+largest positive integer is returned.  Otherwise, if the conversion
+overflows, the largest integer with the same sign as `a' is returned.
+-------------------------------------------------------------------------------
+*/
+int32 floatx80_to_int32( struct roundingData *roundData, floatx80 a )
+{
+    flag aSign;
+    int32 aExp, shiftCount;
+    bits64 aSig;
+
+    aSig = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    aSign = extractFloatx80Sign( a );
+    if ( ( aExp == 0x7FFF ) && (bits64) ( aSig<<1 ) ) aSign = 0;
+    shiftCount = 0x4037 - aExp;
+    if ( shiftCount <= 0 ) shiftCount = 1;
+    shift64RightJamming( aSig, shiftCount, &aSig );
+    return roundAndPackInt32( roundData, aSign, aSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the extended double-precision floating-
+point value `a' to the 32-bit two's complement integer format.  The
+conversion is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic, except that the conversion is always rounded
+toward zero.  If `a' is a NaN, the largest positive integer is returned.
+Otherwise, if the conversion overflows, the largest integer with the same
+sign as `a' is returned.
+-------------------------------------------------------------------------------
+*/
+int32 floatx80_to_int32_round_to_zero( floatx80 a )
+{
+    flag aSign;
+    int32 aExp, shiftCount;
+    bits64 aSig, savedASig;
+    int32 z;
+
+    aSig = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    aSign = extractFloatx80Sign( a );
+    shiftCount = 0x403E - aExp;
+    if ( shiftCount < 32 ) {
+        if ( ( aExp == 0x7FFF ) && (bits64) ( aSig<<1 ) ) aSign = 0;
+        goto invalid;
+    }
+    else if ( 63 < shiftCount ) {
+        if ( aExp || aSig ) float_raise( float_flag_inexact );
+        return 0;
+    }
+    savedASig = aSig;
+    aSig >>= shiftCount;
+    z = aSig;
+    if ( aSign ) z = - z;
+    if ( ( z < 0 ) ^ aSign ) {
+ invalid:
+        float_raise( float_flag_invalid );
+        return aSign ? 0x80000000 : 0x7FFFFFFF;
+    }
+    if ( ( aSig<<shiftCount ) != savedASig ) {
+        float_raise( float_flag_inexact );
+    }
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the extended double-precision floating-
+point value `a' to the single-precision floating-point format.  The
+conversion is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float32 floatx80_to_float32( struct roundingData *roundData, floatx80 a )
+{
+    flag aSign;
+    int32 aExp;
+    bits64 aSig;
+
+    aSig = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    aSign = extractFloatx80Sign( a );
+    if ( aExp == 0x7FFF ) {
+        if ( (bits64) ( aSig<<1 ) ) {
+            return commonNaNToFloat32( floatx80ToCommonNaN( a ) );
+        }
+        return packFloat32( aSign, 0xFF, 0 );
+    }
+    shift64RightJamming( aSig, 33, &aSig );
+    if ( aExp || aSig ) aExp -= 0x3F81;
+    return roundAndPackFloat32( roundData, aSign, aExp, aSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of converting the extended double-precision floating-
+point value `a' to the double-precision floating-point format.  The
+conversion is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+float64 floatx80_to_float64( struct roundingData *roundData, floatx80 a )
+{
+    flag aSign;
+    int32 aExp;
+    bits64 aSig, zSig;
+
+    aSig = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    aSign = extractFloatx80Sign( a );
+    if ( aExp == 0x7FFF ) {
+        if ( (bits64) ( aSig<<1 ) ) {
+            return commonNaNToFloat64( floatx80ToCommonNaN( a ) );
+        }
+        return packFloat64( aSign, 0x7FF, 0 );
+    }
+    shift64RightJamming( aSig, 1, &zSig );
+    if ( aExp || aSig ) aExp -= 0x3C01;
+    return roundAndPackFloat64( roundData, aSign, aExp, zSig );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Rounds the extended double-precision floating-point value `a' to an integer,
+and returns the result as an extended quadruple-precision floating-point
+value.  The operation is performed according to the IEC/IEEE Standard for
+Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 floatx80_round_to_int( struct roundingData *roundData, floatx80 a )
+{
+    flag aSign;
+    int32 aExp;
+    bits64 lastBitMask, roundBitsMask;
+    int8 roundingMode;
+    floatx80 z;
+
+    aExp = extractFloatx80Exp( a );
+    if ( 0x403E <= aExp ) {
+        if ( ( aExp == 0x7FFF ) && (bits64) ( extractFloatx80Frac( a )<<1 ) ) {
+            return propagateFloatx80NaN( a, a );
+        }
+        return a;
+    }
+    if ( aExp <= 0x3FFE ) {
+        if (    ( aExp == 0 )
+             && ( (bits64) ( extractFloatx80Frac( a )<<1 ) == 0 ) ) {
+            return a;
+        }
+        roundData->exception |= float_flag_inexact;
+        aSign = extractFloatx80Sign( a );
+        switch ( roundData->mode ) {
+         case float_round_nearest_even:
+            if ( ( aExp == 0x3FFE ) && (bits64) ( extractFloatx80Frac( a )<<1 )
+               ) {
+                return
+                    packFloatx80( aSign, 0x3FFF, LIT64( 0x8000000000000000 ) );
+            }
+            break;
+         case float_round_down:
+            return
+                  aSign ?
+                      packFloatx80( 1, 0x3FFF, LIT64( 0x8000000000000000 ) )
+                : packFloatx80( 0, 0, 0 );
+         case float_round_up:
+            return
+                  aSign ? packFloatx80( 1, 0, 0 )
+                : packFloatx80( 0, 0x3FFF, LIT64( 0x8000000000000000 ) );
+        }
+        return packFloatx80( aSign, 0, 0 );
+    }
+    lastBitMask = 1;
+    lastBitMask <<= 0x403E - aExp;
+    roundBitsMask = lastBitMask - 1;
+    z = a;
+    roundingMode = roundData->mode;
+    if ( roundingMode == float_round_nearest_even ) {
+        z.low += lastBitMask>>1;
+        if ( ( z.low & roundBitsMask ) == 0 ) z.low &= ~ lastBitMask;
+    }
+    else if ( roundingMode != float_round_to_zero ) {
+        if ( extractFloatx80Sign( z ) ^ ( roundingMode == float_round_up ) ) {
+            z.low += roundBitsMask;
+        }
+    }
+    z.low &= ~ roundBitsMask;
+    if ( z.low == 0 ) {
+        ++z.high;
+        z.low = LIT64( 0x8000000000000000 );
+    }
+    if ( z.low != a.low ) roundData->exception |= float_flag_inexact;
+    return z;
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of adding the absolute values of the extended double-
+precision floating-point values `a' and `b'.  If `zSign' is true, the sum is
+negated before being returned.  `zSign' is ignored if the result is a NaN.
+The addition is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static floatx80 addFloatx80Sigs( struct roundingData *roundData, floatx80 a, floatx80 b, flag zSign )
+{
+    int32 aExp, bExp, zExp;
+    bits64 aSig, bSig, zSig0, zSig1;
+    int32 expDiff;
+
+    aSig = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    bSig = extractFloatx80Frac( b );
+    bExp = extractFloatx80Exp( b );
+    expDiff = aExp - bExp;
+    if ( 0 < expDiff ) {
+        if ( aExp == 0x7FFF ) {
+            if ( (bits64) ( aSig<<1 ) ) return propagateFloatx80NaN( a, b );
+            return a;
+        }
+        if ( bExp == 0 ) --expDiff;
+        shift64ExtraRightJamming( bSig, 0, expDiff, &bSig, &zSig1 );
+        zExp = aExp;
+    }
+    else if ( expDiff < 0 ) {
+        if ( bExp == 0x7FFF ) {
+            if ( (bits64) ( bSig<<1 ) ) return propagateFloatx80NaN( a, b );
+            return packFloatx80( zSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
+        }
+        if ( aExp == 0 ) ++expDiff;
+        shift64ExtraRightJamming( aSig, 0, - expDiff, &aSig, &zSig1 );
+        zExp = bExp;
+    }
+    else {
+        if ( aExp == 0x7FFF ) {
+            if ( (bits64) ( ( aSig | bSig )<<1 ) ) {
+                return propagateFloatx80NaN( a, b );
+            }
+            return a;
+        }
+        zSig1 = 0;
+        zSig0 = aSig + bSig;
+        if ( aExp == 0 ) {
+            normalizeFloatx80Subnormal( zSig0, &zExp, &zSig0 );
+            goto roundAndPack;
+        }
+        zExp = aExp;
+        goto shiftRight1;
+    }
+    
+    zSig0 = aSig + bSig;
+
+    if ( (sbits64) zSig0 < 0 ) goto roundAndPack; 
+ shiftRight1:
+    shift64ExtraRightJamming( zSig0, zSig1, 1, &zSig0, &zSig1 );
+    zSig0 |= LIT64( 0x8000000000000000 );
+    ++zExp;
+ roundAndPack:
+    return
+        roundAndPackFloatx80(
+            roundData, zSign, zExp, zSig0, zSig1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of subtracting the absolute values of the extended
+double-precision floating-point values `a' and `b'.  If `zSign' is true,
+the difference is negated before being returned.  `zSign' is ignored if the
+result is a NaN.  The subtraction is performed according to the IEC/IEEE
+Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+static floatx80 subFloatx80Sigs( struct roundingData *roundData, floatx80 a, floatx80 b, flag zSign )
+{
+    int32 aExp, bExp, zExp;
+    bits64 aSig, bSig, zSig0, zSig1;
+    int32 expDiff;
+    floatx80 z;
+
+    aSig = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    bSig = extractFloatx80Frac( b );
+    bExp = extractFloatx80Exp( b );
+    expDiff = aExp - bExp;
+    if ( 0 < expDiff ) goto aExpBigger;
+    if ( expDiff < 0 ) goto bExpBigger;
+    if ( aExp == 0x7FFF ) {
+        if ( (bits64) ( ( aSig | bSig )<<1 ) ) {
+            return propagateFloatx80NaN( a, b );
+        }
+        roundData->exception |= float_flag_invalid;
+        z.low = floatx80_default_nan_low;
+        z.high = floatx80_default_nan_high;
+        z.__padding = 0;
+        return z;
+    }
+    if ( aExp == 0 ) {
+        aExp = 1;
+        bExp = 1;
+    }
+    zSig1 = 0;
+    if ( bSig < aSig ) goto aBigger;
+    if ( aSig < bSig ) goto bBigger;
+    return packFloatx80( roundData->mode == float_round_down, 0, 0 );
+ bExpBigger:
+    if ( bExp == 0x7FFF ) {
+        if ( (bits64) ( bSig<<1 ) ) return propagateFloatx80NaN( a, b );
+        return packFloatx80( zSign ^ 1, 0x7FFF, LIT64( 0x8000000000000000 ) );
+    }
+    if ( aExp == 0 ) ++expDiff;
+    shift128RightJamming( aSig, 0, - expDiff, &aSig, &zSig1 );
+ bBigger:
+    sub128( bSig, 0, aSig, zSig1, &zSig0, &zSig1 );
+    zExp = bExp;
+    zSign ^= 1;
+    goto normalizeRoundAndPack;
+ aExpBigger:
+    if ( aExp == 0x7FFF ) {
+        if ( (bits64) ( aSig<<1 ) ) return propagateFloatx80NaN( a, b );
+        return a;
+    }
+    if ( bExp == 0 ) --expDiff;
+    shift128RightJamming( bSig, 0, expDiff, &bSig, &zSig1 );
+ aBigger:
+    sub128( aSig, 0, bSig, zSig1, &zSig0, &zSig1 );
+    zExp = aExp;
+ normalizeRoundAndPack:
+    return
+        normalizeRoundAndPackFloatx80(
+            roundData, zSign, zExp, zSig0, zSig1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of adding the extended double-precision floating-point
+values `a' and `b'.  The operation is performed according to the IEC/IEEE
+Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 floatx80_add( struct roundingData *roundData, floatx80 a, floatx80 b )
+{
+    flag aSign, bSign;
+    
+    aSign = extractFloatx80Sign( a );
+    bSign = extractFloatx80Sign( b );
+    if ( aSign == bSign ) {
+        return addFloatx80Sigs( roundData, a, b, aSign );
+    }
+    else {
+        return subFloatx80Sigs( roundData, a, b, aSign );
+    }
+    
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of subtracting the extended double-precision floating-
+point values `a' and `b'.  The operation is performed according to the
+IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 floatx80_sub( struct roundingData *roundData, floatx80 a, floatx80 b )
+{
+    flag aSign, bSign;
+
+    aSign = extractFloatx80Sign( a );
+    bSign = extractFloatx80Sign( b );
+    if ( aSign == bSign ) {
+        return subFloatx80Sigs( roundData, a, b, aSign );
+    }
+    else {
+        return addFloatx80Sigs( roundData, a, b, aSign );
+    }
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of multiplying the extended double-precision floating-
+point values `a' and `b'.  The operation is performed according to the
+IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 floatx80_mul( struct roundingData *roundData, floatx80 a, floatx80 b )
+{
+    flag aSign, bSign, zSign;
+    int32 aExp, bExp, zExp;
+    bits64 aSig, bSig, zSig0, zSig1;
+    floatx80 z;
+
+    aSig = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    aSign = extractFloatx80Sign( a );
+    bSig = extractFloatx80Frac( b );
+    bExp = extractFloatx80Exp( b );
+    bSign = extractFloatx80Sign( b );
+    zSign = aSign ^ bSign;
+    if ( aExp == 0x7FFF ) {
+        if (    (bits64) ( aSig<<1 )
+             || ( ( bExp == 0x7FFF ) && (bits64) ( bSig<<1 ) ) ) {
+            return propagateFloatx80NaN( a, b );
+        }
+        if ( ( bExp | bSig ) == 0 ) goto invalid;
+        return packFloatx80( zSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
+    }
+    if ( bExp == 0x7FFF ) {
+        if ( (bits64) ( bSig<<1 ) ) return propagateFloatx80NaN( a, b );
+        if ( ( aExp | aSig ) == 0 ) {
+ invalid:
+            roundData->exception |= float_flag_invalid;
+            z.low = floatx80_default_nan_low;
+            z.high = floatx80_default_nan_high;
+            z.__padding = 0;
+            return z;
+        }
+        return packFloatx80( zSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloatx80( zSign, 0, 0 );
+        normalizeFloatx80Subnormal( aSig, &aExp, &aSig );
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) return packFloatx80( zSign, 0, 0 );
+        normalizeFloatx80Subnormal( bSig, &bExp, &bSig );
+    }
+    zExp = aExp + bExp - 0x3FFE;
+    mul64To128( aSig, bSig, &zSig0, &zSig1 );
+    if ( 0 < (sbits64) zSig0 ) {
+        shortShift128Left( zSig0, zSig1, 1, &zSig0, &zSig1 );
+        --zExp;
+    }
+    return
+        roundAndPackFloatx80(
+            roundData, zSign, zExp, zSig0, zSig1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the result of dividing the extended double-precision floating-point
+value `a' by the corresponding value `b'.  The operation is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 floatx80_div( struct roundingData *roundData, floatx80 a, floatx80 b )
+{
+    flag aSign, bSign, zSign;
+    int32 aExp, bExp, zExp;
+    bits64 aSig, bSig, zSig0, zSig1;
+    bits64 rem0, rem1, rem2, term0, term1, term2;
+    floatx80 z;
+
+    aSig = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    aSign = extractFloatx80Sign( a );
+    bSig = extractFloatx80Frac( b );
+    bExp = extractFloatx80Exp( b );
+    bSign = extractFloatx80Sign( b );
+    zSign = aSign ^ bSign;
+    if ( aExp == 0x7FFF ) {
+        if ( (bits64) ( aSig<<1 ) ) return propagateFloatx80NaN( a, b );
+        if ( bExp == 0x7FFF ) {
+            if ( (bits64) ( bSig<<1 ) ) return propagateFloatx80NaN( a, b );
+            goto invalid;
+        }
+        return packFloatx80( zSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
+    }
+    if ( bExp == 0x7FFF ) {
+        if ( (bits64) ( bSig<<1 ) ) return propagateFloatx80NaN( a, b );
+        return packFloatx80( zSign, 0, 0 );
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) {
+            if ( ( aExp | aSig ) == 0 ) {
+ invalid:
+                roundData->exception |= float_flag_invalid;
+                z.low = floatx80_default_nan_low;
+                z.high = floatx80_default_nan_high;
+                z.__padding = 0;
+                return z;
+            }
+            roundData->exception |= float_flag_divbyzero;
+            return packFloatx80( zSign, 0x7FFF, LIT64( 0x8000000000000000 ) );
+        }
+        normalizeFloatx80Subnormal( bSig, &bExp, &bSig );
+    }
+    if ( aExp == 0 ) {
+        if ( aSig == 0 ) return packFloatx80( zSign, 0, 0 );
+        normalizeFloatx80Subnormal( aSig, &aExp, &aSig );
+    }
+    zExp = aExp - bExp + 0x3FFE;
+    rem1 = 0;
+    if ( bSig <= aSig ) {
+        shift128Right( aSig, 0, 1, &aSig, &rem1 );
+        ++zExp;
+    }
+    zSig0 = estimateDiv128To64( aSig, rem1, bSig );
+    mul64To128( bSig, zSig0, &term0, &term1 );
+    sub128( aSig, rem1, term0, term1, &rem0, &rem1 );
+    while ( (sbits64) rem0 < 0 ) {
+        --zSig0;
+        add128( rem0, rem1, 0, bSig, &rem0, &rem1 );
+    }
+    zSig1 = estimateDiv128To64( rem1, 0, bSig );
+    if ( (bits64) ( zSig1<<1 ) <= 8 ) {
+        mul64To128( bSig, zSig1, &term1, &term2 );
+        sub128( rem1, 0, term1, term2, &rem1, &rem2 );
+        while ( (sbits64) rem1 < 0 ) {
+            --zSig1;
+            add128( rem1, rem2, 0, bSig, &rem1, &rem2 );
+        }
+        zSig1 |= ( ( rem1 | rem2 ) != 0 );
+    }
+    return
+        roundAndPackFloatx80(
+            roundData, zSign, zExp, zSig0, zSig1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the remainder of the extended double-precision floating-point value
+`a' with respect to the corresponding value `b'.  The operation is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 floatx80_rem( struct roundingData *roundData, floatx80 a, floatx80 b )
+{
+    flag aSign, bSign, zSign;
+    int32 aExp, bExp, expDiff;
+    bits64 aSig0, aSig1, bSig;
+    bits64 q, term0, term1, alternateASig0, alternateASig1;
+    floatx80 z;
+
+    aSig0 = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    aSign = extractFloatx80Sign( a );
+    bSig = extractFloatx80Frac( b );
+    bExp = extractFloatx80Exp( b );
+    bSign = extractFloatx80Sign( b );
+    if ( aExp == 0x7FFF ) {
+        if (    (bits64) ( aSig0<<1 )
+             || ( ( bExp == 0x7FFF ) && (bits64) ( bSig<<1 ) ) ) {
+            return propagateFloatx80NaN( a, b );
+        }
+        goto invalid;
+    }
+    if ( bExp == 0x7FFF ) {
+        if ( (bits64) ( bSig<<1 ) ) return propagateFloatx80NaN( a, b );
+        return a;
+    }
+    if ( bExp == 0 ) {
+        if ( bSig == 0 ) {
+ invalid:
+            roundData->exception |= float_flag_invalid;
+            z.low = floatx80_default_nan_low;
+            z.high = floatx80_default_nan_high;
+            z.__padding = 0;
+            return z;
+        }
+        normalizeFloatx80Subnormal( bSig, &bExp, &bSig );
+    }
+    if ( aExp == 0 ) {
+        if ( (bits64) ( aSig0<<1 ) == 0 ) return a;
+        normalizeFloatx80Subnormal( aSig0, &aExp, &aSig0 );
+    }
+    bSig |= LIT64( 0x8000000000000000 );
+    zSign = aSign;
+    expDiff = aExp - bExp;
+    aSig1 = 0;
+    if ( expDiff < 0 ) {
+        if ( expDiff < -1 ) return a;
+        shift128Right( aSig0, 0, 1, &aSig0, &aSig1 );
+        expDiff = 0;
+    }
+    q = ( bSig <= aSig0 );
+    if ( q ) aSig0 -= bSig;
+    expDiff -= 64;
+    while ( 0 < expDiff ) {
+        q = estimateDiv128To64( aSig0, aSig1, bSig );
+        q = ( 2 < q ) ? q - 2 : 0;
+        mul64To128( bSig, q, &term0, &term1 );
+        sub128( aSig0, aSig1, term0, term1, &aSig0, &aSig1 );
+        shortShift128Left( aSig0, aSig1, 62, &aSig0, &aSig1 );
+        expDiff -= 62;
+    }
+    expDiff += 64;
+    if ( 0 < expDiff ) {
+        q = estimateDiv128To64( aSig0, aSig1, bSig );
+        q = ( 2 < q ) ? q - 2 : 0;
+        q >>= 64 - expDiff;
+        mul64To128( bSig, q<<( 64 - expDiff ), &term0, &term1 );
+        sub128( aSig0, aSig1, term0, term1, &aSig0, &aSig1 );
+        shortShift128Left( 0, bSig, 64 - expDiff, &term0, &term1 );
+        while ( le128( term0, term1, aSig0, aSig1 ) ) {
+            ++q;
+            sub128( aSig0, aSig1, term0, term1, &aSig0, &aSig1 );
+        }
+    }
+    else {
+        term1 = 0;
+        term0 = bSig;
+    }
+    sub128( term0, term1, aSig0, aSig1, &alternateASig0, &alternateASig1 );
+    if (    lt128( alternateASig0, alternateASig1, aSig0, aSig1 )
+         || (    eq128( alternateASig0, alternateASig1, aSig0, aSig1 )
+              && ( q & 1 ) )
+       ) {
+        aSig0 = alternateASig0;
+        aSig1 = alternateASig1;
+        zSign = ! zSign;
+    }
+
+    return
+        normalizeRoundAndPackFloatx80(
+            roundData, zSign, bExp + expDiff, aSig0, aSig1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns the square root of the extended double-precision floating-point
+value `a'.  The operation is performed according to the IEC/IEEE Standard
+for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+floatx80 floatx80_sqrt( struct roundingData *roundData, floatx80 a )
+{
+    flag aSign;
+    int32 aExp, zExp;
+    bits64 aSig0, aSig1, zSig0, zSig1;
+    bits64 rem0, rem1, rem2, rem3, term0, term1, term2, term3;
+    bits64 shiftedRem0, shiftedRem1;
+    floatx80 z;
+
+    aSig0 = extractFloatx80Frac( a );
+    aExp = extractFloatx80Exp( a );
+    aSign = extractFloatx80Sign( a );
+    if ( aExp == 0x7FFF ) {
+        if ( (bits64) ( aSig0<<1 ) ) return propagateFloatx80NaN( a, a );
+        if ( ! aSign ) return a;
+        goto invalid;
+    }
+    if ( aSign ) {
+        if ( ( aExp | aSig0 ) == 0 ) return a;
+ invalid:
+        roundData->exception |= float_flag_invalid;
+        z.low = floatx80_default_nan_low;
+        z.high = floatx80_default_nan_high;
+        z.__padding = 0;
+        return z;
+    }
+    if ( aExp == 0 ) {
+        if ( aSig0 == 0 ) return packFloatx80( 0, 0, 0 );
+        normalizeFloatx80Subnormal( aSig0, &aExp, &aSig0 );
+    }
+    zExp = ( ( aExp - 0x3FFF )>>1 ) + 0x3FFF;
+    zSig0 = estimateSqrt32( aExp, aSig0>>32 );
+    zSig0 <<= 31;
+    aSig1 = 0;
+    shift128Right( aSig0, 0, ( aExp & 1 ) + 2, &aSig0, &aSig1 );
+    zSig0 = estimateDiv128To64( aSig0, aSig1, zSig0 ) + zSig0 + 4;
+    if ( 0 <= (sbits64) zSig0 ) zSig0 = LIT64( 0xFFFFFFFFFFFFFFFF );
+    shortShift128Left( aSig0, aSig1, 2, &aSig0, &aSig1 );
+    mul64To128( zSig0, zSig0, &term0, &term1 );
+    sub128( aSig0, aSig1, term0, term1, &rem0, &rem1 );
+    while ( (sbits64) rem0 < 0 ) {
+        --zSig0;
+        shortShift128Left( 0, zSig0, 1, &term0, &term1 );
+        term1 |= 1;
+        add128( rem0, rem1, term0, term1, &rem0, &rem1 );
+    }
+    shortShift128Left( rem0, rem1, 63, &shiftedRem0, &shiftedRem1 );
+    zSig1 = estimateDiv128To64( shiftedRem0, shiftedRem1, zSig0 );
+    if ( (bits64) ( zSig1<<1 ) <= 10 ) {
+        if ( zSig1 == 0 ) zSig1 = 1;
+        mul64To128( zSig0, zSig1, &term1, &term2 );
+        shortShift128Left( term1, term2, 1, &term1, &term2 );
+        sub128( rem1, 0, term1, term2, &rem1, &rem2 );
+        mul64To128( zSig1, zSig1, &term2, &term3 );
+        sub192( rem1, rem2, 0, 0, term2, term3, &rem1, &rem2, &rem3 );
+        while ( (sbits64) rem1 < 0 ) {
+            --zSig1;
+            shortShift192Left( 0, zSig0, zSig1, 1, &term1, &term2, &term3 );
+            term3 |= 1;
+            add192(
+                rem1, rem2, rem3, term1, term2, term3, &rem1, &rem2, &rem3 );
+        }
+        zSig1 |= ( ( rem1 | rem2 | rem3 ) != 0 );
+    }
+    return
+        roundAndPackFloatx80(
+            roundData, 0, zExp, zSig0, zSig1 );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the extended double-precision floating-point value `a' is
+equal to the corresponding value `b', and 0 otherwise.  The comparison is
+performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag floatx80_eq( floatx80 a, floatx80 b )
+{
+
+    if (    (    ( extractFloatx80Exp( a ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( a )<<1 ) )
+         || (    ( extractFloatx80Exp( b ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( b )<<1 ) )
+       ) {
+        if (    floatx80_is_signaling_nan( a )
+             || floatx80_is_signaling_nan( b ) ) {
+            float_raise( float_flag_invalid );
+        }
+        return 0;
+    }
+    return
+           ( a.low == b.low )
+        && (    ( a.high == b.high )
+             || (    ( a.low == 0 )
+                  && ( (bits16) ( ( a.high | b.high )<<1 ) == 0 ) )
+           );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the extended double-precision floating-point value `a' is
+less than or equal to the corresponding value `b', and 0 otherwise.  The
+comparison is performed according to the IEC/IEEE Standard for Binary
+Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag floatx80_le( floatx80 a, floatx80 b )
+{
+    flag aSign, bSign;
+
+    if (    (    ( extractFloatx80Exp( a ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( a )<<1 ) )
+         || (    ( extractFloatx80Exp( b ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( b )<<1 ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    aSign = extractFloatx80Sign( a );
+    bSign = extractFloatx80Sign( b );
+    if ( aSign != bSign ) {
+        return
+               aSign
+            || (    ( ( (bits16) ( ( a.high | b.high )<<1 ) ) | a.low | b.low )
+                 == 0 );
+    }
+    return
+          aSign ? le128( b.high, b.low, a.high, a.low )
+        : le128( a.high, a.low, b.high, b.low );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the extended double-precision floating-point value `a' is
+less than the corresponding value `b', and 0 otherwise.  The comparison
+is performed according to the IEC/IEEE Standard for Binary Floating-point
+Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag floatx80_lt( floatx80 a, floatx80 b )
+{
+    flag aSign, bSign;
+
+    if (    (    ( extractFloatx80Exp( a ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( a )<<1 ) )
+         || (    ( extractFloatx80Exp( b ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( b )<<1 ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    aSign = extractFloatx80Sign( a );
+    bSign = extractFloatx80Sign( b );
+    if ( aSign != bSign ) {
+        return
+               aSign
+            && (    ( ( (bits16) ( ( a.high | b.high )<<1 ) ) | a.low | b.low )
+                 != 0 );
+    }
+    return
+          aSign ? lt128( b.high, b.low, a.high, a.low )
+        : lt128( a.high, a.low, b.high, b.low );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the extended double-precision floating-point value `a' is equal
+to the corresponding value `b', and 0 otherwise.  The invalid exception is
+raised if either operand is a NaN.  Otherwise, the comparison is performed
+according to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag floatx80_eq_signaling( floatx80 a, floatx80 b )
+{
+
+    if (    (    ( extractFloatx80Exp( a ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( a )<<1 ) )
+         || (    ( extractFloatx80Exp( b ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( b )<<1 ) )
+       ) {
+        float_raise( float_flag_invalid );
+        return 0;
+    }
+    return
+           ( a.low == b.low )
+        && (    ( a.high == b.high )
+             || (    ( a.low == 0 )
+                  && ( (bits16) ( ( a.high | b.high )<<1 ) == 0 ) )
+           );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the extended double-precision floating-point value `a' is less
+than or equal to the corresponding value `b', and 0 otherwise.  Quiet NaNs
+do not cause an exception.  Otherwise, the comparison is performed according
+to the IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag floatx80_le_quiet( floatx80 a, floatx80 b )
+{
+    flag aSign, bSign;
+
+    if (    (    ( extractFloatx80Exp( a ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( a )<<1 ) )
+         || (    ( extractFloatx80Exp( b ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( b )<<1 ) )
+       ) {
+        /* Do nothing, even if NaN as we're quiet */
+        return 0;
+    }
+    aSign = extractFloatx80Sign( a );
+    bSign = extractFloatx80Sign( b );
+    if ( aSign != bSign ) {
+        return
+               aSign
+            || (    ( ( (bits16) ( ( a.high | b.high )<<1 ) ) | a.low | b.low )
+                 == 0 );
+    }
+    return
+          aSign ? le128( b.high, b.low, a.high, a.low )
+        : le128( a.high, a.low, b.high, b.low );
+
+}
+
+/*
+-------------------------------------------------------------------------------
+Returns 1 if the extended double-precision floating-point value `a' is less
+than the corresponding value `b', and 0 otherwise.  Quiet NaNs do not cause
+an exception.  Otherwise, the comparison is performed according to the
+IEC/IEEE Standard for Binary Floating-point Arithmetic.
+-------------------------------------------------------------------------------
+*/
+flag floatx80_lt_quiet( floatx80 a, floatx80 b )
+{
+    flag aSign, bSign;
+
+    if (    (    ( extractFloatx80Exp( a ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( a )<<1 ) )
+         || (    ( extractFloatx80Exp( b ) == 0x7FFF )
+              && (bits64) ( extractFloatx80Frac( b )<<1 ) )
+       ) {
+        /* Do nothing, even if NaN as we're quiet */
+        return 0;
+    }
+    aSign = extractFloatx80Sign( a );
+    bSign = extractFloatx80Sign( b );
+    if ( aSign != bSign ) {
+        return
+               aSign
+            && (    ( ( (bits16) ( ( a.high | b.high )<<1 ) ) | a.low | b.low )
+                 != 0 );
+    }
+    return
+          aSign ? lt128( b.high, b.low, a.high, a.low )
+        : lt128( a.high, a.low, b.high, b.low );
+
+}
+
+#endif
+

arch/arm/nwfpe/softfloat.h

@@ -0,0 +1,281 @@
+
+/*
+===============================================================================
+
+This C header file is part of the SoftFloat IEC/IEEE Floating-point
+Arithmetic Package, Release 2.
+
+Written by John R. Hauser.  This work was made possible in part by the
+International Computer Science Institute, located at Suite 600, 1947 Center
+Street, Berkeley, California 94704.  Funding was partially provided by the
+National Science Foundation under grant MIP-9311980.  The original version
+of this code was written as part of a project to build a fixed-point vector
+processor in collaboration with the University of California at Berkeley,
+overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
+is available through the Web page
+http://www.jhauser.us/arithmetic/SoftFloat-2b/SoftFloat-source.txt
+
+THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort
+has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT
+TIMES RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO
+PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ANY
+AND ALL LOSSES, COSTS, OR OTHER PROBLEMS ARISING FROM ITS USE.
+
+Derivative works are acceptable, even for commercial purposes, so long as
+(1) they include prominent notice that the work is derivative, and (2) they
+include prominent notice akin to these three paragraphs for those parts of
+this code that are retained.
+
+===============================================================================
+*/
+
+#ifndef __SOFTFLOAT_H__
+#define __SOFTFLOAT_H__
+
+
+/*
+-------------------------------------------------------------------------------
+The macro `FLOATX80' must be defined to enable the extended double-precision
+floating-point format `floatx80'.  If this macro is not defined, the
+`floatx80' type will not be defined, and none of the functions that either
+input or output the `floatx80' type will be defined.
+-------------------------------------------------------------------------------
+*/
+#ifdef CONFIG_FPE_NWFPE_XP
+#define FLOATX80
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE floating-point types.
+-------------------------------------------------------------------------------
+*/
+typedef u32 float32;
+typedef u64 float64;
+typedef struct {
+#ifdef __ARMEB__
+    u16 __padding;
+    u16 high;
+#else
+    u16 high;
+    u16 __padding;
+#endif
+    u64 low;
+}  __attribute__ ((packed,aligned(4))) floatx80;
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE floating-point underflow tininess-detection mode.
+-------------------------------------------------------------------------------
+*/
+extern signed char float_detect_tininess;
+enum {
+    float_tininess_after_rounding  = 0,
+    float_tininess_before_rounding = 1
+};
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE floating-point rounding mode.
+-------------------------------------------------------------------------------
+*/
+//extern int8 float_rounding_mode;
+enum {
+    float_round_nearest_even = 0,
+    float_round_to_zero      = 1,
+    float_round_down         = 2,
+    float_round_up           = 3
+};
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE floating-point exception flags.
+-------------------------------------------------------------------------------
+enum {
+    float_flag_inexact   =  1,
+    float_flag_underflow =  2,
+    float_flag_overflow  =  4,
+    float_flag_divbyzero =  8,
+    float_flag_invalid   = 16
+};
+
+ScottB: November 4, 1998
+Changed the enumeration to match the bit order in the FPA11.
+*/
+
+enum {
+    float_flag_invalid   =  1,
+    float_flag_divbyzero =  2,
+    float_flag_overflow  =  4,
+    float_flag_underflow =  8,
+    float_flag_inexact   = 16
+};
+
+/*
+-------------------------------------------------------------------------------
+Routine to raise any or all of the software IEC/IEEE floating-point
+exception flags.
+-------------------------------------------------------------------------------
+*/
+void float_raise( signed char );
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE integer-to-floating-point conversion routines.
+-------------------------------------------------------------------------------
+*/
+float32 int32_to_float32( struct roundingData *, signed int );
+float64 int32_to_float64( signed int );
+#ifdef FLOATX80
+floatx80 int32_to_floatx80( signed int );
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE single-precision conversion routines.
+-------------------------------------------------------------------------------
+*/
+signed int float32_to_int32( struct roundingData *, float32 );
+signed int float32_to_int32_round_to_zero( float32 );
+float64 float32_to_float64( float32 );
+#ifdef FLOATX80
+floatx80 float32_to_floatx80( float32 );
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE single-precision operations.
+-------------------------------------------------------------------------------
+*/
+float32 float32_round_to_int( struct roundingData*, float32 );
+float32 float32_add( struct roundingData *, float32, float32 );
+float32 float32_sub( struct roundingData *, float32, float32 );
+float32 float32_mul( struct roundingData *, float32, float32 );
+float32 float32_div( struct roundingData *, float32, float32 );
+float32 float32_rem( struct roundingData *, float32, float32 );
+float32 float32_sqrt( struct roundingData*, float32 );
+char float32_eq( float32, float32 );
+char float32_le( float32, float32 );
+char float32_lt( float32, float32 );
+char float32_eq_signaling( float32, float32 );
+char float32_le_quiet( float32, float32 );
+char float32_lt_quiet( float32, float32 );
+char float32_is_signaling_nan( float32 );
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE double-precision conversion routines.
+-------------------------------------------------------------------------------
+*/
+signed int float64_to_int32( struct roundingData *, float64 );
+signed int float64_to_int32_round_to_zero( float64 );
+float32 float64_to_float32( struct roundingData *, float64 );
+#ifdef FLOATX80
+floatx80 float64_to_floatx80( float64 );
+#endif
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE double-precision operations.
+-------------------------------------------------------------------------------
+*/
+float64 float64_round_to_int( struct roundingData *, float64 );
+float64 float64_add( struct roundingData *, float64, float64 );
+float64 float64_sub( struct roundingData *, float64, float64 );
+float64 float64_mul( struct roundingData *, float64, float64 );
+float64 float64_div( struct roundingData *, float64, float64 );
+float64 float64_rem( struct roundingData *, float64, float64 );
+float64 float64_sqrt( struct roundingData *, float64 );
+char float64_eq( float64, float64 );
+char float64_le( float64, float64 );
+char float64_lt( float64, float64 );
+char float64_eq_signaling( float64, float64 );
+char float64_le_quiet( float64, float64 );
+char float64_lt_quiet( float64, float64 );
+char float64_is_signaling_nan( float64 );
+
+#ifdef FLOATX80
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE extended double-precision conversion routines.
+-------------------------------------------------------------------------------
+*/
+signed int floatx80_to_int32( struct roundingData *, floatx80 );
+signed int floatx80_to_int32_round_to_zero( floatx80 );
+float32 floatx80_to_float32( struct roundingData *, floatx80 );
+float64 floatx80_to_float64( struct roundingData *, floatx80 );
+
+/*
+-------------------------------------------------------------------------------
+Software IEC/IEEE extended double-precision operations.
+-------------------------------------------------------------------------------
+*/
+floatx80 floatx80_round_to_int( struct roundingData *, floatx80 );
+floatx80 floatx80_add( struct roundingData *, floatx80, floatx80 );
+floatx80 floatx80_sub( struct roundingData *, floatx80, floatx80 );
+floatx80 floatx80_mul( struct roundingData *, floatx80, floatx80 );
+floatx80 floatx80_div( struct roundingData *, floatx80, floatx80 );
+floatx80 floatx80_rem( struct roundingData *, floatx80, floatx80 );
+floatx80 floatx80_sqrt( struct roundingData *, floatx80 );
+char floatx80_eq( floatx80, floatx80 );
+char floatx80_le( floatx80, floatx80 );
+char floatx80_lt( floatx80, floatx80 );
+char floatx80_eq_signaling( floatx80, floatx80 );
+char floatx80_le_quiet( floatx80, floatx80 );
+char floatx80_lt_quiet( floatx80, floatx80 );
+char floatx80_is_signaling_nan( floatx80 );
+
+extern flag floatx80_is_nan(floatx80);
+
+#endif
+
+static inline flag extractFloat32Sign(float32 a)
+{
+	return a >> 31;
+}
+
+static inline flag float32_eq_nocheck(float32 a, float32 b)
+{
+	return (a == b) || ((bits32) ((a | b) << 1) == 0);
+}
+
+static inline flag float32_lt_nocheck(float32 a, float32 b)
+{
+	flag aSign, bSign;
+
+	aSign = extractFloat32Sign(a);
+	bSign = extractFloat32Sign(b);
+	if (aSign != bSign)
+		return aSign && ((bits32) ((a | b) << 1) != 0);
+	return (a != b) && (aSign ^ (a < b));
+}
+
+static inline flag extractFloat64Sign(float64 a)
+{
+	return a >> 63;
+}
+
+static inline flag float64_eq_nocheck(float64 a, float64 b)
+{
+	return (a == b) || ((bits64) ((a | b) << 1) == 0);
+}
+
+static inline flag float64_lt_nocheck(float64 a, float64 b)
+{
+	flag aSign, bSign;
+
+	aSign = extractFloat64Sign(a);
+	bSign = extractFloat64Sign(b);
+	if (aSign != bSign)
+		return aSign && ((bits64) ((a | b) << 1) != 0);
+	return (a != b) && (aSign ^ (a < b));
+}
+
+extern flag float32_is_nan( float32 a );
+extern flag float64_is_nan( float64 a );
+
+extern int32 float64_to_uint32( struct roundingData *roundData, float64 a );
+extern int32 float64_to_uint32_round_to_zero( float64 a );
+
+#endif

arch/arm/oprofile/Makefile

@@ -0,0 +1,13 @@
+obj-$(CONFIG_OPROFILE) += oprofile.o
+
+DRIVER_OBJS = $(addprefix ../../../drivers/oprofile/, \
+		oprof.o cpu_buffer.o buffer_sync.o \
+		event_buffer.o oprofile_files.o \
+		oprofilefs.o oprofile_stats.o \
+		timer_int.o )
+
+ifeq ($(CONFIG_HW_PERF_EVENTS),y)
+DRIVER_OBJS += $(addprefix ../../../drivers/oprofile/, oprofile_perf.o)
+endif
+
+oprofile-y				:= $(DRIVER_OBJS) common.o

arch/arm/oprofile/common.c

@@ -0,0 +1,132 @@
+/**
+ * @file common.c
+ *
+ * @remark Copyright 2004 Oprofile Authors
+ * @remark Copyright 2010 ARM Ltd.
+ * @remark Read the file COPYING
+ *
+ * @author Zwane Mwaikambo
+ * @author Will Deacon [move to perf]
+ */
+
+#include <linux/cpumask.h>
+#include <linux/init.h>
+#include <linux/mutex.h>
+#include <linux/oprofile.h>
+#include <linux/perf_event.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <asm/stacktrace.h>
+#include <linux/uaccess.h>
+
+#include <asm/perf_event.h>
+#include <asm/ptrace.h>
+
+#ifdef CONFIG_HW_PERF_EVENTS
+
+/*
+ * OProfile has a curious naming scheme for the ARM PMUs, but they are
+ * part of the user ABI so we need to map from the perf PMU name for
+ * supported PMUs.
+ */
+static struct op_perf_name {
+	char *perf_name;
+	char *op_name;
+} op_perf_name_map[] = {
+	{ "armv5_xscale1",	"arm/xscale1"	},
+	{ "armv5_xscale2",	"arm/xscale2"	},
+	{ "armv6_1136",		"arm/armv6"	},
+	{ "armv6_1156",		"arm/armv6"	},
+	{ "armv6_1176",		"arm/armv6"	},
+	{ "armv6_11mpcore",	"arm/mpcore"	},
+	{ "armv7_cortex_a8",	"arm/armv7"	},
+	{ "armv7_cortex_a9",	"arm/armv7-ca9"	},
+};
+
+char *op_name_from_perf_id(void)
+{
+	int i;
+	struct op_perf_name names;
+	const char *perf_name = perf_pmu_name();
+
+	for (i = 0; i < ARRAY_SIZE(op_perf_name_map); ++i) {
+		names = op_perf_name_map[i];
+		if (!strcmp(names.perf_name, perf_name))
+			return names.op_name;
+	}
+
+	return NULL;
+}
+#endif
+
+static int report_trace(struct stackframe *frame, void *d)
+{
+	unsigned int *depth = d;
+
+	if (*depth) {
+		oprofile_add_trace(frame->pc);
+		(*depth)--;
+	}
+
+	return *depth == 0;
+}
+
+/*
+ * The registers we're interested in are at the end of the variable
+ * length saved register structure. The fp points at the end of this
+ * structure so the address of this struct is:
+ * (struct frame_tail *)(xxx->fp)-1
+ */
+struct frame_tail {
+	struct frame_tail *fp;
+	unsigned long sp;
+	unsigned long lr;
+} __attribute__((packed));
+
+static struct frame_tail* user_backtrace(struct frame_tail *tail)
+{
+	struct frame_tail buftail[2];
+
+	/* Also check accessibility of one struct frame_tail beyond */
+	if (!access_ok(VERIFY_READ, tail, sizeof(buftail)))
+		return NULL;
+	if (__copy_from_user_inatomic(buftail, tail, sizeof(buftail)))
+		return NULL;
+
+	oprofile_add_trace(buftail[0].lr);
+
+	/* frame pointers should strictly progress back up the stack
+	 * (towards higher addresses) */
+	if (tail + 1 >= buftail[0].fp)
+		return NULL;
+
+	return buftail[0].fp-1;
+}
+
+static void arm_backtrace(struct pt_regs * const regs, unsigned int depth)
+{
+	struct frame_tail *tail = ((struct frame_tail *) regs->ARM_fp) - 1;
+
+	if (!user_mode(regs)) {
+		struct stackframe frame;
+		arm_get_current_stackframe(regs, &frame);
+		walk_stackframe(&frame, report_trace, &depth);
+		return;
+	}
+
+	while (depth-- && tail && !((unsigned long) tail & 3))
+		tail = user_backtrace(tail);
+}
+
+int __init oprofile_arch_init(struct oprofile_operations *ops)
+{
+	/* provide backtrace support also in timer mode: */
+	ops->backtrace		= arm_backtrace;
+
+	return oprofile_perf_init(ops);
+}
+
+void oprofile_arch_exit(void)
+{
+	oprofile_perf_exit();
+}

arch/arm/plat-omap/Kconfig

@@ -0,0 +1,161 @@
+config ARCH_OMAP
+	bool
+
+if ARCH_OMAP
+
+menu "TI OMAP Common Features"
+
+config ARCH_OMAP_OTG
+	bool
+
+comment "OMAP Feature Selections"
+
+config OMAP_DEBUG_DEVICES
+	bool
+	help
+	  For debug cards on TI reference boards.
+
+config OMAP_DEBUG_LEDS
+	def_bool y if NEW_LEDS
+	depends on OMAP_DEBUG_DEVICES
+	select LEDS_CLASS
+
+config POWER_AVS_OMAP
+	bool "AVS(Adaptive Voltage Scaling) support for OMAP IP versions 1&2"
+	depends on POWER_AVS && (ARCH_OMAP3 || ARCH_OMAP4) && PM
+	select POWER_SUPPLY
+	help
+	  Say Y to enable AVS(Adaptive Voltage Scaling)
+	  support on OMAP containing the version 1 or
+	  version 2 of the SmartReflex IP.
+	  V1 is the 65nm version used in OMAP3430.
+	  V2 is the update for the 45nm version of the IP used in OMAP3630
+	  and OMAP4430
+
+	  Please note, that by default SmartReflex is only
+	  initialized and not enabled. To enable the automatic voltage
+	  compensation for vdd mpu and vdd core from user space,
+	  user must write 1 to
+		/debug/smartreflex/sr_<X>/autocomp,
+	  where X is mpu_iva or core for OMAP3.
+	  Optionally autocompensation can be enabled in the kernel
+	  by default during system init via the enable_on_init flag
+	  which an be passed as platform data to the smartreflex driver.
+
+config POWER_AVS_OMAP_CLASS3
+	bool "Class 3 mode of Smartreflex Implementation"
+	depends on POWER_AVS_OMAP && TWL4030_CORE
+	help
+	  Say Y to enable Class 3 implementation of Smartreflex
+
+	  Class 3 implementation of Smartreflex employs continuous hardware
+	  voltage calibration.
+
+config OMAP_RESET_CLOCKS
+	bool "Reset unused clocks during boot"
+	depends on ARCH_OMAP
+	help
+	  Say Y if you want to reset unused clocks during boot.
+	  This option saves power, but assumes all drivers are
+	  using the clock framework. Broken drivers that do not
+	  yet use clock framework may not work with this option.
+	  If you are booting from another operating system, you
+	  probably do not want this option enabled until your
+	  device drivers work properly.
+
+config OMAP_MUX
+	bool "OMAP multiplexing support"
+	depends on ARCH_OMAP
+	default y
+	help
+	  Pin multiplexing support for OMAP boards. If your bootloader
+	  sets the multiplexing correctly, say N. Otherwise, or if unsure,
+	  say Y.
+
+config OMAP_MUX_DEBUG
+	bool "Multiplexing debug output"
+	depends on OMAP_MUX
+	help
+	  Makes the multiplexing functions print out a lot of debug info.
+	  This is useful if you want to find out the correct values of the
+	  multiplexing registers.
+
+config OMAP_MUX_WARNINGS
+	bool "Warn about pins the bootloader didn't set up"
+	depends on OMAP_MUX
+	default y
+	help
+	  Choose Y here to warn whenever driver initialization logic needs
+	  to change the pin multiplexing setup.	 When there are no warnings
+	  printed, it's safe to deselect OMAP_MUX for your product.
+
+config OMAP_MPU_TIMER
+	bool "Use mpu timer"
+	depends on ARCH_OMAP1
+	help
+	  Select this option if you want to use the OMAP mpu timer. This
+	  timer provides more intra-tick resolution than the 32KHz timer,
+	  but consumes more power.
+
+config OMAP_32K_TIMER
+	bool "Use 32KHz timer"
+	depends on ARCH_OMAP16XX || ARCH_OMAP2PLUS
+	default y if (ARCH_OMAP16XX || ARCH_OMAP2PLUS)
+	help
+	  Select this option if you want to enable the OMAP 32KHz timer.
+	  This timer saves power compared to the OMAP_MPU_TIMER, and has
+	  support for no tick during idle. The 32KHz timer provides less
+	  intra-tick resolution than OMAP_MPU_TIMER. The 32KHz timer is
+	  currently only available for OMAP16XX, 24XX, 34XX, OMAP4/5 and DRA7XX.
+
+	  On OMAP2PLUS this value is only used for CONFIG_HZ and
+	  CLOCK_TICK_RATE compile time calculation.
+	  The actual timer selection is done in the board file
+	  through the (DT_)MACHINE_START structure.
+
+
+config OMAP3_L2_AUX_SECURE_SAVE_RESTORE
+	bool "OMAP3 HS/EMU save and restore for L2 AUX control register"
+	depends on ARCH_OMAP3 && PM
+	default n
+	help
+	  Without this option, L2 Auxiliary control register contents are
+	  lost during off-mode entry on HS/EMU devices. This feature
+	  requires support from PPA / boot-loader in HS/EMU devices, which
+	  currently does not exist by default.
+
+config OMAP3_L2_AUX_SECURE_SERVICE_SET_ID
+	int "Service ID for the support routine to set L2 AUX control"
+	depends on OMAP3_L2_AUX_SECURE_SAVE_RESTORE
+	default 43
+	help
+	  PPA routine service ID for setting L2 auxiliary control register.
+
+config OMAP_DM_TIMER
+	bool "Use dual-mode timer"
+	depends on ARCH_OMAP16XX || ARCH_OMAP2PLUS
+	help
+	 Select this option if you want to use OMAP Dual-Mode timers.
+
+config OMAP_SERIAL_WAKE
+	bool "Enable wake-up events for serial ports"
+	depends on ARCH_OMAP1 && OMAP_MUX
+	default y
+	help
+	  Select this option if you want to have your system wake up
+	  to data on the serial RX line. This allows you to wake the
+	  system from serial console.
+
+choice
+	prompt "OMAP PM layer selection"
+	depends on ARCH_OMAP
+	default OMAP_PM_NOOP
+
+config OMAP_PM_NOOP
+	bool "No-op/debug PM layer"
+
+endchoice
+
+endmenu
+
+endif

arch/arm/plat-omap/Makefile

@@ -0,0 +1,16 @@
+#
+# Makefile for the linux kernel.
+#
+
+ccflags-$(CONFIG_ARCH_MULTIPLATFORM) := -I$(srctree)/arch/arm/plat-omap/include
+
+# Common support
+obj-y := sram.o dma.o counter_32k.o
+
+# omap_device support (OMAP2+ only at the moment)
+
+obj-$(CONFIG_OMAP_DM_TIMER) += dmtimer.o
+obj-$(CONFIG_OMAP_DEBUG_LEDS) += debug-leds.o
+i2c-omap-$(CONFIG_I2C_OMAP) := i2c.o
+obj-y += $(i2c-omap-m) $(i2c-omap-y)
+

arch/arm/plat-omap/counter_32k.c

@@ -0,0 +1,117 @@
+/*
+ * OMAP 32ksynctimer/counter_32k-related code
+ *
+ * Copyright (C) 2009 Texas Instruments
+ * Copyright (C) 2010 Nokia Corporation
+ * Tony Lindgren <tony@atomide.com>
+ * Added OMAP4 support - Santosh Shilimkar <santosh.shilimkar@ti.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * NOTE: This timer is not the same timer as the old OMAP1 MPU timer.
+ */
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/clk.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/clocksource.h>
+#include <linux/sched_clock.h>
+
+#include <asm/mach/time.h>
+
+#include <plat/counter-32k.h>
+
+/* OMAP2_32KSYNCNT_CR_OFF: offset of 32ksync counter register */
+#define OMAP2_32KSYNCNT_REV_OFF		0x0
+#define OMAP2_32KSYNCNT_REV_SCHEME	(0x3 << 30)
+#define OMAP2_32KSYNCNT_CR_OFF_LOW	0x10
+#define OMAP2_32KSYNCNT_CR_OFF_HIGH	0x30
+
+/*
+ * 32KHz clocksource ... always available, on pretty most chips except
+ * OMAP 730 and 1510.  Other timers could be used as clocksources, with
+ * higher resolution in free-running counter modes (e.g. 12 MHz xtal),
+ * but systems won't necessarily want to spend resources that way.
+ */
+static void __iomem *sync32k_cnt_reg;
+
+static u64 notrace omap_32k_read_sched_clock(void)
+{
+	return sync32k_cnt_reg ? readl_relaxed(sync32k_cnt_reg) : 0;
+}
+
+/**
+ * omap_read_persistent_clock64 -  Return time from a persistent clock.
+ *
+ * Reads the time from a source which isn't disabled during PM, the
+ * 32k sync timer.  Convert the cycles elapsed since last read into
+ * nsecs and adds to a monotonically increasing timespec64.
+ */
+static struct timespec64 persistent_ts;
+static cycles_t cycles;
+static unsigned int persistent_mult, persistent_shift;
+
+static void omap_read_persistent_clock64(struct timespec64 *ts)
+{
+	unsigned long long nsecs;
+	cycles_t last_cycles;
+
+	last_cycles = cycles;
+	cycles = sync32k_cnt_reg ? readl_relaxed(sync32k_cnt_reg) : 0;
+
+	nsecs = clocksource_cyc2ns(cycles - last_cycles,
+					persistent_mult, persistent_shift);
+
+	timespec64_add_ns(&persistent_ts, nsecs);
+
+	*ts = persistent_ts;
+}
+
+/**
+ * omap_init_clocksource_32k - setup and register counter 32k as a
+ * kernel clocksource
+ * @pbase: base addr of counter_32k module
+ * @size: size of counter_32k to map
+ *
+ * Returns 0 upon success or negative error code upon failure.
+ *
+ */
+int __init omap_init_clocksource_32k(void __iomem *vbase)
+{
+	int ret;
+
+	/*
+	 * 32k sync Counter IP register offsets vary between the
+	 * highlander version and the legacy ones.
+	 * The 'SCHEME' bits(30-31) of the revision register is used
+	 * to identify the version.
+	 */
+	if (readl_relaxed(vbase + OMAP2_32KSYNCNT_REV_OFF) &
+						OMAP2_32KSYNCNT_REV_SCHEME)
+		sync32k_cnt_reg = vbase + OMAP2_32KSYNCNT_CR_OFF_HIGH;
+	else
+		sync32k_cnt_reg = vbase + OMAP2_32KSYNCNT_CR_OFF_LOW;
+
+	/*
+	 * 120000 rough estimate from the calculations in
+	 * __clocksource_update_freq_scale.
+	 */
+	clocks_calc_mult_shift(&persistent_mult, &persistent_shift,
+			32768, NSEC_PER_SEC, 120000);
+
+	ret = clocksource_mmio_init(sync32k_cnt_reg, "32k_counter", 32768,
+				250, 32, clocksource_mmio_readl_up);
+	if (ret) {
+		pr_err("32k_counter: can't register clocksource\n");
+		return ret;
+	}
+
+	sched_clock_register(omap_32k_read_sched_clock, 32, 32768);
+	register_persistent_clock(NULL, omap_read_persistent_clock64);
+	pr_info("OMAP clocksource: 32k_counter at 32768 Hz\n");
+
+	return 0;
+}

arch/arm/plat-omap/debug-leds.c

@@ -0,0 +1,174 @@
+/*
+ * linux/arch/arm/plat-omap/debug-leds.c
+ *
+ * Copyright 2011 by Bryan Wu <bryan.wu@canonical.com>
+ * Copyright 2003 by Texas Instruments Incorporated
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/platform_device.h>
+#include <linux/leds.h>
+#include <linux/io.h>
+#include <linux/platform_data/gpio-omap.h>
+#include <linux/slab.h>
+
+#include <asm/mach-types.h>
+
+/* Many OMAP development platforms reuse the same "debug board"; these
+ * platforms include H2, H3, H4, and Perseus2.  There are 16 LEDs on the
+ * debug board (all green), accessed through FPGA registers.
+ */
+
+/* NOTE:  most boards don't have a static mapping for the FPGA ... */
+struct h2p2_dbg_fpga {
+	/* offset 0x00 */
+	u16		smc91x[8];
+	/* offset 0x10 */
+	u16		fpga_rev;
+	u16		board_rev;
+	u16		gpio_outputs;
+	u16		leds;
+	/* offset 0x18 */
+	u16		misc_inputs;
+	u16		lan_status;
+	u16		lan_reset;
+	u16		reserved0;
+	/* offset 0x20 */
+	u16		ps2_data;
+	u16		ps2_ctrl;
+	/* plus also 4 rs232 ports ... */
+};
+
+static struct h2p2_dbg_fpga __iomem *fpga;
+
+static u16 fpga_led_state;
+
+struct dbg_led {
+	struct led_classdev	cdev;
+	u16			mask;
+};
+
+static const struct {
+	const char *name;
+	const char *trigger;
+} dbg_leds[] = {
+	{ "dbg:d4", "heartbeat", },
+	{ "dbg:d5", "cpu0", },
+	{ "dbg:d6", "default-on", },
+	{ "dbg:d7", },
+	{ "dbg:d8", },
+	{ "dbg:d9", },
+	{ "dbg:d10", },
+	{ "dbg:d11", },
+	{ "dbg:d12", },
+	{ "dbg:d13", },
+	{ "dbg:d14", },
+	{ "dbg:d15", },
+	{ "dbg:d16", },
+	{ "dbg:d17", },
+	{ "dbg:d18", },
+	{ "dbg:d19", },
+};
+
+/*
+ * The triggers lines up below will only be used if the
+ * LED triggers are compiled in.
+ */
+static void dbg_led_set(struct led_classdev *cdev,
+			      enum led_brightness b)
+{
+	struct dbg_led *led = container_of(cdev, struct dbg_led, cdev);
+	u16 reg;
+
+	reg = readw_relaxed(&fpga->leds);
+	if (b != LED_OFF)
+		reg |= led->mask;
+	else
+		reg &= ~led->mask;
+	writew_relaxed(reg, &fpga->leds);
+}
+
+static enum led_brightness dbg_led_get(struct led_classdev *cdev)
+{
+	struct dbg_led *led = container_of(cdev, struct dbg_led, cdev);
+	u16 reg;
+
+	reg = readw_relaxed(&fpga->leds);
+	return (reg & led->mask) ? LED_FULL : LED_OFF;
+}
+
+static int fpga_probe(struct platform_device *pdev)
+{
+	struct resource	*iomem;
+	int i;
+
+	iomem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!iomem)
+		return -ENODEV;
+
+	fpga = ioremap(iomem->start, resource_size(iomem));
+	writew_relaxed(0xff, &fpga->leds);
+
+	for (i = 0; i < ARRAY_SIZE(dbg_leds); i++) {
+		struct dbg_led *led;
+
+		led = kzalloc(sizeof(*led), GFP_KERNEL);
+		if (!led)
+			break;
+
+		led->cdev.name = dbg_leds[i].name;
+		led->cdev.brightness_set = dbg_led_set;
+		led->cdev.brightness_get = dbg_led_get;
+		led->cdev.default_trigger = dbg_leds[i].trigger;
+		led->mask = BIT(i);
+
+		if (led_classdev_register(NULL, &led->cdev) < 0) {
+			kfree(led);
+			break;
+		}
+	}
+
+	return 0;
+}
+
+static int fpga_suspend_noirq(struct device *dev)
+{
+	fpga_led_state = readw_relaxed(&fpga->leds);
+	writew_relaxed(0xff, &fpga->leds);
+
+	return 0;
+}
+
+static int fpga_resume_noirq(struct device *dev)
+{
+	writew_relaxed(~fpga_led_state, &fpga->leds);
+	return 0;
+}
+
+static const struct dev_pm_ops fpga_dev_pm_ops = {
+	.suspend_noirq = fpga_suspend_noirq,
+	.resume_noirq = fpga_resume_noirq,
+};
+
+static struct platform_driver led_driver = {
+	.driver.name	= "omap_dbg_led",
+	.driver.pm	= &fpga_dev_pm_ops,
+	.probe		= fpga_probe,
+};
+
+static int __init fpga_init(void)
+{
+	if (machine_is_omap_h4()
+			|| machine_is_omap_h3()
+			|| machine_is_omap_h2()
+			|| machine_is_omap_perseus2()
+			)
+		return platform_driver_register(&led_driver);
+	return 0;
+}
+fs_initcall(fpga_init);

arch/arm/plat-omap/dma.c

@@ -0,0 +1,1470 @@
+/*
+ * linux/arch/arm/plat-omap/dma.c
+ *
+ * Copyright (C) 2003 - 2008 Nokia Corporation
+ * Author: Juha Yrjölä <juha.yrjola@nokia.com>
+ * DMA channel linking for 1610 by Samuel Ortiz <samuel.ortiz@nokia.com>
+ * Graphics DMA and LCD DMA graphics tranformations
+ * by Imre Deak <imre.deak@nokia.com>
+ * OMAP2/3 support Copyright (C) 2004-2007 Texas Instruments, Inc.
+ * Merged to support both OMAP1 and OMAP2 by Tony Lindgren <tony@atomide.com>
+ * Some functions based on earlier dma-omap.c Copyright (C) 2001 RidgeRun, Inc.
+ *
+ * Copyright (C) 2009 Texas Instruments
+ * Added OMAP4 support - Santosh Shilimkar <santosh.shilimkar@ti.com>
+ *
+ * Support functions for the OMAP internal DMA channels.
+ *
+ * Copyright (C) 2010 Texas Instruments Incorporated - http://www.ti.com/
+ * Converted DMA library into DMA platform driver.
+ *	- G, Manjunath Kondaiah <manjugk@ti.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/sched.h>
+#include <linux/spinlock.h>
+#include <linux/errno.h>
+#include <linux/interrupt.h>
+#include <linux/irq.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+
+#include <linux/omap-dma.h>
+
+#ifdef CONFIG_ARCH_OMAP1
+#include <mach/soc.h>
+#endif
+
+/*
+ * MAX_LOGICAL_DMA_CH_COUNT: the maximum number of logical DMA
+ * channels that an instance of the SDMA IP block can support.  Used
+ * to size arrays.  (The actual maximum on a particular SoC may be less
+ * than this -- for example, OMAP1 SDMA instances only support 17 logical
+ * DMA channels.)
+ */
+#define MAX_LOGICAL_DMA_CH_COUNT		32
+
+#undef DEBUG
+
+#ifndef CONFIG_ARCH_OMAP1
+enum { DMA_CH_ALLOC_DONE, DMA_CH_PARAMS_SET_DONE, DMA_CH_STARTED,
+	DMA_CH_QUEUED, DMA_CH_NOTSTARTED, DMA_CH_PAUSED, DMA_CH_LINK_ENABLED
+};
+
+enum { DMA_CHAIN_STARTED, DMA_CHAIN_NOTSTARTED };
+#endif
+
+#define OMAP_DMA_ACTIVE			0x01
+#define OMAP2_DMA_CSR_CLEAR_MASK	0xffffffff
+
+#define OMAP_FUNC_MUX_ARM_BASE		(0xfffe1000 + 0xec)
+
+static struct omap_system_dma_plat_info *p;
+static struct omap_dma_dev_attr *d;
+static void omap_clear_dma(int lch);
+static int omap_dma_set_prio_lch(int lch, unsigned char read_prio,
+				 unsigned char write_prio);
+static int enable_1510_mode;
+static u32 errata;
+
+static struct omap_dma_global_context_registers {
+	u32 dma_irqenable_l0;
+	u32 dma_irqenable_l1;
+	u32 dma_ocp_sysconfig;
+	u32 dma_gcr;
+} omap_dma_global_context;
+
+struct dma_link_info {
+	int *linked_dmach_q;
+	int no_of_lchs_linked;
+
+	int q_count;
+	int q_tail;
+	int q_head;
+
+	int chain_state;
+	int chain_mode;
+
+};
+
+static struct dma_link_info *dma_linked_lch;
+
+#ifndef CONFIG_ARCH_OMAP1
+
+/* Chain handling macros */
+#define OMAP_DMA_CHAIN_QINIT(chain_id)					\
+	do {								\
+		dma_linked_lch[chain_id].q_head =			\
+		dma_linked_lch[chain_id].q_tail =			\
+		dma_linked_lch[chain_id].q_count = 0;			\
+	} while (0)
+#define OMAP_DMA_CHAIN_QFULL(chain_id)					\
+		(dma_linked_lch[chain_id].no_of_lchs_linked ==		\
+		dma_linked_lch[chain_id].q_count)
+#define OMAP_DMA_CHAIN_QLAST(chain_id)					\
+	do {								\
+		((dma_linked_lch[chain_id].no_of_lchs_linked-1) ==	\
+		dma_linked_lch[chain_id].q_count)			\
+	} while (0)
+#define OMAP_DMA_CHAIN_QEMPTY(chain_id)					\
+		(0 == dma_linked_lch[chain_id].q_count)
+#define __OMAP_DMA_CHAIN_INCQ(end)					\
+	((end) = ((end)+1) % dma_linked_lch[chain_id].no_of_lchs_linked)
+#define OMAP_DMA_CHAIN_INCQHEAD(chain_id)				\
+	do {								\
+		__OMAP_DMA_CHAIN_INCQ(dma_linked_lch[chain_id].q_head);	\
+		dma_linked_lch[chain_id].q_count--;			\
+	} while (0)
+
+#define OMAP_DMA_CHAIN_INCQTAIL(chain_id)				\
+	do {								\
+		__OMAP_DMA_CHAIN_INCQ(dma_linked_lch[chain_id].q_tail);	\
+		dma_linked_lch[chain_id].q_count++; \
+	} while (0)
+#endif
+
+static int dma_lch_count;
+static int dma_chan_count;
+static int omap_dma_reserve_channels;
+
+static spinlock_t dma_chan_lock;
+static struct omap_dma_lch *dma_chan;
+
+static inline void disable_lnk(int lch);
+static void omap_disable_channel_irq(int lch);
+static inline void omap_enable_channel_irq(int lch);
+
+#define REVISIT_24XX()		printk(KERN_ERR "FIXME: no %s on 24xx\n", \
+						__func__);
+
+#ifdef CONFIG_ARCH_OMAP15XX
+/* Returns 1 if the DMA module is in OMAP1510-compatible mode, 0 otherwise */
+static int omap_dma_in_1510_mode(void)
+{
+	return enable_1510_mode;
+}
+#else
+#define omap_dma_in_1510_mode()		0
+#endif
+
+#ifdef CONFIG_ARCH_OMAP1
+static inline void set_gdma_dev(int req, int dev)
+{
+	u32 reg = OMAP_FUNC_MUX_ARM_BASE + ((req - 1) / 5) * 4;
+	int shift = ((req - 1) % 5) * 6;
+	u32 l;
+
+	l = omap_readl(reg);
+	l &= ~(0x3f << shift);
+	l |= (dev - 1) << shift;
+	omap_writel(l, reg);
+}
+#else
+#define set_gdma_dev(req, dev)	do {} while (0)
+#define omap_readl(reg)		0
+#define omap_writel(val, reg)	do {} while (0)
+#endif
+
+#ifdef CONFIG_ARCH_OMAP1
+void omap_set_dma_priority(int lch, int dst_port, int priority)
+{
+	unsigned long reg;
+	u32 l;
+
+	if (dma_omap1()) {
+		switch (dst_port) {
+		case OMAP_DMA_PORT_OCP_T1:	/* FFFECC00 */
+			reg = OMAP_TC_OCPT1_PRIOR;
+			break;
+		case OMAP_DMA_PORT_OCP_T2:	/* FFFECCD0 */
+			reg = OMAP_TC_OCPT2_PRIOR;
+			break;
+		case OMAP_DMA_PORT_EMIFF:	/* FFFECC08 */
+			reg = OMAP_TC_EMIFF_PRIOR;
+			break;
+		case OMAP_DMA_PORT_EMIFS:	/* FFFECC04 */
+			reg = OMAP_TC_EMIFS_PRIOR;
+			break;
+		default:
+			BUG();
+			return;
+		}
+		l = omap_readl(reg);
+		l &= ~(0xf << 8);
+		l |= (priority & 0xf) << 8;
+		omap_writel(l, reg);
+	}
+}
+#endif
+
+#ifdef CONFIG_ARCH_OMAP2PLUS
+void omap_set_dma_priority(int lch, int dst_port, int priority)
+{
+	u32 ccr;
+
+	ccr = p->dma_read(CCR, lch);
+	if (priority)
+		ccr |= (1 << 6);
+	else
+		ccr &= ~(1 << 6);
+	p->dma_write(ccr, CCR, lch);
+}
+#endif
+EXPORT_SYMBOL(omap_set_dma_priority);
+
+void omap_set_dma_transfer_params(int lch, int data_type, int elem_count,
+				  int frame_count, int sync_mode,
+				  int dma_trigger, int src_or_dst_synch)
+{
+	u32 l;
+
+	l = p->dma_read(CSDP, lch);
+	l &= ~0x03;
+	l |= data_type;
+	p->dma_write(l, CSDP, lch);
+
+	if (dma_omap1()) {
+		u16 ccr;
+
+		ccr = p->dma_read(CCR, lch);
+		ccr &= ~(1 << 5);
+		if (sync_mode == OMAP_DMA_SYNC_FRAME)
+			ccr |= 1 << 5;
+		p->dma_write(ccr, CCR, lch);
+
+		ccr = p->dma_read(CCR2, lch);
+		ccr &= ~(1 << 2);
+		if (sync_mode == OMAP_DMA_SYNC_BLOCK)
+			ccr |= 1 << 2;
+		p->dma_write(ccr, CCR2, lch);
+	}
+
+	if (dma_omap2plus() && dma_trigger) {
+		u32 val;
+
+		val = p->dma_read(CCR, lch);
+
+		/* DMA_SYNCHRO_CONTROL_UPPER depends on the channel number */
+		val &= ~((1 << 23) | (3 << 19) | 0x1f);
+		val |= (dma_trigger & ~0x1f) << 14;
+		val |= dma_trigger & 0x1f;
+
+		if (sync_mode & OMAP_DMA_SYNC_FRAME)
+			val |= 1 << 5;
+		else
+			val &= ~(1 << 5);
+
+		if (sync_mode & OMAP_DMA_SYNC_BLOCK)
+			val |= 1 << 18;
+		else
+			val &= ~(1 << 18);
+
+		if (src_or_dst_synch == OMAP_DMA_DST_SYNC_PREFETCH) {
+			val &= ~(1 << 24);	/* dest synch */
+			val |= (1 << 23);	/* Prefetch */
+		} else if (src_or_dst_synch) {
+			val |= 1 << 24;		/* source synch */
+		} else {
+			val &= ~(1 << 24);	/* dest synch */
+		}
+		p->dma_write(val, CCR, lch);
+	}
+
+	p->dma_write(elem_count, CEN, lch);
+	p->dma_write(frame_count, CFN, lch);
+}
+EXPORT_SYMBOL(omap_set_dma_transfer_params);
+
+void omap_set_dma_write_mode(int lch, enum omap_dma_write_mode mode)
+{
+	if (dma_omap2plus()) {
+		u32 csdp;
+
+		csdp = p->dma_read(CSDP, lch);
+		csdp &= ~(0x3 << 16);
+		csdp |= (mode << 16);
+		p->dma_write(csdp, CSDP, lch);
+	}
+}
+EXPORT_SYMBOL(omap_set_dma_write_mode);
+
+void omap_set_dma_channel_mode(int lch, enum omap_dma_channel_mode mode)
+{
+	if (dma_omap1() && !dma_omap15xx()) {
+		u32 l;
+
+		l = p->dma_read(LCH_CTRL, lch);
+		l &= ~0x7;
+		l |= mode;
+		p->dma_write(l, LCH_CTRL, lch);
+	}
+}
+EXPORT_SYMBOL(omap_set_dma_channel_mode);
+
+/* Note that src_port is only for omap1 */
+void omap_set_dma_src_params(int lch, int src_port, int src_amode,
+			     unsigned long src_start,
+			     int src_ei, int src_fi)
+{
+	u32 l;
+
+	if (dma_omap1()) {
+		u16 w;
+
+		w = p->dma_read(CSDP, lch);
+		w &= ~(0x1f << 2);
+		w |= src_port << 2;
+		p->dma_write(w, CSDP, lch);
+	}
+
+	l = p->dma_read(CCR, lch);
+	l &= ~(0x03 << 12);
+	l |= src_amode << 12;
+	p->dma_write(l, CCR, lch);
+
+	p->dma_write(src_start, CSSA, lch);
+
+	p->dma_write(src_ei, CSEI, lch);
+	p->dma_write(src_fi, CSFI, lch);
+}
+EXPORT_SYMBOL(omap_set_dma_src_params);
+
+void omap_set_dma_params(int lch, struct omap_dma_channel_params *params)
+{
+	omap_set_dma_transfer_params(lch, params->data_type,
+				     params->elem_count, params->frame_count,
+				     params->sync_mode, params->trigger,
+				     params->src_or_dst_synch);
+	omap_set_dma_src_params(lch, params->src_port,
+				params->src_amode, params->src_start,
+				params->src_ei, params->src_fi);
+
+	omap_set_dma_dest_params(lch, params->dst_port,
+				 params->dst_amode, params->dst_start,
+				 params->dst_ei, params->dst_fi);
+	if (params->read_prio || params->write_prio)
+		omap_dma_set_prio_lch(lch, params->read_prio,
+				      params->write_prio);
+}
+EXPORT_SYMBOL(omap_set_dma_params);
+
+void omap_set_dma_src_data_pack(int lch, int enable)
+{
+	u32 l;
+
+	l = p->dma_read(CSDP, lch);
+	l &= ~(1 << 6);
+	if (enable)
+		l |= (1 << 6);
+	p->dma_write(l, CSDP, lch);
+}
+EXPORT_SYMBOL(omap_set_dma_src_data_pack);
+
+void omap_set_dma_src_burst_mode(int lch, enum omap_dma_burst_mode burst_mode)
+{
+	unsigned int burst = 0;
+	u32 l;
+
+	l = p->dma_read(CSDP, lch);
+	l &= ~(0x03 << 7);
+
+	switch (burst_mode) {
+	case OMAP_DMA_DATA_BURST_DIS:
+		break;
+	case OMAP_DMA_DATA_BURST_4:
+		if (dma_omap2plus())
+			burst = 0x1;
+		else
+			burst = 0x2;
+		break;
+	case OMAP_DMA_DATA_BURST_8:
+		if (dma_omap2plus()) {
+			burst = 0x2;
+			break;
+		}
+		/*
+		 * not supported by current hardware on OMAP1
+		 * w |= (0x03 << 7);
+		 * fall through
+		 */
+	case OMAP_DMA_DATA_BURST_16:
+		if (dma_omap2plus()) {
+			burst = 0x3;
+			break;
+		}
+		/*
+		 * OMAP1 don't support burst 16
+		 * fall through
+		 */
+	default:
+		BUG();
+	}
+
+	l |= (burst << 7);
+	p->dma_write(l, CSDP, lch);
+}
+EXPORT_SYMBOL(omap_set_dma_src_burst_mode);
+
+/* Note that dest_port is only for OMAP1 */
+void omap_set_dma_dest_params(int lch, int dest_port, int dest_amode,
+			      unsigned long dest_start,
+			      int dst_ei, int dst_fi)
+{
+	u32 l;
+
+	if (dma_omap1()) {
+		l = p->dma_read(CSDP, lch);
+		l &= ~(0x1f << 9);
+		l |= dest_port << 9;
+		p->dma_write(l, CSDP, lch);
+	}
+
+	l = p->dma_read(CCR, lch);
+	l &= ~(0x03 << 14);
+	l |= dest_amode << 14;
+	p->dma_write(l, CCR, lch);
+
+	p->dma_write(dest_start, CDSA, lch);
+
+	p->dma_write(dst_ei, CDEI, lch);
+	p->dma_write(dst_fi, CDFI, lch);
+}
+EXPORT_SYMBOL(omap_set_dma_dest_params);
+
+void omap_set_dma_dest_data_pack(int lch, int enable)
+{
+	u32 l;
+
+	l = p->dma_read(CSDP, lch);
+	l &= ~(1 << 13);
+	if (enable)
+		l |= 1 << 13;
+	p->dma_write(l, CSDP, lch);
+}
+EXPORT_SYMBOL(omap_set_dma_dest_data_pack);
+
+void omap_set_dma_dest_burst_mode(int lch, enum omap_dma_burst_mode burst_mode)
+{
+	unsigned int burst = 0;
+	u32 l;
+
+	l = p->dma_read(CSDP, lch);
+	l &= ~(0x03 << 14);
+
+	switch (burst_mode) {
+	case OMAP_DMA_DATA_BURST_DIS:
+		break;
+	case OMAP_DMA_DATA_BURST_4:
+		if (dma_omap2plus())
+			burst = 0x1;
+		else
+			burst = 0x2;
+		break;
+	case OMAP_DMA_DATA_BURST_8:
+		if (dma_omap2plus())
+			burst = 0x2;
+		else
+			burst = 0x3;
+		break;
+	case OMAP_DMA_DATA_BURST_16:
+		if (dma_omap2plus()) {
+			burst = 0x3;
+			break;
+		}
+		/*
+		 * OMAP1 don't support burst 16
+		 * fall through
+		 */
+	default:
+		printk(KERN_ERR "Invalid DMA burst mode\n");
+		BUG();
+		return;
+	}
+	l |= (burst << 14);
+	p->dma_write(l, CSDP, lch);
+}
+EXPORT_SYMBOL(omap_set_dma_dest_burst_mode);
+
+static inline void omap_enable_channel_irq(int lch)
+{
+	/* Clear CSR */
+	if (dma_omap1())
+		p->dma_read(CSR, lch);
+	else
+		p->dma_write(OMAP2_DMA_CSR_CLEAR_MASK, CSR, lch);
+
+	/* Enable some nice interrupts. */
+	p->dma_write(dma_chan[lch].enabled_irqs, CICR, lch);
+}
+
+static inline void omap_disable_channel_irq(int lch)
+{
+	/* disable channel interrupts */
+	p->dma_write(0, CICR, lch);
+	/* Clear CSR */
+	if (dma_omap1())
+		p->dma_read(CSR, lch);
+	else
+		p->dma_write(OMAP2_DMA_CSR_CLEAR_MASK, CSR, lch);
+}
+
+void omap_enable_dma_irq(int lch, u16 bits)
+{
+	dma_chan[lch].enabled_irqs |= bits;
+}
+EXPORT_SYMBOL(omap_enable_dma_irq);
+
+void omap_disable_dma_irq(int lch, u16 bits)
+{
+	dma_chan[lch].enabled_irqs &= ~bits;
+}
+EXPORT_SYMBOL(omap_disable_dma_irq);
+
+static inline void enable_lnk(int lch)
+{
+	u32 l;
+
+	l = p->dma_read(CLNK_CTRL, lch);
+
+	if (dma_omap1())
+		l &= ~(1 << 14);
+
+	/* Set the ENABLE_LNK bits */
+	if (dma_chan[lch].next_lch != -1)
+		l = dma_chan[lch].next_lch | (1 << 15);
+
+#ifndef CONFIG_ARCH_OMAP1
+	if (dma_omap2plus())
+		if (dma_chan[lch].next_linked_ch != -1)
+			l = dma_chan[lch].next_linked_ch | (1 << 15);
+#endif
+
+	p->dma_write(l, CLNK_CTRL, lch);
+}
+
+static inline void disable_lnk(int lch)
+{
+	u32 l;
+
+	l = p->dma_read(CLNK_CTRL, lch);
+
+	/* Disable interrupts */
+	omap_disable_channel_irq(lch);
+
+	if (dma_omap1()) {
+		/* Set the STOP_LNK bit */
+		l |= 1 << 14;
+	}
+
+	if (dma_omap2plus()) {
+		/* Clear the ENABLE_LNK bit */
+		l &= ~(1 << 15);
+	}
+
+	p->dma_write(l, CLNK_CTRL, lch);
+	dma_chan[lch].flags &= ~OMAP_DMA_ACTIVE;
+}
+
+static inline void omap2_enable_irq_lch(int lch)
+{
+	u32 val;
+	unsigned long flags;
+
+	if (dma_omap1())
+		return;
+
+	spin_lock_irqsave(&dma_chan_lock, flags);
+	/* clear IRQ STATUS */
+	p->dma_write(1 << lch, IRQSTATUS_L0, lch);
+	/* Enable interrupt */
+	val = p->dma_read(IRQENABLE_L0, lch);
+	val |= 1 << lch;
+	p->dma_write(val, IRQENABLE_L0, lch);
+	spin_unlock_irqrestore(&dma_chan_lock, flags);
+}
+
+static inline void omap2_disable_irq_lch(int lch)
+{
+	u32 val;
+	unsigned long flags;
+
+	if (dma_omap1())
+		return;
+
+	spin_lock_irqsave(&dma_chan_lock, flags);
+	/* Disable interrupt */
+	val = p->dma_read(IRQENABLE_L0, lch);
+	val &= ~(1 << lch);
+	p->dma_write(val, IRQENABLE_L0, lch);
+	/* clear IRQ STATUS */
+	p->dma_write(1 << lch, IRQSTATUS_L0, lch);
+	spin_unlock_irqrestore(&dma_chan_lock, flags);
+}
+
+int omap_request_dma(int dev_id, const char *dev_name,
+		     void (*callback)(int lch, u16 ch_status, void *data),
+		     void *data, int *dma_ch_out)
+{
+	int ch, free_ch = -1;
+	unsigned long flags;
+	struct omap_dma_lch *chan;
+
+	WARN(strcmp(dev_name, "DMA engine"), "Using deprecated platform DMA API - please update to DMA engine");
+
+	spin_lock_irqsave(&dma_chan_lock, flags);
+	for (ch = 0; ch < dma_chan_count; ch++) {
+		if (free_ch == -1 && dma_chan[ch].dev_id == -1) {
+			free_ch = ch;
+			/* Exit after first free channel found */
+			break;
+		}
+	}
+	if (free_ch == -1) {
+		spin_unlock_irqrestore(&dma_chan_lock, flags);
+		return -EBUSY;
+	}
+	chan = dma_chan + free_ch;
+	chan->dev_id = dev_id;
+
+	if (p->clear_lch_regs)
+		p->clear_lch_regs(free_ch);
+
+	if (dma_omap2plus())
+		omap_clear_dma(free_ch);
+
+	spin_unlock_irqrestore(&dma_chan_lock, flags);
+
+	chan->dev_name = dev_name;
+	chan->callback = callback;
+	chan->data = data;
+	chan->flags = 0;
+
+#ifndef CONFIG_ARCH_OMAP1
+	if (dma_omap2plus()) {
+		chan->chain_id = -1;
+		chan->next_linked_ch = -1;
+	}
+#endif
+
+	chan->enabled_irqs = OMAP_DMA_DROP_IRQ | OMAP_DMA_BLOCK_IRQ;
+
+	if (dma_omap1())
+		chan->enabled_irqs |= OMAP1_DMA_TOUT_IRQ;
+	else if (dma_omap2plus())
+		chan->enabled_irqs |= OMAP2_DMA_MISALIGNED_ERR_IRQ |
+			OMAP2_DMA_TRANS_ERR_IRQ;
+
+	if (dma_omap16xx()) {
+		/* If the sync device is set, configure it dynamically. */
+		if (dev_id != 0) {
+			set_gdma_dev(free_ch + 1, dev_id);
+			dev_id = free_ch + 1;
+		}
+		/*
+		 * Disable the 1510 compatibility mode and set the sync device
+		 * id.
+		 */
+		p->dma_write(dev_id | (1 << 10), CCR, free_ch);
+	} else if (dma_omap1()) {
+		p->dma_write(dev_id, CCR, free_ch);
+	}
+
+	if (dma_omap2plus()) {
+		omap_enable_channel_irq(free_ch);
+		omap2_enable_irq_lch(free_ch);
+	}
+
+	*dma_ch_out = free_ch;
+
+	return 0;
+}
+EXPORT_SYMBOL(omap_request_dma);
+
+void omap_free_dma(int lch)
+{
+	unsigned long flags;
+
+	if (dma_chan[lch].dev_id == -1) {
+		pr_err("omap_dma: trying to free unallocated DMA channel %d\n",
+		       lch);
+		return;
+	}
+
+	/* Disable interrupt for logical channel */
+	if (dma_omap2plus())
+		omap2_disable_irq_lch(lch);
+
+	/* Disable all DMA interrupts for the channel. */
+	omap_disable_channel_irq(lch);
+
+	/* Make sure the DMA transfer is stopped. */
+	p->dma_write(0, CCR, lch);
+
+	/* Clear registers */
+	if (dma_omap2plus())
+		omap_clear_dma(lch);
+
+	spin_lock_irqsave(&dma_chan_lock, flags);
+	dma_chan[lch].dev_id = -1;
+	dma_chan[lch].next_lch = -1;
+	dma_chan[lch].callback = NULL;
+	spin_unlock_irqrestore(&dma_chan_lock, flags);
+}
+EXPORT_SYMBOL(omap_free_dma);
+
+/**
+ * @brief omap_dma_set_global_params : Set global priority settings for dma
+ *
+ * @param arb_rate
+ * @param max_fifo_depth
+ * @param tparams - Number of threads to reserve : DMA_THREAD_RESERVE_NORM
+ * 						   DMA_THREAD_RESERVE_ONET
+ * 						   DMA_THREAD_RESERVE_TWOT
+ * 						   DMA_THREAD_RESERVE_THREET
+ */
+void
+omap_dma_set_global_params(int arb_rate, int max_fifo_depth, int tparams)
+{
+	u32 reg;
+
+	if (dma_omap1()) {
+		printk(KERN_ERR "FIXME: no %s on 15xx/16xx\n", __func__);
+		return;
+	}
+
+	if (max_fifo_depth == 0)
+		max_fifo_depth = 1;
+	if (arb_rate == 0)
+		arb_rate = 1;
+
+	reg = 0xff & max_fifo_depth;
+	reg |= (0x3 & tparams) << 12;
+	reg |= (arb_rate & 0xff) << 16;
+
+	p->dma_write(reg, GCR, 0);
+}
+EXPORT_SYMBOL(omap_dma_set_global_params);
+
+/**
+ * @brief omap_dma_set_prio_lch : Set channel wise priority settings
+ *
+ * @param lch
+ * @param read_prio - Read priority
+ * @param write_prio - Write priority
+ * Both of the above can be set with one of the following values :
+ * 	DMA_CH_PRIO_HIGH/DMA_CH_PRIO_LOW
+ */
+static int
+omap_dma_set_prio_lch(int lch, unsigned char read_prio,
+		      unsigned char write_prio)
+{
+	u32 l;
+
+	if (unlikely((lch < 0 || lch >= dma_lch_count))) {
+		printk(KERN_ERR "Invalid channel id\n");
+		return -EINVAL;
+	}
+	l = p->dma_read(CCR, lch);
+	l &= ~((1 << 6) | (1 << 26));
+	if (d->dev_caps & IS_RW_PRIORITY)
+		l |= ((read_prio & 0x1) << 6) | ((write_prio & 0x1) << 26);
+	else
+		l |= ((read_prio & 0x1) << 6);
+
+	p->dma_write(l, CCR, lch);
+
+	return 0;
+}
+
+
+/*
+ * Clears any DMA state so the DMA engine is ready to restart with new buffers
+ * through omap_start_dma(). Any buffers in flight are discarded.
+ */
+static void omap_clear_dma(int lch)
+{
+	unsigned long flags;
+
+	local_irq_save(flags);
+	p->clear_dma(lch);
+	local_irq_restore(flags);
+}
+
+void omap_start_dma(int lch)
+{
+	u32 l;
+
+	/*
+	 * The CPC/CDAC register needs to be initialized to zero
+	 * before starting dma transfer.
+	 */
+	if (dma_omap15xx())
+		p->dma_write(0, CPC, lch);
+	else
+		p->dma_write(0, CDAC, lch);
+
+	if (!omap_dma_in_1510_mode() && dma_chan[lch].next_lch != -1) {
+		int next_lch, cur_lch;
+		char dma_chan_link_map[MAX_LOGICAL_DMA_CH_COUNT];
+
+		/* Set the link register of the first channel */
+		enable_lnk(lch);
+
+		memset(dma_chan_link_map, 0, sizeof(dma_chan_link_map));
+		dma_chan_link_map[lch] = 1;
+
+		cur_lch = dma_chan[lch].next_lch;
+		do {
+			next_lch = dma_chan[cur_lch].next_lch;
+
+			/* The loop case: we've been here already */
+			if (dma_chan_link_map[cur_lch])
+				break;
+			/* Mark the current channel */
+			dma_chan_link_map[cur_lch] = 1;
+
+			enable_lnk(cur_lch);
+			omap_enable_channel_irq(cur_lch);
+
+			cur_lch = next_lch;
+		} while (next_lch != -1);
+	} else if (IS_DMA_ERRATA(DMA_ERRATA_PARALLEL_CHANNELS))
+		p->dma_write(lch, CLNK_CTRL, lch);
+
+	omap_enable_channel_irq(lch);
+
+	l = p->dma_read(CCR, lch);
+
+	if (IS_DMA_ERRATA(DMA_ERRATA_IFRAME_BUFFERING))
+			l |= OMAP_DMA_CCR_BUFFERING_DISABLE;
+	l |= OMAP_DMA_CCR_EN;
+
+	/*
+	 * As dma_write() uses IO accessors which are weakly ordered, there
+	 * is no guarantee that data in coherent DMA memory will be visible
+	 * to the DMA device.  Add a memory barrier here to ensure that any
+	 * such data is visible prior to enabling DMA.
+	 */
+	mb();
+	p->dma_write(l, CCR, lch);
+
+	dma_chan[lch].flags |= OMAP_DMA_ACTIVE;
+}
+EXPORT_SYMBOL(omap_start_dma);
+
+void omap_stop_dma(int lch)
+{
+	u32 l;
+
+	/* Disable all interrupts on the channel */
+	omap_disable_channel_irq(lch);
+
+	l = p->dma_read(CCR, lch);
+	if (IS_DMA_ERRATA(DMA_ERRATA_i541) &&
+			(l & OMAP_DMA_CCR_SEL_SRC_DST_SYNC)) {
+		int i = 0;
+		u32 sys_cf;
+
+		/* Configure No-Standby */
+		l = p->dma_read(OCP_SYSCONFIG, lch);
+		sys_cf = l;
+		l &= ~DMA_SYSCONFIG_MIDLEMODE_MASK;
+		l |= DMA_SYSCONFIG_MIDLEMODE(DMA_IDLEMODE_NO_IDLE);
+		p->dma_write(l , OCP_SYSCONFIG, 0);
+
+		l = p->dma_read(CCR, lch);
+		l &= ~OMAP_DMA_CCR_EN;
+		p->dma_write(l, CCR, lch);
+
+		/* Wait for sDMA FIFO drain */
+		l = p->dma_read(CCR, lch);
+		while (i < 100 && (l & (OMAP_DMA_CCR_RD_ACTIVE |
+					OMAP_DMA_CCR_WR_ACTIVE))) {
+			udelay(5);
+			i++;
+			l = p->dma_read(CCR, lch);
+		}
+		if (i >= 100)
+			pr_err("DMA drain did not complete on lch %d\n", lch);
+		/* Restore OCP_SYSCONFIG */
+		p->dma_write(sys_cf, OCP_SYSCONFIG, lch);
+	} else {
+		l &= ~OMAP_DMA_CCR_EN;
+		p->dma_write(l, CCR, lch);
+	}
+
+	/*
+	 * Ensure that data transferred by DMA is visible to any access
+	 * after DMA has been disabled.  This is important for coherent
+	 * DMA regions.
+	 */
+	mb();
+
+	if (!omap_dma_in_1510_mode() && dma_chan[lch].next_lch != -1) {
+		int next_lch, cur_lch = lch;
+		char dma_chan_link_map[MAX_LOGICAL_DMA_CH_COUNT];
+
+		memset(dma_chan_link_map, 0, sizeof(dma_chan_link_map));
+		do {
+			/* The loop case: we've been here already */
+			if (dma_chan_link_map[cur_lch])
+				break;
+			/* Mark the current channel */
+			dma_chan_link_map[cur_lch] = 1;
+
+			disable_lnk(cur_lch);
+
+			next_lch = dma_chan[cur_lch].next_lch;
+			cur_lch = next_lch;
+		} while (next_lch != -1);
+	}
+
+	dma_chan[lch].flags &= ~OMAP_DMA_ACTIVE;
+}
+EXPORT_SYMBOL(omap_stop_dma);
+
+/*
+ * Allows changing the DMA callback function or data. This may be needed if
+ * the driver shares a single DMA channel for multiple dma triggers.
+ */
+int omap_set_dma_callback(int lch,
+			  void (*callback)(int lch, u16 ch_status, void *data),
+			  void *data)
+{
+	unsigned long flags;
+
+	if (lch < 0)
+		return -ENODEV;
+
+	spin_lock_irqsave(&dma_chan_lock, flags);
+	if (dma_chan[lch].dev_id == -1) {
+		printk(KERN_ERR "DMA callback for not set for free channel\n");
+		spin_unlock_irqrestore(&dma_chan_lock, flags);
+		return -EINVAL;
+	}
+	dma_chan[lch].callback = callback;
+	dma_chan[lch].data = data;
+	spin_unlock_irqrestore(&dma_chan_lock, flags);
+
+	return 0;
+}
+EXPORT_SYMBOL(omap_set_dma_callback);
+
+/*
+ * Returns current physical source address for the given DMA channel.
+ * If the channel is running the caller must disable interrupts prior calling
+ * this function and process the returned value before re-enabling interrupt to
+ * prevent races with the interrupt handler. Note that in continuous mode there
+ * is a chance for CSSA_L register overflow between the two reads resulting
+ * in incorrect return value.
+ */
+dma_addr_t omap_get_dma_src_pos(int lch)
+{
+	dma_addr_t offset = 0;
+
+	if (dma_omap15xx())
+		offset = p->dma_read(CPC, lch);
+	else
+		offset = p->dma_read(CSAC, lch);
+
+	if (IS_DMA_ERRATA(DMA_ERRATA_3_3) && offset == 0)
+		offset = p->dma_read(CSAC, lch);
+
+	if (!dma_omap15xx()) {
+		/*
+		 * CDAC == 0 indicates that the DMA transfer on the channel has
+		 * not been started (no data has been transferred so far).
+		 * Return the programmed source start address in this case.
+		 */
+		if (likely(p->dma_read(CDAC, lch)))
+			offset = p->dma_read(CSAC, lch);
+		else
+			offset = p->dma_read(CSSA, lch);
+	}
+
+	if (dma_omap1())
+		offset |= (p->dma_read(CSSA, lch) & 0xFFFF0000);
+
+	return offset;
+}
+EXPORT_SYMBOL(omap_get_dma_src_pos);
+
+/*
+ * Returns current physical destination address for the given DMA channel.
+ * If the channel is running the caller must disable interrupts prior calling
+ * this function and process the returned value before re-enabling interrupt to
+ * prevent races with the interrupt handler. Note that in continuous mode there
+ * is a chance for CDSA_L register overflow between the two reads resulting
+ * in incorrect return value.
+ */
+dma_addr_t omap_get_dma_dst_pos(int lch)
+{
+	dma_addr_t offset = 0;
+
+	if (dma_omap15xx())
+		offset = p->dma_read(CPC, lch);
+	else
+		offset = p->dma_read(CDAC, lch);
+
+	/*
+	 * omap 3.2/3.3 erratum: sometimes 0 is returned if CSAC/CDAC is
+	 * read before the DMA controller finished disabling the channel.
+	 */
+	if (!dma_omap15xx() && offset == 0) {
+		offset = p->dma_read(CDAC, lch);
+		/*
+		 * CDAC == 0 indicates that the DMA transfer on the channel has
+		 * not been started (no data has been transferred so far).
+		 * Return the programmed destination start address in this case.
+		 */
+		if (unlikely(!offset))
+			offset = p->dma_read(CDSA, lch);
+	}
+
+	if (dma_omap1())
+		offset |= (p->dma_read(CDSA, lch) & 0xFFFF0000);
+
+	return offset;
+}
+EXPORT_SYMBOL(omap_get_dma_dst_pos);
+
+int omap_get_dma_active_status(int lch)
+{
+	return (p->dma_read(CCR, lch) & OMAP_DMA_CCR_EN) != 0;
+}
+EXPORT_SYMBOL(omap_get_dma_active_status);
+
+int omap_dma_running(void)
+{
+	int lch;
+
+	if (dma_omap1())
+		if (omap_lcd_dma_running())
+			return 1;
+
+	for (lch = 0; lch < dma_chan_count; lch++)
+		if (p->dma_read(CCR, lch) & OMAP_DMA_CCR_EN)
+			return 1;
+
+	return 0;
+}
+
+/*
+ * lch_queue DMA will start right after lch_head one is finished.
+ * For this DMA link to start, you still need to start (see omap_start_dma)
+ * the first one. That will fire up the entire queue.
+ */
+void omap_dma_link_lch(int lch_head, int lch_queue)
+{
+	if (omap_dma_in_1510_mode()) {
+		if (lch_head == lch_queue) {
+			p->dma_write(p->dma_read(CCR, lch_head) | (3 << 8),
+								CCR, lch_head);
+			return;
+		}
+		printk(KERN_ERR "DMA linking is not supported in 1510 mode\n");
+		BUG();
+		return;
+	}
+
+	if ((dma_chan[lch_head].dev_id == -1) ||
+	    (dma_chan[lch_queue].dev_id == -1)) {
+		pr_err("omap_dma: trying to link non requested channels\n");
+		dump_stack();
+	}
+
+	dma_chan[lch_head].next_lch = lch_queue;
+}
+EXPORT_SYMBOL(omap_dma_link_lch);
+
+/*----------------------------------------------------------------------------*/
+
+#ifdef CONFIG_ARCH_OMAP1
+
+static int omap1_dma_handle_ch(int ch)
+{
+	u32 csr;
+
+	if (enable_1510_mode && ch >= 6) {
+		csr = dma_chan[ch].saved_csr;
+		dma_chan[ch].saved_csr = 0;
+	} else
+		csr = p->dma_read(CSR, ch);
+	if (enable_1510_mode && ch <= 2 && (csr >> 7) != 0) {
+		dma_chan[ch + 6].saved_csr = csr >> 7;
+		csr &= 0x7f;
+	}
+	if ((csr & 0x3f) == 0)
+		return 0;
+	if (unlikely(dma_chan[ch].dev_id == -1)) {
+		pr_warn("Spurious interrupt from DMA channel %d (CSR %04x)\n",
+			ch, csr);
+		return 0;
+	}
+	if (unlikely(csr & OMAP1_DMA_TOUT_IRQ))
+		pr_warn("DMA timeout with device %d\n", dma_chan[ch].dev_id);
+	if (unlikely(csr & OMAP_DMA_DROP_IRQ))
+		pr_warn("DMA synchronization event drop occurred with device %d\n",
+			dma_chan[ch].dev_id);
+	if (likely(csr & OMAP_DMA_BLOCK_IRQ))
+		dma_chan[ch].flags &= ~OMAP_DMA_ACTIVE;
+	if (likely(dma_chan[ch].callback != NULL))
+		dma_chan[ch].callback(ch, csr, dma_chan[ch].data);
+
+	return 1;
+}
+
+static irqreturn_t omap1_dma_irq_handler(int irq, void *dev_id)
+{
+	int ch = ((int) dev_id) - 1;
+	int handled = 0;
+
+	for (;;) {
+		int handled_now = 0;
+
+		handled_now += omap1_dma_handle_ch(ch);
+		if (enable_1510_mode && dma_chan[ch + 6].saved_csr)
+			handled_now += omap1_dma_handle_ch(ch + 6);
+		if (!handled_now)
+			break;
+		handled += handled_now;
+	}
+
+	return handled ? IRQ_HANDLED : IRQ_NONE;
+}
+
+#else
+#define omap1_dma_irq_handler	NULL
+#endif
+
+#ifdef CONFIG_ARCH_OMAP2PLUS
+
+static int omap2_dma_handle_ch(int ch)
+{
+	u32 status = p->dma_read(CSR, ch);
+
+	if (!status) {
+		if (printk_ratelimit())
+			pr_warn("Spurious DMA IRQ for lch %d\n", ch);
+		p->dma_write(1 << ch, IRQSTATUS_L0, ch);
+		return 0;
+	}
+	if (unlikely(dma_chan[ch].dev_id == -1)) {
+		if (printk_ratelimit())
+			pr_warn("IRQ %04x for non-allocated DMA channel %d\n",
+				status, ch);
+		return 0;
+	}
+	if (unlikely(status & OMAP_DMA_DROP_IRQ))
+		pr_info("DMA synchronization event drop occurred with device %d\n",
+			dma_chan[ch].dev_id);
+	if (unlikely(status & OMAP2_DMA_TRANS_ERR_IRQ)) {
+		printk(KERN_INFO "DMA transaction error with device %d\n",
+		       dma_chan[ch].dev_id);
+		if (IS_DMA_ERRATA(DMA_ERRATA_i378)) {
+			u32 ccr;
+
+			ccr = p->dma_read(CCR, ch);
+			ccr &= ~OMAP_DMA_CCR_EN;
+			p->dma_write(ccr, CCR, ch);
+			dma_chan[ch].flags &= ~OMAP_DMA_ACTIVE;
+		}
+	}
+	if (unlikely(status & OMAP2_DMA_SECURE_ERR_IRQ))
+		printk(KERN_INFO "DMA secure error with device %d\n",
+		       dma_chan[ch].dev_id);
+	if (unlikely(status & OMAP2_DMA_MISALIGNED_ERR_IRQ))
+		printk(KERN_INFO "DMA misaligned error with device %d\n",
+		       dma_chan[ch].dev_id);
+
+	p->dma_write(status, CSR, ch);
+	p->dma_write(1 << ch, IRQSTATUS_L0, ch);
+	/* read back the register to flush the write */
+	p->dma_read(IRQSTATUS_L0, ch);
+
+	/* If the ch is not chained then chain_id will be -1 */
+	if (dma_chan[ch].chain_id != -1) {
+		int chain_id = dma_chan[ch].chain_id;
+		dma_chan[ch].state = DMA_CH_NOTSTARTED;
+		if (p->dma_read(CLNK_CTRL, ch) & (1 << 15))
+			dma_chan[dma_chan[ch].next_linked_ch].state =
+							DMA_CH_STARTED;
+		if (dma_linked_lch[chain_id].chain_mode ==
+						OMAP_DMA_DYNAMIC_CHAIN)
+			disable_lnk(ch);
+
+		if (!OMAP_DMA_CHAIN_QEMPTY(chain_id))
+			OMAP_DMA_CHAIN_INCQHEAD(chain_id);
+
+		status = p->dma_read(CSR, ch);
+		p->dma_write(status, CSR, ch);
+	}
+
+	if (likely(dma_chan[ch].callback != NULL))
+		dma_chan[ch].callback(ch, status, dma_chan[ch].data);
+
+	return 0;
+}
+
+/* STATUS register count is from 1-32 while our is 0-31 */
+static irqreturn_t omap2_dma_irq_handler(int irq, void *dev_id)
+{
+	u32 val, enable_reg;
+	int i;
+
+	val = p->dma_read(IRQSTATUS_L0, 0);
+	if (val == 0) {
+		if (printk_ratelimit())
+			printk(KERN_WARNING "Spurious DMA IRQ\n");
+		return IRQ_HANDLED;
+	}
+	enable_reg = p->dma_read(IRQENABLE_L0, 0);
+	val &= enable_reg; /* Dispatch only relevant interrupts */
+	for (i = 0; i < dma_lch_count && val != 0; i++) {
+		if (val & 1)
+			omap2_dma_handle_ch(i);
+		val >>= 1;
+	}
+
+	return IRQ_HANDLED;
+}
+
+static struct irqaction omap24xx_dma_irq = {
+	.name = "DMA",
+	.handler = omap2_dma_irq_handler,
+};
+
+#else
+static struct irqaction omap24xx_dma_irq;
+#endif
+
+/*----------------------------------------------------------------------------*/
+
+/*
+ * Note that we are currently using only IRQENABLE_L0 and L1.
+ * As the DSP may be using IRQENABLE_L2 and L3, let's not
+ * touch those for now.
+ */
+void omap_dma_global_context_save(void)
+{
+	omap_dma_global_context.dma_irqenable_l0 =
+		p->dma_read(IRQENABLE_L0, 0);
+	omap_dma_global_context.dma_irqenable_l1 =
+		p->dma_read(IRQENABLE_L1, 0);
+	omap_dma_global_context.dma_ocp_sysconfig =
+		p->dma_read(OCP_SYSCONFIG, 0);
+	omap_dma_global_context.dma_gcr = p->dma_read(GCR, 0);
+}
+
+void omap_dma_global_context_restore(void)
+{
+	int ch;
+
+	p->dma_write(omap_dma_global_context.dma_gcr, GCR, 0);
+	p->dma_write(omap_dma_global_context.dma_ocp_sysconfig,
+		OCP_SYSCONFIG, 0);
+	p->dma_write(omap_dma_global_context.dma_irqenable_l0,
+		IRQENABLE_L0, 0);
+	p->dma_write(omap_dma_global_context.dma_irqenable_l1,
+		IRQENABLE_L1, 0);
+
+	if (IS_DMA_ERRATA(DMA_ROMCODE_BUG))
+		p->dma_write(0x3 , IRQSTATUS_L0, 0);
+
+	for (ch = 0; ch < dma_chan_count; ch++)
+		if (dma_chan[ch].dev_id != -1)
+			omap_clear_dma(ch);
+}
+
+struct omap_system_dma_plat_info *omap_get_plat_info(void)
+{
+	return p;
+}
+EXPORT_SYMBOL_GPL(omap_get_plat_info);
+
+static int omap_system_dma_probe(struct platform_device *pdev)
+{
+	int ch, ret = 0;
+	int dma_irq;
+	char irq_name[4];
+	int irq_rel;
+
+	p = pdev->dev.platform_data;
+	if (!p) {
+		dev_err(&pdev->dev,
+			"%s: System DMA initialized without platform data\n",
+			__func__);
+		return -EINVAL;
+	}
+
+	d			= p->dma_attr;
+	errata			= p->errata;
+
+	if ((d->dev_caps & RESERVE_CHANNEL) && omap_dma_reserve_channels
+			&& (omap_dma_reserve_channels < d->lch_count))
+		d->lch_count	= omap_dma_reserve_channels;
+
+	dma_lch_count		= d->lch_count;
+	dma_chan_count		= dma_lch_count;
+	enable_1510_mode	= d->dev_caps & ENABLE_1510_MODE;
+
+	dma_chan = devm_kcalloc(&pdev->dev, dma_lch_count,
+				sizeof(struct omap_dma_lch), GFP_KERNEL);
+	if (!dma_chan) {
+		dev_err(&pdev->dev, "%s: kzalloc fail\n", __func__);
+		return -ENOMEM;
+	}
+
+
+	if (dma_omap2plus()) {
+		dma_linked_lch = kzalloc(sizeof(struct dma_link_info) *
+						dma_lch_count, GFP_KERNEL);
+		if (!dma_linked_lch) {
+			ret = -ENOMEM;
+			goto exit_dma_lch_fail;
+		}
+	}
+
+	spin_lock_init(&dma_chan_lock);
+	for (ch = 0; ch < dma_chan_count; ch++) {
+		omap_clear_dma(ch);
+		if (dma_omap2plus())
+			omap2_disable_irq_lch(ch);
+
+		dma_chan[ch].dev_id = -1;
+		dma_chan[ch].next_lch = -1;
+
+		if (ch >= 6 && enable_1510_mode)
+			continue;
+
+		if (dma_omap1()) {
+			/*
+			 * request_irq() doesn't like dev_id (ie. ch) being
+			 * zero, so we have to kludge around this.
+			 */
+			sprintf(&irq_name[0], "%d", ch);
+			dma_irq = platform_get_irq_byname(pdev, irq_name);
+
+			if (dma_irq < 0) {
+				ret = dma_irq;
+				goto exit_dma_irq_fail;
+			}
+
+			/* INT_DMA_LCD is handled in lcd_dma.c */
+			if (dma_irq == INT_DMA_LCD)
+				continue;
+
+			ret = request_irq(dma_irq,
+					omap1_dma_irq_handler, 0, "DMA",
+					(void *) (ch + 1));
+			if (ret != 0)
+				goto exit_dma_irq_fail;
+		}
+	}
+
+	if (d->dev_caps & IS_RW_PRIORITY)
+		omap_dma_set_global_params(DMA_DEFAULT_ARB_RATE,
+				DMA_DEFAULT_FIFO_DEPTH, 0);
+
+	if (dma_omap2plus() && !(d->dev_caps & DMA_ENGINE_HANDLE_IRQ)) {
+		strcpy(irq_name, "0");
+		dma_irq = platform_get_irq_byname(pdev, irq_name);
+		if (dma_irq < 0) {
+			dev_err(&pdev->dev, "failed: request IRQ %d", dma_irq);
+			ret = dma_irq;
+			goto exit_dma_lch_fail;
+		}
+		ret = setup_irq(dma_irq, &omap24xx_dma_irq);
+		if (ret) {
+			dev_err(&pdev->dev, "set_up failed for IRQ %d for DMA (error %d)\n",
+				dma_irq, ret);
+			goto exit_dma_lch_fail;
+		}
+	}
+
+	/* reserve dma channels 0 and 1 in high security devices on 34xx */
+	if (d->dev_caps & HS_CHANNELS_RESERVED) {
+		pr_info("Reserving DMA channels 0 and 1 for HS ROM code\n");
+		dma_chan[0].dev_id = 0;
+		dma_chan[1].dev_id = 1;
+	}
+	p->show_dma_caps();
+	return 0;
+
+exit_dma_irq_fail:
+	dev_err(&pdev->dev, "unable to request IRQ %d for DMA (error %d)\n",
+		dma_irq, ret);
+	for (irq_rel = 0; irq_rel < ch;	irq_rel++) {
+		dma_irq = platform_get_irq(pdev, irq_rel);
+		free_irq(dma_irq, (void *)(irq_rel + 1));
+	}
+
+exit_dma_lch_fail:
+	return ret;
+}
+
+static int omap_system_dma_remove(struct platform_device *pdev)
+{
+	int dma_irq;
+
+	if (dma_omap2plus()) {
+		char irq_name[4];
+		strcpy(irq_name, "0");
+		dma_irq = platform_get_irq_byname(pdev, irq_name);
+		if (dma_irq >= 0)
+			remove_irq(dma_irq, &omap24xx_dma_irq);
+	} else {
+		int irq_rel = 0;
+		for ( ; irq_rel < dma_chan_count; irq_rel++) {
+			dma_irq = platform_get_irq(pdev, irq_rel);
+			free_irq(dma_irq, (void *)(irq_rel + 1));
+		}
+	}
+	return 0;
+}
+
+static struct platform_driver omap_system_dma_driver = {
+	.probe		= omap_system_dma_probe,
+	.remove		= omap_system_dma_remove,
+	.driver		= {
+		.name	= "omap_dma_system"
+	},
+};
+
+static int __init omap_system_dma_init(void)
+{
+	return platform_driver_register(&omap_system_dma_driver);
+}
+arch_initcall(omap_system_dma_init);
+
+static void __exit omap_system_dma_exit(void)
+{
+	platform_driver_unregister(&omap_system_dma_driver);
+}
+
+MODULE_DESCRIPTION("OMAP SYSTEM DMA DRIVER");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS("platform:" DRIVER_NAME);
+MODULE_AUTHOR("Texas Instruments Inc");
+
+/*
+ * Reserve the omap SDMA channels using cmdline bootarg
+ * "omap_dma_reserve_ch=". The valid range is 1 to 32
+ */
+static int __init omap_dma_cmdline_reserve_ch(char *str)
+{
+	if (get_option(&str, &omap_dma_reserve_channels) != 1)
+		omap_dma_reserve_channels = 0;
+	return 1;
+}
+
+__setup("omap_dma_reserve_ch=", omap_dma_cmdline_reserve_ch);
+
+

arch/arm/plat-omap/dmtimer.c

@@ -0,0 +1,1002 @@
+/*
+ * linux/arch/arm/plat-omap/dmtimer.c
+ *
+ * OMAP Dual-Mode Timers
+ *
+ * Copyright (C) 2010 Texas Instruments Incorporated - http://www.ti.com/
+ * Tarun Kanti DebBarma <tarun.kanti@ti.com>
+ * Thara Gopinath <thara@ti.com>
+ *
+ * dmtimer adaptation to platform_driver.
+ *
+ * Copyright (C) 2005 Nokia Corporation
+ * OMAP2 support by Juha Yrjola
+ * API improvements and OMAP2 clock framework support by Timo Teras
+ *
+ * Copyright (C) 2009 Texas Instruments
+ * Added OMAP4 support - Santosh Shilimkar <santosh.shilimkar@ti.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2 of the License, or (at your
+ * option) any later version.
+ *
+ * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN
+ * NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
+ * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * You should have received a copy of the  GNU General Public License along
+ * with this program; if not, write  to the Free Software Foundation, Inc.,
+ * 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/clk.h>
+#include <linux/clk-provider.h>
+#include <linux/module.h>
+#include <linux/io.h>
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/pm_runtime.h>
+#include <linux/of.h>
+#include <linux/of_device.h>
+#include <linux/platform_device.h>
+#include <linux/platform_data/dmtimer-omap.h>
+
+#include <plat/dmtimer.h>
+
+static u32 omap_reserved_systimers;
+static LIST_HEAD(omap_timer_list);
+static DEFINE_SPINLOCK(dm_timer_lock);
+
+enum {
+	REQUEST_ANY = 0,
+	REQUEST_BY_ID,
+	REQUEST_BY_CAP,
+	REQUEST_BY_NODE,
+};
+
+/**
+ * omap_dm_timer_read_reg - read timer registers in posted and non-posted mode
+ * @timer:      timer pointer over which read operation to perform
+ * @reg:        lowest byte holds the register offset
+ *
+ * The posted mode bit is encoded in reg. Note that in posted mode write
+ * pending bit must be checked. Otherwise a read of a non completed write
+ * will produce an error.
+ */
+static inline u32 omap_dm_timer_read_reg(struct omap_dm_timer *timer, u32 reg)
+{
+	WARN_ON((reg & 0xff) < _OMAP_TIMER_WAKEUP_EN_OFFSET);
+	return __omap_dm_timer_read(timer, reg, timer->posted);
+}
+
+/**
+ * omap_dm_timer_write_reg - write timer registers in posted and non-posted mode
+ * @timer:      timer pointer over which write operation is to perform
+ * @reg:        lowest byte holds the register offset
+ * @value:      data to write into the register
+ *
+ * The posted mode bit is encoded in reg. Note that in posted mode the write
+ * pending bit must be checked. Otherwise a write on a register which has a
+ * pending write will be lost.
+ */
+static void omap_dm_timer_write_reg(struct omap_dm_timer *timer, u32 reg,
+						u32 value)
+{
+	WARN_ON((reg & 0xff) < _OMAP_TIMER_WAKEUP_EN_OFFSET);
+	__omap_dm_timer_write(timer, reg, value, timer->posted);
+}
+
+static void omap_timer_restore_context(struct omap_dm_timer *timer)
+{
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_WAKEUP_EN_REG,
+				timer->context.twer);
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_COUNTER_REG,
+				timer->context.tcrr);
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_LOAD_REG,
+				timer->context.tldr);
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_MATCH_REG,
+				timer->context.tmar);
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_IF_CTRL_REG,
+				timer->context.tsicr);
+	writel_relaxed(timer->context.tier, timer->irq_ena);
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_CTRL_REG,
+				timer->context.tclr);
+}
+
+static int omap_dm_timer_reset(struct omap_dm_timer *timer)
+{
+	u32 l, timeout = 100000;
+
+	if (timer->revision != 1)
+		return -EINVAL;
+
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_IF_CTRL_REG, 0x06);
+
+	do {
+		l = __omap_dm_timer_read(timer,
+					 OMAP_TIMER_V1_SYS_STAT_OFFSET, 0);
+	} while (!l && timeout--);
+
+	if (!timeout) {
+		dev_err(&timer->pdev->dev, "Timer failed to reset\n");
+		return -ETIMEDOUT;
+	}
+
+	/* Configure timer for smart-idle mode */
+	l = __omap_dm_timer_read(timer, OMAP_TIMER_OCP_CFG_OFFSET, 0);
+	l |= 0x2 << 0x3;
+	__omap_dm_timer_write(timer, OMAP_TIMER_OCP_CFG_OFFSET, l, 0);
+
+	timer->posted = 0;
+
+	return 0;
+}
+
+static int omap_dm_timer_of_set_source(struct omap_dm_timer *timer)
+{
+	int ret;
+	struct clk *parent;
+
+	/*
+	 * FIXME: OMAP1 devices do not use the clock framework for dmtimers so
+	 * do not call clk_get() for these devices.
+	 */
+	if (!timer->fclk)
+		return -ENODEV;
+
+	parent = clk_get(&timer->pdev->dev, NULL);
+	if (IS_ERR(parent))
+		return -ENODEV;
+
+	ret = clk_set_parent(timer->fclk, parent);
+	if (ret < 0)
+		pr_err("%s: failed to set parent\n", __func__);
+
+	clk_put(parent);
+
+	return ret;
+}
+
+static int omap_dm_timer_prepare(struct omap_dm_timer *timer)
+{
+	int rc;
+
+	/*
+	 * FIXME: OMAP1 devices do not use the clock framework for dmtimers so
+	 * do not call clk_get() for these devices.
+	 */
+	if (!(timer->capability & OMAP_TIMER_NEEDS_RESET)) {
+		timer->fclk = clk_get(&timer->pdev->dev, "fck");
+		if (WARN_ON_ONCE(IS_ERR(timer->fclk))) {
+			dev_err(&timer->pdev->dev, ": No fclk handle.\n");
+			return -EINVAL;
+		}
+	}
+
+	omap_dm_timer_enable(timer);
+
+	if (timer->capability & OMAP_TIMER_NEEDS_RESET) {
+		rc = omap_dm_timer_reset(timer);
+		if (rc) {
+			omap_dm_timer_disable(timer);
+			return rc;
+		}
+	}
+
+	__omap_dm_timer_enable_posted(timer);
+	omap_dm_timer_disable(timer);
+
+	rc = omap_dm_timer_of_set_source(timer);
+	if (rc == -ENODEV)
+		return omap_dm_timer_set_source(timer, OMAP_TIMER_SRC_32_KHZ);
+
+	return rc;
+}
+
+static inline u32 omap_dm_timer_reserved_systimer(int id)
+{
+	return (omap_reserved_systimers & (1 << (id - 1))) ? 1 : 0;
+}
+
+int omap_dm_timer_reserve_systimer(int id)
+{
+	if (omap_dm_timer_reserved_systimer(id))
+		return -ENODEV;
+
+	omap_reserved_systimers |= (1 << (id - 1));
+
+	return 0;
+}
+
+static struct omap_dm_timer *_omap_dm_timer_request(int req_type, void *data)
+{
+	struct omap_dm_timer *timer = NULL, *t;
+	struct device_node *np = NULL;
+	unsigned long flags;
+	u32 cap = 0;
+	int id = 0;
+
+	switch (req_type) {
+	case REQUEST_BY_ID:
+		id = *(int *)data;
+		break;
+	case REQUEST_BY_CAP:
+		cap = *(u32 *)data;
+		break;
+	case REQUEST_BY_NODE:
+		np = (struct device_node *)data;
+		break;
+	default:
+		/* REQUEST_ANY */
+		break;
+	}
+
+	spin_lock_irqsave(&dm_timer_lock, flags);
+	list_for_each_entry(t, &omap_timer_list, node) {
+		if (t->reserved)
+			continue;
+
+		switch (req_type) {
+		case REQUEST_BY_ID:
+			if (id == t->pdev->id) {
+				timer = t;
+				timer->reserved = 1;
+				goto found;
+			}
+			break;
+		case REQUEST_BY_CAP:
+			if (cap == (t->capability & cap)) {
+				/*
+				 * If timer is not NULL, we have already found
+				 * one timer but it was not an exact match
+				 * because it had more capabilites that what
+				 * was required. Therefore, unreserve the last
+				 * timer found and see if this one is a better
+				 * match.
+				 */
+				if (timer)
+					timer->reserved = 0;
+				timer = t;
+				timer->reserved = 1;
+
+				/* Exit loop early if we find an exact match */
+				if (t->capability == cap)
+					goto found;
+			}
+			break;
+		case REQUEST_BY_NODE:
+			if (np == t->pdev->dev.of_node) {
+				timer = t;
+				timer->reserved = 1;
+				goto found;
+			}
+			break;
+		default:
+			/* REQUEST_ANY */
+			timer = t;
+			timer->reserved = 1;
+			goto found;
+		}
+	}
+found:
+	spin_unlock_irqrestore(&dm_timer_lock, flags);
+
+	if (timer && omap_dm_timer_prepare(timer)) {
+		timer->reserved = 0;
+		timer = NULL;
+	}
+
+	if (!timer)
+		pr_debug("%s: timer request failed!\n", __func__);
+
+	return timer;
+}
+
+struct omap_dm_timer *omap_dm_timer_request(void)
+{
+	return _omap_dm_timer_request(REQUEST_ANY, NULL);
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_request);
+
+struct omap_dm_timer *omap_dm_timer_request_specific(int id)
+{
+	/* Requesting timer by ID is not supported when device tree is used */
+	if (of_have_populated_dt()) {
+		pr_warn("%s: Please use omap_dm_timer_request_by_cap/node()\n",
+			__func__);
+		return NULL;
+	}
+
+	return _omap_dm_timer_request(REQUEST_BY_ID, &id);
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_request_specific);
+
+/**
+ * omap_dm_timer_request_by_cap - Request a timer by capability
+ * @cap:	Bit mask of capabilities to match
+ *
+ * Find a timer based upon capabilities bit mask. Callers of this function
+ * should use the definitions found in the plat/dmtimer.h file under the
+ * comment "timer capabilities used in hwmod database". Returns pointer to
+ * timer handle on success and a NULL pointer on failure.
+ */
+struct omap_dm_timer *omap_dm_timer_request_by_cap(u32 cap)
+{
+	return _omap_dm_timer_request(REQUEST_BY_CAP, &cap);
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_request_by_cap);
+
+/**
+ * omap_dm_timer_request_by_node - Request a timer by device-tree node
+ * @np:		Pointer to device-tree timer node
+ *
+ * Request a timer based upon a device node pointer. Returns pointer to
+ * timer handle on success and a NULL pointer on failure.
+ */
+struct omap_dm_timer *omap_dm_timer_request_by_node(struct device_node *np)
+{
+	if (!np)
+		return NULL;
+
+	return _omap_dm_timer_request(REQUEST_BY_NODE, np);
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_request_by_node);
+
+int omap_dm_timer_free(struct omap_dm_timer *timer)
+{
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	clk_put(timer->fclk);
+
+	WARN_ON(!timer->reserved);
+	timer->reserved = 0;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_free);
+
+void omap_dm_timer_enable(struct omap_dm_timer *timer)
+{
+	int c;
+
+	pm_runtime_get_sync(&timer->pdev->dev);
+
+	if (!(timer->capability & OMAP_TIMER_ALWON)) {
+		if (timer->get_context_loss_count) {
+			c = timer->get_context_loss_count(&timer->pdev->dev);
+			if (c != timer->ctx_loss_count) {
+				omap_timer_restore_context(timer);
+				timer->ctx_loss_count = c;
+			}
+		} else {
+			omap_timer_restore_context(timer);
+		}
+	}
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_enable);
+
+void omap_dm_timer_disable(struct omap_dm_timer *timer)
+{
+	pm_runtime_put_sync(&timer->pdev->dev);
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_disable);
+
+int omap_dm_timer_get_irq(struct omap_dm_timer *timer)
+{
+	if (timer)
+		return timer->irq;
+	return -EINVAL;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_get_irq);
+
+#if defined(CONFIG_ARCH_OMAP1)
+#include <mach/hardware.h>
+/**
+ * omap_dm_timer_modify_idlect_mask - Check if any running timers use ARMXOR
+ * @inputmask: current value of idlect mask
+ */
+__u32 omap_dm_timer_modify_idlect_mask(__u32 inputmask)
+{
+	int i = 0;
+	struct omap_dm_timer *timer = NULL;
+	unsigned long flags;
+
+	/* If ARMXOR cannot be idled this function call is unnecessary */
+	if (!(inputmask & (1 << 1)))
+		return inputmask;
+
+	/* If any active timer is using ARMXOR return modified mask */
+	spin_lock_irqsave(&dm_timer_lock, flags);
+	list_for_each_entry(timer, &omap_timer_list, node) {
+		u32 l;
+
+		l = omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG);
+		if (l & OMAP_TIMER_CTRL_ST) {
+			if (((omap_readl(MOD_CONF_CTRL_1) >> (i * 2)) & 0x03) == 0)
+				inputmask &= ~(1 << 1);
+			else
+				inputmask &= ~(1 << 2);
+		}
+		i++;
+	}
+	spin_unlock_irqrestore(&dm_timer_lock, flags);
+
+	return inputmask;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_modify_idlect_mask);
+
+#else
+
+struct clk *omap_dm_timer_get_fclk(struct omap_dm_timer *timer)
+{
+	if (timer && !IS_ERR(timer->fclk))
+		return timer->fclk;
+	return NULL;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_get_fclk);
+
+__u32 omap_dm_timer_modify_idlect_mask(__u32 inputmask)
+{
+	BUG();
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_modify_idlect_mask);
+
+#endif
+
+int omap_dm_timer_trigger(struct omap_dm_timer *timer)
+{
+	if (unlikely(!timer || pm_runtime_suspended(&timer->pdev->dev))) {
+		pr_err("%s: timer not available or enabled.\n", __func__);
+		return -EINVAL;
+	}
+
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_TRIGGER_REG, 0);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_trigger);
+
+int omap_dm_timer_start(struct omap_dm_timer *timer)
+{
+	u32 l;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	omap_dm_timer_enable(timer);
+
+	l = omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG);
+	if (!(l & OMAP_TIMER_CTRL_ST)) {
+		l |= OMAP_TIMER_CTRL_ST;
+		omap_dm_timer_write_reg(timer, OMAP_TIMER_CTRL_REG, l);
+	}
+
+	/* Save the context */
+	timer->context.tclr = l;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_start);
+
+int omap_dm_timer_stop(struct omap_dm_timer *timer)
+{
+	unsigned long rate = 0;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	if (!(timer->capability & OMAP_TIMER_NEEDS_RESET))
+		rate = clk_get_rate(timer->fclk);
+
+	__omap_dm_timer_stop(timer, timer->posted, rate);
+
+	/*
+	 * Since the register values are computed and written within
+	 * __omap_dm_timer_stop, we need to use read to retrieve the
+	 * context.
+	 */
+	timer->context.tclr =
+			omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG);
+	omap_dm_timer_disable(timer);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_stop);
+
+int omap_dm_timer_set_source(struct omap_dm_timer *timer, int source)
+{
+	int ret;
+	char *parent_name = NULL;
+	struct clk *parent;
+	struct dmtimer_platform_data *pdata;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	pdata = timer->pdev->dev.platform_data;
+
+	if (source < 0 || source >= 3)
+		return -EINVAL;
+
+	/*
+	 * FIXME: Used for OMAP1 devices only because they do not currently
+	 * use the clock framework to set the parent clock. To be removed
+	 * once OMAP1 migrated to using clock framework for dmtimers
+	 */
+	if (pdata && pdata->set_timer_src)
+		return pdata->set_timer_src(timer->pdev, source);
+
+	if (IS_ERR(timer->fclk))
+		return -EINVAL;
+
+#if defined(CONFIG_COMMON_CLK)
+	/* Check if the clock has configurable parents */
+	if (clk_hw_get_num_parents(__clk_get_hw(timer->fclk)) < 2)
+		return 0;
+#endif
+
+	switch (source) {
+	case OMAP_TIMER_SRC_SYS_CLK:
+		parent_name = "timer_sys_ck";
+		break;
+
+	case OMAP_TIMER_SRC_32_KHZ:
+		parent_name = "timer_32k_ck";
+		break;
+
+	case OMAP_TIMER_SRC_EXT_CLK:
+		parent_name = "timer_ext_ck";
+		break;
+	}
+
+	parent = clk_get(&timer->pdev->dev, parent_name);
+	if (IS_ERR(parent)) {
+		pr_err("%s: %s not found\n", __func__, parent_name);
+		return -EINVAL;
+	}
+
+	ret = clk_set_parent(timer->fclk, parent);
+	if (ret < 0)
+		pr_err("%s: failed to set %s as parent\n", __func__,
+			parent_name);
+
+	clk_put(parent);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_set_source);
+
+int omap_dm_timer_set_load(struct omap_dm_timer *timer, int autoreload,
+			    unsigned int load)
+{
+	u32 l;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	omap_dm_timer_enable(timer);
+	l = omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG);
+	if (autoreload)
+		l |= OMAP_TIMER_CTRL_AR;
+	else
+		l &= ~OMAP_TIMER_CTRL_AR;
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_CTRL_REG, l);
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_LOAD_REG, load);
+
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_TRIGGER_REG, 0);
+	/* Save the context */
+	timer->context.tclr = l;
+	timer->context.tldr = load;
+	omap_dm_timer_disable(timer);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_set_load);
+
+/* Optimized set_load which removes costly spin wait in timer_start */
+int omap_dm_timer_set_load_start(struct omap_dm_timer *timer, int autoreload,
+                            unsigned int load)
+{
+	u32 l;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	omap_dm_timer_enable(timer);
+
+	l = omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG);
+	if (autoreload) {
+		l |= OMAP_TIMER_CTRL_AR;
+		omap_dm_timer_write_reg(timer, OMAP_TIMER_LOAD_REG, load);
+	} else {
+		l &= ~OMAP_TIMER_CTRL_AR;
+	}
+	l |= OMAP_TIMER_CTRL_ST;
+
+	__omap_dm_timer_load_start(timer, l, load, timer->posted);
+
+	/* Save the context */
+	timer->context.tclr = l;
+	timer->context.tldr = load;
+	timer->context.tcrr = load;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_set_load_start);
+
+int omap_dm_timer_set_match(struct omap_dm_timer *timer, int enable,
+			     unsigned int match)
+{
+	u32 l;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	omap_dm_timer_enable(timer);
+	l = omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG);
+	if (enable)
+		l |= OMAP_TIMER_CTRL_CE;
+	else
+		l &= ~OMAP_TIMER_CTRL_CE;
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_MATCH_REG, match);
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_CTRL_REG, l);
+
+	/* Save the context */
+	timer->context.tclr = l;
+	timer->context.tmar = match;
+	omap_dm_timer_disable(timer);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_set_match);
+
+int omap_dm_timer_set_pwm(struct omap_dm_timer *timer, int def_on,
+			   int toggle, int trigger)
+{
+	u32 l;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	omap_dm_timer_enable(timer);
+	l = omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG);
+	l &= ~(OMAP_TIMER_CTRL_GPOCFG | OMAP_TIMER_CTRL_SCPWM |
+	       OMAP_TIMER_CTRL_PT | (0x03 << 10));
+	if (def_on)
+		l |= OMAP_TIMER_CTRL_SCPWM;
+	if (toggle)
+		l |= OMAP_TIMER_CTRL_PT;
+	l |= trigger << 10;
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_CTRL_REG, l);
+
+	/* Save the context */
+	timer->context.tclr = l;
+	omap_dm_timer_disable(timer);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_set_pwm);
+
+int omap_dm_timer_set_prescaler(struct omap_dm_timer *timer, int prescaler)
+{
+	u32 l;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	omap_dm_timer_enable(timer);
+	l = omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG);
+	l &= ~(OMAP_TIMER_CTRL_PRE | (0x07 << 2));
+	if (prescaler >= 0x00 && prescaler <= 0x07) {
+		l |= OMAP_TIMER_CTRL_PRE;
+		l |= prescaler << 2;
+	}
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_CTRL_REG, l);
+
+	/* Save the context */
+	timer->context.tclr = l;
+	omap_dm_timer_disable(timer);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_set_prescaler);
+
+int omap_dm_timer_set_int_enable(struct omap_dm_timer *timer,
+				  unsigned int value)
+{
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	omap_dm_timer_enable(timer);
+	__omap_dm_timer_int_enable(timer, value);
+
+	/* Save the context */
+	timer->context.tier = value;
+	timer->context.twer = value;
+	omap_dm_timer_disable(timer);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_set_int_enable);
+
+/**
+ * omap_dm_timer_set_int_disable - disable timer interrupts
+ * @timer:	pointer to timer handle
+ * @mask:	bit mask of interrupts to be disabled
+ *
+ * Disables the specified timer interrupts for a timer.
+ */
+int omap_dm_timer_set_int_disable(struct omap_dm_timer *timer, u32 mask)
+{
+	u32 l = mask;
+
+	if (unlikely(!timer))
+		return -EINVAL;
+
+	omap_dm_timer_enable(timer);
+
+	if (timer->revision == 1)
+		l = readl_relaxed(timer->irq_ena) & ~mask;
+
+	writel_relaxed(l, timer->irq_dis);
+	l = omap_dm_timer_read_reg(timer, OMAP_TIMER_WAKEUP_EN_REG) & ~mask;
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_WAKEUP_EN_REG, l);
+
+	/* Save the context */
+	timer->context.tier &= ~mask;
+	timer->context.twer &= ~mask;
+	omap_dm_timer_disable(timer);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_set_int_disable);
+
+unsigned int omap_dm_timer_read_status(struct omap_dm_timer *timer)
+{
+	unsigned int l;
+
+	if (unlikely(!timer || pm_runtime_suspended(&timer->pdev->dev))) {
+		pr_err("%s: timer not available or enabled.\n", __func__);
+		return 0;
+	}
+
+	l = readl_relaxed(timer->irq_stat);
+
+	return l;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_read_status);
+
+int omap_dm_timer_write_status(struct omap_dm_timer *timer, unsigned int value)
+{
+	if (unlikely(!timer || pm_runtime_suspended(&timer->pdev->dev)))
+		return -EINVAL;
+
+	__omap_dm_timer_write_status(timer, value);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_write_status);
+
+unsigned int omap_dm_timer_read_counter(struct omap_dm_timer *timer)
+{
+	if (unlikely(!timer || pm_runtime_suspended(&timer->pdev->dev))) {
+		pr_err("%s: timer not iavailable or enabled.\n", __func__);
+		return 0;
+	}
+
+	return __omap_dm_timer_read_counter(timer, timer->posted);
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_read_counter);
+
+int omap_dm_timer_write_counter(struct omap_dm_timer *timer, unsigned int value)
+{
+	if (unlikely(!timer || pm_runtime_suspended(&timer->pdev->dev))) {
+		pr_err("%s: timer not available or enabled.\n", __func__);
+		return -EINVAL;
+	}
+
+	omap_dm_timer_write_reg(timer, OMAP_TIMER_COUNTER_REG, value);
+
+	/* Save the context */
+	timer->context.tcrr = value;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timer_write_counter);
+
+int omap_dm_timers_active(void)
+{
+	struct omap_dm_timer *timer;
+
+	list_for_each_entry(timer, &omap_timer_list, node) {
+		if (!timer->reserved)
+			continue;
+
+		if (omap_dm_timer_read_reg(timer, OMAP_TIMER_CTRL_REG) &
+		    OMAP_TIMER_CTRL_ST) {
+			return 1;
+		}
+	}
+	return 0;
+}
+EXPORT_SYMBOL_GPL(omap_dm_timers_active);
+
+static const struct of_device_id omap_timer_match[];
+
+/**
+ * omap_dm_timer_probe - probe function called for every registered device
+ * @pdev:	pointer to current timer platform device
+ *
+ * Called by driver framework at the end of device registration for all
+ * timer devices.
+ */
+static int omap_dm_timer_probe(struct platform_device *pdev)
+{
+	unsigned long flags;
+	struct omap_dm_timer *timer;
+	struct resource *mem, *irq;
+	struct device *dev = &pdev->dev;
+	const struct of_device_id *match;
+	const struct dmtimer_platform_data *pdata;
+	int ret;
+
+	match = of_match_device(of_match_ptr(omap_timer_match), dev);
+	pdata = match ? match->data : dev->platform_data;
+
+	if (!pdata && !dev->of_node) {
+		dev_err(dev, "%s: no platform data.\n", __func__);
+		return -ENODEV;
+	}
+
+	irq = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
+	if (unlikely(!irq)) {
+		dev_err(dev, "%s: no IRQ resource.\n", __func__);
+		return -ENODEV;
+	}
+
+	mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (unlikely(!mem)) {
+		dev_err(dev, "%s: no memory resource.\n", __func__);
+		return -ENODEV;
+	}
+
+	timer = devm_kzalloc(dev, sizeof(struct omap_dm_timer), GFP_KERNEL);
+	if (!timer) {
+		dev_err(dev, "%s: memory alloc failed!\n", __func__);
+		return  -ENOMEM;
+	}
+
+	timer->fclk = ERR_PTR(-ENODEV);
+	timer->io_base = devm_ioremap_resource(dev, mem);
+	if (IS_ERR(timer->io_base))
+		return PTR_ERR(timer->io_base);
+
+	if (dev->of_node) {
+		if (of_find_property(dev->of_node, "ti,timer-alwon", NULL))
+			timer->capability |= OMAP_TIMER_ALWON;
+		if (of_find_property(dev->of_node, "ti,timer-dsp", NULL))
+			timer->capability |= OMAP_TIMER_HAS_DSP_IRQ;
+		if (of_find_property(dev->of_node, "ti,timer-pwm", NULL))
+			timer->capability |= OMAP_TIMER_HAS_PWM;
+		if (of_find_property(dev->of_node, "ti,timer-secure", NULL))
+			timer->capability |= OMAP_TIMER_SECURE;
+	} else {
+		timer->id = pdev->id;
+		timer->capability = pdata->timer_capability;
+		timer->reserved = omap_dm_timer_reserved_systimer(timer->id);
+		timer->get_context_loss_count = pdata->get_context_loss_count;
+	}
+
+	if (pdata)
+		timer->errata = pdata->timer_errata;
+
+	timer->irq = irq->start;
+	timer->pdev = pdev;
+
+	pm_runtime_enable(dev);
+	pm_runtime_irq_safe(dev);
+
+	if (!timer->reserved) {
+		ret = pm_runtime_get_sync(dev);
+		if (ret < 0) {
+			dev_err(dev, "%s: pm_runtime_get_sync failed!\n",
+				__func__);
+			goto err_get_sync;
+		}
+		__omap_dm_timer_init_regs(timer);
+		pm_runtime_put(dev);
+	}
+
+	/* add the timer element to the list */
+	spin_lock_irqsave(&dm_timer_lock, flags);
+	list_add_tail(&timer->node, &omap_timer_list);
+	spin_unlock_irqrestore(&dm_timer_lock, flags);
+
+	dev_dbg(dev, "Device Probed.\n");
+
+	return 0;
+
+err_get_sync:
+	pm_runtime_put_noidle(dev);
+	pm_runtime_disable(dev);
+	return ret;
+}
+
+/**
+ * omap_dm_timer_remove - cleanup a registered timer device
+ * @pdev:	pointer to current timer platform device
+ *
+ * Called by driver framework whenever a timer device is unregistered.
+ * In addition to freeing platform resources it also deletes the timer
+ * entry from the local list.
+ */
+static int omap_dm_timer_remove(struct platform_device *pdev)
+{
+	struct omap_dm_timer *timer;
+	unsigned long flags;
+	int ret = -EINVAL;
+
+	spin_lock_irqsave(&dm_timer_lock, flags);
+	list_for_each_entry(timer, &omap_timer_list, node)
+		if (!strcmp(dev_name(&timer->pdev->dev),
+			    dev_name(&pdev->dev))) {
+			list_del(&timer->node);
+			ret = 0;
+			break;
+		}
+	spin_unlock_irqrestore(&dm_timer_lock, flags);
+
+	pm_runtime_disable(&pdev->dev);
+
+	return ret;
+}
+
+static const struct dmtimer_platform_data omap3plus_pdata = {
+	.timer_errata = OMAP_TIMER_ERRATA_I103_I767,
+};
+
+static const struct of_device_id omap_timer_match[] = {
+	{
+		.compatible = "ti,omap2420-timer",
+	},
+	{
+		.compatible = "ti,omap3430-timer",
+		.data = &omap3plus_pdata,
+	},
+	{
+		.compatible = "ti,omap4430-timer",
+		.data = &omap3plus_pdata,
+	},
+	{
+		.compatible = "ti,omap5430-timer",
+		.data = &omap3plus_pdata,
+	},
+	{
+		.compatible = "ti,am335x-timer",
+		.data = &omap3plus_pdata,
+	},
+	{
+		.compatible = "ti,am335x-timer-1ms",
+		.data = &omap3plus_pdata,
+	},
+	{
+		.compatible = "ti,dm816-timer",
+		.data = &omap3plus_pdata,
+	},
+	{},
+};
+MODULE_DEVICE_TABLE(of, omap_timer_match);
+
+static struct platform_driver omap_dm_timer_driver = {
+	.probe  = omap_dm_timer_probe,
+	.remove = omap_dm_timer_remove,
+	.driver = {
+		.name   = "omap_timer",
+		.of_match_table = of_match_ptr(omap_timer_match),
+	},
+};
+
+early_platform_init("earlytimer", &omap_dm_timer_driver);
+module_platform_driver(omap_dm_timer_driver);
+
+MODULE_DESCRIPTION("OMAP Dual-Mode Timer Driver");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS("platform:" DRIVER_NAME);
+MODULE_AUTHOR("Texas Instruments Inc");

arch/arm/plat-omap/i2c.c

@@ -0,0 +1,116 @@
+/*
+ * linux/arch/arm/plat-omap/i2c.c
+ *
+ * Helper module for board specific I2C bus registration
+ *
+ * Copyright (C) 2007 Nokia Corporation.
+ *
+ * Contact: Jarkko Nikula <jhnikula@gmail.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
+ * 02110-1301 USA
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+#include <linux/i2c.h>
+#include <linux/i2c-omap.h>
+#include <linux/slab.h>
+#include <linux/err.h>
+#include <linux/clk.h>
+
+#include <plat/i2c.h>
+
+#define OMAP_I2C_MAX_CONTROLLERS 4
+static struct omap_i2c_bus_platform_data i2c_pdata[OMAP_I2C_MAX_CONTROLLERS];
+
+#define OMAP_I2C_CMDLINE_SETUP	(BIT(31))
+
+/**
+ * omap_i2c_bus_setup - Process command line options for the I2C bus speed
+ * @str: String of options
+ *
+ * This function allow to override the default I2C bus speed for given I2C
+ * bus with a command line option.
+ *
+ * Format: i2c_bus=bus_id,clkrate (in kHz)
+ *
+ * Returns 1 on success, 0 otherwise.
+ */
+static int __init omap_i2c_bus_setup(char *str)
+{
+	int ints[3];
+
+	get_options(str, 3, ints);
+	if (ints[0] < 2 || ints[1] < 1 ||
+			ints[1] > OMAP_I2C_MAX_CONTROLLERS)
+		return 0;
+	i2c_pdata[ints[1] - 1].clkrate = ints[2];
+	i2c_pdata[ints[1] - 1].clkrate |= OMAP_I2C_CMDLINE_SETUP;
+
+	return 1;
+}
+__setup("i2c_bus=", omap_i2c_bus_setup);
+
+/*
+ * Register busses defined in command line but that are not registered with
+ * omap_register_i2c_bus from board initialization code.
+ */
+int __init omap_register_i2c_bus_cmdline(void)
+{
+	int i, err = 0;
+
+	for (i = 0; i < ARRAY_SIZE(i2c_pdata); i++)
+		if (i2c_pdata[i].clkrate & OMAP_I2C_CMDLINE_SETUP) {
+			i2c_pdata[i].clkrate &= ~OMAP_I2C_CMDLINE_SETUP;
+			err = omap_i2c_add_bus(&i2c_pdata[i], i + 1);
+			if (err)
+				goto out;
+		}
+
+out:
+	return err;
+}
+
+/**
+ * omap_register_i2c_bus - register I2C bus with device descriptors
+ * @bus_id: bus id counting from number 1
+ * @clkrate: clock rate of the bus in kHz
+ * @info: pointer into I2C device descriptor table or NULL
+ * @len: number of descriptors in the table
+ *
+ * Returns 0 on success or an error code.
+ */
+int __init omap_register_i2c_bus(int bus_id, u32 clkrate,
+			  struct i2c_board_info const *info,
+			  unsigned len)
+{
+	int err;
+
+	BUG_ON(bus_id < 1 || bus_id > OMAP_I2C_MAX_CONTROLLERS);
+
+	if (info) {
+		err = i2c_register_board_info(bus_id, info, len);
+		if (err)
+			return err;
+	}
+
+	if (!i2c_pdata[bus_id - 1].clkrate)
+		i2c_pdata[bus_id - 1].clkrate = clkrate;
+
+	i2c_pdata[bus_id - 1].clkrate &= ~OMAP_I2C_CMDLINE_SETUP;
+
+	return omap_i2c_add_bus(&i2c_pdata[bus_id - 1], bus_id);
+}

arch/arm/plat-omap/include/plat/counter-32k.h

@@ -0,0 +1 @@
+int omap_init_clocksource_32k(void __iomem *vbase);

arch/arm/plat-omap/include/plat/cpu.h

@@ -0,0 +1,35 @@
+/*
+ * OMAP cpu type detection
+ *
+ * Copyright (C) 2004, 2008 Nokia Corporation
+ *
+ * Copyright (C) 2009-11 Texas Instruments.
+ *
+ * Written by Tony Lindgren <tony.lindgren@nokia.com>
+ *
+ * Added OMAP4/5 specific defines - Santosh Shilimkar<santosh.shilimkar@ti.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ *
+ */
+
+#ifndef __ASM_ARCH_OMAP_CPU_H
+#define __ASM_ARCH_OMAP_CPU_H
+
+#ifdef CONFIG_ARCH_OMAP1
+#include <mach/soc.h>
+#endif
+
+#endif

arch/arm/plat-omap/include/plat/dmtimer.h

@@ -0,0 +1,418 @@
+/*
+ * arch/arm/plat-omap/include/plat/dmtimer.h
+ *
+ * OMAP Dual-Mode Timers
+ *
+ * Copyright (C) 2010 Texas Instruments Incorporated - http://www.ti.com/
+ * Tarun Kanti DebBarma <tarun.kanti@ti.com>
+ * Thara Gopinath <thara@ti.com>
+ *
+ * Platform device conversion and hwmod support.
+ *
+ * Copyright (C) 2005 Nokia Corporation
+ * Author: Lauri Leukkunen <lauri.leukkunen@nokia.com>
+ * PWM and clock framwork support by Timo Teras.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2 of the License, or (at your
+ * option) any later version.
+ *
+ * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN
+ * NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
+ * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * You should have received a copy of the  GNU General Public License along
+ * with this program; if not, write  to the Free Software Foundation, Inc.,
+ * 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/delay.h>
+#include <linux/io.h>
+#include <linux/platform_device.h>
+
+#ifndef __ASM_ARCH_DMTIMER_H
+#define __ASM_ARCH_DMTIMER_H
+
+/* clock sources */
+#define OMAP_TIMER_SRC_SYS_CLK			0x00
+#define OMAP_TIMER_SRC_32_KHZ			0x01
+#define OMAP_TIMER_SRC_EXT_CLK			0x02
+
+/* timer interrupt enable bits */
+#define OMAP_TIMER_INT_CAPTURE			(1 << 2)
+#define OMAP_TIMER_INT_OVERFLOW			(1 << 1)
+#define OMAP_TIMER_INT_MATCH			(1 << 0)
+
+/* trigger types */
+#define OMAP_TIMER_TRIGGER_NONE			0x00
+#define OMAP_TIMER_TRIGGER_OVERFLOW		0x01
+#define OMAP_TIMER_TRIGGER_OVERFLOW_AND_COMPARE	0x02
+
+/* posted mode types */
+#define OMAP_TIMER_NONPOSTED			0x00
+#define OMAP_TIMER_POSTED			0x01
+
+/* timer capabilities used in hwmod database */
+#define OMAP_TIMER_SECURE				0x80000000
+#define OMAP_TIMER_ALWON				0x40000000
+#define OMAP_TIMER_HAS_PWM				0x20000000
+#define OMAP_TIMER_NEEDS_RESET				0x10000000
+#define OMAP_TIMER_HAS_DSP_IRQ				0x08000000
+
+/*
+ * timer errata flags
+ *
+ * Errata i103/i767 impacts all OMAP3/4/5 devices including AM33xx. This
+ * errata prevents us from using posted mode on these devices, unless the
+ * timer counter register is never read. For more details please refer to
+ * the OMAP3/4/5 errata documents.
+ */
+#define OMAP_TIMER_ERRATA_I103_I767			0x80000000
+
+struct omap_timer_capability_dev_attr {
+	u32 timer_capability;
+};
+
+struct timer_regs {
+	u32 tidr;
+	u32 tier;
+	u32 twer;
+	u32 tclr;
+	u32 tcrr;
+	u32 tldr;
+	u32 ttrg;
+	u32 twps;
+	u32 tmar;
+	u32 tcar1;
+	u32 tsicr;
+	u32 tcar2;
+	u32 tpir;
+	u32 tnir;
+	u32 tcvr;
+	u32 tocr;
+	u32 towr;
+};
+
+struct omap_dm_timer {
+	int id;
+	int irq;
+	struct clk *fclk;
+
+	void __iomem	*io_base;
+	void __iomem	*irq_stat;	/* TISR/IRQSTATUS interrupt status */
+	void __iomem	*irq_ena;	/* irq enable */
+	void __iomem	*irq_dis;	/* irq disable, only on v2 ip */
+	void __iomem	*pend;		/* write pending */
+	void __iomem	*func_base;	/* function register base */
+
+	unsigned long rate;
+	unsigned reserved:1;
+	unsigned posted:1;
+	struct timer_regs context;
+	int (*get_context_loss_count)(struct device *);
+	int ctx_loss_count;
+	int revision;
+	u32 capability;
+	u32 errata;
+	struct platform_device *pdev;
+	struct list_head node;
+};
+
+int omap_dm_timer_reserve_systimer(int id);
+struct omap_dm_timer *omap_dm_timer_request(void);
+struct omap_dm_timer *omap_dm_timer_request_specific(int timer_id);
+struct omap_dm_timer *omap_dm_timer_request_by_cap(u32 cap);
+struct omap_dm_timer *omap_dm_timer_request_by_node(struct device_node *np);
+int omap_dm_timer_free(struct omap_dm_timer *timer);
+void omap_dm_timer_enable(struct omap_dm_timer *timer);
+void omap_dm_timer_disable(struct omap_dm_timer *timer);
+
+int omap_dm_timer_get_irq(struct omap_dm_timer *timer);
+
+u32 omap_dm_timer_modify_idlect_mask(u32 inputmask);
+struct clk *omap_dm_timer_get_fclk(struct omap_dm_timer *timer);
+
+int omap_dm_timer_trigger(struct omap_dm_timer *timer);
+int omap_dm_timer_start(struct omap_dm_timer *timer);
+int omap_dm_timer_stop(struct omap_dm_timer *timer);
+
+int omap_dm_timer_set_source(struct omap_dm_timer *timer, int source);
+int omap_dm_timer_set_load(struct omap_dm_timer *timer, int autoreload, unsigned int value);
+int omap_dm_timer_set_load_start(struct omap_dm_timer *timer, int autoreload, unsigned int value);
+int omap_dm_timer_set_match(struct omap_dm_timer *timer, int enable, unsigned int match);
+int omap_dm_timer_set_pwm(struct omap_dm_timer *timer, int def_on, int toggle, int trigger);
+int omap_dm_timer_set_prescaler(struct omap_dm_timer *timer, int prescaler);
+
+int omap_dm_timer_set_int_enable(struct omap_dm_timer *timer, unsigned int value);
+int omap_dm_timer_set_int_disable(struct omap_dm_timer *timer, u32 mask);
+
+unsigned int omap_dm_timer_read_status(struct omap_dm_timer *timer);
+int omap_dm_timer_write_status(struct omap_dm_timer *timer, unsigned int value);
+unsigned int omap_dm_timer_read_counter(struct omap_dm_timer *timer);
+int omap_dm_timer_write_counter(struct omap_dm_timer *timer, unsigned int value);
+
+int omap_dm_timers_active(void);
+
+/*
+ * Do not use the defines below, they are not needed. They should be only
+ * used by dmtimer.c and sys_timer related code.
+ */
+
+/*
+ * The interrupt registers are different between v1 and v2 ip.
+ * These registers are offsets from timer->iobase.
+ */
+#define OMAP_TIMER_ID_OFFSET		0x00
+#define OMAP_TIMER_OCP_CFG_OFFSET	0x10
+
+#define OMAP_TIMER_V1_SYS_STAT_OFFSET	0x14
+#define OMAP_TIMER_V1_STAT_OFFSET	0x18
+#define OMAP_TIMER_V1_INT_EN_OFFSET	0x1c
+
+#define OMAP_TIMER_V2_IRQSTATUS_RAW	0x24
+#define OMAP_TIMER_V2_IRQSTATUS		0x28
+#define OMAP_TIMER_V2_IRQENABLE_SET	0x2c
+#define OMAP_TIMER_V2_IRQENABLE_CLR	0x30
+
+/*
+ * The functional registers have a different base on v1 and v2 ip.
+ * These registers are offsets from timer->func_base. The func_base
+ * is samae as io_base for v1 and io_base + 0x14 for v2 ip.
+ *
+ */
+#define OMAP_TIMER_V2_FUNC_OFFSET		0x14
+
+#define _OMAP_TIMER_WAKEUP_EN_OFFSET	0x20
+#define _OMAP_TIMER_CTRL_OFFSET		0x24
+#define		OMAP_TIMER_CTRL_GPOCFG		(1 << 14)
+#define		OMAP_TIMER_CTRL_CAPTMODE	(1 << 13)
+#define		OMAP_TIMER_CTRL_PT		(1 << 12)
+#define		OMAP_TIMER_CTRL_TCM_LOWTOHIGH	(0x1 << 8)
+#define		OMAP_TIMER_CTRL_TCM_HIGHTOLOW	(0x2 << 8)
+#define		OMAP_TIMER_CTRL_TCM_BOTHEDGES	(0x3 << 8)
+#define		OMAP_TIMER_CTRL_SCPWM		(1 << 7)
+#define		OMAP_TIMER_CTRL_CE		(1 << 6) /* compare enable */
+#define		OMAP_TIMER_CTRL_PRE		(1 << 5) /* prescaler enable */
+#define		OMAP_TIMER_CTRL_PTV_SHIFT	2 /* prescaler value shift */
+#define		OMAP_TIMER_CTRL_POSTED		(1 << 2)
+#define		OMAP_TIMER_CTRL_AR		(1 << 1) /* auto-reload enable */
+#define		OMAP_TIMER_CTRL_ST		(1 << 0) /* start timer */
+#define _OMAP_TIMER_COUNTER_OFFSET	0x28
+#define _OMAP_TIMER_LOAD_OFFSET		0x2c
+#define _OMAP_TIMER_TRIGGER_OFFSET	0x30
+#define _OMAP_TIMER_WRITE_PEND_OFFSET	0x34
+#define		WP_NONE			0	/* no write pending bit */
+#define		WP_TCLR			(1 << 0)
+#define		WP_TCRR			(1 << 1)
+#define		WP_TLDR			(1 << 2)
+#define		WP_TTGR			(1 << 3)
+#define		WP_TMAR			(1 << 4)
+#define		WP_TPIR			(1 << 5)
+#define		WP_TNIR			(1 << 6)
+#define		WP_TCVR			(1 << 7)
+#define		WP_TOCR			(1 << 8)
+#define		WP_TOWR			(1 << 9)
+#define _OMAP_TIMER_MATCH_OFFSET	0x38
+#define _OMAP_TIMER_CAPTURE_OFFSET	0x3c
+#define _OMAP_TIMER_IF_CTRL_OFFSET	0x40
+#define _OMAP_TIMER_CAPTURE2_OFFSET		0x44	/* TCAR2, 34xx only */
+#define _OMAP_TIMER_TICK_POS_OFFSET		0x48	/* TPIR, 34xx only */
+#define _OMAP_TIMER_TICK_NEG_OFFSET		0x4c	/* TNIR, 34xx only */
+#define _OMAP_TIMER_TICK_COUNT_OFFSET		0x50	/* TCVR, 34xx only */
+#define _OMAP_TIMER_TICK_INT_MASK_SET_OFFSET	0x54	/* TOCR, 34xx only */
+#define _OMAP_TIMER_TICK_INT_MASK_COUNT_OFFSET	0x58	/* TOWR, 34xx only */
+
+/* register offsets with the write pending bit encoded */
+#define	WPSHIFT					16
+
+#define OMAP_TIMER_WAKEUP_EN_REG		(_OMAP_TIMER_WAKEUP_EN_OFFSET \
+							| (WP_NONE << WPSHIFT))
+
+#define OMAP_TIMER_CTRL_REG			(_OMAP_TIMER_CTRL_OFFSET \
+							| (WP_TCLR << WPSHIFT))
+
+#define OMAP_TIMER_COUNTER_REG			(_OMAP_TIMER_COUNTER_OFFSET \
+							| (WP_TCRR << WPSHIFT))
+
+#define OMAP_TIMER_LOAD_REG			(_OMAP_TIMER_LOAD_OFFSET \
+							| (WP_TLDR << WPSHIFT))
+
+#define OMAP_TIMER_TRIGGER_REG			(_OMAP_TIMER_TRIGGER_OFFSET \
+							| (WP_TTGR << WPSHIFT))
+
+#define OMAP_TIMER_WRITE_PEND_REG		(_OMAP_TIMER_WRITE_PEND_OFFSET \
+							| (WP_NONE << WPSHIFT))
+
+#define OMAP_TIMER_MATCH_REG			(_OMAP_TIMER_MATCH_OFFSET \
+							| (WP_TMAR << WPSHIFT))
+
+#define OMAP_TIMER_CAPTURE_REG			(_OMAP_TIMER_CAPTURE_OFFSET \
+							| (WP_NONE << WPSHIFT))
+
+#define OMAP_TIMER_IF_CTRL_REG			(_OMAP_TIMER_IF_CTRL_OFFSET \
+							| (WP_NONE << WPSHIFT))
+
+#define OMAP_TIMER_CAPTURE2_REG			(_OMAP_TIMER_CAPTURE2_OFFSET \
+							| (WP_NONE << WPSHIFT))
+
+#define OMAP_TIMER_TICK_POS_REG			(_OMAP_TIMER_TICK_POS_OFFSET \
+							| (WP_TPIR << WPSHIFT))
+
+#define OMAP_TIMER_TICK_NEG_REG			(_OMAP_TIMER_TICK_NEG_OFFSET \
+							| (WP_TNIR << WPSHIFT))
+
+#define OMAP_TIMER_TICK_COUNT_REG		(_OMAP_TIMER_TICK_COUNT_OFFSET \
+							| (WP_TCVR << WPSHIFT))
+
+#define OMAP_TIMER_TICK_INT_MASK_SET_REG				\
+		(_OMAP_TIMER_TICK_INT_MASK_SET_OFFSET | (WP_TOCR << WPSHIFT))
+
+#define OMAP_TIMER_TICK_INT_MASK_COUNT_REG				\
+		(_OMAP_TIMER_TICK_INT_MASK_COUNT_OFFSET | (WP_TOWR << WPSHIFT))
+
+static inline u32 __omap_dm_timer_read(struct omap_dm_timer *timer, u32 reg,
+						int posted)
+{
+	if (posted)
+		while (readl_relaxed(timer->pend) & (reg >> WPSHIFT))
+			cpu_relax();
+
+	return readl_relaxed(timer->func_base + (reg & 0xff));
+}
+
+static inline void __omap_dm_timer_write(struct omap_dm_timer *timer,
+					u32 reg, u32 val, int posted)
+{
+	if (posted)
+		while (readl_relaxed(timer->pend) & (reg >> WPSHIFT))
+			cpu_relax();
+
+	writel_relaxed(val, timer->func_base + (reg & 0xff));
+}
+
+static inline void __omap_dm_timer_init_regs(struct omap_dm_timer *timer)
+{
+	u32 tidr;
+
+	/* Assume v1 ip if bits [31:16] are zero */
+	tidr = readl_relaxed(timer->io_base);
+	if (!(tidr >> 16)) {
+		timer->revision = 1;
+		timer->irq_stat = timer->io_base + OMAP_TIMER_V1_STAT_OFFSET;
+		timer->irq_ena = timer->io_base + OMAP_TIMER_V1_INT_EN_OFFSET;
+		timer->irq_dis = timer->io_base + OMAP_TIMER_V1_INT_EN_OFFSET;
+		timer->pend = timer->io_base + _OMAP_TIMER_WRITE_PEND_OFFSET;
+		timer->func_base = timer->io_base;
+	} else {
+		timer->revision = 2;
+		timer->irq_stat = timer->io_base + OMAP_TIMER_V2_IRQSTATUS;
+		timer->irq_ena = timer->io_base + OMAP_TIMER_V2_IRQENABLE_SET;
+		timer->irq_dis = timer->io_base + OMAP_TIMER_V2_IRQENABLE_CLR;
+		timer->pend = timer->io_base +
+			_OMAP_TIMER_WRITE_PEND_OFFSET +
+				OMAP_TIMER_V2_FUNC_OFFSET;
+		timer->func_base = timer->io_base + OMAP_TIMER_V2_FUNC_OFFSET;
+	}
+}
+
+/*
+ * __omap_dm_timer_enable_posted - enables write posted mode
+ * @timer:      pointer to timer instance handle
+ *
+ * Enables the write posted mode for the timer. When posted mode is enabled
+ * writes to certain timer registers are immediately acknowledged by the
+ * internal bus and hence prevents stalling the CPU waiting for the write to
+ * complete. Enabling this feature can improve performance for writing to the
+ * timer registers.
+ */
+static inline void __omap_dm_timer_enable_posted(struct omap_dm_timer *timer)
+{
+	if (timer->posted)
+		return;
+
+	if (timer->errata & OMAP_TIMER_ERRATA_I103_I767) {
+		timer->posted = OMAP_TIMER_NONPOSTED;
+		__omap_dm_timer_write(timer, OMAP_TIMER_IF_CTRL_REG, 0, 0);
+		return;
+	}
+
+	__omap_dm_timer_write(timer, OMAP_TIMER_IF_CTRL_REG,
+			      OMAP_TIMER_CTRL_POSTED, 0);
+	timer->context.tsicr = OMAP_TIMER_CTRL_POSTED;
+	timer->posted = OMAP_TIMER_POSTED;
+}
+
+/**
+ * __omap_dm_timer_override_errata - override errata flags for a timer
+ * @timer:      pointer to timer handle
+ * @errata:	errata flags to be ignored
+ *
+ * For a given timer, override a timer errata by clearing the flags
+ * specified by the errata argument. A specific erratum should only be
+ * overridden for a timer if the timer is used in such a way the erratum
+ * has no impact.
+ */
+static inline void __omap_dm_timer_override_errata(struct omap_dm_timer *timer,
+						   u32 errata)
+{
+	timer->errata &= ~errata;
+}
+
+static inline void __omap_dm_timer_stop(struct omap_dm_timer *timer,
+					int posted, unsigned long rate)
+{
+	u32 l;
+
+	l = __omap_dm_timer_read(timer, OMAP_TIMER_CTRL_REG, posted);
+	if (l & OMAP_TIMER_CTRL_ST) {
+		l &= ~0x1;
+		__omap_dm_timer_write(timer, OMAP_TIMER_CTRL_REG, l, posted);
+#ifdef CONFIG_ARCH_OMAP2PLUS
+		/* Readback to make sure write has completed */
+		__omap_dm_timer_read(timer, OMAP_TIMER_CTRL_REG, posted);
+		/*
+		 * Wait for functional clock period x 3.5 to make sure that
+		 * timer is stopped
+		 */
+		udelay(3500000 / rate + 1);
+#endif
+	}
+
+	/* Ack possibly pending interrupt */
+	writel_relaxed(OMAP_TIMER_INT_OVERFLOW, timer->irq_stat);
+}
+
+static inline void __omap_dm_timer_load_start(struct omap_dm_timer *timer,
+						u32 ctrl, unsigned int load,
+						int posted)
+{
+	__omap_dm_timer_write(timer, OMAP_TIMER_COUNTER_REG, load, posted);
+	__omap_dm_timer_write(timer, OMAP_TIMER_CTRL_REG, ctrl, posted);
+}
+
+static inline void __omap_dm_timer_int_enable(struct omap_dm_timer *timer,
+						unsigned int value)
+{
+	writel_relaxed(value, timer->irq_ena);
+	__omap_dm_timer_write(timer, OMAP_TIMER_WAKEUP_EN_REG, value, 0);
+}
+
+static inline unsigned int
+__omap_dm_timer_read_counter(struct omap_dm_timer *timer, int posted)
+{
+	return __omap_dm_timer_read(timer, OMAP_TIMER_COUNTER_REG, posted);
+}
+
+static inline void __omap_dm_timer_write_status(struct omap_dm_timer *timer,
+						unsigned int value)
+{
+	writel_relaxed(value, timer->irq_stat);
+}
+
+#endif /* __ASM_ARCH_DMTIMER_H */

arch/arm/plat-omap/include/plat/i2c.h

@@ -0,0 +1,53 @@
+/*
+ * Helper module for board specific I2C bus registration
+ *
+ * Copyright (C) 2009 Nokia Corporation.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
+ * 02110-1301 USA
+ *
+ */
+
+#ifndef __PLAT_OMAP_I2C_H
+#define __PLAT_OMAP_I2C_H
+
+struct i2c_board_info;
+struct omap_i2c_bus_platform_data;
+
+int omap_i2c_add_bus(struct omap_i2c_bus_platform_data *i2c_pdata,
+			int bus_id);
+
+#if defined(CONFIG_I2C_OMAP) || defined(CONFIG_I2C_OMAP_MODULE)
+extern int omap_register_i2c_bus(int bus_id, u32 clkrate,
+				 struct i2c_board_info const *info,
+				 unsigned len);
+extern int omap_register_i2c_bus_cmdline(void);
+#else
+static inline int omap_register_i2c_bus(int bus_id, u32 clkrate,
+				 struct i2c_board_info const *info,
+				 unsigned len)
+{
+	return 0;
+}
+
+static inline int omap_register_i2c_bus_cmdline(void)
+{
+	return 0;
+}
+#endif
+
+struct omap_hwmod;
+int omap_i2c_reset(struct omap_hwmod *oh);
+
+#endif /* __PLAT_OMAP_I2C_H */

arch/arm/plat-omap/include/plat/sram.h

@@ -0,0 +1,16 @@
+int omap_sram_init(void);
+
+void omap_map_sram(unsigned long start, unsigned long size,
+			unsigned long skip, int cached);
+void omap_sram_reset(void);
+
+extern void *omap_sram_push_address(unsigned long size);
+
+/* Macro to push a function to the internal SRAM, using the fncpy API */
+#define omap_sram_push(funcp, size) ({				\
+	typeof(&(funcp)) _res = NULL;				\
+	void *_sram_address = omap_sram_push_address(size);	\
+	if (_sram_address)					\
+		_res = fncpy(_sram_address, &(funcp), size);	\
+	_res;							\
+})

arch/arm/plat-omap/sram.c

@@ -0,0 +1,98 @@
+/*
+ * linux/arch/arm/plat-omap/sram.c
+ *
+ * OMAP SRAM detection and management
+ *
+ * Copyright (C) 2005 Nokia Corporation
+ * Written by Tony Lindgren <tony@atomide.com>
+ *
+ * Copyright (C) 2009-2012 Texas Instruments
+ * Added OMAP4/5 support - Santosh Shilimkar <santosh.shilimkar@ti.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+#undef DEBUG
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/io.h>
+
+#include <asm/fncpy.h>
+#include <asm/tlb.h>
+#include <asm/cacheflush.h>
+
+#include <asm/mach/map.h>
+
+#include <plat/sram.h>
+
+#define ROUND_DOWN(value,boundary)	((value) & (~((boundary)-1)))
+
+static void __iomem *omap_sram_base;
+static unsigned long omap_sram_skip;
+static unsigned long omap_sram_size;
+static void __iomem *omap_sram_ceil;
+
+/*
+ * Memory allocator for SRAM: calculates the new ceiling address
+ * for pushing a function using the fncpy API.
+ *
+ * Note that fncpy requires the returned address to be aligned
+ * to an 8-byte boundary.
+ */
+void *omap_sram_push_address(unsigned long size)
+{
+	unsigned long available, new_ceil = (unsigned long)omap_sram_ceil;
+
+	available = omap_sram_ceil - (omap_sram_base + omap_sram_skip);
+
+	if (size > available) {
+		pr_err("Not enough space in SRAM\n");
+		return NULL;
+	}
+
+	new_ceil -= size;
+	new_ceil = ROUND_DOWN(new_ceil, FNCPY_ALIGN);
+	omap_sram_ceil = IOMEM(new_ceil);
+
+	return (void *)omap_sram_ceil;
+}
+
+/*
+ * The SRAM context is lost during off-idle and stack
+ * needs to be reset.
+ */
+void omap_sram_reset(void)
+{
+	omap_sram_ceil = omap_sram_base + omap_sram_size;
+}
+
+/*
+ * Note that we cannot use ioremap for SRAM, as clock init needs SRAM early.
+ */
+void __init omap_map_sram(unsigned long start, unsigned long size,
+				 unsigned long skip, int cached)
+{
+	if (size == 0)
+		return;
+
+	start = ROUND_DOWN(start, PAGE_SIZE);
+	omap_sram_size = size;
+	omap_sram_skip = skip;
+	omap_sram_base = __arm_ioremap_exec(start, size, cached);
+	if (!omap_sram_base) {
+		pr_err("SRAM: Could not map\n");
+		return;
+	}
+
+	omap_sram_reset();
+
+	/*
+	 * Looks like we need to preserve some bootloader code at the
+	 * beginning of SRAM for jumping to flash for reboot to work...
+	 */
+	memset_io(omap_sram_base + omap_sram_skip, 0,
+		  omap_sram_size - omap_sram_skip);
+}

arch/arm/plat-samsung/Kconfig

@@ -0,0 +1,309 @@
+# arch/arm/plat-samsung/Kconfig
+#
+# Copyright 2009 Simtec Electronics
+#
+# Licensed under GPLv2
+
+config PLAT_SAMSUNG
+	bool
+	depends on PLAT_S3C24XX || ARCH_S3C64XX || ARCH_EXYNOS || ARCH_S5PV210
+	default y
+	select GENERIC_IRQ_CHIP
+	select NO_IOPORT_MAP
+	help
+	  Base platform code for all Samsung SoC based systems
+
+config SAMSUNG_PM
+	bool
+	depends on PM && (PLAT_S3C24XX || ARCH_S3C64XX)
+	default y
+	help
+	  Base platform power management code for samsung code
+
+if PLAT_SAMSUNG
+menu "Samsung Common options"
+
+# boot configurations
+
+comment "Boot options"
+
+config S3C_LOWLEVEL_UART_PORT
+	int "S3C UART to use for low-level messages"
+	depends on ARCH_S3C64XX
+	default 0
+	help
+	  Choice of which UART port to use for the low-level messages,
+	  such as the `Uncompressing...` at start time. The value of
+	  this configuration should be between zero and two. The port
+	  must have been initialised by the boot-loader before use.
+
+config SAMSUNG_ATAGS
+	def_bool n
+	depends on ATAGS
+	help
+	   This option enables ATAGS based boot support code for
+	   Samsung platforms, including static platform devices, legacy
+	   clock, timer and interrupt initialization, etc.
+
+	   Platforms that support only DT based boot need not to select
+	   this option.
+
+if SAMSUNG_ATAGS
+
+config S3C_GPIO_SPACE
+	int "Space between gpio banks"
+	default 0
+	help
+	  Add a number of spare GPIO entries between each bank for debugging
+	  purposes. This allows any problems where an counter overflows from
+	  one bank to another to be caught, at the expense of using a little
+	  more memory.
+
+config S3C_GPIO_TRACK
+	bool
+	help
+	  Internal configuration option to enable the s3c specific gpio
+	  chip tracking if the platform requires it.
+
+# ADC driver
+
+config S3C_ADC
+	bool "ADC common driver support"
+	depends on !ARCH_MULTIPLATFORM
+	help
+	  Core support for the ADC block found in the Samsung SoC systems
+	  for drivers such as the touchscreen and hwmon to use to share
+	  this resource.
+
+# device definitions to compile in
+
+config S3C_DEV_HSMMC
+	bool
+	help
+	  Compile in platform device definitions for HSMMC code
+
+config S3C_DEV_HSMMC1
+	bool
+	help
+	  Compile in platform device definitions for HSMMC channel 1
+
+config S3C_DEV_HSMMC2
+	bool
+	help
+	  Compile in platform device definitions for HSMMC channel 2
+
+config S3C_DEV_HSMMC3
+	bool
+	help
+	  Compile in platform device definitions for HSMMC channel 3
+
+config S3C_DEV_HWMON
+	bool
+	help
+	    Compile in platform device definitions for HWMON
+
+config S3C_DEV_I2C1
+	bool
+	help
+	  Compile in platform device definitions for I2C channel 1
+
+config S3C_DEV_I2C2
+	bool
+	help
+	  Compile in platform device definitions for I2C channel 2
+
+config S3C_DEV_I2C3
+	bool
+	help
+	  Compile in platform device definition for I2C controller 3
+
+config S3C_DEV_I2C4
+	bool
+	help
+	  Compile in platform device definition for I2C controller 4
+
+config S3C_DEV_I2C5
+	bool
+	help
+	  Compile in platform device definition for I2C controller 5
+
+config S3C_DEV_I2C6
+	bool
+	help
+	  Compile in platform device definition for I2C controller 6
+
+config S3C_DEV_I2C7
+	bool
+	help
+	  Compile in platform device definition for I2C controller 7
+
+config S3C_DEV_FB
+	bool
+	help
+	  Compile in platform device definition for framebuffer
+
+config S3C_DEV_USB_HOST
+	bool
+	help
+	  Compile in platform device definition for USB host.
+
+config S3C_DEV_USB_HSOTG
+	bool
+	help
+	  Compile in platform device definition for USB high-speed OtG
+
+config S3C_DEV_WDT
+	bool
+	default y if ARCH_S3C24XX
+	help
+	  Complie in platform device definition for Watchdog Timer
+
+config S3C_DEV_NAND
+	bool
+	help
+	  Compile in platform device definition for NAND controller
+
+config S3C_DEV_ONENAND
+	bool
+	help
+	  Compile in platform device definition for OneNAND controller
+
+config S3C_DEV_RTC
+	bool
+	help
+	  Complie in platform device definition for RTC
+
+config SAMSUNG_DEV_ADC
+	bool
+	help
+	  Compile in platform device definition for ADC controller
+
+config SAMSUNG_DEV_IDE
+	bool
+	help
+	  Compile in platform device definitions for IDE
+
+config S3C64XX_DEV_SPI0
+	bool
+	help
+	  Compile in platform device definitions for S3C64XX's type
+	  SPI controller 0
+
+config S3C64XX_DEV_SPI1
+	bool
+	help
+	  Compile in platform device definitions for S3C64XX's type
+	  SPI controller 1
+
+config S3C64XX_DEV_SPI2
+	bool
+	help
+	  Compile in platform device definitions for S3C64XX's type
+	  SPI controller 2
+
+config SAMSUNG_DEV_TS
+	bool
+	help
+	    Common in platform device definitions for touchscreen device
+
+config SAMSUNG_DEV_KEYPAD
+	bool
+	help
+	  Compile in platform device definitions for keypad
+
+config SAMSUNG_DEV_PWM
+	bool
+	default y if ARCH_S3C24XX
+	help
+	  Compile in platform device definition for PWM Timer
+
+config S3C24XX_PWM
+	bool "PWM device support"
+	select PWM
+	select PWM_SAMSUNG
+	help
+	  Support for exporting the PWM timer blocks via the pwm device
+	  system
+
+config GPIO_SAMSUNG
+	def_bool y
+
+config SAMSUNG_PM_GPIO
+	bool
+	default y if GPIO_SAMSUNG && PM
+	help
+	  Include legacy GPIO power management code for platforms not using
+	  pinctrl-samsung driver.
+endif
+
+comment "Power management"
+
+config SAMSUNG_PM_DEBUG
+	bool "Samsung PM Suspend debug"
+	depends on PM && DEBUG_KERNEL
+	depends on DEBUG_EXYNOS_UART || DEBUG_S3C24XX_UART || DEBUG_S3C2410_UART
+	help
+	  Say Y here if you want verbose debugging from the PM Suspend and
+	  Resume code. See <file:Documentation/arm/Samsung-S3C24XX/Suspend.txt>
+	  for more information.
+
+config S3C_PM_DEBUG_LED_SMDK
+       bool "SMDK LED suspend/resume debugging"
+       depends on PM && (MACH_SMDK6410)
+       help
+         Say Y here to enable the use of the SMDK LEDs on the baseboard
+	 for debugging of the state of the suspend and resume process.
+
+	 Note, this currently only works for S3C64XX based SMDK boards.
+
+config SAMSUNG_PM_CHECK
+	bool "S3C2410 PM Suspend Memory CRC"
+	depends on PM
+	select CRC32
+	help
+	  Enable the PM code's memory area checksum over sleep. This option
+	  will generate CRCs of all blocks of memory, and store them before
+	  going to sleep. The blocks are then checked on resume for any
+	  errors.
+
+	  Note, this can take several seconds depending on memory size
+	  and CPU speed.
+
+	  See <file:Documentation/arm/Samsung-S3C24XX/Suspend.txt>
+
+config SAMSUNG_PM_CHECK_CHUNKSIZE
+	int "S3C2410 PM Suspend CRC Chunksize (KiB)"
+	depends on PM && SAMSUNG_PM_CHECK
+	default 64
+	help
+	  Set the chunksize in Kilobytes of the CRC for checking memory
+	  corruption over suspend and resume. A smaller value will mean that
+	  the CRC data block will take more memory, but wil identify any
+	  faults with better precision.
+
+	  See <file:Documentation/arm/Samsung-S3C24XX/Suspend.txt>
+
+config SAMSUNG_WAKEMASK
+	bool
+	depends on PM
+	help
+	  Compile support for wakeup-mask controls found on the S3C6400
+	  and above. This code allows a set of interrupt to wakeup-mask
+	  mappings. See <plat/wakeup-mask.h>
+
+config SAMSUNG_WDT_RESET
+	bool
+	help
+	  Compile support for system restart by triggering watchdog reset.
+	  Used on SoCs that do not provide dedicated reset control.
+
+config DEBUG_S3C_UART
+	depends on PLAT_SAMSUNG
+	int
+	default "0" if DEBUG_S3C_UART0
+	default "1" if DEBUG_S3C_UART1
+	default "2" if DEBUG_S3C_UART2
+	default "3" if DEBUG_S3C_UART3
+
+endmenu
+endif

arch/arm/plat-samsung/Makefile

@@ -0,0 +1,37 @@
+# arch/arm/plat-samsung/Makefile
+#
+# Copyright 2009 Simtec Electronics
+#
+# Licensed under GPLv2
+
+ccflags-$(CONFIG_ARCH_S3C64XX) := -I$(srctree)/arch/arm/mach-s3c64xx/include
+ccflags-$(CONFIG_ARCH_MULTIPLATFORM) += -I$(srctree)/$(src)/include
+
+# Objects we always build independent of SoC choice
+
+obj-y				+= init.o cpu.o
+
+# ADC
+
+obj-$(CONFIG_S3C_ADC)	+= adc.o
+
+# devices
+
+obj-$(CONFIG_SAMSUNG_ATAGS)	+= platformdata.o
+
+obj-$(CONFIG_SAMSUNG_ATAGS)	+= devs.o
+obj-$(CONFIG_SAMSUNG_ATAGS)	+= dev-uart.o
+
+obj-$(CONFIG_GPIO_SAMSUNG)     += gpio-samsung.o
+
+# PM support
+
+obj-$(CONFIG_PM_SLEEP)		+= pm-common.o
+obj-$(CONFIG_EXYNOS_CPU_SUSPEND) += pm-common.o
+obj-$(CONFIG_SAMSUNG_PM)	+= pm.o
+obj-$(CONFIG_SAMSUNG_PM_GPIO)	+= pm-gpio.o
+obj-$(CONFIG_SAMSUNG_PM_CHECK)	+= pm-check.o
+obj-$(CONFIG_SAMSUNG_PM_DEBUG)	+= pm-debug.o
+
+obj-$(CONFIG_SAMSUNG_WAKEMASK)	+= wakeup-mask.o
+obj-$(CONFIG_SAMSUNG_WDT_RESET)	+= watchdog-reset.o

arch/arm/plat-samsung/adc.c

@@ -0,0 +1,523 @@
+/* arch/arm/plat-samsung/adc.c
+ *
+ * Copyright (c) 2008 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>, <ben-linux@fluff.org>
+ *
+ * Samsung ADC device core
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License.
+*/
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+#include <linux/sched.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/err.h>
+#include <linux/clk.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/regulator/consumer.h>
+
+#include <plat/regs-adc.h>
+#include <plat/adc.h>
+
+/* This driver is designed to control the usage of the ADC block between
+ * the touchscreen and any other drivers that may need to use it, such as
+ * the hwmon driver.
+ *
+ * Priority will be given to the touchscreen driver, but as this itself is
+ * rate limited it should not starve other requests which are processed in
+ * order that they are received.
+ *
+ * Each user registers to get a client block which uniquely identifies it
+ * and stores information such as the necessary functions to callback when
+ * action is required.
+ */
+
+enum s3c_cpu_type {
+	TYPE_ADCV1, /* S3C24XX */
+	TYPE_ADCV11, /* S3C2443 */
+	TYPE_ADCV12, /* S3C2416, S3C2450 */
+	TYPE_ADCV2, /* S3C64XX */
+	TYPE_ADCV3, /* S5PV210, S5PC110, EXYNOS4210 */
+};
+
+struct s3c_adc_client {
+	struct platform_device	*pdev;
+	struct list_head	 pend;
+	wait_queue_head_t	*wait;
+
+	unsigned int		 nr_samples;
+	int			 result;
+	unsigned char		 is_ts;
+	unsigned char		 channel;
+
+	void	(*select_cb)(struct s3c_adc_client *c, unsigned selected);
+	void	(*convert_cb)(struct s3c_adc_client *c,
+			      unsigned val1, unsigned val2,
+			      unsigned *samples_left);
+};
+
+struct adc_device {
+	struct platform_device	*pdev;
+	struct platform_device	*owner;
+	struct clk		*clk;
+	struct s3c_adc_client	*cur;
+	struct s3c_adc_client	*ts_pend;
+	void __iomem		*regs;
+	spinlock_t		 lock;
+
+	unsigned int		 prescale;
+
+	int			 irq;
+	struct regulator	*vdd;
+};
+
+static struct adc_device *adc_dev;
+
+static LIST_HEAD(adc_pending);	/* protected by adc_device.lock */
+
+#define adc_dbg(_adc, msg...) dev_dbg(&(_adc)->pdev->dev, msg)
+
+static inline void s3c_adc_convert(struct adc_device *adc)
+{
+	unsigned con = readl(adc->regs + S3C2410_ADCCON);
+
+	con |= S3C2410_ADCCON_ENABLE_START;
+	writel(con, adc->regs + S3C2410_ADCCON);
+}
+
+static inline void s3c_adc_select(struct adc_device *adc,
+				  struct s3c_adc_client *client)
+{
+	unsigned con = readl(adc->regs + S3C2410_ADCCON);
+	enum s3c_cpu_type cpu = platform_get_device_id(adc->pdev)->driver_data;
+
+	client->select_cb(client, 1);
+
+	if (cpu == TYPE_ADCV1 || cpu == TYPE_ADCV2)
+		con &= ~S3C2410_ADCCON_MUXMASK;
+	con &= ~S3C2410_ADCCON_STDBM;
+	con &= ~S3C2410_ADCCON_STARTMASK;
+
+	if (!client->is_ts) {
+		if (cpu == TYPE_ADCV3)
+			writel(client->channel & 0xf, adc->regs + S5P_ADCMUX);
+		else if (cpu == TYPE_ADCV11 || cpu == TYPE_ADCV12)
+			writel(client->channel & 0xf,
+						adc->regs + S3C2443_ADCMUX);
+		else
+			con |= S3C2410_ADCCON_SELMUX(client->channel);
+	}
+
+	writel(con, adc->regs + S3C2410_ADCCON);
+}
+
+static void s3c_adc_dbgshow(struct adc_device *adc)
+{
+	adc_dbg(adc, "CON=%08x, TSC=%08x, DLY=%08x\n",
+		readl(adc->regs + S3C2410_ADCCON),
+		readl(adc->regs + S3C2410_ADCTSC),
+		readl(adc->regs + S3C2410_ADCDLY));
+}
+
+static void s3c_adc_try(struct adc_device *adc)
+{
+	struct s3c_adc_client *next = adc->ts_pend;
+
+	if (!next && !list_empty(&adc_pending)) {
+		next = list_first_entry(&adc_pending,
+					struct s3c_adc_client, pend);
+		list_del(&next->pend);
+	} else
+		adc->ts_pend = NULL;
+
+	if (next) {
+		adc_dbg(adc, "new client is %p\n", next);
+		adc->cur = next;
+		s3c_adc_select(adc, next);
+		s3c_adc_convert(adc);
+		s3c_adc_dbgshow(adc);
+	}
+}
+
+int s3c_adc_start(struct s3c_adc_client *client,
+		  unsigned int channel, unsigned int nr_samples)
+{
+	struct adc_device *adc = adc_dev;
+	unsigned long flags;
+
+	if (!adc) {
+		printk(KERN_ERR "%s: failed to find adc\n", __func__);
+		return -EINVAL;
+	}
+
+	spin_lock_irqsave(&adc->lock, flags);
+
+	if (client->is_ts && adc->ts_pend) {
+		spin_unlock_irqrestore(&adc->lock, flags);
+		return -EAGAIN;
+	}
+
+	client->channel = channel;
+	client->nr_samples = nr_samples;
+
+	if (client->is_ts)
+		adc->ts_pend = client;
+	else
+		list_add_tail(&client->pend, &adc_pending);
+
+	if (!adc->cur)
+		s3c_adc_try(adc);
+
+	spin_unlock_irqrestore(&adc->lock, flags);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(s3c_adc_start);
+
+static void s3c_convert_done(struct s3c_adc_client *client,
+			     unsigned v, unsigned u, unsigned *left)
+{
+	client->result = v;
+	wake_up(client->wait);
+}
+
+int s3c_adc_read(struct s3c_adc_client *client, unsigned int ch)
+{
+	DECLARE_WAIT_QUEUE_HEAD_ONSTACK(wake);
+	int ret;
+
+	client->convert_cb = s3c_convert_done;
+	client->wait = &wake;
+	client->result = -1;
+
+	ret = s3c_adc_start(client, ch, 1);
+	if (ret < 0)
+		goto err;
+
+	ret = wait_event_timeout(wake, client->result >= 0, HZ / 2);
+	if (client->result < 0) {
+		ret = -ETIMEDOUT;
+		goto err;
+	}
+
+	client->convert_cb = NULL;
+	return client->result;
+
+err:
+	return ret;
+}
+EXPORT_SYMBOL_GPL(s3c_adc_read);
+
+static void s3c_adc_default_select(struct s3c_adc_client *client,
+				   unsigned select)
+{
+}
+
+struct s3c_adc_client *s3c_adc_register(struct platform_device *pdev,
+					void (*select)(struct s3c_adc_client *client,
+						       unsigned int selected),
+					void (*conv)(struct s3c_adc_client *client,
+						     unsigned d0, unsigned d1,
+						     unsigned *samples_left),
+					unsigned int is_ts)
+{
+	struct s3c_adc_client *client;
+
+	WARN_ON(!pdev);
+
+	if (!select)
+		select = s3c_adc_default_select;
+
+	if (!pdev)
+		return ERR_PTR(-EINVAL);
+
+	client = kzalloc(sizeof(struct s3c_adc_client), GFP_KERNEL);
+	if (!client) {
+		dev_err(&pdev->dev, "no memory for adc client\n");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	client->pdev = pdev;
+	client->is_ts = is_ts;
+	client->select_cb = select;
+	client->convert_cb = conv;
+
+	return client;
+}
+EXPORT_SYMBOL_GPL(s3c_adc_register);
+
+void s3c_adc_release(struct s3c_adc_client *client)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adc_dev->lock, flags);
+
+	/* We should really check that nothing is in progress. */
+	if (adc_dev->cur == client)
+		adc_dev->cur = NULL;
+	if (adc_dev->ts_pend == client)
+		adc_dev->ts_pend = NULL;
+	else {
+		struct list_head *p, *n;
+		struct s3c_adc_client *tmp;
+
+		list_for_each_safe(p, n, &adc_pending) {
+			tmp = list_entry(p, struct s3c_adc_client, pend);
+			if (tmp == client)
+				list_del(&tmp->pend);
+		}
+	}
+
+	if (adc_dev->cur == NULL)
+		s3c_adc_try(adc_dev);
+
+	spin_unlock_irqrestore(&adc_dev->lock, flags);
+	kfree(client);
+}
+EXPORT_SYMBOL_GPL(s3c_adc_release);
+
+static irqreturn_t s3c_adc_irq(int irq, void *pw)
+{
+	struct adc_device *adc = pw;
+	struct s3c_adc_client *client = adc->cur;
+	enum s3c_cpu_type cpu = platform_get_device_id(adc->pdev)->driver_data;
+	unsigned data0, data1;
+
+	if (!client) {
+		dev_warn(&adc->pdev->dev, "%s: no adc pending\n", __func__);
+		goto exit;
+	}
+
+	data0 = readl(adc->regs + S3C2410_ADCDAT0);
+	data1 = readl(adc->regs + S3C2410_ADCDAT1);
+	adc_dbg(adc, "read %d: 0x%04x, 0x%04x\n", client->nr_samples, data0, data1);
+
+	client->nr_samples--;
+
+	if (cpu == TYPE_ADCV1 || cpu == TYPE_ADCV11) {
+		data0 &= 0x3ff;
+		data1 &= 0x3ff;
+	} else {
+		/* S3C2416/S3C64XX/S5P ADC resolution is 12-bit */
+		data0 &= 0xfff;
+		data1 &= 0xfff;
+	}
+
+	if (client->convert_cb)
+		(client->convert_cb)(client, data0, data1, &client->nr_samples);
+
+	if (client->nr_samples > 0) {
+		/* fire another conversion for this */
+
+		client->select_cb(client, 1);
+		s3c_adc_convert(adc);
+	} else {
+		spin_lock(&adc->lock);
+		(client->select_cb)(client, 0);
+		adc->cur = NULL;
+
+		s3c_adc_try(adc);
+		spin_unlock(&adc->lock);
+	}
+
+exit:
+	if (cpu == TYPE_ADCV2 || cpu == TYPE_ADCV3) {
+		/* Clear ADC interrupt */
+		writel(0, adc->regs + S3C64XX_ADCCLRINT);
+	}
+	return IRQ_HANDLED;
+}
+
+static int s3c_adc_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct adc_device *adc;
+	struct resource *regs;
+	enum s3c_cpu_type cpu = platform_get_device_id(pdev)->driver_data;
+	int ret;
+	unsigned tmp;
+
+	adc = devm_kzalloc(dev, sizeof(struct adc_device), GFP_KERNEL);
+	if (adc == NULL) {
+		dev_err(dev, "failed to allocate adc_device\n");
+		return -ENOMEM;
+	}
+
+	spin_lock_init(&adc->lock);
+
+	adc->pdev = pdev;
+	adc->prescale = S3C2410_ADCCON_PRSCVL(49);
+
+	adc->vdd = devm_regulator_get(dev, "vdd");
+	if (IS_ERR(adc->vdd)) {
+		dev_err(dev, "operating without regulator \"vdd\" .\n");
+		return PTR_ERR(adc->vdd);
+	}
+
+	adc->irq = platform_get_irq(pdev, 1);
+	if (adc->irq <= 0) {
+		dev_err(dev, "failed to get adc irq\n");
+		return -ENOENT;
+	}
+
+	ret = devm_request_irq(dev, adc->irq, s3c_adc_irq, 0, dev_name(dev),
+				adc);
+	if (ret < 0) {
+		dev_err(dev, "failed to attach adc irq\n");
+		return ret;
+	}
+
+	adc->clk = devm_clk_get(dev, "adc");
+	if (IS_ERR(adc->clk)) {
+		dev_err(dev, "failed to get adc clock\n");
+		return PTR_ERR(adc->clk);
+	}
+
+	regs = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	adc->regs = devm_ioremap_resource(dev, regs);
+	if (IS_ERR(adc->regs))
+		return PTR_ERR(adc->regs);
+
+	ret = regulator_enable(adc->vdd);
+	if (ret)
+		return ret;
+
+	clk_prepare_enable(adc->clk);
+
+	tmp = adc->prescale | S3C2410_ADCCON_PRSCEN;
+
+	/* Enable 12-bit ADC resolution */
+	if (cpu == TYPE_ADCV12)
+		tmp |= S3C2416_ADCCON_RESSEL;
+	if (cpu == TYPE_ADCV2 || cpu == TYPE_ADCV3)
+		tmp |= S3C64XX_ADCCON_RESSEL;
+
+	writel(tmp, adc->regs + S3C2410_ADCCON);
+
+	dev_info(dev, "attached adc driver\n");
+
+	platform_set_drvdata(pdev, adc);
+	adc_dev = adc;
+
+	return 0;
+}
+
+static int s3c_adc_remove(struct platform_device *pdev)
+{
+	struct adc_device *adc = platform_get_drvdata(pdev);
+
+	clk_disable_unprepare(adc->clk);
+	regulator_disable(adc->vdd);
+
+	return 0;
+}
+
+#ifdef CONFIG_PM
+static int s3c_adc_suspend(struct device *dev)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct adc_device *adc = platform_get_drvdata(pdev);
+	unsigned long flags;
+	u32 con;
+
+	spin_lock_irqsave(&adc->lock, flags);
+
+	con = readl(adc->regs + S3C2410_ADCCON);
+	con |= S3C2410_ADCCON_STDBM;
+	writel(con, adc->regs + S3C2410_ADCCON);
+
+	disable_irq(adc->irq);
+	spin_unlock_irqrestore(&adc->lock, flags);
+	clk_disable(adc->clk);
+	regulator_disable(adc->vdd);
+
+	return 0;
+}
+
+static int s3c_adc_resume(struct device *dev)
+{
+	struct platform_device *pdev = to_platform_device(dev);
+	struct adc_device *adc = platform_get_drvdata(pdev);
+	enum s3c_cpu_type cpu = platform_get_device_id(pdev)->driver_data;
+	int ret;
+	unsigned long tmp;
+
+	ret = regulator_enable(adc->vdd);
+	if (ret)
+		return ret;
+	clk_enable(adc->clk);
+	enable_irq(adc->irq);
+
+	tmp = adc->prescale | S3C2410_ADCCON_PRSCEN;
+
+	/* Enable 12-bit ADC resolution */
+	if (cpu == TYPE_ADCV12)
+		tmp |= S3C2416_ADCCON_RESSEL;
+	if (cpu == TYPE_ADCV2 || cpu == TYPE_ADCV3)
+		tmp |= S3C64XX_ADCCON_RESSEL;
+
+	writel(tmp, adc->regs + S3C2410_ADCCON);
+
+	return 0;
+}
+
+#else
+#define s3c_adc_suspend NULL
+#define s3c_adc_resume NULL
+#endif
+
+static const struct platform_device_id s3c_adc_driver_ids[] = {
+	{
+		.name           = "s3c24xx-adc",
+		.driver_data    = TYPE_ADCV1,
+	}, {
+		.name		= "s3c2443-adc",
+		.driver_data	= TYPE_ADCV11,
+	}, {
+		.name		= "s3c2416-adc",
+		.driver_data	= TYPE_ADCV12,
+	}, {
+		.name           = "s3c64xx-adc",
+		.driver_data    = TYPE_ADCV2,
+	}, {
+		.name		= "samsung-adc-v3",
+		.driver_data	= TYPE_ADCV3,
+	},
+	{ }
+};
+MODULE_DEVICE_TABLE(platform, s3c_adc_driver_ids);
+
+static const struct dev_pm_ops adc_pm_ops = {
+	.suspend	= s3c_adc_suspend,
+	.resume		= s3c_adc_resume,
+};
+
+static struct platform_driver s3c_adc_driver = {
+	.id_table	= s3c_adc_driver_ids,
+	.driver		= {
+		.name	= "s3c-adc",
+		.pm	= &adc_pm_ops,
+	},
+	.probe		= s3c_adc_probe,
+	.remove		= s3c_adc_remove,
+};
+
+static int __init adc_init(void)
+{
+	int ret;
+
+	ret = platform_driver_register(&s3c_adc_driver);
+	if (ret)
+		printk(KERN_ERR "%s: failed to add adc driver\n", __func__);
+
+	return ret;
+}
+
+module_init(adc_init);

arch/arm/plat-samsung/cpu.c

@@ -0,0 +1,53 @@
+/* linux/arch/arm/plat-samsung/cpu.c
+ *
+ * Copyright (c) 2009-2011 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com
+ *
+ * Samsung CPU Support
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/io.h>
+
+#include <plat/map-base.h>
+#include <plat/cpu.h>
+
+unsigned long samsung_cpu_id;
+static unsigned int samsung_cpu_rev;
+
+unsigned int samsung_rev(void)
+{
+	return samsung_cpu_rev;
+}
+EXPORT_SYMBOL(samsung_rev);
+
+void __init s3c64xx_init_cpu(void)
+{
+	samsung_cpu_id = readl_relaxed(S3C_VA_SYS + 0x118);
+	if (!samsung_cpu_id) {
+		/*
+		 * S3C6400 has the ID register in a different place,
+		 * and needs a write before it can be read.
+		 */
+		writel_relaxed(0x0, S3C_VA_SYS + 0xA1C);
+		samsung_cpu_id = readl_relaxed(S3C_VA_SYS + 0xA1C);
+	}
+
+	samsung_cpu_rev = 0;
+
+	pr_info("Samsung CPU ID: 0x%08lx\n", samsung_cpu_id);
+}
+
+void __init s5p_init_cpu(const void __iomem *cpuid_addr)
+{
+	samsung_cpu_id = readl_relaxed(cpuid_addr);
+	samsung_cpu_rev = samsung_cpu_id & 0xFF;
+
+	pr_info("Samsung CPU ID: 0x%08lx\n", samsung_cpu_id);
+}

arch/arm/plat-samsung/dev-uart.c

@@ -0,0 +1,46 @@
+/* linux/arch/arm/plat-samsung/dev-uart.c
+ *	originally from arch/arm/plat-s3c24xx/devs.c
+ *x
+ * Copyright (c) 2004 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * Base S3C24XX platform device definitions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+*/
+
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+
+#include <plat/devs.h>
+
+/* uart devices */
+
+static struct platform_device s3c24xx_uart_device0 = {
+	.id		= 0,
+};
+
+static struct platform_device s3c24xx_uart_device1 = {
+	.id		= 1,
+};
+
+static struct platform_device s3c24xx_uart_device2 = {
+	.id		= 2,
+};
+
+static struct platform_device s3c24xx_uart_device3 = {
+	.id		= 3,
+};
+
+struct platform_device *s3c24xx_uart_src[4] = {
+	&s3c24xx_uart_device0,
+	&s3c24xx_uart_device1,
+	&s3c24xx_uart_device2,
+	&s3c24xx_uart_device3,
+};
+
+struct platform_device *s3c24xx_uart_devs[4] = {
+};

arch/arm/plat-samsung/devs.c

@@ -0,0 +1,1226 @@
+/* linux/arch/arm/plat-samsung/devs.c
+ *
+ * Copyright (c) 2011 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com
+ *
+ * Base SAMSUNG platform device definitions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/amba/pl330.h>
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/interrupt.h>
+#include <linux/list.h>
+#include <linux/timer.h>
+#include <linux/init.h>
+#include <linux/serial_core.h>
+#include <linux/serial_s3c.h>
+#include <linux/platform_device.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/string.h>
+#include <linux/dma-mapping.h>
+#include <linux/fb.h>
+#include <linux/gfp.h>
+#include <linux/mtd/mtd.h>
+#include <linux/mtd/onenand.h>
+#include <linux/mtd/partitions.h>
+#include <linux/mmc/host.h>
+#include <linux/ioport.h>
+#include <linux/sizes.h>
+#include <linux/platform_data/s3c-hsudc.h>
+#include <linux/platform_data/s3c-hsotg.h>
+#include <linux/platform_data/dma-s3c24xx.h>
+
+#include <linux/platform_data/media/s5p_hdmi.h>
+
+#include <asm/irq.h>
+#include <asm/mach/arch.h>
+#include <asm/mach/map.h>
+#include <asm/mach/irq.h>
+
+#include <mach/dma.h>
+#include <mach/irqs.h>
+#include <mach/map.h>
+
+#include <plat/cpu.h>
+#include <plat/devs.h>
+#include <plat/adc.h>
+#include <linux/platform_data/ata-samsung_cf.h>
+#include <plat/fb.h>
+#include <plat/fb-s3c2410.h>
+#include <linux/platform_data/hwmon-s3c.h>
+#include <linux/platform_data/i2c-s3c2410.h>
+#include <plat/keypad.h>
+#include <linux/platform_data/mmc-s3cmci.h>
+#include <linux/platform_data/mtd-nand-s3c2410.h>
+#include <plat/pwm-core.h>
+#include <plat/sdhci.h>
+#include <linux/platform_data/touchscreen-s3c2410.h>
+#include <linux/platform_data/usb-s3c2410_udc.h>
+#include <linux/platform_data/usb-ohci-s3c2410.h>
+#include <plat/usb-phy.h>
+#include <plat/regs-spi.h>
+#include <linux/platform_data/asoc-s3c.h>
+#include <linux/platform_data/spi-s3c64xx.h>
+
+#define samsung_device_dma_mask (*((u64[]) { DMA_BIT_MASK(32) }))
+
+/* AC97 */
+#ifdef CONFIG_CPU_S3C2440
+static struct resource s3c_ac97_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C2440_PA_AC97, S3C2440_SZ_AC97),
+	[1] = DEFINE_RES_IRQ(IRQ_S3C244X_AC97),
+};
+
+static struct s3c_audio_pdata s3c_ac97_pdata = {
+#ifdef CONFIG_S3C24XX_DMAC
+	.dma_filter = s3c24xx_dma_filter,
+#endif
+	.dma_playback = (void *)DMACH_PCM_OUT,
+	.dma_capture = (void *)DMACH_PCM_IN,
+	.dma_capture_mic = (void *)DMACH_MIC_IN,
+};
+
+struct platform_device s3c_device_ac97 = {
+	.name		= "samsung-ac97",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_ac97_resource),
+	.resource	= s3c_ac97_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+		.platform_data		= &s3c_ac97_pdata,
+	}
+};
+#endif /* CONFIG_CPU_S3C2440 */
+
+/* ADC */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_adc_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_ADC, S3C24XX_SZ_ADC),
+	[1] = DEFINE_RES_IRQ(IRQ_TC),
+	[2] = DEFINE_RES_IRQ(IRQ_ADC),
+};
+
+struct platform_device s3c_device_adc = {
+	.name		= "s3c24xx-adc",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_adc_resource),
+	.resource	= s3c_adc_resource,
+};
+#endif /* CONFIG_PLAT_S3C24XX */
+
+#if defined(CONFIG_SAMSUNG_DEV_ADC)
+static struct resource s3c_adc_resource[] = {
+	[0] = DEFINE_RES_MEM(SAMSUNG_PA_ADC, SZ_256),
+	[1] = DEFINE_RES_IRQ(IRQ_ADC),
+	[2] = DEFINE_RES_IRQ(IRQ_TC),
+};
+
+struct platform_device s3c_device_adc = {
+	.name		= "exynos-adc",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_adc_resource),
+	.resource	= s3c_adc_resource,
+};
+#endif /* CONFIG_SAMSUNG_DEV_ADC */
+
+/* Camif Controller */
+
+#ifdef CONFIG_CPU_S3C2440
+static struct resource s3c_camif_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C2440_PA_CAMIF, S3C2440_SZ_CAMIF),
+	[1] = DEFINE_RES_IRQ(IRQ_S3C2440_CAM_C),
+	[2] = DEFINE_RES_IRQ(IRQ_S3C2440_CAM_P),
+};
+
+struct platform_device s3c_device_camif = {
+	.name		= "s3c2440-camif",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_camif_resource),
+	.resource	= s3c_camif_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	}
+};
+#endif /* CONFIG_CPU_S3C2440 */
+
+/* FB */
+
+#ifdef CONFIG_S3C_DEV_FB
+static struct resource s3c_fb_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_FB, SZ_16K),
+	[1] = DEFINE_RES_IRQ(IRQ_LCD_VSYNC),
+	[2] = DEFINE_RES_IRQ(IRQ_LCD_FIFO),
+	[3] = DEFINE_RES_IRQ(IRQ_LCD_SYSTEM),
+};
+
+struct platform_device s3c_device_fb = {
+	.name		= "s3c-fb",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_fb_resource),
+	.resource	= s3c_fb_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	},
+};
+
+void __init s3c_fb_set_platdata(struct s3c_fb_platdata *pd)
+{
+	s3c_set_platdata(pd, sizeof(struct s3c_fb_platdata),
+			 &s3c_device_fb);
+}
+#endif /* CONFIG_S3C_DEV_FB */
+
+/* HWMON */
+
+#ifdef CONFIG_S3C_DEV_HWMON
+struct platform_device s3c_device_hwmon = {
+	.name		= "s3c-hwmon",
+	.id		= -1,
+	.dev.parent	= &s3c_device_adc.dev,
+};
+
+void __init s3c_hwmon_set_platdata(struct s3c_hwmon_pdata *pd)
+{
+	s3c_set_platdata(pd, sizeof(struct s3c_hwmon_pdata),
+			 &s3c_device_hwmon);
+}
+#endif /* CONFIG_S3C_DEV_HWMON */
+
+/* HSMMC */
+
+#ifdef CONFIG_S3C_DEV_HSMMC
+static struct resource s3c_hsmmc_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_HSMMC0, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_HSMMC0),
+};
+
+struct s3c_sdhci_platdata s3c_hsmmc0_def_platdata = {
+	.max_width	= 4,
+	.host_caps	= (MMC_CAP_4_BIT_DATA |
+			   MMC_CAP_MMC_HIGHSPEED | MMC_CAP_SD_HIGHSPEED),
+};
+
+struct platform_device s3c_device_hsmmc0 = {
+	.name		= "s3c-sdhci",
+	.id		= 0,
+	.num_resources	= ARRAY_SIZE(s3c_hsmmc_resource),
+	.resource	= s3c_hsmmc_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+		.platform_data		= &s3c_hsmmc0_def_platdata,
+	},
+};
+
+void s3c_sdhci0_set_platdata(struct s3c_sdhci_platdata *pd)
+{
+	s3c_sdhci_set_platdata(pd, &s3c_hsmmc0_def_platdata);
+}
+#endif /* CONFIG_S3C_DEV_HSMMC */
+
+#ifdef CONFIG_S3C_DEV_HSMMC1
+static struct resource s3c_hsmmc1_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_HSMMC1, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_HSMMC1),
+};
+
+struct s3c_sdhci_platdata s3c_hsmmc1_def_platdata = {
+	.max_width	= 4,
+	.host_caps	= (MMC_CAP_4_BIT_DATA |
+			   MMC_CAP_MMC_HIGHSPEED | MMC_CAP_SD_HIGHSPEED),
+};
+
+struct platform_device s3c_device_hsmmc1 = {
+	.name		= "s3c-sdhci",
+	.id		= 1,
+	.num_resources	= ARRAY_SIZE(s3c_hsmmc1_resource),
+	.resource	= s3c_hsmmc1_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+		.platform_data		= &s3c_hsmmc1_def_platdata,
+	},
+};
+
+void s3c_sdhci1_set_platdata(struct s3c_sdhci_platdata *pd)
+{
+	s3c_sdhci_set_platdata(pd, &s3c_hsmmc1_def_platdata);
+}
+#endif /* CONFIG_S3C_DEV_HSMMC1 */
+
+/* HSMMC2 */
+
+#ifdef CONFIG_S3C_DEV_HSMMC2
+static struct resource s3c_hsmmc2_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_HSMMC2, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_HSMMC2),
+};
+
+struct s3c_sdhci_platdata s3c_hsmmc2_def_platdata = {
+	.max_width	= 4,
+	.host_caps	= (MMC_CAP_4_BIT_DATA |
+			   MMC_CAP_MMC_HIGHSPEED | MMC_CAP_SD_HIGHSPEED),
+};
+
+struct platform_device s3c_device_hsmmc2 = {
+	.name		= "s3c-sdhci",
+	.id		= 2,
+	.num_resources	= ARRAY_SIZE(s3c_hsmmc2_resource),
+	.resource	= s3c_hsmmc2_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+		.platform_data		= &s3c_hsmmc2_def_platdata,
+	},
+};
+
+void s3c_sdhci2_set_platdata(struct s3c_sdhci_platdata *pd)
+{
+	s3c_sdhci_set_platdata(pd, &s3c_hsmmc2_def_platdata);
+}
+#endif /* CONFIG_S3C_DEV_HSMMC2 */
+
+#ifdef CONFIG_S3C_DEV_HSMMC3
+static struct resource s3c_hsmmc3_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_HSMMC3, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_HSMMC3),
+};
+
+struct s3c_sdhci_platdata s3c_hsmmc3_def_platdata = {
+	.max_width	= 4,
+	.host_caps	= (MMC_CAP_4_BIT_DATA |
+			   MMC_CAP_MMC_HIGHSPEED | MMC_CAP_SD_HIGHSPEED),
+};
+
+struct platform_device s3c_device_hsmmc3 = {
+	.name		= "s3c-sdhci",
+	.id		= 3,
+	.num_resources	= ARRAY_SIZE(s3c_hsmmc3_resource),
+	.resource	= s3c_hsmmc3_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+		.platform_data		= &s3c_hsmmc3_def_platdata,
+	},
+};
+
+void s3c_sdhci3_set_platdata(struct s3c_sdhci_platdata *pd)
+{
+	s3c_sdhci_set_platdata(pd, &s3c_hsmmc3_def_platdata);
+}
+#endif /* CONFIG_S3C_DEV_HSMMC3 */
+
+/* I2C */
+
+static struct resource s3c_i2c0_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_IIC, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_IIC),
+};
+
+struct platform_device s3c_device_i2c0 = {
+	.name		= "s3c2410-i2c",
+	.id		= 0,
+	.num_resources	= ARRAY_SIZE(s3c_i2c0_resource),
+	.resource	= s3c_i2c0_resource,
+};
+
+struct s3c2410_platform_i2c default_i2c_data __initdata = {
+	.flags		= 0,
+	.slave_addr	= 0x10,
+	.frequency	= 100*1000,
+	.sda_delay	= 100,
+};
+
+void __init s3c_i2c0_set_platdata(struct s3c2410_platform_i2c *pd)
+{
+	struct s3c2410_platform_i2c *npd;
+
+	if (!pd) {
+		pd = &default_i2c_data;
+		pd->bus_num = 0;
+	}
+
+	npd = s3c_set_platdata(pd, sizeof(struct s3c2410_platform_i2c),
+			       &s3c_device_i2c0);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = s3c_i2c0_cfg_gpio;
+}
+
+#ifdef CONFIG_S3C_DEV_I2C1
+static struct resource s3c_i2c1_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_IIC1, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_IIC1),
+};
+
+struct platform_device s3c_device_i2c1 = {
+	.name		= "s3c2410-i2c",
+	.id		= 1,
+	.num_resources	= ARRAY_SIZE(s3c_i2c1_resource),
+	.resource	= s3c_i2c1_resource,
+};
+
+void __init s3c_i2c1_set_platdata(struct s3c2410_platform_i2c *pd)
+{
+	struct s3c2410_platform_i2c *npd;
+
+	if (!pd) {
+		pd = &default_i2c_data;
+		pd->bus_num = 1;
+	}
+
+	npd = s3c_set_platdata(pd, sizeof(struct s3c2410_platform_i2c),
+			       &s3c_device_i2c1);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = s3c_i2c1_cfg_gpio;
+}
+#endif /* CONFIG_S3C_DEV_I2C1 */
+
+#ifdef CONFIG_S3C_DEV_I2C2
+static struct resource s3c_i2c2_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_IIC2, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_IIC2),
+};
+
+struct platform_device s3c_device_i2c2 = {
+	.name		= "s3c2410-i2c",
+	.id		= 2,
+	.num_resources	= ARRAY_SIZE(s3c_i2c2_resource),
+	.resource	= s3c_i2c2_resource,
+};
+
+void __init s3c_i2c2_set_platdata(struct s3c2410_platform_i2c *pd)
+{
+	struct s3c2410_platform_i2c *npd;
+
+	if (!pd) {
+		pd = &default_i2c_data;
+		pd->bus_num = 2;
+	}
+
+	npd = s3c_set_platdata(pd, sizeof(struct s3c2410_platform_i2c),
+			       &s3c_device_i2c2);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = s3c_i2c2_cfg_gpio;
+}
+#endif /* CONFIG_S3C_DEV_I2C2 */
+
+#ifdef CONFIG_S3C_DEV_I2C3
+static struct resource s3c_i2c3_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_IIC3, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_IIC3),
+};
+
+struct platform_device s3c_device_i2c3 = {
+	.name		= "s3c2440-i2c",
+	.id		= 3,
+	.num_resources	= ARRAY_SIZE(s3c_i2c3_resource),
+	.resource	= s3c_i2c3_resource,
+};
+
+void __init s3c_i2c3_set_platdata(struct s3c2410_platform_i2c *pd)
+{
+	struct s3c2410_platform_i2c *npd;
+
+	if (!pd) {
+		pd = &default_i2c_data;
+		pd->bus_num = 3;
+	}
+
+	npd = s3c_set_platdata(pd, sizeof(struct s3c2410_platform_i2c),
+			       &s3c_device_i2c3);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = s3c_i2c3_cfg_gpio;
+}
+#endif /*CONFIG_S3C_DEV_I2C3 */
+
+#ifdef CONFIG_S3C_DEV_I2C4
+static struct resource s3c_i2c4_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_IIC4, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_IIC4),
+};
+
+struct platform_device s3c_device_i2c4 = {
+	.name		= "s3c2440-i2c",
+	.id		= 4,
+	.num_resources	= ARRAY_SIZE(s3c_i2c4_resource),
+	.resource	= s3c_i2c4_resource,
+};
+
+void __init s3c_i2c4_set_platdata(struct s3c2410_platform_i2c *pd)
+{
+	struct s3c2410_platform_i2c *npd;
+
+	if (!pd) {
+		pd = &default_i2c_data;
+		pd->bus_num = 4;
+	}
+
+	npd = s3c_set_platdata(pd, sizeof(struct s3c2410_platform_i2c),
+			       &s3c_device_i2c4);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = s3c_i2c4_cfg_gpio;
+}
+#endif /*CONFIG_S3C_DEV_I2C4 */
+
+#ifdef CONFIG_S3C_DEV_I2C5
+static struct resource s3c_i2c5_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_IIC5, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_IIC5),
+};
+
+struct platform_device s3c_device_i2c5 = {
+	.name		= "s3c2440-i2c",
+	.id		= 5,
+	.num_resources	= ARRAY_SIZE(s3c_i2c5_resource),
+	.resource	= s3c_i2c5_resource,
+};
+
+void __init s3c_i2c5_set_platdata(struct s3c2410_platform_i2c *pd)
+{
+	struct s3c2410_platform_i2c *npd;
+
+	if (!pd) {
+		pd = &default_i2c_data;
+		pd->bus_num = 5;
+	}
+
+	npd = s3c_set_platdata(pd, sizeof(struct s3c2410_platform_i2c),
+			       &s3c_device_i2c5);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = s3c_i2c5_cfg_gpio;
+}
+#endif /*CONFIG_S3C_DEV_I2C5 */
+
+#ifdef CONFIG_S3C_DEV_I2C6
+static struct resource s3c_i2c6_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_IIC6, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_IIC6),
+};
+
+struct platform_device s3c_device_i2c6 = {
+	.name		= "s3c2440-i2c",
+	.id		= 6,
+	.num_resources	= ARRAY_SIZE(s3c_i2c6_resource),
+	.resource	= s3c_i2c6_resource,
+};
+
+void __init s3c_i2c6_set_platdata(struct s3c2410_platform_i2c *pd)
+{
+	struct s3c2410_platform_i2c *npd;
+
+	if (!pd) {
+		pd = &default_i2c_data;
+		pd->bus_num = 6;
+	}
+
+	npd = s3c_set_platdata(pd, sizeof(struct s3c2410_platform_i2c),
+			       &s3c_device_i2c6);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = s3c_i2c6_cfg_gpio;
+}
+#endif /* CONFIG_S3C_DEV_I2C6 */
+
+#ifdef CONFIG_S3C_DEV_I2C7
+static struct resource s3c_i2c7_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_IIC7, SZ_4K),
+	[1] = DEFINE_RES_IRQ(IRQ_IIC7),
+};
+
+struct platform_device s3c_device_i2c7 = {
+	.name		= "s3c2440-i2c",
+	.id		= 7,
+	.num_resources	= ARRAY_SIZE(s3c_i2c7_resource),
+	.resource	= s3c_i2c7_resource,
+};
+
+void __init s3c_i2c7_set_platdata(struct s3c2410_platform_i2c *pd)
+{
+	struct s3c2410_platform_i2c *npd;
+
+	if (!pd) {
+		pd = &default_i2c_data;
+		pd->bus_num = 7;
+	}
+
+	npd = s3c_set_platdata(pd, sizeof(struct s3c2410_platform_i2c),
+			       &s3c_device_i2c7);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = s3c_i2c7_cfg_gpio;
+}
+#endif /* CONFIG_S3C_DEV_I2C7 */
+
+/* I2S */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_iis_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_IIS, S3C24XX_SZ_IIS),
+};
+
+static struct s3c_audio_pdata s3c_iis_platdata = {
+#ifdef CONFIG_S3C24XX_DMAC
+	.dma_filter = s3c24xx_dma_filter,
+#endif
+	.dma_playback = (void *)DMACH_I2S_OUT,
+	.dma_capture = (void *)DMACH_I2S_IN,
+};
+
+struct platform_device s3c_device_iis = {
+	.name		= "s3c24xx-iis",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_iis_resource),
+	.resource	= s3c_iis_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+		.platform_data		= &s3c_iis_platdata,
+	}
+};
+#endif /* CONFIG_PLAT_S3C24XX */
+
+/* IDE CFCON */
+
+#ifdef CONFIG_SAMSUNG_DEV_IDE
+static struct resource s3c_cfcon_resource[] = {
+	[0] = DEFINE_RES_MEM(SAMSUNG_PA_CFCON, SZ_16K),
+	[1] = DEFINE_RES_IRQ(IRQ_CFCON),
+};
+
+struct platform_device s3c_device_cfcon = {
+	.id		= 0,
+	.num_resources	= ARRAY_SIZE(s3c_cfcon_resource),
+	.resource	= s3c_cfcon_resource,
+};
+
+void __init s3c_ide_set_platdata(struct s3c_ide_platdata *pdata)
+{
+	s3c_set_platdata(pdata, sizeof(struct s3c_ide_platdata),
+			 &s3c_device_cfcon);
+}
+#endif /* CONFIG_SAMSUNG_DEV_IDE */
+
+/* KEYPAD */
+
+#ifdef CONFIG_SAMSUNG_DEV_KEYPAD
+static struct resource samsung_keypad_resources[] = {
+	[0] = DEFINE_RES_MEM(SAMSUNG_PA_KEYPAD, SZ_32),
+	[1] = DEFINE_RES_IRQ(IRQ_KEYPAD),
+};
+
+struct platform_device samsung_device_keypad = {
+	.name		= "samsung-keypad",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(samsung_keypad_resources),
+	.resource	= samsung_keypad_resources,
+};
+
+void __init samsung_keypad_set_platdata(struct samsung_keypad_platdata *pd)
+{
+	struct samsung_keypad_platdata *npd;
+
+	npd = s3c_set_platdata(pd, sizeof(struct samsung_keypad_platdata),
+			&samsung_device_keypad);
+
+	if (!npd->cfg_gpio)
+		npd->cfg_gpio = samsung_keypad_cfg_gpio;
+}
+#endif /* CONFIG_SAMSUNG_DEV_KEYPAD */
+
+/* LCD Controller */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_lcd_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_LCD, S3C24XX_SZ_LCD),
+	[1] = DEFINE_RES_IRQ(IRQ_LCD),
+};
+
+struct platform_device s3c_device_lcd = {
+	.name		= "s3c2410-lcd",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_lcd_resource),
+	.resource	= s3c_lcd_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	}
+};
+
+void __init s3c24xx_fb_set_platdata(struct s3c2410fb_mach_info *pd)
+{
+	struct s3c2410fb_mach_info *npd;
+
+	npd = s3c_set_platdata(pd, sizeof(*npd), &s3c_device_lcd);
+	if (npd) {
+		npd->displays = kmemdup(pd->displays,
+			sizeof(struct s3c2410fb_display) * npd->num_displays,
+			GFP_KERNEL);
+		if (!npd->displays)
+			printk(KERN_ERR "no memory for LCD display data\n");
+	} else {
+		printk(KERN_ERR "no memory for LCD platform data\n");
+	}
+}
+#endif /* CONFIG_PLAT_S3C24XX */
+
+/* NAND */
+
+#ifdef CONFIG_S3C_DEV_NAND
+static struct resource s3c_nand_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_NAND, SZ_1M),
+};
+
+struct platform_device s3c_device_nand = {
+	.name		= "s3c2410-nand",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_nand_resource),
+	.resource	= s3c_nand_resource,
+};
+
+/*
+ * s3c_nand_copy_set() - copy nand set data
+ * @set: The new structure, directly copied from the old.
+ *
+ * Copy all the fields from the NAND set field from what is probably __initdata
+ * to new kernel memory. The code returns 0 if the copy happened correctly or
+ * an error code for the calling function to display.
+ *
+ * Note, we currently do not try and look to see if we've already copied the
+ * data in a previous set.
+ */
+static int __init s3c_nand_copy_set(struct s3c2410_nand_set *set)
+{
+	void *ptr;
+	int size;
+
+	size = sizeof(struct mtd_partition) * set->nr_partitions;
+	if (size) {
+		ptr = kmemdup(set->partitions, size, GFP_KERNEL);
+		set->partitions = ptr;
+
+		if (!ptr)
+			return -ENOMEM;
+	}
+
+	if (set->nr_map && set->nr_chips) {
+		size = sizeof(int) * set->nr_chips;
+		ptr = kmemdup(set->nr_map, size, GFP_KERNEL);
+		set->nr_map = ptr;
+
+		if (!ptr)
+			return -ENOMEM;
+	}
+
+	return 0;
+}
+
+void __init s3c_nand_set_platdata(struct s3c2410_platform_nand *nand)
+{
+	struct s3c2410_platform_nand *npd;
+	int size;
+	int ret;
+
+	/* note, if we get a failure in allocation, we simply drop out of the
+	 * function. If there is so little memory available at initialisation
+	 * time then there is little chance the system is going to run.
+	 */
+
+	npd = s3c_set_platdata(nand, sizeof(struct s3c2410_platform_nand),
+				&s3c_device_nand);
+	if (!npd)
+		return;
+
+	/* now see if we need to copy any of the nand set data */
+
+	size = sizeof(struct s3c2410_nand_set) * npd->nr_sets;
+	if (size) {
+		struct s3c2410_nand_set *from = npd->sets;
+		struct s3c2410_nand_set *to;
+		int i;
+
+		to = kmemdup(from, size, GFP_KERNEL);
+		npd->sets = to;	/* set, even if we failed */
+
+		if (!to) {
+			printk(KERN_ERR "%s: no memory for sets\n", __func__);
+			return;
+		}
+
+		for (i = 0; i < npd->nr_sets; i++) {
+			ret = s3c_nand_copy_set(to);
+			if (ret) {
+				printk(KERN_ERR "%s: failed to copy set %d\n",
+				__func__, i);
+				return;
+			}
+			to++;
+		}
+	}
+}
+#endif /* CONFIG_S3C_DEV_NAND */
+
+/* ONENAND */
+
+#ifdef CONFIG_S3C_DEV_ONENAND
+static struct resource s3c_onenand_resources[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_ONENAND, SZ_1K),
+	[1] = DEFINE_RES_MEM(S3C_PA_ONENAND_BUF, S3C_SZ_ONENAND_BUF),
+	[2] = DEFINE_RES_IRQ(IRQ_ONENAND),
+};
+
+struct platform_device s3c_device_onenand = {
+	.name		= "samsung-onenand",
+	.id		= 0,
+	.num_resources	= ARRAY_SIZE(s3c_onenand_resources),
+	.resource	= s3c_onenand_resources,
+};
+#endif /* CONFIG_S3C_DEV_ONENAND */
+
+#ifdef CONFIG_S3C64XX_DEV_ONENAND1
+static struct resource s3c64xx_onenand1_resources[] = {
+	[0] = DEFINE_RES_MEM(S3C64XX_PA_ONENAND1, SZ_1K),
+	[1] = DEFINE_RES_MEM(S3C64XX_PA_ONENAND1_BUF, S3C64XX_SZ_ONENAND1_BUF),
+	[2] = DEFINE_RES_IRQ(IRQ_ONENAND1),
+};
+
+struct platform_device s3c64xx_device_onenand1 = {
+	.name		= "samsung-onenand",
+	.id		= 1,
+	.num_resources	= ARRAY_SIZE(s3c64xx_onenand1_resources),
+	.resource	= s3c64xx_onenand1_resources,
+};
+
+void __init s3c64xx_onenand1_set_platdata(struct onenand_platform_data *pdata)
+{
+	s3c_set_platdata(pdata, sizeof(struct onenand_platform_data),
+			 &s3c64xx_device_onenand1);
+}
+#endif /* CONFIG_S3C64XX_DEV_ONENAND1 */
+
+/* PWM Timer */
+
+#ifdef CONFIG_SAMSUNG_DEV_PWM
+static struct resource samsung_pwm_resource[] = {
+	DEFINE_RES_MEM(SAMSUNG_PA_TIMER, SZ_4K),
+};
+
+struct platform_device samsung_device_pwm = {
+	.name		= "samsung-pwm",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(samsung_pwm_resource),
+	.resource	= samsung_pwm_resource,
+};
+
+void __init samsung_pwm_set_platdata(struct samsung_pwm_variant *pd)
+{
+	samsung_device_pwm.dev.platform_data = pd;
+}
+#endif /* CONFIG_SAMSUNG_DEV_PWM */
+
+/* RTC */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_rtc_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_RTC, SZ_256),
+	[1] = DEFINE_RES_IRQ(IRQ_RTC),
+	[2] = DEFINE_RES_IRQ(IRQ_TICK),
+};
+
+struct platform_device s3c_device_rtc = {
+	.name		= "s3c2410-rtc",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_rtc_resource),
+	.resource	= s3c_rtc_resource,
+};
+#endif /* CONFIG_PLAT_S3C24XX */
+
+#ifdef CONFIG_S3C_DEV_RTC
+static struct resource s3c_rtc_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_RTC, SZ_256),
+	[1] = DEFINE_RES_IRQ(IRQ_RTC_ALARM),
+	[2] = DEFINE_RES_IRQ(IRQ_RTC_TIC),
+};
+
+struct platform_device s3c_device_rtc = {
+	.name		= "s3c64xx-rtc",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_rtc_resource),
+	.resource	= s3c_rtc_resource,
+};
+#endif /* CONFIG_S3C_DEV_RTC */
+
+/* SDI */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_sdi_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_SDI, S3C24XX_SZ_SDI),
+	[1] = DEFINE_RES_IRQ(IRQ_SDI),
+};
+
+struct platform_device s3c_device_sdi = {
+	.name		= "s3c2410-sdi",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_sdi_resource),
+	.resource	= s3c_sdi_resource,
+};
+
+void __init s3c24xx_mci_set_platdata(struct s3c24xx_mci_pdata *pdata)
+{
+	s3c_set_platdata(pdata, sizeof(struct s3c24xx_mci_pdata),
+			 &s3c_device_sdi);
+}
+#endif /* CONFIG_PLAT_S3C24XX */
+
+/* SPI */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_spi0_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_SPI, SZ_32),
+	[1] = DEFINE_RES_IRQ(IRQ_SPI0),
+};
+
+struct platform_device s3c_device_spi0 = {
+	.name		= "s3c2410-spi",
+	.id		= 0,
+	.num_resources	= ARRAY_SIZE(s3c_spi0_resource),
+	.resource	= s3c_spi0_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	}
+};
+
+static struct resource s3c_spi1_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_SPI1, SZ_32),
+	[1] = DEFINE_RES_IRQ(IRQ_SPI1),
+};
+
+struct platform_device s3c_device_spi1 = {
+	.name		= "s3c2410-spi",
+	.id		= 1,
+	.num_resources	= ARRAY_SIZE(s3c_spi1_resource),
+	.resource	= s3c_spi1_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	}
+};
+#endif /* CONFIG_PLAT_S3C24XX */
+
+/* Touchscreen */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_ts_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_ADC, S3C24XX_SZ_ADC),
+	[1] = DEFINE_RES_IRQ(IRQ_TC),
+};
+
+struct platform_device s3c_device_ts = {
+	.name		= "s3c2410-ts",
+	.id		= -1,
+	.dev.parent	= &s3c_device_adc.dev,
+	.num_resources	= ARRAY_SIZE(s3c_ts_resource),
+	.resource	= s3c_ts_resource,
+};
+
+void __init s3c24xx_ts_set_platdata(struct s3c2410_ts_mach_info *hard_s3c2410ts_info)
+{
+	s3c_set_platdata(hard_s3c2410ts_info,
+			 sizeof(struct s3c2410_ts_mach_info), &s3c_device_ts);
+}
+#endif /* CONFIG_PLAT_S3C24XX */
+
+#ifdef CONFIG_SAMSUNG_DEV_TS
+static struct s3c2410_ts_mach_info default_ts_data __initdata = {
+	.delay			= 10000,
+	.presc			= 49,
+	.oversampling_shift	= 2,
+};
+
+void __init s3c64xx_ts_set_platdata(struct s3c2410_ts_mach_info *pd)
+{
+	if (!pd)
+		pd = &default_ts_data;
+
+	s3c_set_platdata(pd, sizeof(struct s3c2410_ts_mach_info),
+			 &s3c_device_adc);
+}
+#endif /* CONFIG_SAMSUNG_DEV_TS */
+
+/* USB */
+
+#ifdef CONFIG_S3C_DEV_USB_HOST
+static struct resource s3c_usb_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_USBHOST, SZ_256),
+	[1] = DEFINE_RES_IRQ(IRQ_USBH),
+};
+
+struct platform_device s3c_device_ohci = {
+	.name		= "s3c2410-ohci",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_usb_resource),
+	.resource	= s3c_usb_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	}
+};
+
+/*
+ * s3c_ohci_set_platdata - initialise OHCI device platform data
+ * @info: The platform data.
+ *
+ * This call copies the @info passed in and sets the device .platform_data
+ * field to that copy. The @info is copied so that the original can be marked
+ * __initdata.
+ */
+
+void __init s3c_ohci_set_platdata(struct s3c2410_hcd_info *info)
+{
+	s3c_set_platdata(info, sizeof(struct s3c2410_hcd_info),
+			 &s3c_device_ohci);
+}
+#endif /* CONFIG_S3C_DEV_USB_HOST */
+
+/* USB Device (Gadget) */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_usbgadget_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C24XX_PA_USBDEV, S3C24XX_SZ_USBDEV),
+	[1] = DEFINE_RES_IRQ(IRQ_USBD),
+};
+
+struct platform_device s3c_device_usbgadget = {
+	.name		= "s3c2410-usbgadget",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_usbgadget_resource),
+	.resource	= s3c_usbgadget_resource,
+};
+
+void __init s3c24xx_udc_set_platdata(struct s3c2410_udc_mach_info *pd)
+{
+	s3c_set_platdata(pd, sizeof(*pd), &s3c_device_usbgadget);
+}
+#endif /* CONFIG_PLAT_S3C24XX */
+
+/* USB HSOTG */
+
+#ifdef CONFIG_S3C_DEV_USB_HSOTG
+static struct resource s3c_usb_hsotg_resources[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_USB_HSOTG, SZ_128K),
+	[1] = DEFINE_RES_IRQ(IRQ_OTG),
+};
+
+struct platform_device s3c_device_usb_hsotg = {
+	.name		= "s3c-hsotg",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_usb_hsotg_resources),
+	.resource	= s3c_usb_hsotg_resources,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	},
+};
+
+void __init dwc2_hsotg_set_platdata(struct dwc2_hsotg_plat *pd)
+{
+	struct dwc2_hsotg_plat *npd;
+
+	npd = s3c_set_platdata(pd, sizeof(struct dwc2_hsotg_plat),
+			&s3c_device_usb_hsotg);
+
+	if (!npd->phy_init)
+		npd->phy_init = s5p_usb_phy_init;
+	if (!npd->phy_exit)
+		npd->phy_exit = s5p_usb_phy_exit;
+}
+#endif /* CONFIG_S3C_DEV_USB_HSOTG */
+
+/* USB High Spped 2.0 Device (Gadget) */
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct resource s3c_hsudc_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C2416_PA_HSUDC, S3C2416_SZ_HSUDC),
+	[1] = DEFINE_RES_IRQ(IRQ_USBD),
+};
+
+struct platform_device s3c_device_usb_hsudc = {
+	.name		= "s3c-hsudc",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_hsudc_resource),
+	.resource	= s3c_hsudc_resource,
+	.dev		= {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	},
+};
+
+void __init s3c24xx_hsudc_set_platdata(struct s3c24xx_hsudc_platdata *pd)
+{
+	s3c_set_platdata(pd, sizeof(*pd), &s3c_device_usb_hsudc);
+}
+#endif /* CONFIG_PLAT_S3C24XX */
+
+/* WDT */
+
+#ifdef CONFIG_S3C_DEV_WDT
+static struct resource s3c_wdt_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_WDT, SZ_1K),
+	[1] = DEFINE_RES_IRQ(IRQ_WDT),
+};
+
+struct platform_device s3c_device_wdt = {
+	.name		= "s3c2410-wdt",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(s3c_wdt_resource),
+	.resource	= s3c_wdt_resource,
+};
+#endif /* CONFIG_S3C_DEV_WDT */
+
+#ifdef CONFIG_S3C64XX_DEV_SPI0
+static struct resource s3c64xx_spi0_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_SPI0, SZ_256),
+	[1] = DEFINE_RES_IRQ(IRQ_SPI0),
+};
+
+struct platform_device s3c64xx_device_spi0 = {
+	.name		= "s3c6410-spi",
+	.id		= 0,
+	.num_resources	= ARRAY_SIZE(s3c64xx_spi0_resource),
+	.resource	= s3c64xx_spi0_resource,
+	.dev = {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	},
+};
+
+void __init s3c64xx_spi0_set_platdata(int (*cfg_gpio)(void), int src_clk_nr,
+						int num_cs)
+{
+	struct s3c64xx_spi_info pd;
+
+	/* Reject invalid configuration */
+	if (!num_cs || src_clk_nr < 0) {
+		pr_err("%s: Invalid SPI configuration\n", __func__);
+		return;
+	}
+
+	pd.num_cs = num_cs;
+	pd.src_clk_nr = src_clk_nr;
+	pd.cfg_gpio = (cfg_gpio) ? cfg_gpio : s3c64xx_spi0_cfg_gpio;
+	pd.dma_tx = (void *)DMACH_SPI0_TX;
+	pd.dma_rx = (void *)DMACH_SPI0_RX;
+#if defined(CONFIG_PL330_DMA)
+	pd.filter = pl330_filter;
+#elif defined(CONFIG_S3C64XX_PL080)
+	pd.filter = pl08x_filter_id;
+#elif defined(CONFIG_S3C24XX_DMAC)
+	pd.filter = s3c24xx_dma_filter;
+#endif
+
+	s3c_set_platdata(&pd, sizeof(pd), &s3c64xx_device_spi0);
+}
+#endif /* CONFIG_S3C64XX_DEV_SPI0 */
+
+#ifdef CONFIG_S3C64XX_DEV_SPI1
+static struct resource s3c64xx_spi1_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_SPI1, SZ_256),
+	[1] = DEFINE_RES_IRQ(IRQ_SPI1),
+};
+
+struct platform_device s3c64xx_device_spi1 = {
+	.name		= "s3c6410-spi",
+	.id		= 1,
+	.num_resources	= ARRAY_SIZE(s3c64xx_spi1_resource),
+	.resource	= s3c64xx_spi1_resource,
+	.dev = {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	},
+};
+
+void __init s3c64xx_spi1_set_platdata(int (*cfg_gpio)(void), int src_clk_nr,
+						int num_cs)
+{
+	struct s3c64xx_spi_info pd;
+
+	/* Reject invalid configuration */
+	if (!num_cs || src_clk_nr < 0) {
+		pr_err("%s: Invalid SPI configuration\n", __func__);
+		return;
+	}
+
+	pd.num_cs = num_cs;
+	pd.src_clk_nr = src_clk_nr;
+	pd.cfg_gpio = (cfg_gpio) ? cfg_gpio : s3c64xx_spi1_cfg_gpio;
+	pd.dma_tx = (void *)DMACH_SPI1_TX;
+	pd.dma_rx = (void *)DMACH_SPI1_RX;
+#if defined(CONFIG_PL330_DMA)
+	pd.filter = pl330_filter;
+#elif defined(CONFIG_S3C64XX_PL080)
+	pd.filter = pl08x_filter_id;
+#endif
+
+
+	s3c_set_platdata(&pd, sizeof(pd), &s3c64xx_device_spi1);
+}
+#endif /* CONFIG_S3C64XX_DEV_SPI1 */
+
+#ifdef CONFIG_S3C64XX_DEV_SPI2
+static struct resource s3c64xx_spi2_resource[] = {
+	[0] = DEFINE_RES_MEM(S3C_PA_SPI2, SZ_256),
+	[1] = DEFINE_RES_IRQ(IRQ_SPI2),
+};
+
+struct platform_device s3c64xx_device_spi2 = {
+	.name		= "s3c6410-spi",
+	.id		= 2,
+	.num_resources	= ARRAY_SIZE(s3c64xx_spi2_resource),
+	.resource	= s3c64xx_spi2_resource,
+	.dev = {
+		.dma_mask		= &samsung_device_dma_mask,
+		.coherent_dma_mask	= DMA_BIT_MASK(32),
+	},
+};
+
+void __init s3c64xx_spi2_set_platdata(int (*cfg_gpio)(void), int src_clk_nr,
+						int num_cs)
+{
+	struct s3c64xx_spi_info pd;
+
+	/* Reject invalid configuration */
+	if (!num_cs || src_clk_nr < 0) {
+		pr_err("%s: Invalid SPI configuration\n", __func__);
+		return;
+	}
+
+	pd.num_cs = num_cs;
+	pd.src_clk_nr = src_clk_nr;
+	pd.cfg_gpio = (cfg_gpio) ? cfg_gpio : s3c64xx_spi2_cfg_gpio;
+	pd.dma_tx = (void *)DMACH_SPI2_TX;
+	pd.dma_rx = (void *)DMACH_SPI2_RX;
+#if defined(CONFIG_PL330_DMA)
+	pd.filter = pl330_filter;
+#elif defined(CONFIG_S3C64XX_PL080)
+	pd.filter = pl08x_filter_id;
+#endif
+
+	s3c_set_platdata(&pd, sizeof(pd), &s3c64xx_device_spi2);
+}
+#endif /* CONFIG_S3C64XX_DEV_SPI2 */

arch/arm/plat-samsung/gpio-samsung.c

@@ -0,0 +1,1328 @@
+/*
+ * Copyright (c) 2009-2011 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com/
+ *
+ * Copyright 2008 Openmoko, Inc.
+ * Copyright 2008 Simtec Electronics
+ *      Ben Dooks <ben@simtec.co.uk>
+ *      http://armlinux.simtec.co.uk/
+ *
+ * SAMSUNG - GPIOlib support
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/kernel.h>
+#include <linux/irq.h>
+#include <linux/io.h>
+#include <linux/gpio.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/module.h>
+#include <linux/interrupt.h>
+#include <linux/device.h>
+#include <linux/ioport.h>
+#include <linux/of.h>
+#include <linux/slab.h>
+#include <linux/of_address.h>
+
+#include <asm/irq.h>
+
+#include <mach/irqs.h>
+#include <mach/map.h>
+#include <mach/regs-gpio.h>
+#include <mach/gpio-samsung.h>
+
+#include <plat/cpu.h>
+#include <plat/gpio-core.h>
+#include <plat/gpio-cfg.h>
+#include <plat/gpio-cfg-helpers.h>
+#include <plat/pm.h>
+
+int samsung_gpio_setpull_updown(struct samsung_gpio_chip *chip,
+				unsigned int off, samsung_gpio_pull_t pull)
+{
+	void __iomem *reg = chip->base + 0x08;
+	int shift = off * 2;
+	u32 pup;
+
+	pup = __raw_readl(reg);
+	pup &= ~(3 << shift);
+	pup |= pull << shift;
+	__raw_writel(pup, reg);
+
+	return 0;
+}
+
+samsung_gpio_pull_t samsung_gpio_getpull_updown(struct samsung_gpio_chip *chip,
+						unsigned int off)
+{
+	void __iomem *reg = chip->base + 0x08;
+	int shift = off * 2;
+	u32 pup = __raw_readl(reg);
+
+	pup >>= shift;
+	pup &= 0x3;
+
+	return (__force samsung_gpio_pull_t)pup;
+}
+
+int s3c2443_gpio_setpull(struct samsung_gpio_chip *chip,
+			 unsigned int off, samsung_gpio_pull_t pull)
+{
+	switch (pull) {
+	case S3C_GPIO_PULL_NONE:
+		pull = 0x01;
+		break;
+	case S3C_GPIO_PULL_UP:
+		pull = 0x00;
+		break;
+	case S3C_GPIO_PULL_DOWN:
+		pull = 0x02;
+		break;
+	}
+	return samsung_gpio_setpull_updown(chip, off, pull);
+}
+
+samsung_gpio_pull_t s3c2443_gpio_getpull(struct samsung_gpio_chip *chip,
+					 unsigned int off)
+{
+	samsung_gpio_pull_t pull;
+
+	pull = samsung_gpio_getpull_updown(chip, off);
+
+	switch (pull) {
+	case 0x00:
+		pull = S3C_GPIO_PULL_UP;
+		break;
+	case 0x01:
+	case 0x03:
+		pull = S3C_GPIO_PULL_NONE;
+		break;
+	case 0x02:
+		pull = S3C_GPIO_PULL_DOWN;
+		break;
+	}
+
+	return pull;
+}
+
+static int s3c24xx_gpio_setpull_1(struct samsung_gpio_chip *chip,
+				  unsigned int off, samsung_gpio_pull_t pull,
+				  samsung_gpio_pull_t updown)
+{
+	void __iomem *reg = chip->base + 0x08;
+	u32 pup = __raw_readl(reg);
+
+	if (pull == updown)
+		pup &= ~(1 << off);
+	else if (pull == S3C_GPIO_PULL_NONE)
+		pup |= (1 << off);
+	else
+		return -EINVAL;
+
+	__raw_writel(pup, reg);
+	return 0;
+}
+
+static samsung_gpio_pull_t s3c24xx_gpio_getpull_1(struct samsung_gpio_chip *chip,
+						  unsigned int off,
+						  samsung_gpio_pull_t updown)
+{
+	void __iomem *reg = chip->base + 0x08;
+	u32 pup = __raw_readl(reg);
+
+	pup &= (1 << off);
+	return pup ? S3C_GPIO_PULL_NONE : updown;
+}
+
+samsung_gpio_pull_t s3c24xx_gpio_getpull_1up(struct samsung_gpio_chip *chip,
+					     unsigned int off)
+{
+	return s3c24xx_gpio_getpull_1(chip, off, S3C_GPIO_PULL_UP);
+}
+
+int s3c24xx_gpio_setpull_1up(struct samsung_gpio_chip *chip,
+			     unsigned int off, samsung_gpio_pull_t pull)
+{
+	return s3c24xx_gpio_setpull_1(chip, off, pull, S3C_GPIO_PULL_UP);
+}
+
+samsung_gpio_pull_t s3c24xx_gpio_getpull_1down(struct samsung_gpio_chip *chip,
+					       unsigned int off)
+{
+	return s3c24xx_gpio_getpull_1(chip, off, S3C_GPIO_PULL_DOWN);
+}
+
+int s3c24xx_gpio_setpull_1down(struct samsung_gpio_chip *chip,
+			       unsigned int off, samsung_gpio_pull_t pull)
+{
+	return s3c24xx_gpio_setpull_1(chip, off, pull, S3C_GPIO_PULL_DOWN);
+}
+
+/*
+ * samsung_gpio_setcfg_2bit - Samsung 2bit style GPIO configuration.
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ * @cfg: The configuration value to set.
+ *
+ * This helper deal with the GPIO cases where the control register
+ * has two bits of configuration per gpio, which have the following
+ * functions:
+ *	00 = input
+ *	01 = output
+ *	1x = special function
+ */
+
+static int samsung_gpio_setcfg_2bit(struct samsung_gpio_chip *chip,
+				    unsigned int off, unsigned int cfg)
+{
+	void __iomem *reg = chip->base;
+	unsigned int shift = off * 2;
+	u32 con;
+
+	if (samsung_gpio_is_cfg_special(cfg)) {
+		cfg &= 0xf;
+		if (cfg > 3)
+			return -EINVAL;
+
+		cfg <<= shift;
+	}
+
+	con = __raw_readl(reg);
+	con &= ~(0x3 << shift);
+	con |= cfg;
+	__raw_writel(con, reg);
+
+	return 0;
+}
+
+/*
+ * samsung_gpio_getcfg_2bit - Samsung 2bit style GPIO configuration read.
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ *
+ * The reverse of samsung_gpio_setcfg_2bit(). Will return a value which
+ * could be directly passed back to samsung_gpio_setcfg_2bit(), from the
+ * S3C_GPIO_SPECIAL() macro.
+ */
+
+static unsigned int samsung_gpio_getcfg_2bit(struct samsung_gpio_chip *chip,
+					     unsigned int off)
+{
+	u32 con;
+
+	con = __raw_readl(chip->base);
+	con >>= off * 2;
+	con &= 3;
+
+	/* this conversion works for IN and OUT as well as special mode */
+	return S3C_GPIO_SPECIAL(con);
+}
+
+/*
+ * samsung_gpio_setcfg_4bit - Samsung 4bit single register GPIO config.
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ * @cfg: The configuration value to set.
+ *
+ * This helper deal with the GPIO cases where the control register has 4 bits
+ * of control per GPIO, generally in the form of:
+ *	0000 = Input
+ *	0001 = Output
+ *	others = Special functions (dependent on bank)
+ *
+ * Note, since the code to deal with the case where there are two control
+ * registers instead of one, we do not have a separate set of functions for
+ * each case.
+ */
+
+static int samsung_gpio_setcfg_4bit(struct samsung_gpio_chip *chip,
+				    unsigned int off, unsigned int cfg)
+{
+	void __iomem *reg = chip->base;
+	unsigned int shift = (off & 7) * 4;
+	u32 con;
+
+	if (off < 8 && chip->chip.ngpio > 8)
+		reg -= 4;
+
+	if (samsung_gpio_is_cfg_special(cfg)) {
+		cfg &= 0xf;
+		cfg <<= shift;
+	}
+
+	con = __raw_readl(reg);
+	con &= ~(0xf << shift);
+	con |= cfg;
+	__raw_writel(con, reg);
+
+	return 0;
+}
+
+/*
+ * samsung_gpio_getcfg_4bit - Samsung 4bit single register GPIO config read.
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ *
+ * The reverse of samsung_gpio_setcfg_4bit(), turning a gpio configuration
+ * register setting into a value the software can use, such as could be passed
+ * to samsung_gpio_setcfg_4bit().
+ *
+ * @sa samsung_gpio_getcfg_2bit
+ */
+
+static unsigned samsung_gpio_getcfg_4bit(struct samsung_gpio_chip *chip,
+					 unsigned int off)
+{
+	void __iomem *reg = chip->base;
+	unsigned int shift = (off & 7) * 4;
+	u32 con;
+
+	if (off < 8 && chip->chip.ngpio > 8)
+		reg -= 4;
+
+	con = __raw_readl(reg);
+	con >>= shift;
+	con &= 0xf;
+
+	/* this conversion works for IN and OUT as well as special mode */
+	return S3C_GPIO_SPECIAL(con);
+}
+
+#ifdef CONFIG_PLAT_S3C24XX
+/*
+ * s3c24xx_gpio_setcfg_abank - S3C24XX style GPIO configuration (Bank A)
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ * @cfg: The configuration value to set.
+ *
+ * This helper deal with the GPIO cases where the control register
+ * has one bit of configuration for the gpio, where setting the bit
+ * means the pin is in special function mode and unset means output.
+ */
+
+static int s3c24xx_gpio_setcfg_abank(struct samsung_gpio_chip *chip,
+				     unsigned int off, unsigned int cfg)
+{
+	void __iomem *reg = chip->base;
+	unsigned int shift = off;
+	u32 con;
+
+	if (samsung_gpio_is_cfg_special(cfg)) {
+		cfg &= 0xf;
+
+		/* Map output to 0, and SFN2 to 1 */
+		cfg -= 1;
+		if (cfg > 1)
+			return -EINVAL;
+
+		cfg <<= shift;
+	}
+
+	con = __raw_readl(reg);
+	con &= ~(0x1 << shift);
+	con |= cfg;
+	__raw_writel(con, reg);
+
+	return 0;
+}
+
+/*
+ * s3c24xx_gpio_getcfg_abank - S3C24XX style GPIO configuration read (Bank A)
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ *
+ * The reverse of s3c24xx_gpio_setcfg_abank() turning an GPIO into a usable
+ * GPIO configuration value.
+ *
+ * @sa samsung_gpio_getcfg_2bit
+ * @sa samsung_gpio_getcfg_4bit
+ */
+
+static unsigned s3c24xx_gpio_getcfg_abank(struct samsung_gpio_chip *chip,
+					  unsigned int off)
+{
+	u32 con;
+
+	con = __raw_readl(chip->base);
+	con >>= off;
+	con &= 1;
+	con++;
+
+	return S3C_GPIO_SFN(con);
+}
+#endif
+
+static void __init samsung_gpiolib_set_cfg(struct samsung_gpio_cfg *chipcfg,
+					   int nr_chips)
+{
+	for (; nr_chips > 0; nr_chips--, chipcfg++) {
+		if (!chipcfg->set_config)
+			chipcfg->set_config = samsung_gpio_setcfg_4bit;
+		if (!chipcfg->get_config)
+			chipcfg->get_config = samsung_gpio_getcfg_4bit;
+		if (!chipcfg->set_pull)
+			chipcfg->set_pull = samsung_gpio_setpull_updown;
+		if (!chipcfg->get_pull)
+			chipcfg->get_pull = samsung_gpio_getpull_updown;
+	}
+}
+
+struct samsung_gpio_cfg s3c24xx_gpiocfg_default = {
+	.set_config	= samsung_gpio_setcfg_2bit,
+	.get_config	= samsung_gpio_getcfg_2bit,
+};
+
+#ifdef CONFIG_PLAT_S3C24XX
+static struct samsung_gpio_cfg s3c24xx_gpiocfg_banka = {
+	.set_config	= s3c24xx_gpio_setcfg_abank,
+	.get_config	= s3c24xx_gpio_getcfg_abank,
+};
+#endif
+
+static struct samsung_gpio_cfg samsung_gpio_cfgs[] = {
+	[0] = {
+		.cfg_eint	= 0x0,
+	},
+	[1] = {
+		.cfg_eint	= 0x3,
+	},
+	[2] = {
+		.cfg_eint	= 0x7,
+	},
+	[3] = {
+		.cfg_eint	= 0xF,
+	},
+	[4] = {
+		.cfg_eint	= 0x0,
+		.set_config	= samsung_gpio_setcfg_2bit,
+		.get_config	= samsung_gpio_getcfg_2bit,
+	},
+	[5] = {
+		.cfg_eint	= 0x2,
+		.set_config	= samsung_gpio_setcfg_2bit,
+		.get_config	= samsung_gpio_getcfg_2bit,
+	},
+	[6] = {
+		.cfg_eint	= 0x3,
+		.set_config	= samsung_gpio_setcfg_2bit,
+		.get_config	= samsung_gpio_getcfg_2bit,
+	},
+	[7] = {
+		.set_config	= samsung_gpio_setcfg_2bit,
+		.get_config	= samsung_gpio_getcfg_2bit,
+	},
+};
+
+/*
+ * Default routines for controlling GPIO, based on the original S3C24XX
+ * GPIO functions which deal with the case where each gpio bank of the
+ * chip is as following:
+ *
+ * base + 0x00: Control register, 2 bits per gpio
+ *	        gpio n: 2 bits starting at (2*n)
+ *		00 = input, 01 = output, others mean special-function
+ * base + 0x04: Data register, 1 bit per gpio
+ *		bit n: data bit n
+*/
+
+static int samsung_gpiolib_2bit_input(struct gpio_chip *chip, unsigned offset)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	void __iomem *base = ourchip->base;
+	unsigned long flags;
+	unsigned long con;
+
+	samsung_gpio_lock(ourchip, flags);
+
+	con = __raw_readl(base + 0x00);
+	con &= ~(3 << (offset * 2));
+
+	__raw_writel(con, base + 0x00);
+
+	samsung_gpio_unlock(ourchip, flags);
+	return 0;
+}
+
+static int samsung_gpiolib_2bit_output(struct gpio_chip *chip,
+				       unsigned offset, int value)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	void __iomem *base = ourchip->base;
+	unsigned long flags;
+	unsigned long dat;
+	unsigned long con;
+
+	samsung_gpio_lock(ourchip, flags);
+
+	dat = __raw_readl(base + 0x04);
+	dat &= ~(1 << offset);
+	if (value)
+		dat |= 1 << offset;
+	__raw_writel(dat, base + 0x04);
+
+	con = __raw_readl(base + 0x00);
+	con &= ~(3 << (offset * 2));
+	con |= 1 << (offset * 2);
+
+	__raw_writel(con, base + 0x00);
+	__raw_writel(dat, base + 0x04);
+
+	samsung_gpio_unlock(ourchip, flags);
+	return 0;
+}
+
+/*
+ * The samsung_gpiolib_4bit routines are to control the gpio banks where
+ * the gpio configuration register (GPxCON) has 4 bits per GPIO, as the
+ * following example:
+ *
+ * base + 0x00: Control register, 4 bits per gpio
+ *		gpio n: 4 bits starting at (4*n)
+ *		0000 = input, 0001 = output, others mean special-function
+ * base + 0x04: Data register, 1 bit per gpio
+ *		bit n: data bit n
+ *
+ * Note, since the data register is one bit per gpio and is at base + 0x4
+ * we can use samsung_gpiolib_get and samsung_gpiolib_set to change the
+ * state of the output.
+ */
+
+static int samsung_gpiolib_4bit_input(struct gpio_chip *chip,
+				      unsigned int offset)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	void __iomem *base = ourchip->base;
+	unsigned long con;
+
+	con = __raw_readl(base + GPIOCON_OFF);
+	if (ourchip->bitmap_gpio_int & BIT(offset))
+		con |= 0xf << con_4bit_shift(offset);
+	else
+		con &= ~(0xf << con_4bit_shift(offset));
+	__raw_writel(con, base + GPIOCON_OFF);
+
+	pr_debug("%s: %p: CON now %08lx\n", __func__, base, con);
+
+	return 0;
+}
+
+static int samsung_gpiolib_4bit_output(struct gpio_chip *chip,
+				       unsigned int offset, int value)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	void __iomem *base = ourchip->base;
+	unsigned long con;
+	unsigned long dat;
+
+	con = __raw_readl(base + GPIOCON_OFF);
+	con &= ~(0xf << con_4bit_shift(offset));
+	con |= 0x1 << con_4bit_shift(offset);
+
+	dat = __raw_readl(base + GPIODAT_OFF);
+
+	if (value)
+		dat |= 1 << offset;
+	else
+		dat &= ~(1 << offset);
+
+	__raw_writel(dat, base + GPIODAT_OFF);
+	__raw_writel(con, base + GPIOCON_OFF);
+	__raw_writel(dat, base + GPIODAT_OFF);
+
+	pr_debug("%s: %p: CON %08lx, DAT %08lx\n", __func__, base, con, dat);
+
+	return 0;
+}
+
+/*
+ * The next set of routines are for the case where the GPIO configuration
+ * registers are 4 bits per GPIO but there is more than one register (the
+ * bank has more than 8 GPIOs.
+ *
+ * This case is the similar to the 4 bit case, but the registers are as
+ * follows:
+ *
+ * base + 0x00: Control register, 4 bits per gpio (lower 8 GPIOs)
+ *		gpio n: 4 bits starting at (4*n)
+ *		0000 = input, 0001 = output, others mean special-function
+ * base + 0x04: Control register, 4 bits per gpio (up to 8 additions GPIOs)
+ *		gpio n: 4 bits starting at (4*n)
+ *		0000 = input, 0001 = output, others mean special-function
+ * base + 0x08: Data register, 1 bit per gpio
+ *		bit n: data bit n
+ *
+ * To allow us to use the samsung_gpiolib_get and samsung_gpiolib_set
+ * routines we store the 'base + 0x4' address so that these routines see
+ * the data register at ourchip->base + 0x04.
+ */
+
+static int samsung_gpiolib_4bit2_input(struct gpio_chip *chip,
+				       unsigned int offset)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	void __iomem *base = ourchip->base;
+	void __iomem *regcon = base;
+	unsigned long con;
+
+	if (offset > 7)
+		offset -= 8;
+	else
+		regcon -= 4;
+
+	con = __raw_readl(regcon);
+	con &= ~(0xf << con_4bit_shift(offset));
+	__raw_writel(con, regcon);
+
+	pr_debug("%s: %p: CON %08lx\n", __func__, base, con);
+
+	return 0;
+}
+
+static int samsung_gpiolib_4bit2_output(struct gpio_chip *chip,
+					unsigned int offset, int value)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	void __iomem *base = ourchip->base;
+	void __iomem *regcon = base;
+	unsigned long con;
+	unsigned long dat;
+	unsigned con_offset = offset;
+
+	if (con_offset > 7)
+		con_offset -= 8;
+	else
+		regcon -= 4;
+
+	con = __raw_readl(regcon);
+	con &= ~(0xf << con_4bit_shift(con_offset));
+	con |= 0x1 << con_4bit_shift(con_offset);
+
+	dat = __raw_readl(base + GPIODAT_OFF);
+
+	if (value)
+		dat |= 1 << offset;
+	else
+		dat &= ~(1 << offset);
+
+	__raw_writel(dat, base + GPIODAT_OFF);
+	__raw_writel(con, regcon);
+	__raw_writel(dat, base + GPIODAT_OFF);
+
+	pr_debug("%s: %p: CON %08lx, DAT %08lx\n", __func__, base, con, dat);
+
+	return 0;
+}
+
+#ifdef CONFIG_PLAT_S3C24XX
+/* The next set of routines are for the case of s3c24xx bank a */
+
+static int s3c24xx_gpiolib_banka_input(struct gpio_chip *chip, unsigned offset)
+{
+	return -EINVAL;
+}
+
+static int s3c24xx_gpiolib_banka_output(struct gpio_chip *chip,
+					unsigned offset, int value)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	void __iomem *base = ourchip->base;
+	unsigned long flags;
+	unsigned long dat;
+	unsigned long con;
+
+	local_irq_save(flags);
+
+	con = __raw_readl(base + 0x00);
+	dat = __raw_readl(base + 0x04);
+
+	dat &= ~(1 << offset);
+	if (value)
+		dat |= 1 << offset;
+
+	__raw_writel(dat, base + 0x04);
+
+	con &= ~(1 << offset);
+
+	__raw_writel(con, base + 0x00);
+	__raw_writel(dat, base + 0x04);
+
+	local_irq_restore(flags);
+	return 0;
+}
+#endif
+
+static void samsung_gpiolib_set(struct gpio_chip *chip,
+				unsigned offset, int value)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	void __iomem *base = ourchip->base;
+	unsigned long flags;
+	unsigned long dat;
+
+	samsung_gpio_lock(ourchip, flags);
+
+	dat = __raw_readl(base + 0x04);
+	dat &= ~(1 << offset);
+	if (value)
+		dat |= 1 << offset;
+	__raw_writel(dat, base + 0x04);
+
+	samsung_gpio_unlock(ourchip, flags);
+}
+
+static int samsung_gpiolib_get(struct gpio_chip *chip, unsigned offset)
+{
+	struct samsung_gpio_chip *ourchip = to_samsung_gpio(chip);
+	unsigned long val;
+
+	val = __raw_readl(ourchip->base + 0x04);
+	val >>= offset;
+	val &= 1;
+
+	return val;
+}
+
+/*
+ * CONFIG_S3C_GPIO_TRACK enables the tracking of the s3c specific gpios
+ * for use with the configuration calls, and other parts of the s3c gpiolib
+ * support code.
+ *
+ * Not all s3c support code will need this, as some configurations of cpu
+ * may only support one or two different configuration options and have an
+ * easy gpio to samsung_gpio_chip mapping function. If this is the case, then
+ * the machine support file should provide its own samsung_gpiolib_getchip()
+ * and any other necessary functions.
+ */
+
+#ifdef CONFIG_S3C_GPIO_TRACK
+struct samsung_gpio_chip *s3c_gpios[S3C_GPIO_END];
+
+static __init void s3c_gpiolib_track(struct samsung_gpio_chip *chip)
+{
+	unsigned int gpn;
+	int i;
+
+	gpn = chip->chip.base;
+	for (i = 0; i < chip->chip.ngpio; i++, gpn++) {
+		BUG_ON(gpn >= ARRAY_SIZE(s3c_gpios));
+		s3c_gpios[gpn] = chip;
+	}
+}
+#endif /* CONFIG_S3C_GPIO_TRACK */
+
+/*
+ * samsung_gpiolib_add() - add the Samsung gpio_chip.
+ * @chip: The chip to register
+ *
+ * This is a wrapper to gpiochip_add() that takes our specific gpio chip
+ * information and makes the necessary alterations for the platform and
+ * notes the information for use with the configuration systems and any
+ * other parts of the system.
+ */
+
+static void __init samsung_gpiolib_add(struct samsung_gpio_chip *chip)
+{
+	struct gpio_chip *gc = &chip->chip;
+	int ret;
+
+	BUG_ON(!chip->base);
+	BUG_ON(!gc->label);
+	BUG_ON(!gc->ngpio);
+
+	spin_lock_init(&chip->lock);
+
+	if (!gc->direction_input)
+		gc->direction_input = samsung_gpiolib_2bit_input;
+	if (!gc->direction_output)
+		gc->direction_output = samsung_gpiolib_2bit_output;
+	if (!gc->set)
+		gc->set = samsung_gpiolib_set;
+	if (!gc->get)
+		gc->get = samsung_gpiolib_get;
+
+#ifdef CONFIG_PM
+	if (chip->pm != NULL) {
+		if (!chip->pm->save || !chip->pm->resume)
+			pr_err("gpio: %s has missing PM functions\n",
+			       gc->label);
+	} else
+		pr_err("gpio: %s has no PM function\n", gc->label);
+#endif
+
+	/* gpiochip_add() prints own failure message on error. */
+	ret = gpiochip_add_data(gc, chip);
+	if (ret >= 0)
+		s3c_gpiolib_track(chip);
+}
+
+static void __init s3c24xx_gpiolib_add_chips(struct samsung_gpio_chip *chip,
+					     int nr_chips, void __iomem *base)
+{
+	int i;
+	struct gpio_chip *gc = &chip->chip;
+
+	for (i = 0 ; i < nr_chips; i++, chip++) {
+		/* skip banks not present on SoC */
+		if (chip->chip.base >= S3C_GPIO_END)
+			continue;
+
+		if (!chip->config)
+			chip->config = &s3c24xx_gpiocfg_default;
+		if (!chip->pm)
+			chip->pm = __gpio_pm(&samsung_gpio_pm_2bit);
+		if ((base != NULL) && (chip->base == NULL))
+			chip->base = base + ((i) * 0x10);
+
+		if (!gc->direction_input)
+			gc->direction_input = samsung_gpiolib_2bit_input;
+		if (!gc->direction_output)
+			gc->direction_output = samsung_gpiolib_2bit_output;
+
+		samsung_gpiolib_add(chip);
+	}
+}
+
+static void __init samsung_gpiolib_add_2bit_chips(struct samsung_gpio_chip *chip,
+						  int nr_chips, void __iomem *base,
+						  unsigned int offset)
+{
+	int i;
+
+	for (i = 0 ; i < nr_chips; i++, chip++) {
+		chip->chip.direction_input = samsung_gpiolib_2bit_input;
+		chip->chip.direction_output = samsung_gpiolib_2bit_output;
+
+		if (!chip->config)
+			chip->config = &samsung_gpio_cfgs[7];
+		if (!chip->pm)
+			chip->pm = __gpio_pm(&samsung_gpio_pm_2bit);
+		if ((base != NULL) && (chip->base == NULL))
+			chip->base = base + ((i) * offset);
+
+		samsung_gpiolib_add(chip);
+	}
+}
+
+/*
+ * samsung_gpiolib_add_4bit_chips - 4bit single register GPIO config.
+ * @chip: The gpio chip that is being configured.
+ * @nr_chips: The no of chips (gpio ports) for the GPIO being configured.
+ *
+ * This helper deal with the GPIO cases where the control register has 4 bits
+ * of control per GPIO, generally in the form of:
+ * 0000 = Input
+ * 0001 = Output
+ * others = Special functions (dependent on bank)
+ *
+ * Note, since the code to deal with the case where there are two control
+ * registers instead of one, we do not have a separate set of function
+ * (samsung_gpiolib_add_4bit2_chips)for each case.
+ */
+
+static void __init samsung_gpiolib_add_4bit_chips(struct samsung_gpio_chip *chip,
+						  int nr_chips, void __iomem *base)
+{
+	int i;
+
+	for (i = 0 ; i < nr_chips; i++, chip++) {
+		chip->chip.direction_input = samsung_gpiolib_4bit_input;
+		chip->chip.direction_output = samsung_gpiolib_4bit_output;
+
+		if (!chip->config)
+			chip->config = &samsung_gpio_cfgs[2];
+		if (!chip->pm)
+			chip->pm = __gpio_pm(&samsung_gpio_pm_4bit);
+		if ((base != NULL) && (chip->base == NULL))
+			chip->base = base + ((i) * 0x20);
+
+		chip->bitmap_gpio_int = 0;
+
+		samsung_gpiolib_add(chip);
+	}
+}
+
+static void __init samsung_gpiolib_add_4bit2_chips(struct samsung_gpio_chip *chip,
+						   int nr_chips)
+{
+	for (; nr_chips > 0; nr_chips--, chip++) {
+		chip->chip.direction_input = samsung_gpiolib_4bit2_input;
+		chip->chip.direction_output = samsung_gpiolib_4bit2_output;
+
+		if (!chip->config)
+			chip->config = &samsung_gpio_cfgs[2];
+		if (!chip->pm)
+			chip->pm = __gpio_pm(&samsung_gpio_pm_4bit);
+
+		samsung_gpiolib_add(chip);
+	}
+}
+
+int samsung_gpiolib_to_irq(struct gpio_chip *chip, unsigned int offset)
+{
+	struct samsung_gpio_chip *samsung_chip = gpiochip_get_data(chip);
+
+	return samsung_chip->irq_base + offset;
+}
+
+#ifdef CONFIG_PLAT_S3C24XX
+static int s3c24xx_gpiolib_fbank_to_irq(struct gpio_chip *chip, unsigned offset)
+{
+	if (offset < 4) {
+		if (soc_is_s3c2412())
+			return IRQ_EINT0_2412 + offset;
+		else
+			return IRQ_EINT0 + offset;
+	}
+
+	if (offset < 8)
+		return IRQ_EINT4 + offset - 4;
+
+	return -EINVAL;
+}
+#endif
+
+#ifdef CONFIG_ARCH_S3C64XX
+static int s3c64xx_gpiolib_mbank_to_irq(struct gpio_chip *chip, unsigned pin)
+{
+	return pin < 5 ? IRQ_EINT(23) + pin : -ENXIO;
+}
+
+static int s3c64xx_gpiolib_lbank_to_irq(struct gpio_chip *chip, unsigned pin)
+{
+	return pin >= 8 ? IRQ_EINT(16) + pin - 8 : -ENXIO;
+}
+#endif
+
+struct samsung_gpio_chip s3c24xx_gpios[] = {
+#ifdef CONFIG_PLAT_S3C24XX
+	{
+		.config	= &s3c24xx_gpiocfg_banka,
+		.chip	= {
+			.base			= S3C2410_GPA(0),
+			.owner			= THIS_MODULE,
+			.label			= "GPIOA",
+			.ngpio			= 27,
+			.direction_input	= s3c24xx_gpiolib_banka_input,
+			.direction_output	= s3c24xx_gpiolib_banka_output,
+		},
+	}, {
+		.chip	= {
+			.base	= S3C2410_GPB(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOB",
+			.ngpio	= 11,
+		},
+	}, {
+		.chip	= {
+			.base	= S3C2410_GPC(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOC",
+			.ngpio	= 16,
+		},
+	}, {
+		.chip	= {
+			.base	= S3C2410_GPD(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOD",
+			.ngpio	= 16,
+		},
+	}, {
+		.chip	= {
+			.base	= S3C2410_GPE(0),
+			.label	= "GPIOE",
+			.owner	= THIS_MODULE,
+			.ngpio	= 16,
+		},
+	}, {
+		.chip	= {
+			.base	= S3C2410_GPF(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOF",
+			.ngpio	= 8,
+			.to_irq	= s3c24xx_gpiolib_fbank_to_irq,
+		},
+	}, {
+		.irq_base = IRQ_EINT8,
+		.chip	= {
+			.base	= S3C2410_GPG(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOG",
+			.ngpio	= 16,
+			.to_irq	= samsung_gpiolib_to_irq,
+		},
+	}, {
+		.chip	= {
+			.base	= S3C2410_GPH(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOH",
+			.ngpio	= 15,
+		},
+	},
+		/* GPIOS for the S3C2443 and later devices. */
+	{
+		.base	= S3C2440_GPJCON,
+		.chip	= {
+			.base	= S3C2410_GPJ(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOJ",
+			.ngpio	= 16,
+		},
+	}, {
+		.base	= S3C2443_GPKCON,
+		.chip	= {
+			.base	= S3C2410_GPK(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOK",
+			.ngpio	= 16,
+		},
+	}, {
+		.base	= S3C2443_GPLCON,
+		.chip	= {
+			.base	= S3C2410_GPL(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOL",
+			.ngpio	= 15,
+		},
+	}, {
+		.base	= S3C2443_GPMCON,
+		.chip	= {
+			.base	= S3C2410_GPM(0),
+			.owner	= THIS_MODULE,
+			.label	= "GPIOM",
+			.ngpio	= 2,
+		},
+	},
+#endif
+};
+
+/*
+ * GPIO bank summary:
+ *
+ * Bank	GPIOs	Style	SlpCon	ExtInt Group
+ * A	8	4Bit	Yes	1
+ * B	7	4Bit	Yes	1
+ * C	8	4Bit	Yes	2
+ * D	5	4Bit	Yes	3
+ * E	5	4Bit	Yes	None
+ * F	16	2Bit	Yes	4 [1]
+ * G	7	4Bit	Yes	5
+ * H	10	4Bit[2]	Yes	6
+ * I	16	2Bit	Yes	None
+ * J	12	2Bit	Yes	None
+ * K	16	4Bit[2]	No	None
+ * L	15	4Bit[2] No	None
+ * M	6	4Bit	No	IRQ_EINT
+ * N	16	2Bit	No	IRQ_EINT
+ * O	16	2Bit	Yes	7
+ * P	15	2Bit	Yes	8
+ * Q	9	2Bit	Yes	9
+ *
+ * [1] BANKF pins 14,15 do not form part of the external interrupt sources
+ * [2] BANK has two control registers, GPxCON0 and GPxCON1
+ */
+
+static struct samsung_gpio_chip s3c64xx_gpios_4bit[] = {
+#ifdef CONFIG_ARCH_S3C64XX
+	{
+		.chip	= {
+			.base	= S3C64XX_GPA(0),
+			.ngpio	= S3C64XX_GPIO_A_NR,
+			.label	= "GPA",
+		},
+	}, {
+		.chip	= {
+			.base	= S3C64XX_GPB(0),
+			.ngpio	= S3C64XX_GPIO_B_NR,
+			.label	= "GPB",
+		},
+	}, {
+		.chip	= {
+			.base	= S3C64XX_GPC(0),
+			.ngpio	= S3C64XX_GPIO_C_NR,
+			.label	= "GPC",
+		},
+	}, {
+		.chip	= {
+			.base	= S3C64XX_GPD(0),
+			.ngpio	= S3C64XX_GPIO_D_NR,
+			.label	= "GPD",
+		},
+	}, {
+		.config	= &samsung_gpio_cfgs[0],
+		.chip	= {
+			.base	= S3C64XX_GPE(0),
+			.ngpio	= S3C64XX_GPIO_E_NR,
+			.label	= "GPE",
+		},
+	}, {
+		.base	= S3C64XX_GPG_BASE,
+		.chip	= {
+			.base	= S3C64XX_GPG(0),
+			.ngpio	= S3C64XX_GPIO_G_NR,
+			.label	= "GPG",
+		},
+	}, {
+		.base	= S3C64XX_GPM_BASE,
+		.config	= &samsung_gpio_cfgs[1],
+		.chip	= {
+			.base	= S3C64XX_GPM(0),
+			.ngpio	= S3C64XX_GPIO_M_NR,
+			.label	= "GPM",
+			.to_irq = s3c64xx_gpiolib_mbank_to_irq,
+		},
+	},
+#endif
+};
+
+static struct samsung_gpio_chip s3c64xx_gpios_4bit2[] = {
+#ifdef CONFIG_ARCH_S3C64XX
+	{
+		.base	= S3C64XX_GPH_BASE + 0x4,
+		.chip	= {
+			.base	= S3C64XX_GPH(0),
+			.ngpio	= S3C64XX_GPIO_H_NR,
+			.label	= "GPH",
+		},
+	}, {
+		.base	= S3C64XX_GPK_BASE + 0x4,
+		.config	= &samsung_gpio_cfgs[0],
+		.chip	= {
+			.base	= S3C64XX_GPK(0),
+			.ngpio	= S3C64XX_GPIO_K_NR,
+			.label	= "GPK",
+		},
+	}, {
+		.base	= S3C64XX_GPL_BASE + 0x4,
+		.config	= &samsung_gpio_cfgs[1],
+		.chip	= {
+			.base	= S3C64XX_GPL(0),
+			.ngpio	= S3C64XX_GPIO_L_NR,
+			.label	= "GPL",
+			.to_irq = s3c64xx_gpiolib_lbank_to_irq,
+		},
+	},
+#endif
+};
+
+static struct samsung_gpio_chip s3c64xx_gpios_2bit[] = {
+#ifdef CONFIG_ARCH_S3C64XX
+	{
+		.base	= S3C64XX_GPF_BASE,
+		.config	= &samsung_gpio_cfgs[6],
+		.chip	= {
+			.base	= S3C64XX_GPF(0),
+			.ngpio	= S3C64XX_GPIO_F_NR,
+			.label	= "GPF",
+		},
+	}, {
+		.config	= &samsung_gpio_cfgs[7],
+		.chip	= {
+			.base	= S3C64XX_GPI(0),
+			.ngpio	= S3C64XX_GPIO_I_NR,
+			.label	= "GPI",
+		},
+	}, {
+		.config	= &samsung_gpio_cfgs[7],
+		.chip	= {
+			.base	= S3C64XX_GPJ(0),
+			.ngpio	= S3C64XX_GPIO_J_NR,
+			.label	= "GPJ",
+		},
+	}, {
+		.config	= &samsung_gpio_cfgs[6],
+		.chip	= {
+			.base	= S3C64XX_GPO(0),
+			.ngpio	= S3C64XX_GPIO_O_NR,
+			.label	= "GPO",
+		},
+	}, {
+		.config	= &samsung_gpio_cfgs[6],
+		.chip	= {
+			.base	= S3C64XX_GPP(0),
+			.ngpio	= S3C64XX_GPIO_P_NR,
+			.label	= "GPP",
+		},
+	}, {
+		.config	= &samsung_gpio_cfgs[6],
+		.chip	= {
+			.base	= S3C64XX_GPQ(0),
+			.ngpio	= S3C64XX_GPIO_Q_NR,
+			.label	= "GPQ",
+		},
+	}, {
+		.base	= S3C64XX_GPN_BASE,
+		.irq_base = IRQ_EINT(0),
+		.config	= &samsung_gpio_cfgs[5],
+		.chip	= {
+			.base	= S3C64XX_GPN(0),
+			.ngpio	= S3C64XX_GPIO_N_NR,
+			.label	= "GPN",
+			.to_irq = samsung_gpiolib_to_irq,
+		},
+	},
+#endif
+};
+
+/* TODO: cleanup soc_is_* */
+static __init int samsung_gpiolib_init(void)
+{
+	/*
+	 * Currently there are two drivers that can provide GPIO support for
+	 * Samsung SoCs. For device tree enabled platforms, the new
+	 * pinctrl-samsung driver is used, providing both GPIO and pin control
+	 * interfaces. For legacy (non-DT) platforms this driver is used.
+	 */
+	if (of_have_populated_dt())
+		return 0;
+
+	if (soc_is_s3c24xx()) {
+		samsung_gpiolib_set_cfg(samsung_gpio_cfgs,
+				ARRAY_SIZE(samsung_gpio_cfgs));
+		s3c24xx_gpiolib_add_chips(s3c24xx_gpios,
+				ARRAY_SIZE(s3c24xx_gpios), S3C24XX_VA_GPIO);
+	} else if (soc_is_s3c64xx()) {
+		samsung_gpiolib_set_cfg(samsung_gpio_cfgs,
+				ARRAY_SIZE(samsung_gpio_cfgs));
+		samsung_gpiolib_add_2bit_chips(s3c64xx_gpios_2bit,
+				ARRAY_SIZE(s3c64xx_gpios_2bit),
+				S3C64XX_VA_GPIO + 0xE0, 0x20);
+		samsung_gpiolib_add_4bit_chips(s3c64xx_gpios_4bit,
+				ARRAY_SIZE(s3c64xx_gpios_4bit),
+				S3C64XX_VA_GPIO);
+		samsung_gpiolib_add_4bit2_chips(s3c64xx_gpios_4bit2,
+				ARRAY_SIZE(s3c64xx_gpios_4bit2));
+	}
+
+	return 0;
+}
+core_initcall(samsung_gpiolib_init);
+
+int s3c_gpio_cfgpin(unsigned int pin, unsigned int config)
+{
+	struct samsung_gpio_chip *chip = samsung_gpiolib_getchip(pin);
+	unsigned long flags;
+	int offset;
+	int ret;
+
+	if (!chip)
+		return -EINVAL;
+
+	offset = pin - chip->chip.base;
+
+	samsung_gpio_lock(chip, flags);
+	ret = samsung_gpio_do_setcfg(chip, offset, config);
+	samsung_gpio_unlock(chip, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(s3c_gpio_cfgpin);
+
+int s3c_gpio_cfgpin_range(unsigned int start, unsigned int nr,
+			  unsigned int cfg)
+{
+	int ret;
+
+	for (; nr > 0; nr--, start++) {
+		ret = s3c_gpio_cfgpin(start, cfg);
+		if (ret != 0)
+			return ret;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(s3c_gpio_cfgpin_range);
+
+int s3c_gpio_cfgall_range(unsigned int start, unsigned int nr,
+			  unsigned int cfg, samsung_gpio_pull_t pull)
+{
+	int ret;
+
+	for (; nr > 0; nr--, start++) {
+		s3c_gpio_setpull(start, pull);
+		ret = s3c_gpio_cfgpin(start, cfg);
+		if (ret != 0)
+			return ret;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(s3c_gpio_cfgall_range);
+
+unsigned s3c_gpio_getcfg(unsigned int pin)
+{
+	struct samsung_gpio_chip *chip = samsung_gpiolib_getchip(pin);
+	unsigned long flags;
+	unsigned ret = 0;
+	int offset;
+
+	if (chip) {
+		offset = pin - chip->chip.base;
+
+		samsung_gpio_lock(chip, flags);
+		ret = samsung_gpio_do_getcfg(chip, offset);
+		samsung_gpio_unlock(chip, flags);
+	}
+
+	return ret;
+}
+EXPORT_SYMBOL(s3c_gpio_getcfg);
+
+int s3c_gpio_setpull(unsigned int pin, samsung_gpio_pull_t pull)
+{
+	struct samsung_gpio_chip *chip = samsung_gpiolib_getchip(pin);
+	unsigned long flags;
+	int offset, ret;
+
+	if (!chip)
+		return -EINVAL;
+
+	offset = pin - chip->chip.base;
+
+	samsung_gpio_lock(chip, flags);
+	ret = samsung_gpio_do_setpull(chip, offset, pull);
+	samsung_gpio_unlock(chip, flags);
+
+	return ret;
+}
+EXPORT_SYMBOL(s3c_gpio_setpull);
+
+samsung_gpio_pull_t s3c_gpio_getpull(unsigned int pin)
+{
+	struct samsung_gpio_chip *chip = samsung_gpiolib_getchip(pin);
+	unsigned long flags;
+	int offset;
+	u32 pup = 0;
+
+	if (chip) {
+		offset = pin - chip->chip.base;
+
+		samsung_gpio_lock(chip, flags);
+		pup = samsung_gpio_do_getpull(chip, offset);
+		samsung_gpio_unlock(chip, flags);
+	}
+
+	return (__force samsung_gpio_pull_t)pup;
+}
+EXPORT_SYMBOL(s3c_gpio_getpull);
+
+#ifdef CONFIG_PLAT_S3C24XX
+unsigned int s3c2410_modify_misccr(unsigned int clear, unsigned int change)
+{
+	unsigned long flags;
+	unsigned long misccr;
+
+	local_irq_save(flags);
+	misccr = __raw_readl(S3C24XX_MISCCR);
+	misccr &= ~clear;
+	misccr ^= change;
+	__raw_writel(misccr, S3C24XX_MISCCR);
+	local_irq_restore(flags);
+
+	return misccr;
+}
+EXPORT_SYMBOL(s3c2410_modify_misccr);
+#endif

arch/arm/plat-samsung/include/plat/adc-core.h

@@ -0,0 +1,28 @@
+/* linux/arch/arm/plat-samsung/include/plat/adc-core.h
+ *
+ * Copyright (c) 2010 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com/
+ *
+ * Samsung ADC Controller core functions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __ASM_PLAT_ADC_CORE_H
+#define __ASM_PLAT_ADC_CORE_H __FILE__
+
+/* These functions are only for use with the core support code, such as
+ * the cpu specific initialisation code
+ */
+
+/* re-define device name depending on support. */
+static inline void s3c_adc_setname(char *name)
+{
+#if defined(CONFIG_SAMSUNG_DEV_ADC) || defined(CONFIG_PLAT_S3C24XX)
+	s3c_device_adc.name = name;
+#endif
+}
+
+#endif /* __ASM_PLAT_ADC_CORE_H */

arch/arm/plat-samsung/include/plat/adc.h

@@ -0,0 +1,36 @@
+/* arch/arm/plat-samsung/include/plat/adc.h
+ *
+ * Copyright (c) 2008 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/	
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C ADC driver information
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __ASM_PLAT_ADC_H
+#define __ASM_PLAT_ADC_H __FILE__
+
+struct s3c_adc_client;
+struct platform_device;
+
+extern int s3c_adc_start(struct s3c_adc_client *client,
+			 unsigned int channel, unsigned int nr_samples);
+
+extern int s3c_adc_read(struct s3c_adc_client *client, unsigned int ch);
+
+extern struct s3c_adc_client *
+	s3c_adc_register(struct platform_device *pdev,
+			 void (*select)(struct s3c_adc_client *client,
+					unsigned selected),
+			 void (*conv)(struct s3c_adc_client *client,
+				      unsigned d0, unsigned d1,
+				      unsigned *samples_left),
+			 unsigned int is_ts);
+
+extern void s3c_adc_release(struct s3c_adc_client *client);
+
+#endif /* __ASM_PLAT_ADC_H */

arch/arm/plat-samsung/include/plat/cpu-freq-core.h

@@ -0,0 +1,291 @@
+/* arch/arm/plat-samsung/include/plat/cpu-freq-core.h
+ *
+ * Copyright (c) 2006-2009 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C CPU frequency scaling support - core support
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <plat/cpu-freq.h>
+
+struct seq_file;
+
+#define MAX_BANKS (8)
+#define S3C2412_MAX_IO	(8)
+
+/**
+ * struct s3c2410_iobank_timing - IO bank timings for S3C2410 style timings
+ * @bankcon: The cached version of settings in this structure.
+ * @tacp:
+ * @tacs: Time from address valid to nCS asserted.
+ * @tcos: Time from nCS asserted to nOE or nWE asserted.
+ * @tacc: Time that nOE or nWE is asserted.
+ * @tcoh: Time nCS is held after nOE or nWE are released.
+ * @tcah: Time address is held for after
+ * @nwait_en: Whether nWAIT is enabled for this bank.
+ *
+ * This structure represents the IO timings for a S3C2410 style IO bank
+ * used by the CPU frequency support if it needs to change the settings
+ * of the IO.
+ */
+struct s3c2410_iobank_timing {
+	unsigned long	bankcon;
+	unsigned int	tacp;
+	unsigned int	tacs;
+	unsigned int	tcos;
+	unsigned int	tacc;
+	unsigned int	tcoh;		/* nCS hold after nOE/nWE */
+	unsigned int	tcah;		/* Address hold after nCS */
+	unsigned char	nwait_en;	/* nWait enabled for bank. */
+};
+
+/**
+ * struct s3c2412_iobank_timing - io timings for PL092 (S3C2412) style IO
+ * @idcy: The idle cycle time between transactions.
+ * @wstrd: nCS release to end of read cycle.
+ * @wstwr: nCS release to end of write cycle.
+ * @wstoen: nCS assertion to nOE assertion time.
+ * @wstwen: nCS assertion to nWE assertion time.
+ * @wstbrd: Burst ready delay.
+ * @smbidcyr: Register cache for smbidcyr value.
+ * @smbwstrd: Register cache for smbwstrd value.
+ * @smbwstwr: Register cache for smbwstwr value.
+ * @smbwstoen: Register cache for smbwstoen value.
+ * @smbwstwen: Register cache for smbwstwen value.
+ * @smbwstbrd: Register cache for smbwstbrd value.
+ *
+ * Timing information for a IO bank on an S3C2412 or similar system which
+ * uses a PL093 block.
+ */
+struct s3c2412_iobank_timing {
+	unsigned int	idcy;
+	unsigned int	wstrd;
+	unsigned int	wstwr;
+	unsigned int	wstoen;
+	unsigned int	wstwen;
+	unsigned int	wstbrd;
+
+	/* register cache */
+	unsigned char	smbidcyr;
+	unsigned char	smbwstrd;
+	unsigned char	smbwstwr;
+	unsigned char	smbwstoen;
+	unsigned char	smbwstwen;
+	unsigned char	smbwstbrd;
+};
+
+union s3c_iobank {
+	struct s3c2410_iobank_timing	*io_2410;
+	struct s3c2412_iobank_timing	*io_2412;
+};
+
+/**
+ * struct s3c_iotimings - Chip IO timings holder
+ * @bank: The timings for each IO bank.
+ */
+struct s3c_iotimings {
+	union s3c_iobank	bank[MAX_BANKS];
+};
+
+/**
+ * struct s3c_plltab - PLL table information.
+ * @vals: List of PLL values.
+ * @size: Size of the PLL table @vals.
+ */
+struct s3c_plltab {
+	struct s3c_pllval	*vals;
+	int			 size;
+};
+
+/**
+ * struct s3c_cpufreq_config - current cpu frequency configuration
+ * @freq: The current settings for the core clocks.
+ * @max: Maxium settings, derived from core, board and user settings.
+ * @pll: The PLL table entry for the current PLL settings.
+ * @divs: The divisor settings for the core clocks.
+ * @info: The current core driver information.
+ * @board: The information for the board we are running on.
+ * @lock_pll: Set if the PLL settings cannot be changed.
+ *
+ * This is for the core drivers that need to know information about
+ * the current settings and values. It should not be needed by any
+ * device drivers.
+*/
+struct s3c_cpufreq_config {
+	struct s3c_freq		freq;
+	struct s3c_freq		max;
+	struct clk		*mpll;
+	struct cpufreq_frequency_table pll;
+	struct s3c_clkdivs	divs;
+	struct s3c_cpufreq_info *info;	/* for core, not drivers */
+	struct s3c_cpufreq_board *board;
+
+	unsigned int	lock_pll:1;
+};
+
+/**
+ * struct s3c_cpufreq_info - Information for the CPU frequency driver.
+ * @name: The name of this implementation.
+ * @max: The maximum frequencies for the system.
+ * @latency: Transition latency to give to cpufreq.
+ * @locktime_m: The lock-time in uS for the MPLL.
+ * @locktime_u: The lock-time in uS for the UPLL.
+ * @locttime_bits: The number of bits each LOCKTIME field.
+ * @need_pll: Set if this driver needs to change the PLL values to achieve
+ *	any frequency changes. This is really only need by devices like the
+ *	S3C2410 where there is no or limited divider between the PLL and the
+ *	ARMCLK.
+ * @get_iotiming: Get the current IO timing data, mainly for use at start.
+ * @set_iotiming: Update the IO timings from the cached copies calculated
+ *	from the @calc_iotiming entry when changing the frequency.
+ * @calc_iotiming: Calculate and update the cached copies of the IO timings
+ *	from the newly calculated frequencies.
+ * @calc_freqtable: Calculate (fill in) the given frequency table from the
+ *	current frequency configuration. If the table passed in is NULL,
+ *	then the return is the number of elements to be filled for allocation
+ *	of the table.
+ * @set_refresh: Set the memory refresh configuration.
+ * @set_fvco: Set the PLL frequencies.
+ * @set_divs: Update the clock divisors.
+ * @calc_divs: Calculate the clock divisors.
+ */
+struct s3c_cpufreq_info {
+	const char		*name;
+	struct s3c_freq		max;
+
+	unsigned int		latency;
+
+	unsigned int		locktime_m;
+	unsigned int		locktime_u;
+	unsigned char		locktime_bits;
+
+	unsigned int		need_pll:1;
+
+	/* driver routines */
+
+	int		(*get_iotiming)(struct s3c_cpufreq_config *cfg,
+					struct s3c_iotimings *timings);
+
+	void		(*set_iotiming)(struct s3c_cpufreq_config *cfg,
+					struct s3c_iotimings *timings);
+
+	int		(*calc_iotiming)(struct s3c_cpufreq_config *cfg,
+					 struct s3c_iotimings *timings);
+
+	int		(*calc_freqtable)(struct s3c_cpufreq_config *cfg,
+					  struct cpufreq_frequency_table *t,
+					  size_t table_size);
+
+	void		(*debug_io_show)(struct seq_file *seq,
+					 struct s3c_cpufreq_config *cfg,
+					 union s3c_iobank *iob);
+
+	void		(*set_refresh)(struct s3c_cpufreq_config *cfg);
+	void		(*set_fvco)(struct s3c_cpufreq_config *cfg);
+	void		(*set_divs)(struct s3c_cpufreq_config *cfg);
+	int		(*calc_divs)(struct s3c_cpufreq_config *cfg);
+};
+
+extern int s3c_cpufreq_register(struct s3c_cpufreq_info *info);
+
+extern int s3c_plltab_register(struct cpufreq_frequency_table *plls,
+			       unsigned int plls_no);
+
+/* exports and utilities for debugfs */
+extern struct s3c_cpufreq_config *s3c_cpufreq_getconfig(void);
+extern struct s3c_iotimings *s3c_cpufreq_getiotimings(void);
+
+#ifdef CONFIG_ARM_S3C24XX_CPUFREQ_DEBUGFS
+#define s3c_cpufreq_debugfs_call(x) x
+#else
+#define s3c_cpufreq_debugfs_call(x) NULL
+#endif
+
+/* Useful utility functions. */
+
+extern struct clk *s3c_cpufreq_clk_get(struct device *, const char *);
+
+/* S3C2410 and compatible exported functions */
+
+extern void s3c2410_cpufreq_setrefresh(struct s3c_cpufreq_config *cfg);
+extern void s3c2410_set_fvco(struct s3c_cpufreq_config *cfg);
+
+#ifdef CONFIG_S3C2410_IOTIMING
+extern void s3c2410_iotiming_debugfs(struct seq_file *seq,
+				     struct s3c_cpufreq_config *cfg,
+				     union s3c_iobank *iob);
+
+extern int s3c2410_iotiming_calc(struct s3c_cpufreq_config *cfg,
+				 struct s3c_iotimings *iot);
+
+extern int s3c2410_iotiming_get(struct s3c_cpufreq_config *cfg,
+				struct s3c_iotimings *timings);
+
+extern void s3c2410_iotiming_set(struct s3c_cpufreq_config *cfg,
+				 struct s3c_iotimings *iot);
+#else
+#define s3c2410_iotiming_debugfs NULL
+#define s3c2410_iotiming_calc NULL
+#define s3c2410_iotiming_get NULL
+#define s3c2410_iotiming_set NULL
+#endif /* CONFIG_S3C2410_IOTIMING */
+
+/* S3C2412 compatible routines */
+
+#ifdef CONFIG_S3C2412_IOTIMING
+extern void s3c2412_iotiming_debugfs(struct seq_file *seq,
+				     struct s3c_cpufreq_config *cfg,
+				     union s3c_iobank *iob);
+
+extern int s3c2412_iotiming_get(struct s3c_cpufreq_config *cfg,
+				struct s3c_iotimings *timings);
+
+extern int s3c2412_iotiming_calc(struct s3c_cpufreq_config *cfg,
+				 struct s3c_iotimings *iot);
+
+extern void s3c2412_iotiming_set(struct s3c_cpufreq_config *cfg,
+				 struct s3c_iotimings *iot);
+#else
+#define s3c2412_iotiming_debugfs NULL
+#define s3c2412_iotiming_calc NULL
+#define s3c2412_iotiming_get NULL
+#define s3c2412_iotiming_set NULL
+#endif /* CONFIG_S3C2412_IOTIMING */
+
+#ifdef CONFIG_ARM_S3C24XX_CPUFREQ_DEBUG
+#define s3c_freq_dbg(x...) printk(KERN_INFO x)
+#else
+#define s3c_freq_dbg(x...) do { if (0) printk(x); } while (0)
+#endif /* CONFIG_ARM_S3C24XX_CPUFREQ_DEBUG */
+
+#ifdef CONFIG_ARM_S3C24XX_CPUFREQ_IODEBUG
+#define s3c_freq_iodbg(x...) printk(KERN_INFO x)
+#else
+#define s3c_freq_iodbg(x...) do { if (0) printk(x); } while (0)
+#endif /* CONFIG_ARM_S3C24XX_CPUFREQ_IODEBUG */
+
+static inline int s3c_cpufreq_addfreq(struct cpufreq_frequency_table *table,
+				      int index, size_t table_size,
+				      unsigned int freq)
+{
+	if (index < 0)
+		return index;
+
+	if (table) {
+		if (index >= table_size)
+			return -ENOMEM;
+
+		s3c_freq_dbg("%s: { %d = %u kHz }\n",
+			     __func__, index, freq);
+
+		table[index].driver_data = index;
+		table[index].frequency = freq;
+	}
+
+	return index + 1;
+}

arch/arm/plat-samsung/include/plat/cpu-freq.h

@@ -0,0 +1,145 @@
+/* arch/arm/plat-samsung/include/plat/cpu-freq.h
+ *
+ * Copyright (c) 2006-2007 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C CPU frequency scaling support - driver and board
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/cpufreq.h>
+
+struct s3c_cpufreq_info;
+struct s3c_cpufreq_board;
+struct s3c_iotimings;
+
+/**
+ * struct s3c_freq - frequency information (mainly for core drivers)
+ * @fclk: The FCLK frequency in Hz.
+ * @armclk: The ARMCLK frequency in Hz.
+ * @hclk_tns: HCLK cycle time in 10ths of nano-seconds.
+ * @hclk: The HCLK frequency in Hz.
+ * @pclk: The PCLK frequency in Hz.
+ *
+ * This contains the frequency information about the current configuration
+ * mainly for the core drivers to ensure we do not end up passing about
+ * a large number of parameters.
+ *
+ * The @hclk_tns field is a useful cache for the parts of the drivers that
+ * need to calculate IO timings and suchlike.
+ */
+struct s3c_freq {
+	unsigned long	fclk;
+	unsigned long	armclk;
+	unsigned long	hclk_tns;	/* in 10ths of ns */
+	unsigned long	hclk;
+	unsigned long	pclk;
+};
+
+/**
+ * struct s3c_cpufreq_freqs - s3c cpufreq notification information.
+ * @freqs: The cpufreq setting information.
+ * @old: The old clock settings.
+ * @new: The new clock settings.
+ * @pll_changing: Set if the PLL is changing.
+ *
+ * Wrapper 'struct cpufreq_freqs' so that any drivers receiving the
+ * notification can use this information that is not provided by just
+ * having the core frequency alone.
+ *
+ * The pll_changing flag is used to indicate if the PLL itself is
+ * being set during this change. This is important as the clocks
+ * will temporarily be set to the XTAL clock during this time, so
+ * drivers may want to close down their output during this time.
+ *
+ * Note, this is not being used by any current drivers and therefore
+ * may be removed in the future.
+ */
+struct s3c_cpufreq_freqs {
+	struct cpufreq_freqs	freqs;
+	struct s3c_freq		old;
+	struct s3c_freq		new;
+
+	unsigned int		pll_changing:1;
+};
+
+#define to_s3c_cpufreq(_cf) container_of(_cf, struct s3c_cpufreq_freqs, freqs)
+
+/**
+ * struct s3c_clkdivs - clock divisor information
+ * @p_divisor: Divisor from FCLK to PCLK.
+ * @h_divisor: Divisor from FCLK to HCLK.
+ * @arm_divisor: Divisor from FCLK to ARMCLK (not all CPUs).
+ * @dvs: Non-zero if using DVS mode for ARMCLK.
+ *
+ * Divisor settings for the core clocks.
+ */
+struct s3c_clkdivs {
+	int		p_divisor;
+	int		h_divisor;
+	int		arm_divisor;
+	unsigned char	dvs;
+};
+
+#define PLLVAL(_m, _p, _s) (((_m) << 12) | ((_p) << 4) | (_s))
+
+/**
+ * struct s3c_pllval - PLL value entry.
+ * @freq: The frequency for this entry in Hz.
+ * @pll_reg: The PLL register setting for this PLL value.
+ */
+struct s3c_pllval {
+	unsigned long		freq;
+	unsigned long		pll_reg;
+};
+
+/**
+ * struct s3c_cpufreq_board - per-board cpu frequency informatin
+ * @refresh: The SDRAM refresh period in nanoseconds.
+ * @auto_io: Set if the IO timing settings should be generated from the
+ *	initialisation time hardware registers.
+ * @need_io: Set if the board has external IO on any of the chipselect
+ *	lines that will require the hardware timing registers to be
+ *	updated on a clock change.
+ * @max: The maxium frequency limits for the system. Any field that
+ *	is left at zero will use the CPU's settings.
+ *
+ * This contains the board specific settings that affect how the CPU
+ * drivers chose settings. These include the memory refresh and IO
+ * timing information.
+ *
+ * Registration depends on the driver being used, the ARMCLK only
+ * implementation does not currently need this but the older style
+ * driver requires this to be available.
+ */
+struct s3c_cpufreq_board {
+	unsigned int	refresh;
+	unsigned int	auto_io:1;	/* automatically init io timings. */
+	unsigned int	need_io:1;	/* set if needs io timing support. */
+
+	/* any non-zero field in here is taken as an upper limit. */
+	struct s3c_freq	max;	/* frequency limits */
+};
+
+/* Things depending on frequency scaling. */
+#ifdef CONFIG_ARM_S3C_CPUFREQ
+#define __init_or_cpufreq
+#else
+#define __init_or_cpufreq __init
+#endif
+
+/* Board functions */
+
+#ifdef CONFIG_ARM_S3C_CPUFREQ
+extern int s3c_cpufreq_setboard(struct s3c_cpufreq_board *board);
+#else
+
+static inline int s3c_cpufreq_setboard(struct s3c_cpufreq_board *board)
+{
+	return 0;
+}
+#endif  /* CONFIG_ARM_S3C_CPUFREQ */

arch/arm/plat-samsung/include/plat/cpu.h

@@ -0,0 +1,144 @@
+/* linux/arch/arm/plat-samsung/include/plat/cpu.h
+ *
+ * Copyright (c) 2011 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com/
+ *
+ * Copyright (c) 2004-2005 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * Header file for Samsung CPU support
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+/* todo - fix when rmk changes iodescs to use `void __iomem *` */
+
+#ifndef __SAMSUNG_PLAT_CPU_H
+#define __SAMSUNG_PLAT_CPU_H
+
+extern unsigned long samsung_cpu_id;
+
+#define S3C2410_CPU_ID		0x32410000
+#define S3C2410_CPU_MASK	0xFFFFFFFF
+
+#define S3C24XX_CPU_ID		0x32400000
+#define S3C24XX_CPU_MASK	0xFFF00000
+
+#define S3C2412_CPU_ID		0x32412000
+#define S3C2412_CPU_MASK	0xFFFFF000
+
+#define S3C6400_CPU_ID		0x36400000
+#define S3C6410_CPU_ID		0x36410000
+#define S3C64XX_CPU_MASK	0xFFFFF000
+
+#define S5PV210_CPU_ID		0x43110000
+#define S5PV210_CPU_MASK	0xFFFFF000
+
+#define IS_SAMSUNG_CPU(name, id, mask)		\
+static inline int is_samsung_##name(void)	\
+{						\
+	return ((samsung_cpu_id & mask) == (id & mask));	\
+}
+
+IS_SAMSUNG_CPU(s3c2410, S3C2410_CPU_ID, S3C2410_CPU_MASK)
+IS_SAMSUNG_CPU(s3c24xx, S3C24XX_CPU_ID, S3C24XX_CPU_MASK)
+IS_SAMSUNG_CPU(s3c2412, S3C2412_CPU_ID, S3C2412_CPU_MASK)
+IS_SAMSUNG_CPU(s3c6400, S3C6400_CPU_ID, S3C64XX_CPU_MASK)
+IS_SAMSUNG_CPU(s3c6410, S3C6410_CPU_ID, S3C64XX_CPU_MASK)
+
+#if defined(CONFIG_CPU_S3C2410) || defined(CONFIG_CPU_S3C2412) || \
+    defined(CONFIG_CPU_S3C2416) || defined(CONFIG_CPU_S3C2440) || \
+    defined(CONFIG_CPU_S3C2442) || defined(CONFIG_CPU_S3C244X) || \
+    defined(CONFIG_CPU_S3C2443)
+# define soc_is_s3c24xx()	is_samsung_s3c24xx()
+# define soc_is_s3c2410()	is_samsung_s3c2410()
+#else
+# define soc_is_s3c24xx()	0
+# define soc_is_s3c2410()	0
+#endif
+
+#if defined(CONFIG_CPU_S3C2412)
+# define soc_is_s3c2412()	is_samsung_s3c2412()
+#else
+# define soc_is_s3c2412()	0
+#endif
+
+#if defined(CONFIG_CPU_S3C6400) || defined(CONFIG_CPU_S3C6410)
+# define soc_is_s3c6400()	is_samsung_s3c6400()
+# define soc_is_s3c6410()	is_samsung_s3c6410()
+# define soc_is_s3c64xx()	(is_samsung_s3c6400() || is_samsung_s3c6410())
+#else
+# define soc_is_s3c6400()	0
+# define soc_is_s3c6410()	0
+# define soc_is_s3c64xx()	0
+#endif
+
+#define IODESC_ENT(x) { (unsigned long)S3C24XX_VA_##x, __phys_to_pfn(S3C24XX_PA_##x), S3C24XX_SZ_##x, MT_DEVICE }
+
+#ifndef KHZ
+#define KHZ (1000)
+#endif
+
+#ifndef MHZ
+#define MHZ (1000*1000)
+#endif
+
+#define print_mhz(m) ((m) / MHZ), (((m) / 1000) % 1000)
+
+/* forward declaration */
+struct s3c24xx_uart_resources;
+struct platform_device;
+struct s3c2410_uartcfg;
+struct map_desc;
+
+/* per-cpu initialisation function table. */
+
+struct cpu_table {
+	unsigned long	idcode;
+	unsigned long	idmask;
+	void		(*map_io)(void);
+	void		(*init_uarts)(struct s3c2410_uartcfg *cfg, int no);
+	void		(*init_clocks)(int xtal);
+	int		(*init)(void);
+	const char	*name;
+};
+
+extern void s3c_init_cpu(unsigned long idcode,
+			 struct cpu_table *cpus, unsigned int cputab_size);
+
+/* core initialisation functions */
+
+extern void s3c24xx_init_io(struct map_desc *mach_desc, int size);
+
+extern void s3c64xx_init_cpu(void);
+extern void s5p_init_cpu(const void __iomem *cpuid_addr);
+
+extern unsigned int samsung_rev(void);
+
+extern void s3c24xx_init_uarts(struct s3c2410_uartcfg *cfg, int no);
+
+extern void s3c24xx_init_clocks(int xtal);
+
+extern void s3c24xx_init_uartdevs(char *name,
+				  struct s3c24xx_uart_resources *res,
+				  struct s3c2410_uartcfg *cfg, int no);
+
+extern struct syscore_ops s3c2410_pm_syscore_ops;
+extern struct syscore_ops s3c2412_pm_syscore_ops;
+extern struct syscore_ops s3c2416_pm_syscore_ops;
+extern struct syscore_ops s3c244x_pm_syscore_ops;
+
+/* system device subsystems */
+
+extern struct bus_type s3c2410_subsys;
+extern struct bus_type s3c2410a_subsys;
+extern struct bus_type s3c2412_subsys;
+extern struct bus_type s3c2416_subsys;
+extern struct bus_type s3c2440_subsys;
+extern struct bus_type s3c2442_subsys;
+extern struct bus_type s3c2443_subsys;
+extern struct bus_type s3c6410_subsys;
+
+#endif

arch/arm/plat-samsung/include/plat/devs.h

@@ -0,0 +1,100 @@
+/* arch/arm/plat-samsung/include/plat/devs.h
+ *
+ * Copyright (c) 2011 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com
+ *
+ * Copyright (c) 2004 Simtec Electronics
+ * Ben Dooks <ben@simtec.co.uk>
+ *
+ * Header file for s3c2410 standard platform devices
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __PLAT_DEVS_H
+#define __PLAT_DEVS_H __FILE__
+
+#include <linux/platform_device.h>
+
+struct s3c24xx_uart_resources {
+	struct resource		*resources;
+	unsigned long		 nr_resources;
+};
+
+extern struct s3c24xx_uart_resources s3c2410_uart_resources[];
+extern struct s3c24xx_uart_resources s3c64xx_uart_resources[];
+
+extern struct platform_device *s3c24xx_uart_devs[];
+extern struct platform_device *s3c24xx_uart_src[];
+
+extern struct platform_device s3c64xx_device_ac97;
+extern struct platform_device s3c64xx_device_iis0;
+extern struct platform_device s3c64xx_device_iis1;
+extern struct platform_device s3c64xx_device_iisv4;
+extern struct platform_device s3c64xx_device_onenand1;
+extern struct platform_device s3c64xx_device_pcm0;
+extern struct platform_device s3c64xx_device_pcm1;
+extern struct platform_device s3c64xx_device_spi0;
+extern struct platform_device s3c64xx_device_spi1;
+extern struct platform_device s3c64xx_device_spi2;
+
+extern struct platform_device s3c_device_adc;
+extern struct platform_device s3c_device_cfcon;
+extern struct platform_device s3c_device_fb;
+extern struct platform_device s3c_device_hwmon;
+extern struct platform_device s3c_device_hsmmc0;
+extern struct platform_device s3c_device_hsmmc1;
+extern struct platform_device s3c_device_hsmmc2;
+extern struct platform_device s3c_device_hsmmc3;
+extern struct platform_device s3c_device_i2c0;
+extern struct platform_device s3c_device_i2c1;
+extern struct platform_device s3c_device_i2c2;
+extern struct platform_device s3c_device_i2c3;
+extern struct platform_device s3c_device_i2c4;
+extern struct platform_device s3c_device_i2c5;
+extern struct platform_device s3c_device_i2c6;
+extern struct platform_device s3c_device_i2c7;
+extern struct platform_device s3c_device_iis;
+extern struct platform_device s3c_device_lcd;
+extern struct platform_device s3c_device_nand;
+extern struct platform_device s3c_device_ohci;
+extern struct platform_device s3c_device_onenand;
+extern struct platform_device s3c_device_rtc;
+extern struct platform_device s3c_device_sdi;
+extern struct platform_device s3c_device_spi0;
+extern struct platform_device s3c_device_spi1;
+extern struct platform_device s3c_device_ts;
+extern struct platform_device s3c_device_timer[];
+extern struct platform_device s3c_device_usbgadget;
+extern struct platform_device s3c_device_usb_hsotg;
+extern struct platform_device s3c_device_usb_hsudc;
+extern struct platform_device s3c_device_wdt;
+
+extern struct platform_device samsung_asoc_idma;
+extern struct platform_device samsung_device_keypad;
+extern struct platform_device samsung_device_pwm;
+
+/* s3c2440 specific devices */
+
+#ifdef CONFIG_CPU_S3C2440
+
+extern struct platform_device s3c_device_camif;
+extern struct platform_device s3c_device_ac97;
+
+#endif
+
+/**
+ * s3c_set_platdata() - helper for setting platform data
+ * @pd: The default platform data for this device.
+ * @pdsize: The size of the platform data.
+ * @pdev: Pointer to the device to fill in.
+ *
+ * This helper replaces a number of calls that copy and then set the
+ * platform data of the device.
+ */
+extern void *s3c_set_platdata(void *pd, size_t pdsize,
+			      struct platform_device *pdev);
+
+#endif /* __PLAT_DEVS_H */

arch/arm/plat-samsung/include/plat/fb-s3c2410.h

@@ -0,0 +1,72 @@
+/* arch/arm/plat-samsung/include/plat/fb-s3c2410.h
+ *
+ * Copyright (c) 2004 Arnaud Patard <arnaud.patard@rtp-net.org>
+ *
+ * Inspired by pxafb.h
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __ASM_PLAT_FB_S3C2410_H
+#define __ASM_PLAT_FB_S3C2410_H __FILE__
+
+struct s3c2410fb_hw {
+	unsigned long	lcdcon1;
+	unsigned long	lcdcon2;
+	unsigned long	lcdcon3;
+	unsigned long	lcdcon4;
+	unsigned long	lcdcon5;
+};
+
+/* LCD description */
+struct s3c2410fb_display {
+	/* LCD type */
+	unsigned type;
+
+	/* Screen size */
+	unsigned short width;
+	unsigned short height;
+
+	/* Screen info */
+	unsigned short xres;
+	unsigned short yres;
+	unsigned short bpp;
+
+	unsigned pixclock;		/* pixclock in picoseconds */
+	unsigned short left_margin;  /* value in pixels (TFT) or HCLKs (STN) */
+	unsigned short right_margin; /* value in pixels (TFT) or HCLKs (STN) */
+	unsigned short hsync_len;    /* value in pixels (TFT) or HCLKs (STN) */
+	unsigned short upper_margin;	/* value in lines (TFT) or 0 (STN) */
+	unsigned short lower_margin;	/* value in lines (TFT) or 0 (STN) */
+	unsigned short vsync_len;	/* value in lines (TFT) or 0 (STN) */
+
+	/* lcd configuration registers */
+	unsigned long	lcdcon5;
+};
+
+struct s3c2410fb_mach_info {
+
+	struct s3c2410fb_display *displays;	/* attached displays info */
+	unsigned num_displays;			/* number of defined displays */
+	unsigned default_display;
+
+	/* GPIOs */
+
+	unsigned long	gpcup;
+	unsigned long	gpcup_mask;
+	unsigned long	gpccon;
+	unsigned long	gpccon_mask;
+	unsigned long	gpdup;
+	unsigned long	gpdup_mask;
+	unsigned long	gpdcon;
+	unsigned long	gpdcon_mask;
+
+	/* lpc3600 control register */
+	unsigned long	lpcsel;
+};
+
+extern void __init s3c24xx_fb_set_platdata(struct s3c2410fb_mach_info *);
+
+#endif /* __ASM_PLAT_FB_S3C2410_H */

arch/arm/plat-samsung/include/plat/fb.h

@@ -0,0 +1,35 @@
+/* arch/arm/plat-samsung/include/plat/fb.h
+ *
+ * Copyright 2008 Openmoko, Inc.
+ * Copyright 2008 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C - FB platform data definitions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __PLAT_S3C_FB_H
+#define __PLAT_S3C_FB_H __FILE__
+
+#include <linux/platform_data/video_s3c.h>
+
+/**
+ * s3c_fb_set_platdata() - Setup the FB device with platform data.
+ * @pd: The platform data to set. The data is copied from the passed structure
+ *      so the machine data can mark the data __initdata so that any unused
+ *      machines will end up dumping their data at runtime.
+ */
+extern void s3c_fb_set_platdata(struct s3c_fb_platdata *pd);
+
+/**
+ * s3c64xx_fb_gpio_setup_24bpp() - S3C64XX setup function for 24bpp LCD
+ *
+ * Initialise the GPIO for an 24bpp LCD display on the RGB interface.
+ */
+extern void s3c64xx_fb_gpio_setup_24bpp(void);
+
+#endif /* __PLAT_S3C_FB_H */

arch/arm/plat-samsung/include/plat/gpio-cfg-helpers.h

@@ -0,0 +1,163 @@
+/* linux/arch/arm/plat-samsung/include/plat/gpio-cfg-helper.h
+ *
+ * Copyright 2008 Openmoko, Inc.
+ * Copyright 2008 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * Samsung Platform - GPIO pin configuration helper definitions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+/* This is meant for core cpu support, machine or other driver files
+ * should not be including this header.
+ */
+
+#ifndef __PLAT_GPIO_CFG_HELPERS_H
+#define __PLAT_GPIO_CFG_HELPERS_H __FILE__
+
+/* As a note, all gpio configuration functions are entered exclusively, either
+ * with the relevant lock held or the system prevented from doing anything else
+ * by disabling interrupts.
+*/
+
+static inline int samsung_gpio_do_setcfg(struct samsung_gpio_chip *chip,
+					 unsigned int off, unsigned int config)
+{
+	return (chip->config->set_config)(chip, off, config);
+}
+
+static inline unsigned samsung_gpio_do_getcfg(struct samsung_gpio_chip *chip,
+					      unsigned int off)
+{
+	return (chip->config->get_config)(chip, off);
+}
+
+static inline int samsung_gpio_do_setpull(struct samsung_gpio_chip *chip,
+					  unsigned int off, samsung_gpio_pull_t pull)
+{
+	return (chip->config->set_pull)(chip, off, pull);
+}
+
+static inline samsung_gpio_pull_t samsung_gpio_do_getpull(struct samsung_gpio_chip *chip,
+							  unsigned int off)
+{
+	return chip->config->get_pull(chip, off);
+}
+
+/* Pull-{up,down} resistor controls.
+ *
+ * S3C2410,S3C2440 = Pull-UP,
+ * S3C2412,S3C2413 = Pull-Down
+ * S3C6400,S3C6410 = Pull-Both [None,Down,Up,Undef]
+ * S3C2443 = Pull-Both [not same as S3C6400]
+ */
+
+/**
+ * s3c24xx_gpio_setpull_1up() - Pull configuration for choice of up or none.
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ * @param: pull: The pull mode being requested.
+ *
+ * This is a helper function for the case where we have GPIOs with one
+ * bit configuring the presence of a pull-up resistor.
+ */
+extern int s3c24xx_gpio_setpull_1up(struct samsung_gpio_chip *chip,
+				    unsigned int off, samsung_gpio_pull_t pull);
+
+/**
+ * s3c24xx_gpio_setpull_1down() - Pull configuration for choice of down or none
+ * @chip: The gpio chip that is being configured
+ * @off: The offset for the GPIO being configured
+ * @param: pull: The pull mode being requested
+ *
+ * This is a helper function for the case where we have GPIOs with one
+ * bit configuring the presence of a pull-down resistor.
+ */
+extern int s3c24xx_gpio_setpull_1down(struct samsung_gpio_chip *chip,
+				      unsigned int off, samsung_gpio_pull_t pull);
+
+/**
+ * samsung_gpio_setpull_upown() - Pull configuration for choice of up,
+ * down or none
+ *
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ * @param: pull: The pull mode being requested.
+ *
+ * This is a helper function for the case where we have GPIOs with two
+ * bits configuring the presence of a pull resistor, in the following
+ * order:
+ *	00 = No pull resistor connected
+ *	01 = Pull-up resistor connected
+ *	10 = Pull-down resistor connected
+ */
+extern int samsung_gpio_setpull_updown(struct samsung_gpio_chip *chip,
+				       unsigned int off, samsung_gpio_pull_t pull);
+
+/**
+ * samsung_gpio_getpull_updown() - Get configuration for choice of up,
+ * down or none
+ *
+ * @chip: The gpio chip that the GPIO pin belongs to
+ * @off: The offset to the pin to get the configuration of.
+ *
+ * This helper function reads the state of the pull-{up,down} resistor
+ * for the given GPIO in the same case as samsung_gpio_setpull_upown.
+*/
+extern samsung_gpio_pull_t samsung_gpio_getpull_updown(struct samsung_gpio_chip *chip,
+						       unsigned int off);
+
+/**
+ * s3c24xx_gpio_getpull_1up() - Get configuration for choice of up or none
+ * @chip: The gpio chip that the GPIO pin belongs to
+ * @off: The offset to the pin to get the configuration of.
+ *
+ * This helper function reads the state of the pull-up resistor for the
+ * given GPIO in the same case as s3c24xx_gpio_setpull_1up.
+*/
+extern samsung_gpio_pull_t s3c24xx_gpio_getpull_1up(struct samsung_gpio_chip *chip,
+						    unsigned int off);
+
+/**
+ * s3c24xx_gpio_getpull_1down() - Get configuration for choice of down or none
+ * @chip: The gpio chip that the GPIO pin belongs to
+ * @off: The offset to the pin to get the configuration of.
+ *
+ * This helper function reads the state of the pull-down resistor for the
+ * given GPIO in the same case as s3c24xx_gpio_setpull_1down.
+*/
+extern samsung_gpio_pull_t s3c24xx_gpio_getpull_1down(struct samsung_gpio_chip *chip,
+						      unsigned int off);
+
+/**
+ * s3c2443_gpio_setpull() - Pull configuration for s3c2443.
+ * @chip: The gpio chip that is being configured.
+ * @off: The offset for the GPIO being configured.
+ * @param: pull: The pull mode being requested.
+ *
+ * This is a helper function for the case where we have GPIOs with two
+ * bits configuring the presence of a pull resistor, in the following
+ * order:
+ *	00 = Pull-up resistor connected
+ *	10 = Pull-down resistor connected
+ *	x1 = No pull up resistor
+ */
+extern int s3c2443_gpio_setpull(struct samsung_gpio_chip *chip,
+				unsigned int off, samsung_gpio_pull_t pull);
+
+/**
+ * s3c2443_gpio_getpull() - Get configuration for s3c2443 pull resistors
+ * @chip: The gpio chip that the GPIO pin belongs to.
+ * @off: The offset to the pin to get the configuration of.
+ *
+ * This helper function reads the state of the pull-{up,down} resistor for the
+ * given GPIO in the same case as samsung_gpio_setpull_upown.
+*/
+extern samsung_gpio_pull_t s3c2443_gpio_getpull(struct samsung_gpio_chip *chip,
+						unsigned int off);
+
+#endif /* __PLAT_GPIO_CFG_HELPERS_H */

arch/arm/plat-samsung/include/plat/gpio-cfg.h

@@ -0,0 +1,182 @@
+/* linux/arch/arm/plat-s3c/include/plat/gpio-cfg.h
+ *
+ * Copyright 2008 Openmoko, Inc.
+ * Copyright 2008 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C Platform - GPIO pin configuration
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+/* This file contains the necessary definitions to get the basic gpio
+ * pin configuration done such as setting a pin to input or output or
+ * changing the pull-{up,down} configurations.
+ */
+
+/* Note, this interface is being added to the s3c64xx arch first and will
+ * be added to the s3c24xx systems later.
+ */
+
+#ifndef __PLAT_GPIO_CFG_H
+#define __PLAT_GPIO_CFG_H __FILE__
+
+#include <linux/types.h>
+
+typedef unsigned int __bitwise__ samsung_gpio_pull_t;
+
+/* forward declaration if gpio-core.h hasn't been included */
+struct samsung_gpio_chip;
+
+/**
+ * struct samsung_gpio_cfg GPIO configuration
+ * @cfg_eint: Configuration setting when used for external interrupt source
+ * @get_pull: Read the current pull configuration for the GPIO
+ * @set_pull: Set the current pull configuration for the GPIO
+ * @set_config: Set the current configuration for the GPIO
+ * @get_config: Read the current configuration for the GPIO
+ *
+ * Each chip can have more than one type of GPIO bank available and some
+ * have different capabilites even when they have the same control register
+ * layouts. Provide an point to vector control routine and provide any
+ * per-bank configuration information that other systems such as the
+ * external interrupt code will need.
+ *
+ * @sa samsung_gpio_cfgpin
+ * @sa s3c_gpio_getcfg
+ * @sa s3c_gpio_setpull
+ * @sa s3c_gpio_getpull
+ */
+struct samsung_gpio_cfg {
+	unsigned int	cfg_eint;
+
+	samsung_gpio_pull_t	(*get_pull)(struct samsung_gpio_chip *chip, unsigned offs);
+	int		(*set_pull)(struct samsung_gpio_chip *chip, unsigned offs,
+				    samsung_gpio_pull_t pull);
+
+	unsigned (*get_config)(struct samsung_gpio_chip *chip, unsigned offs);
+	int	 (*set_config)(struct samsung_gpio_chip *chip, unsigned offs,
+			       unsigned config);
+};
+
+#define S3C_GPIO_SPECIAL_MARK	(0xfffffff0)
+#define S3C_GPIO_SPECIAL(x)	(S3C_GPIO_SPECIAL_MARK | (x))
+
+/* Defines for generic pin configurations */
+#define S3C_GPIO_INPUT	(S3C_GPIO_SPECIAL(0))
+#define S3C_GPIO_OUTPUT	(S3C_GPIO_SPECIAL(1))
+#define S3C_GPIO_SFN(x)	(S3C_GPIO_SPECIAL(x))
+
+#define samsung_gpio_is_cfg_special(_cfg) \
+	(((_cfg) & S3C_GPIO_SPECIAL_MARK) == S3C_GPIO_SPECIAL_MARK)
+
+/**
+ * s3c_gpio_cfgpin() - Change the GPIO function of a pin.
+ * @pin pin The pin number to configure.
+ * @to to The configuration for the pin's function.
+ *
+ * Configure which function is actually connected to the external
+ * pin, such as an gpio input, output or some form of special function
+ * connected to an internal peripheral block.
+ *
+ * The @to parameter can be one of the generic S3C_GPIO_INPUT, S3C_GPIO_OUTPUT
+ * or S3C_GPIO_SFN() to indicate one of the possible values that the helper
+ * will then generate the correct bit mask and shift for the configuration.
+ *
+ * If a bank of GPIOs all needs to be set to special-function 2, then
+ * the following code will work:
+ *
+ *	for (gpio = start; gpio < end; gpio++)
+ *		s3c_gpio_cfgpin(gpio, S3C_GPIO_SFN(2));
+ *
+ * The @to parameter can also be a specific value already shifted to the
+ * correct position in the control register, although these are discouraged
+ * in newer kernels and are only being kept for compatibility.
+ */
+extern int s3c_gpio_cfgpin(unsigned int pin, unsigned int to);
+
+/**
+ * s3c_gpio_getcfg - Read the current function for a GPIO pin
+ * @pin: The pin to read the configuration value for.
+ *
+ * Read the configuration state of the given @pin, returning a value that
+ * could be passed back to s3c_gpio_cfgpin().
+ *
+ * @sa s3c_gpio_cfgpin
+ */
+extern unsigned s3c_gpio_getcfg(unsigned int pin);
+
+/**
+ * s3c_gpio_cfgpin_range() - Change the GPIO function for configuring pin range
+ * @start: The pin number to start at
+ * @nr: The number of pins to configure from @start.
+ * @cfg: The configuration for the pin's function
+ *
+ * Call s3c_gpio_cfgpin() for the @nr pins starting at @start.
+ *
+ * @sa s3c_gpio_cfgpin.
+ */
+extern int s3c_gpio_cfgpin_range(unsigned int start, unsigned int nr,
+				 unsigned int cfg);
+
+/* Define values for the pull-{up,down} available for each gpio pin.
+ *
+ * These values control the state of the weak pull-{up,down} resistors
+ * available on most pins on the S3C series. Not all chips support both
+ * up or down settings, and it may be dependent on the chip that is being
+ * used to whether the particular mode is available.
+ */
+#define S3C_GPIO_PULL_NONE	((__force samsung_gpio_pull_t)0x00)
+#define S3C_GPIO_PULL_DOWN	((__force samsung_gpio_pull_t)0x01)
+#define S3C_GPIO_PULL_UP	((__force samsung_gpio_pull_t)0x02)
+
+/**
+ * s3c_gpio_setpull() - set the state of a gpio pin pull resistor
+ * @pin: The pin number to configure the pull resistor.
+ * @pull: The configuration for the pull resistor.
+ *
+ * This function sets the state of the pull-{up,down} resistor for the
+ * specified pin. It will return 0 if successful, or a negative error
+ * code if the pin cannot support the requested pull setting.
+ *
+ * @pull is one of S3C_GPIO_PULL_NONE, S3C_GPIO_PULL_DOWN or S3C_GPIO_PULL_UP.
+*/
+extern int s3c_gpio_setpull(unsigned int pin, samsung_gpio_pull_t pull);
+
+/**
+ * s3c_gpio_getpull() - get the pull resistor state of a gpio pin
+ * @pin: The pin number to get the settings for
+ *
+ * Read the pull resistor value for the specified pin.
+*/
+extern samsung_gpio_pull_t s3c_gpio_getpull(unsigned int pin);
+
+/* configure `all` aspects of an gpio */
+
+/**
+ * s3c_gpio_cfgall_range() - configure range of gpio functtion and pull.
+ * @start: The gpio number to start at.
+ * @nr: The number of gpio to configure from @start.
+ * @cfg: The configuration to use
+ * @pull: The pull setting to use.
+ *
+ * Run s3c_gpio_cfgpin() and s3c_gpio_setpull() over the gpio range starting
+ * @gpio and running for @size.
+ *
+ * @sa s3c_gpio_cfgpin
+ * @sa s3c_gpio_setpull
+ * @sa s3c_gpio_cfgpin_range
+ */
+extern int s3c_gpio_cfgall_range(unsigned int start, unsigned int nr,
+				 unsigned int cfg, samsung_gpio_pull_t pull);
+
+static inline int s3c_gpio_cfgrange_nopull(unsigned int pin, unsigned int size,
+					   unsigned int cfg)
+{
+	return s3c_gpio_cfgall_range(pin, size, cfg, S3C_GPIO_PULL_NONE);
+}
+
+#endif /* __PLAT_GPIO_CFG_H */

arch/arm/plat-samsung/include/plat/gpio-core.h

@@ -0,0 +1,145 @@
+/* linux/arch/arm/plat-s3c/include/plat/gpio-core.h
+ *
+ * Copyright 2008 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C Platform - GPIO core
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __PLAT_SAMSUNG_GPIO_CORE_H
+#define __PLAT_SAMSUNG_GPIO_CORE_H
+
+/* Bring in machine-local definitions, especially S3C_GPIO_END */
+#include <mach/gpio-samsung.h>
+
+#define GPIOCON_OFF	(0x00)
+#define GPIODAT_OFF	(0x04)
+
+#define con_4bit_shift(__off) ((__off) * 4)
+
+/* Define the core gpiolib support functions that the s3c platforms may
+ * need to extend or change depending on the hardware and the s3c chip
+ * selected at build or found at run time.
+ *
+ * These definitions are not intended for driver inclusion, there is
+ * nothing here that should not live outside the platform and core
+ * specific code.
+*/
+
+struct samsung_gpio_chip;
+
+/**
+ * struct samsung_gpio_pm - power management (suspend/resume) information
+ * @save: Routine to save the state of the GPIO block
+ * @resume: Routine to resume the GPIO block.
+ */
+struct samsung_gpio_pm {
+	void (*save)(struct samsung_gpio_chip *chip);
+	void (*resume)(struct samsung_gpio_chip *chip);
+};
+
+struct samsung_gpio_cfg;
+
+/**
+ * struct samsung_gpio_chip - wrapper for specific implementation of gpio
+ * @chip: The chip structure to be exported via gpiolib.
+ * @base: The base pointer to the gpio configuration registers.
+ * @group: The group register number for gpio interrupt support.
+ * @irq_base: The base irq number.
+ * @config: special function and pull-resistor control information.
+ * @lock: Lock for exclusive access to this gpio bank.
+ * @pm_save: Save information for suspend/resume support.
+ * @bitmap_gpio_int: Bitmap for representing GPIO interrupt or not.
+ *
+ * This wrapper provides the necessary information for the Samsung
+ * specific gpios being registered with gpiolib.
+ *
+ * The lock protects each gpio bank from multiple access of the shared
+ * configuration registers, or from reading of data whilst another thread
+ * is writing to the register set.
+ *
+ * Each chip has its own lock to avoid any  contention between different
+ * CPU cores trying to get one lock for different GPIO banks, where each
+ * bank of GPIO has its own register space and configuration registers.
+ */
+struct samsung_gpio_chip {
+	struct gpio_chip	chip;
+	struct samsung_gpio_cfg	*config;
+	struct samsung_gpio_pm	*pm;
+	void __iomem		*base;
+	int			irq_base;
+	int			group;
+	spinlock_t		 lock;
+#ifdef CONFIG_PM
+	u32			pm_save[4];
+#endif
+	u32			bitmap_gpio_int;
+};
+
+static inline struct samsung_gpio_chip *to_samsung_gpio(struct gpio_chip *gpc)
+{
+	return container_of(gpc, struct samsung_gpio_chip, chip);
+}
+
+/**
+ * samsung_gpiolib_to_irq - convert gpio pin to irq number
+ * @chip: The gpio chip that the pin belongs to.
+ * @offset: The offset of the pin in the chip.
+ *
+ * This helper returns the irq number calculated from the chip->irq_base and
+ * the provided offset.
+ */
+extern int samsung_gpiolib_to_irq(struct gpio_chip *chip, unsigned int offset);
+
+/* exported for core SoC support to change */
+extern struct samsung_gpio_cfg s3c24xx_gpiocfg_default;
+
+#ifdef CONFIG_S3C_GPIO_TRACK
+extern struct samsung_gpio_chip *s3c_gpios[S3C_GPIO_END];
+
+static inline struct samsung_gpio_chip *samsung_gpiolib_getchip(unsigned int chip)
+{
+	return (chip < S3C_GPIO_END) ? s3c_gpios[chip] : NULL;
+}
+#else
+/* machine specific code should provide samsung_gpiolib_getchip */
+
+extern struct samsung_gpio_chip s3c24xx_gpios[];
+
+static inline struct samsung_gpio_chip *samsung_gpiolib_getchip(unsigned int pin)
+{
+	struct samsung_gpio_chip *chip;
+
+	if (pin > S3C_GPIO_END)
+		return NULL;
+
+	chip = &s3c24xx_gpios[pin/32];
+	return ((pin - chip->chip.base) < chip->chip.ngpio) ? chip : NULL;
+}
+
+static inline void s3c_gpiolib_track(struct samsung_gpio_chip *chip) { }
+#endif
+
+#ifdef CONFIG_PM
+extern struct samsung_gpio_pm samsung_gpio_pm_1bit;
+extern struct samsung_gpio_pm samsung_gpio_pm_2bit;
+extern struct samsung_gpio_pm samsung_gpio_pm_4bit;
+#define __gpio_pm(x) x
+#else
+#define samsung_gpio_pm_1bit NULL
+#define samsung_gpio_pm_2bit NULL
+#define samsung_gpio_pm_4bit NULL
+#define __gpio_pm(x) NULL
+
+#endif /* CONFIG_PM */
+
+/* locking wrappers to deal with multiple access to the same gpio bank */
+#define samsung_gpio_lock(_oc, _fl) spin_lock_irqsave(&(_oc)->lock, _fl)
+#define samsung_gpio_unlock(_oc, _fl) spin_unlock_irqrestore(&(_oc)->lock, _fl)
+
+#endif /* __PLAT_SAMSUNG_GPIO_CORE_H */

arch/arm/plat-samsung/include/plat/iic-core.h

@@ -0,0 +1,42 @@
+/* arch/arm/mach-s3c2410/include/mach/iic-core.h
+ *
+ * Copyright 2008 Openmoko, Inc.
+ * Copyright 2008 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C - I2C Controller core functions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __ASM_ARCH_IIC_CORE_H
+#define __ASM_ARCH_IIC_CORE_H __FILE__
+
+/* These functions are only for use with the core support code, such as
+ * the cpu specific initialisation code
+ */
+
+/* re-define device name depending on support. */
+static inline void s3c_i2c0_setname(char *name)
+{
+	/* currently this device is always compiled in */
+	s3c_device_i2c0.name = name;
+}
+
+static inline void s3c_i2c1_setname(char *name)
+{
+#ifdef CONFIG_S3C_DEV_I2C1
+	s3c_device_i2c1.name = name;
+#endif
+}
+
+static inline void s3c_i2c2_setname(char *name)
+{
+#ifdef CONFIG_S3C_DEV_I2C2
+	s3c_device_i2c2.name = name;
+#endif
+}
+
+#endif /* __ASM_ARCH_IIC_H */

arch/arm/plat-samsung/include/plat/keypad.h

@@ -0,0 +1,31 @@
+/*
+ * Samsung Platform - Keypad platform data definitions
+ *
+ * Copyright (C) 2010 Samsung Electronics Co.Ltd
+ * Author: Joonyoung Shim <jy0922.shim@samsung.com>
+ *
+ * This program is free software; you can redistribute  it and/or modify it
+ * under  the terms of  the GNU General  Public License as published by the
+ * Free Software Foundation;  either version 2 of the  License, or (at your
+ * option) any later version.
+ */
+
+#ifndef __PLAT_SAMSUNG_KEYPAD_H
+#define __PLAT_SAMSUNG_KEYPAD_H
+
+#include <linux/input/samsung-keypad.h>
+
+/**
+ * samsung_keypad_set_platdata - Set platform data for Samsung Keypad device.
+ * @pd: Platform data to register to device.
+ *
+ * Register the given platform data for use with Samsung Keypad device.
+ * The call will copy the platform data, so the board definitions can
+ * make the structure itself __initdata.
+ */
+extern void samsung_keypad_set_platdata(struct samsung_keypad_platdata *pd);
+
+/* defined by architecture to configure gpio. */
+extern void samsung_keypad_cfg_gpio(unsigned int rows, unsigned int cols);
+
+#endif /* __PLAT_SAMSUNG_KEYPAD_H */

arch/arm/plat-samsung/include/plat/map-base.h

@@ -0,0 +1,46 @@
+/* linux/include/asm-arm/plat-s3c/map.h
+ *
+ * Copyright 2003, 2007 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C - Memory map definitions (virtual addresses)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __ASM_PLAT_MAP_H
+#define __ASM_PLAT_MAP_H __FILE__
+
+/* Fit all our registers in at 0xF6000000 upwards, trying to use as
+ * little of the VA space as possible so vmalloc and friends have a
+ * better chance of getting memory.
+ *
+ * we try to ensure stuff like the IRQ registers are available for
+ * an single MOVS instruction (ie, only 8 bits of set data)
+ */
+
+#define S3C_ADDR_BASE	0xF6000000
+
+#ifndef __ASSEMBLY__
+#define S3C_ADDR(x)	((void __iomem __force *)S3C_ADDR_BASE + (x))
+#else
+#define S3C_ADDR(x)	(S3C_ADDR_BASE + (x))
+#endif
+
+#define S3C_VA_IRQ	S3C_ADDR(0x00000000)	/* irq controller(s) */
+#define S3C_VA_SYS	S3C_ADDR(0x00100000)	/* system control */
+#define S3C_VA_MEM	S3C_ADDR(0x00200000)	/* memory control */
+#define S3C_VA_TIMER	S3C_ADDR(0x00300000)	/* timer block */
+#define S3C_VA_WATCHDOG	S3C_ADDR(0x00400000)	/* watchdog */
+#define S3C_VA_UART	S3C_ADDR(0x01000000)	/* UART */
+
+/* This is used for the CPU specific mappings that may be needed, so that
+ * they do not need to directly used S3C_ADDR() and thus make it easier to
+ * modify the space for mapping.
+ */
+#define S3C_ADDR_CPU(x)	S3C_ADDR(0x00500000 + (x))
+
+#endif /* __ASM_PLAT_MAP_H */

arch/arm/plat-samsung/include/plat/map-s3c.h

@@ -0,0 +1,80 @@
+/* linux/arch/arm/plat-samsung/include/plat/map-s3c.h
+ *
+ * Copyright (c) 2008 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C24XX - Memory map definitions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __ASM_PLAT_MAP_S3C_H
+#define __ASM_PLAT_MAP_S3C_H __FILE__
+
+#define S3C24XX_VA_IRQ		S3C_VA_IRQ
+#define S3C24XX_VA_MEMCTRL	S3C_VA_MEM
+#define S3C24XX_VA_UART		S3C_VA_UART
+
+#define S3C24XX_VA_TIMER	S3C_VA_TIMER
+#define S3C24XX_VA_CLKPWR	S3C_VA_SYS
+#define S3C24XX_VA_WATCHDOG	S3C_VA_WATCHDOG
+
+#define S3C2412_VA_SSMC		S3C_ADDR_CPU(0x00000000)
+#define S3C2412_VA_EBI		S3C_ADDR_CPU(0x00100000)
+
+#define S3C2410_PA_UART		(0x50000000)
+#define S3C24XX_PA_UART		S3C2410_PA_UART
+
+/*
+ * GPIO ports
+ *
+ * the calculation for the VA of this must ensure that
+ * it is the same distance apart from the UART in the
+ * phsyical address space, as the initial mapping for the IO
+ * is done as a 1:1 mapping. This puts it (currently) at
+ * 0xFA800000, which is not in the way of any current mapping
+ * by the base system.
+*/
+
+#define S3C2410_PA_GPIO		(0x56000000)
+#define S3C24XX_PA_GPIO		S3C2410_PA_GPIO
+
+#define S3C24XX_VA_GPIO		((S3C24XX_PA_GPIO - S3C24XX_PA_UART) + S3C24XX_VA_UART)
+#define S3C64XX_VA_GPIO		S3C_ADDR_CPU(0x00000000)
+
+#define S3C64XX_VA_MODEM	S3C_ADDR_CPU(0x00100000)
+#define S3C64XX_VA_USB_HSPHY	S3C_ADDR_CPU(0x00200000)
+
+#define S3C_VA_USB_HSPHY	S3C64XX_VA_USB_HSPHY
+
+/*
+ * ISA style IO, for each machine to sort out mappings for,
+ * if it implements it. We reserve two 16M regions for ISA.
+ */
+
+#define S3C2410_ADDR(x)		S3C_ADDR(x)
+
+#define S3C24XX_VA_ISA_WORD	S3C2410_ADDR(0x02000000)
+#define S3C24XX_VA_ISA_BYTE	S3C2410_ADDR(0x03000000)
+
+/* deal with the registers that move under the 2412/2413 */
+
+#if defined(CONFIG_CPU_S3C2412) || defined(CONFIG_CPU_S3C2413)
+#ifndef __ASSEMBLY__
+extern void __iomem *s3c24xx_va_gpio2;
+#endif
+#ifdef CONFIG_CPU_S3C2412_ONLY
+#define S3C24XX_VA_GPIO2	(S3C24XX_VA_GPIO + 0x10)
+#else
+#define S3C24XX_VA_GPIO2 s3c24xx_va_gpio2
+#endif
+#else
+#define s3c24xx_va_gpio2 S3C24XX_VA_GPIO
+#define S3C24XX_VA_GPIO2 S3C24XX_VA_GPIO
+#endif
+
+#include <plat/map-s5p.h>
+
+#endif /* __ASM_PLAT_MAP_S3C_H */

arch/arm/plat-samsung/include/plat/map-s5p.h

@@ -0,0 +1,30 @@
+/* linux/arch/arm/plat-samsung/include/plat/map-s5p.h
+ *
+ * Copyright (c) 2010 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com/
+ *
+ * S5P - Memory map definitions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __ASM_PLAT_MAP_S5P_H
+#define __ASM_PLAT_MAP_S5P_H __FILE__
+
+#define S5P_VA_CHIPID		S3C_ADDR(0x02000000)
+
+#define S5P_VA_COREPERI_BASE	S3C_ADDR(0x02800000)
+#define S5P_VA_COREPERI(x)	(S5P_VA_COREPERI_BASE + (x))
+#define S5P_VA_SCU		S5P_VA_COREPERI(0x0)
+
+#define VA_VIC(x)		(S3C_VA_IRQ + ((x) * 0x10000))
+#define VA_VIC0			VA_VIC(0)
+#define VA_VIC1			VA_VIC(1)
+#define VA_VIC2			VA_VIC(2)
+#define VA_VIC3			VA_VIC(3)
+
+#include <plat/map-s3c.h>
+
+#endif /* __ASM_PLAT_MAP_S5P_H */

arch/arm/plat-samsung/include/plat/pm-common.h

@@ -0,0 +1,110 @@
+/*
+ * Copyright (C) 2013 Samsung Electronics Co., Ltd.
+ *	Tomasz Figa <t.figa@samsung.com>
+ * Copyright (c) 2004 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Written by Ben Dooks, <ben@simtec.co.uk>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __PLAT_SAMSUNG_PM_COMMON_H
+#define __PLAT_SAMSUNG_PM_COMMON_H __FILE__
+
+#include <linux/irq.h>
+
+/* sleep save info */
+
+/**
+ * struct sleep_save - save information for shared peripherals.
+ * @reg: Pointer to the register to save.
+ * @val: Holder for the value saved from reg.
+ *
+ * This describes a list of registers which is used by the pm core and
+ * other subsystem to save and restore register values over suspend.
+ */
+struct sleep_save {
+	void __iomem	*reg;
+	unsigned long	val;
+};
+
+#define SAVE_ITEM(x) \
+	{ .reg = (x) }
+
+/* helper functions to save/restore lists of registers. */
+
+extern void s3c_pm_do_save(struct sleep_save *ptr, int count);
+extern void s3c_pm_do_restore(const struct sleep_save *ptr, int count);
+extern void s3c_pm_do_restore_core(const struct sleep_save *ptr, int count);
+
+/* PM debug functions */
+
+/**
+ * struct pm_uart_save - save block for core UART
+ * @ulcon: Save value for S3C2410_ULCON
+ * @ucon: Save value for S3C2410_UCON
+ * @ufcon: Save value for S3C2410_UFCON
+ * @umcon: Save value for S3C2410_UMCON
+ * @ubrdiv: Save value for S3C2410_UBRDIV
+ *
+ * Save block for UART registers to be held over sleep and restored if they
+ * are needed (say by debug).
+*/
+struct pm_uart_save {
+	u32	ulcon;
+	u32	ucon;
+	u32	ufcon;
+	u32	umcon;
+	u32	ubrdiv;
+	u32	udivslot;
+};
+
+#ifdef CONFIG_SAMSUNG_PM_DEBUG
+/**
+ * s3c_pm_dbg() - low level debug function for use in suspend/resume.
+ * @msg: The message to print.
+ *
+ * This function is used mainly to debug the resume process before the system
+ * can rely on printk/console output. It uses the low-level debugging output
+ * routine printascii() to do its work.
+ */
+extern void s3c_pm_dbg(const char *msg, ...);
+
+/**
+ * s3c_pm_debug_init() - suspend/resume low level debug initialization.
+ * @base: Virtual base of UART to use for suspend/resume debugging.
+ *
+ * This function needs to be called before S3C_PMDBG() can be used, to set up
+ * UART port base address and configuration.
+ */
+extern void s3c_pm_debug_init(void);
+
+#define S3C_PMDBG(fmt...) s3c_pm_dbg(fmt)
+
+extern void s3c_pm_save_uarts(void);
+extern void s3c_pm_restore_uarts(void);
+#else
+#define S3C_PMDBG(fmt...) pr_debug(fmt)
+#define s3c_pm_debug_init() do { } while (0)
+
+static inline void s3c_pm_save_uarts(void) { }
+static inline void s3c_pm_restore_uarts(void) { }
+#endif
+
+/* suspend memory checking */
+
+#ifdef CONFIG_SAMSUNG_PM_CHECK
+extern void s3c_pm_check_prepare(void);
+extern void s3c_pm_check_restore(void);
+extern void s3c_pm_check_cleanup(void);
+extern void s3c_pm_check_store(void);
+#else
+#define s3c_pm_check_prepare() do { } while (0)
+#define s3c_pm_check_restore() do { } while (0)
+#define s3c_pm_check_cleanup() do { } while (0)
+#define s3c_pm_check_store()   do { } while (0)
+#endif
+
+#endif

arch/arm/plat-samsung/include/plat/pm.h

@@ -0,0 +1,113 @@
+/* arch/arm/plat-samsung/include/plat/pm.h
+ *
+ * Copyright (c) 2004 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Written by Ben Dooks, <ben@simtec.co.uk>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+/* s3c_pm_init
+ *
+ * called from board at initialisation time to setup the power
+ * management
+*/
+
+#include <plat/pm-common.h>
+
+struct device;
+
+#ifdef CONFIG_SAMSUNG_PM
+
+extern __init int s3c_pm_init(void);
+extern __init int s3c64xx_pm_init(void);
+
+#else
+
+static inline int s3c_pm_init(void)
+{
+	return 0;
+}
+
+static inline int s3c64xx_pm_init(void)
+{
+	return 0;
+}
+#endif
+
+/* configuration for the IRQ mask over sleep */
+extern unsigned long s3c_irqwake_intmask;
+extern unsigned long s3c_irqwake_eintmask;
+
+/* per-cpu sleep functions */
+
+extern void (*pm_cpu_prep)(void);
+extern int (*pm_cpu_sleep)(unsigned long);
+
+/* Flags for PM Control */
+
+extern unsigned long s3c_pm_flags;
+
+/* from sleep.S */
+
+extern int s3c2410_cpu_suspend(unsigned long);
+
+#ifdef CONFIG_PM_SLEEP
+extern int s3c_irq_wake(struct irq_data *data, unsigned int state);
+extern void s3c_cpu_resume(void);
+#else
+#define s3c_irq_wake NULL
+#define s3c_cpu_resume NULL
+#endif
+
+#ifdef CONFIG_SAMSUNG_PM
+extern int s3c_irqext_wake(struct irq_data *data, unsigned int state);
+#else
+#define s3c_irqext_wake NULL
+#endif
+
+#ifdef CONFIG_S3C_PM_DEBUG_LED_SMDK
+/**
+ * s3c_pm_debug_smdkled() - Debug PM suspend/resume via SMDK Board LEDs
+ * @set: set bits for the state of the LEDs
+ * @clear: clear bits for the state of the LEDs.
+ */
+extern void s3c_pm_debug_smdkled(u32 set, u32 clear);
+
+#else
+static inline void s3c_pm_debug_smdkled(u32 set, u32 clear) { }
+#endif /* CONFIG_S3C_PM_DEBUG_LED_SMDK */
+
+/**
+ * s3c_pm_configure_extint() - ensure pins are correctly set for IRQ
+ *
+ * Setup all the necessary GPIO pins for waking the system on external
+ * interrupt.
+ */
+extern void s3c_pm_configure_extint(void);
+
+#ifdef CONFIG_GPIO_SAMSUNG
+/**
+ * samsung_pm_restore_gpios() - restore the state of the gpios after sleep.
+ *
+ * Restore the state of the GPIO pins after sleep, which may involve ensuring
+ * that we do not glitch the state of the pins from that the bootloader's
+ * resume code has done.
+*/
+extern void samsung_pm_restore_gpios(void);
+
+/**
+ * samsung_pm_save_gpios() - save the state of the GPIOs for restoring after sleep.
+ *
+ * Save the GPIO states for resotration on resume. See samsung_pm_restore_gpios().
+ */
+extern void samsung_pm_save_gpios(void);
+#else
+static inline void samsung_pm_restore_gpios(void) {}
+static inline void samsung_pm_save_gpios(void) {}
+#endif
+
+extern void s3c_pm_save_core(void);
+extern void s3c_pm_restore_core(void);

arch/arm/plat-samsung/include/plat/pwm-core.h

@@ -0,0 +1,22 @@
+/*
+ * Copyright 2013 Tomasz Figa <tomasz.figa@gmail.com>
+ *
+ * Samsung PWM controller platform data helpers.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARCH_PWM_CORE_H
+#define __ASM_ARCH_PWM_CORE_H __FILE__
+
+#include <clocksource/samsung_pwm.h>
+
+#ifdef CONFIG_SAMSUNG_DEV_PWM
+extern void samsung_pwm_set_platdata(struct samsung_pwm_variant *pd);
+#else
+static inline void samsung_pwm_set_platdata(struct samsung_pwm_variant *pd) { }
+#endif
+
+#endif /* __ASM_ARCH_PWM_CORE_H */

arch/arm/plat-samsung/include/plat/regs-adc.h

@@ -0,0 +1,68 @@
+/* arch/arm/mach-s3c2410/include/mach/regs-adc.h
+ *
+ * Copyright (c) 2004 Shannon Holland <holland@loser.net>
+ *
+ * This program is free software; yosu can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * S3C2410 ADC registers
+*/
+
+#ifndef __ASM_ARCH_REGS_ADC_H
+#define __ASM_ARCH_REGS_ADC_H "regs-adc.h"
+
+#define S3C2410_ADCREG(x) (x)
+
+#define S3C2410_ADCCON	   S3C2410_ADCREG(0x00)
+#define S3C2410_ADCTSC	   S3C2410_ADCREG(0x04)
+#define S3C2410_ADCDLY	   S3C2410_ADCREG(0x08)
+#define S3C2410_ADCDAT0	   S3C2410_ADCREG(0x0C)
+#define S3C2410_ADCDAT1	   S3C2410_ADCREG(0x10)
+#define S3C64XX_ADCUPDN		S3C2410_ADCREG(0x14)
+#define S3C2443_ADCMUX		S3C2410_ADCREG(0x18)
+#define S3C64XX_ADCCLRINT	S3C2410_ADCREG(0x18)
+#define S5P_ADCMUX		S3C2410_ADCREG(0x1C)
+#define S3C64XX_ADCCLRINTPNDNUP	S3C2410_ADCREG(0x20)
+
+
+/* ADCCON Register Bits */
+#define S3C64XX_ADCCON_RESSEL		(1<<16)
+#define S3C2410_ADCCON_ECFLG		(1<<15)
+#define S3C2410_ADCCON_PRSCEN		(1<<14)
+#define S3C2410_ADCCON_PRSCVL(x)	(((x)&0xFF)<<6)
+#define S3C2410_ADCCON_PRSCVLMASK	(0xFF<<6)
+#define S3C2410_ADCCON_SELMUX(x)	(((x)&0x7)<<3)
+#define S3C2410_ADCCON_MUXMASK		(0x7<<3)
+#define S3C2416_ADCCON_RESSEL		(1 << 3)
+#define S3C2410_ADCCON_STDBM		(1<<2)
+#define S3C2410_ADCCON_READ_START	(1<<1)
+#define S3C2410_ADCCON_ENABLE_START	(1<<0)
+#define S3C2410_ADCCON_STARTMASK	(0x3<<0)
+
+
+/* ADCTSC Register Bits */
+#define S3C2443_ADCTSC_UD_SEN		(1 << 8)
+#define S3C2410_ADCTSC_YM_SEN		(1<<7)
+#define S3C2410_ADCTSC_YP_SEN		(1<<6)
+#define S3C2410_ADCTSC_XM_SEN		(1<<5)
+#define S3C2410_ADCTSC_XP_SEN		(1<<4)
+#define S3C2410_ADCTSC_PULL_UP_DISABLE	(1<<3)
+#define S3C2410_ADCTSC_AUTO_PST		(1<<2)
+#define S3C2410_ADCTSC_XY_PST(x)	(((x)&0x3)<<0)
+
+/* ADCDAT0 Bits */
+#define S3C2410_ADCDAT0_UPDOWN		(1<<15)
+#define S3C2410_ADCDAT0_AUTO_PST	(1<<14)
+#define S3C2410_ADCDAT0_XY_PST		(0x3<<12)
+#define S3C2410_ADCDAT0_XPDATA_MASK	(0x03FF)
+
+/* ADCDAT1 Bits */
+#define S3C2410_ADCDAT1_UPDOWN		(1<<15)
+#define S3C2410_ADCDAT1_AUTO_PST	(1<<14)
+#define S3C2410_ADCDAT1_XY_PST		(0x3<<12)
+#define S3C2410_ADCDAT1_YPDATA_MASK	(0x03FF)
+
+#endif /* __ASM_ARCH_REGS_ADC_H */
+
+

arch/arm/plat-samsung/include/plat/regs-irqtype.h

@@ -0,0 +1,21 @@
+/* arch/arm/plat-s3c/include/plat/regs-irqtype.h
+ *
+ * Copyright 2008 Simtec Electronics
+ *      Ben Dooks <ben@simtec.co.uk>
+ *      http://armlinux.simtec.co.uk/
+ *
+ * S3C - IRQ detection types.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+/* values for S3C2410_EXTINT0/1/2 and other cpus in the series, including
+ * the S3C64XX
+*/
+#define S3C2410_EXTINT_LOWLEV	 (0x00)
+#define S3C2410_EXTINT_HILEV	 (0x01)
+#define S3C2410_EXTINT_FALLEDGE	 (0x02)
+#define S3C2410_EXTINT_RISEEDGE	 (0x04)
+#define S3C2410_EXTINT_BOTHEDGE	 (0x06)

arch/arm/plat-samsung/include/plat/regs-spi.h

@@ -0,0 +1,48 @@
+/* arch/arm/plat-samsung/include/plat/regs-spi.h
+ *
+ * Copyright (c) 2004 Fetron GmbH
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * S3C2410 SPI register definition
+*/
+
+#ifndef __ASM_ARCH_REGS_SPI_H
+#define __ASM_ARCH_REGS_SPI_H
+
+#define S3C2410_SPI1		(0x20)
+#define S3C2412_SPI1		(0x100)
+
+#define S3C2410_SPCON		(0x00)
+
+#define S3C2410_SPCON_SMOD_DMA	(2 << 5)	/* DMA mode */
+#define S3C2410_SPCON_SMOD_INT	(1 << 5)	/* interrupt mode */
+#define S3C2410_SPCON_SMOD_POLL	(0 << 5)	/* polling mode */
+#define S3C2410_SPCON_ENSCK	(1 << 4)	/* Enable SCK */
+#define S3C2410_SPCON_MSTR	(1 << 3)	/* Master:1, Slave:0 select */
+#define S3C2410_SPCON_CPOL_HIGH	(1 << 2)	/* Clock polarity select */
+#define S3C2410_SPCON_CPOL_LOW	(0 << 2)	/* Clock polarity select */
+
+#define S3C2410_SPCON_CPHA_FMTB	(1 << 1)	/* Clock Phase Select */
+#define S3C2410_SPCON_CPHA_FMTA	(0 << 1)	/* Clock Phase Select */
+
+#define S3C2410_SPSTA		(0x04)
+
+#define S3C2410_SPSTA_DCOL	(1 << 2)	/* Data Collision Error */
+#define S3C2410_SPSTA_MULD	(1 << 1)	/* Multi Master Error */
+#define S3C2410_SPSTA_READY	(1 << 0)	/* Data Tx/Rx ready */
+#define S3C2412_SPSTA_READY_ORG	(1 << 3)
+
+#define S3C2410_SPPIN		(0x08)
+
+#define S3C2410_SPPIN_ENMUL	(1 << 2)	/* Multi Master Error detect */
+#define S3C2410_SPPIN_RESERVED	(1 << 1)
+#define S3C2410_SPPIN_KEEP	(1 << 0)	/* Master Out keep */
+
+#define S3C2410_SPPRE		(0x0C)
+#define S3C2410_SPTDAT		(0x10)
+#define S3C2410_SPRDAT		(0x14)
+
+#endif /* __ASM_ARCH_REGS_SPI_H */

arch/arm/plat-samsung/include/plat/regs-udc.h

@@ -0,0 +1,151 @@
+/* arch/arm/plat-samsung/include/plat/regs-udc.h
+ *
+ * Copyright (C) 2004 Herbert Poetzl <herbert@13thfloor.at>
+ *
+ * This include file is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+*/
+
+#ifndef __ASM_ARCH_REGS_UDC_H
+#define __ASM_ARCH_REGS_UDC_H
+
+#define S3C2410_USBDREG(x) (x)
+
+#define S3C2410_UDC_FUNC_ADDR_REG	S3C2410_USBDREG(0x0140)
+#define S3C2410_UDC_PWR_REG		S3C2410_USBDREG(0x0144)
+#define S3C2410_UDC_EP_INT_REG		S3C2410_USBDREG(0x0148)
+
+#define S3C2410_UDC_USB_INT_REG		S3C2410_USBDREG(0x0158)
+#define S3C2410_UDC_EP_INT_EN_REG	S3C2410_USBDREG(0x015c)
+
+#define S3C2410_UDC_USB_INT_EN_REG	S3C2410_USBDREG(0x016c)
+
+#define S3C2410_UDC_FRAME_NUM1_REG	S3C2410_USBDREG(0x0170)
+#define S3C2410_UDC_FRAME_NUM2_REG	S3C2410_USBDREG(0x0174)
+
+#define S3C2410_UDC_EP0_FIFO_REG	S3C2410_USBDREG(0x01c0)
+#define S3C2410_UDC_EP1_FIFO_REG	S3C2410_USBDREG(0x01c4)
+#define S3C2410_UDC_EP2_FIFO_REG	S3C2410_USBDREG(0x01c8)
+#define S3C2410_UDC_EP3_FIFO_REG	S3C2410_USBDREG(0x01cc)
+#define S3C2410_UDC_EP4_FIFO_REG	S3C2410_USBDREG(0x01d0)
+
+#define S3C2410_UDC_EP1_DMA_CON		S3C2410_USBDREG(0x0200)
+#define S3C2410_UDC_EP1_DMA_UNIT	S3C2410_USBDREG(0x0204)
+#define S3C2410_UDC_EP1_DMA_FIFO	S3C2410_USBDREG(0x0208)
+#define S3C2410_UDC_EP1_DMA_TTC_L	S3C2410_USBDREG(0x020c)
+#define S3C2410_UDC_EP1_DMA_TTC_M	S3C2410_USBDREG(0x0210)
+#define S3C2410_UDC_EP1_DMA_TTC_H	S3C2410_USBDREG(0x0214)
+
+#define S3C2410_UDC_EP2_DMA_CON		S3C2410_USBDREG(0x0218)
+#define S3C2410_UDC_EP2_DMA_UNIT	S3C2410_USBDREG(0x021c)
+#define S3C2410_UDC_EP2_DMA_FIFO	S3C2410_USBDREG(0x0220)
+#define S3C2410_UDC_EP2_DMA_TTC_L	S3C2410_USBDREG(0x0224)
+#define S3C2410_UDC_EP2_DMA_TTC_M	S3C2410_USBDREG(0x0228)
+#define S3C2410_UDC_EP2_DMA_TTC_H	S3C2410_USBDREG(0x022c)
+
+#define S3C2410_UDC_EP3_DMA_CON		S3C2410_USBDREG(0x0240)
+#define S3C2410_UDC_EP3_DMA_UNIT	S3C2410_USBDREG(0x0244)
+#define S3C2410_UDC_EP3_DMA_FIFO	S3C2410_USBDREG(0x0248)
+#define S3C2410_UDC_EP3_DMA_TTC_L	S3C2410_USBDREG(0x024c)
+#define S3C2410_UDC_EP3_DMA_TTC_M	S3C2410_USBDREG(0x0250)
+#define S3C2410_UDC_EP3_DMA_TTC_H	S3C2410_USBDREG(0x0254)
+
+#define S3C2410_UDC_EP4_DMA_CON		S3C2410_USBDREG(0x0258)
+#define S3C2410_UDC_EP4_DMA_UNIT	S3C2410_USBDREG(0x025c)
+#define S3C2410_UDC_EP4_DMA_FIFO	S3C2410_USBDREG(0x0260)
+#define S3C2410_UDC_EP4_DMA_TTC_L	S3C2410_USBDREG(0x0264)
+#define S3C2410_UDC_EP4_DMA_TTC_M	S3C2410_USBDREG(0x0268)
+#define S3C2410_UDC_EP4_DMA_TTC_H	S3C2410_USBDREG(0x026c)
+
+#define S3C2410_UDC_INDEX_REG		S3C2410_USBDREG(0x0178)
+
+/* indexed registers */
+
+#define S3C2410_UDC_MAXP_REG		S3C2410_USBDREG(0x0180)
+
+#define S3C2410_UDC_EP0_CSR_REG		S3C2410_USBDREG(0x0184)
+
+#define S3C2410_UDC_IN_CSR1_REG		S3C2410_USBDREG(0x0184)
+#define S3C2410_UDC_IN_CSR2_REG		S3C2410_USBDREG(0x0188)
+
+#define S3C2410_UDC_OUT_CSR1_REG	S3C2410_USBDREG(0x0190)
+#define S3C2410_UDC_OUT_CSR2_REG	S3C2410_USBDREG(0x0194)
+#define S3C2410_UDC_OUT_FIFO_CNT1_REG	S3C2410_USBDREG(0x0198)
+#define S3C2410_UDC_OUT_FIFO_CNT2_REG	S3C2410_USBDREG(0x019c)
+
+#define S3C2410_UDC_FUNCADDR_UPDATE	(1 << 7)
+
+#define S3C2410_UDC_PWR_ISOUP		(1 << 7) /* R/W */
+#define S3C2410_UDC_PWR_RESET		(1 << 3) /* R   */
+#define S3C2410_UDC_PWR_RESUME		(1 << 2) /* R/W */
+#define S3C2410_UDC_PWR_SUSPEND		(1 << 1) /* R   */
+#define S3C2410_UDC_PWR_ENSUSPEND	(1 << 0) /* R/W */
+
+#define S3C2410_UDC_PWR_DEFAULT		(0x00)
+
+#define S3C2410_UDC_INT_EP4		(1 << 4) /* R/W (clear only) */
+#define S3C2410_UDC_INT_EP3		(1 << 3) /* R/W (clear only) */
+#define S3C2410_UDC_INT_EP2		(1 << 2) /* R/W (clear only) */
+#define S3C2410_UDC_INT_EP1		(1 << 1) /* R/W (clear only) */
+#define S3C2410_UDC_INT_EP0		(1 << 0) /* R/W (clear only) */
+
+#define S3C2410_UDC_USBINT_RESET	(1 << 2) /* R/W (clear only) */
+#define S3C2410_UDC_USBINT_RESUME	(1 << 1) /* R/W (clear only) */
+#define S3C2410_UDC_USBINT_SUSPEND	(1 << 0) /* R/W (clear only) */
+
+#define S3C2410_UDC_INTE_EP4		(1 << 4) /* R/W */
+#define S3C2410_UDC_INTE_EP3		(1 << 3) /* R/W */
+#define S3C2410_UDC_INTE_EP2		(1 << 2) /* R/W */
+#define S3C2410_UDC_INTE_EP1		(1 << 1) /* R/W */
+#define S3C2410_UDC_INTE_EP0		(1 << 0) /* R/W */
+
+#define S3C2410_UDC_USBINTE_RESET	(1 << 2) /* R/W */
+#define S3C2410_UDC_USBINTE_SUSPEND	(1 << 0) /* R/W */
+
+#define S3C2410_UDC_INDEX_EP0		(0x00)
+#define S3C2410_UDC_INDEX_EP1		(0x01)
+#define S3C2410_UDC_INDEX_EP2		(0x02)
+#define S3C2410_UDC_INDEX_EP3		(0x03)
+#define S3C2410_UDC_INDEX_EP4		(0x04)
+
+#define S3C2410_UDC_ICSR1_CLRDT		(1 << 6) /* R/W */
+#define S3C2410_UDC_ICSR1_SENTSTL	(1 << 5) /* R/W (clear only) */
+#define S3C2410_UDC_ICSR1_SENDSTL	(1 << 4) /* R/W */
+#define S3C2410_UDC_ICSR1_FFLUSH	(1 << 3) /* W   (set only) */
+#define S3C2410_UDC_ICSR1_UNDRUN	(1 << 2) /* R/W (clear only) */
+#define S3C2410_UDC_ICSR1_PKTRDY	(1 << 0) /* R/W (set only) */
+
+#define S3C2410_UDC_ICSR2_AUTOSET	(1 << 7) /* R/W */
+#define S3C2410_UDC_ICSR2_ISO		(1 << 6) /* R/W */
+#define S3C2410_UDC_ICSR2_MODEIN	(1 << 5) /* R/W */
+#define S3C2410_UDC_ICSR2_DMAIEN	(1 << 4) /* R/W */
+
+#define S3C2410_UDC_OCSR1_CLRDT		(1 << 7) /* R/W */
+#define S3C2410_UDC_OCSR1_SENTSTL	(1 << 6) /* R/W (clear only) */
+#define S3C2410_UDC_OCSR1_SENDSTL	(1 << 5) /* R/W */
+#define S3C2410_UDC_OCSR1_FFLUSH	(1 << 4) /* R/W */
+#define S3C2410_UDC_OCSR1_DERROR	(1 << 3) /* R   */
+#define S3C2410_UDC_OCSR1_OVRRUN	(1 << 2) /* R/W (clear only) */
+#define S3C2410_UDC_OCSR1_PKTRDY	(1 << 0) /* R/W (clear only) */
+
+#define S3C2410_UDC_OCSR2_AUTOCLR	(1 << 7) /* R/W */
+#define S3C2410_UDC_OCSR2_ISO		(1 << 6) /* R/W */
+#define S3C2410_UDC_OCSR2_DMAIEN	(1 << 5) /* R/W */
+
+#define S3C2410_UDC_EP0_CSR_OPKRDY	(1 << 0)
+#define S3C2410_UDC_EP0_CSR_IPKRDY	(1 << 1)
+#define S3C2410_UDC_EP0_CSR_SENTSTL	(1 << 2)
+#define S3C2410_UDC_EP0_CSR_DE		(1 << 3)
+#define S3C2410_UDC_EP0_CSR_SE		(1 << 4)
+#define S3C2410_UDC_EP0_CSR_SENDSTL	(1 << 5)
+#define S3C2410_UDC_EP0_CSR_SOPKTRDY	(1 << 6)
+#define S3C2410_UDC_EP0_CSR_SSE		(1 << 7)
+
+#define S3C2410_UDC_MAXP_8		(1 << 0)
+#define S3C2410_UDC_MAXP_16		(1 << 1)
+#define S3C2410_UDC_MAXP_32		(1 << 2)
+#define S3C2410_UDC_MAXP_64		(1 << 3)
+
+#endif

arch/arm/plat-samsung/include/plat/samsung-time.h

@@ -0,0 +1,30 @@
+/* linux/arch/arm/plat-samsung/include/plat/samsung-time.h
+ *
+ * Copyright 2011 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com/
+ *
+ * Header file for samsung s3c and s5p time support
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __ASM_PLAT_SAMSUNG_TIME_H
+#define __ASM_PLAT_SAMSUNG_TIME_H __FILE__
+
+/* SAMSUNG HR-Timer Clock mode */
+enum samsung_timer_mode {
+	SAMSUNG_PWM0,
+	SAMSUNG_PWM1,
+	SAMSUNG_PWM2,
+	SAMSUNG_PWM3,
+	SAMSUNG_PWM4,
+};
+
+extern void __init samsung_set_timer_source(enum samsung_timer_mode event,
+					enum samsung_timer_mode source);
+
+extern void __init samsung_timer_init(void);
+
+#endif /* __ASM_PLAT_SAMSUNG_TIME_H */

arch/arm/plat-samsung/include/plat/sdhci.h

@@ -0,0 +1,166 @@
+/* linux/arch/arm/plat-samsung/include/plat/sdhci.h
+ *
+ * Copyright (c) 2011 Samsung Electronics Co., Ltd.
+ *		http://www.samsung.com
+ *
+ * Copyright 2008 Openmoko, Inc.
+ * Copyright 2008 Simtec Electronics
+ *	http://armlinux.simtec.co.uk/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C Platform - SDHCI (HSMMC) platform data definitions
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#ifndef __PLAT_S3C_SDHCI_H
+#define __PLAT_S3C_SDHCI_H __FILE__
+
+#include <linux/platform_data/mmc-sdhci-s3c.h>
+#include <plat/devs.h>
+
+/* s3c_sdhci_set_platdata() - common helper for setting SDHCI platform data
+ * @pd: The default platform data for this device.
+ * @set: Pointer to the platform data to fill in.
+ */
+extern void s3c_sdhci_set_platdata(struct s3c_sdhci_platdata *pd,
+				    struct s3c_sdhci_platdata *set);
+
+/**
+ * s3c_sdhci0_set_platdata - Set platform data for S3C SDHCI device.
+ * @pd: Platform data to register to device.
+ *
+ * Register the given platform data for use withe S3C SDHCI device.
+ * The call will copy the platform data, so the board definitions can
+ * make the structure itself __initdata.
+ */
+extern void s3c_sdhci0_set_platdata(struct s3c_sdhci_platdata *pd);
+extern void s3c_sdhci1_set_platdata(struct s3c_sdhci_platdata *pd);
+extern void s3c_sdhci2_set_platdata(struct s3c_sdhci_platdata *pd);
+extern void s3c_sdhci3_set_platdata(struct s3c_sdhci_platdata *pd);
+
+/* Default platform data, exported so that per-cpu initialisation can
+ * set the correct one when there are more than one cpu type selected.
+*/
+
+extern struct s3c_sdhci_platdata s3c_hsmmc0_def_platdata;
+extern struct s3c_sdhci_platdata s3c_hsmmc1_def_platdata;
+extern struct s3c_sdhci_platdata s3c_hsmmc2_def_platdata;
+extern struct s3c_sdhci_platdata s3c_hsmmc3_def_platdata;
+
+/* Helper function availability */
+
+extern void s3c2416_setup_sdhci0_cfg_gpio(struct platform_device *, int w);
+extern void s3c2416_setup_sdhci1_cfg_gpio(struct platform_device *, int w);
+extern void s3c64xx_setup_sdhci0_cfg_gpio(struct platform_device *, int w);
+extern void s3c64xx_setup_sdhci1_cfg_gpio(struct platform_device *, int w);
+extern void s3c64xx_setup_sdhci2_cfg_gpio(struct platform_device *, int w);
+
+/* S3C2416 SDHCI setup */
+
+#ifdef CONFIG_S3C2416_SETUP_SDHCI
+static inline void s3c2416_default_sdhci0(void)
+{
+#ifdef CONFIG_S3C_DEV_HSMMC
+	s3c_hsmmc0_def_platdata.cfg_gpio = s3c2416_setup_sdhci0_cfg_gpio;
+#endif /* CONFIG_S3C_DEV_HSMMC */
+}
+
+static inline void s3c2416_default_sdhci1(void)
+{
+#ifdef CONFIG_S3C_DEV_HSMMC1
+	s3c_hsmmc1_def_platdata.cfg_gpio = s3c2416_setup_sdhci1_cfg_gpio;
+#endif /* CONFIG_S3C_DEV_HSMMC1 */
+}
+
+#else
+static inline void s3c2416_default_sdhci0(void) { }
+static inline void s3c2416_default_sdhci1(void) { }
+
+#endif /* CONFIG_S3C2416_SETUP_SDHCI */
+
+/* S3C64XX SDHCI setup */
+
+#ifdef CONFIG_S3C64XX_SETUP_SDHCI
+static inline void s3c6400_default_sdhci0(void)
+{
+#ifdef CONFIG_S3C_DEV_HSMMC
+	s3c_hsmmc0_def_platdata.cfg_gpio = s3c64xx_setup_sdhci0_cfg_gpio;
+#endif
+}
+
+static inline void s3c6400_default_sdhci1(void)
+{
+#ifdef CONFIG_S3C_DEV_HSMMC1
+	s3c_hsmmc1_def_platdata.cfg_gpio = s3c64xx_setup_sdhci1_cfg_gpio;
+#endif
+}
+
+static inline void s3c6400_default_sdhci2(void)
+{
+#ifdef CONFIG_S3C_DEV_HSMMC2
+	s3c_hsmmc2_def_platdata.cfg_gpio = s3c64xx_setup_sdhci2_cfg_gpio;
+#endif
+}
+
+static inline void s3c6410_default_sdhci0(void)
+{
+#ifdef CONFIG_S3C_DEV_HSMMC
+	s3c_hsmmc0_def_platdata.cfg_gpio = s3c64xx_setup_sdhci0_cfg_gpio;
+#endif
+}
+
+static inline void s3c6410_default_sdhci1(void)
+{
+#ifdef CONFIG_S3C_DEV_HSMMC1
+	s3c_hsmmc1_def_platdata.cfg_gpio = s3c64xx_setup_sdhci1_cfg_gpio;
+#endif
+}
+
+static inline void s3c6410_default_sdhci2(void)
+{
+#ifdef CONFIG_S3C_DEV_HSMMC2
+	s3c_hsmmc2_def_platdata.cfg_gpio = s3c64xx_setup_sdhci2_cfg_gpio;
+#endif
+}
+
+#else
+static inline void s3c6410_default_sdhci0(void) { }
+static inline void s3c6410_default_sdhci1(void) { }
+static inline void s3c6410_default_sdhci2(void) { }
+static inline void s3c6400_default_sdhci0(void) { }
+static inline void s3c6400_default_sdhci1(void) { }
+static inline void s3c6400_default_sdhci2(void) { }
+
+#endif /* CONFIG_S3C64XX_SETUP_SDHCI */
+
+static inline void s3c_sdhci_setname(int id, char *name)
+{
+	switch (id) {
+#ifdef CONFIG_S3C_DEV_HSMMC
+	case 0:
+		s3c_device_hsmmc0.name = name;
+		break;
+#endif
+#ifdef CONFIG_S3C_DEV_HSMMC1
+	case 1:
+		s3c_device_hsmmc1.name = name;
+		break;
+#endif
+#ifdef CONFIG_S3C_DEV_HSMMC2
+	case 2:
+		s3c_device_hsmmc2.name = name;
+		break;
+#endif
+#ifdef CONFIG_S3C_DEV_HSMMC3
+	case 3:
+		s3c_device_hsmmc3.name = name;
+		break;
+#endif
+	default:
+		break;
+	}
+}
+#endif /* __PLAT_S3C_SDHCI_H */

arch/arm/plat-samsung/include/plat/usb-phy.h

@@ -0,0 +1,19 @@
+/*
+ * Copyright (C) 2011 Samsung Electronics Co.Ltd
+ * Author: Joonyoung Shim <jy0922.shim@samsung.com>
+ *
+ * This program is free software; you can redistribute  it and/or modify it
+ * under  the terms of  the GNU General  Public License as published by the
+ * Free Software Foundation;  either version 2 of the  License, or (at your
+ * option) any later version.
+ */
+
+#ifndef __PLAT_SAMSUNG_USB_PHY_H
+#define __PLAT_SAMSUNG_USB_PHY_H __FILE__
+
+#include <linux/usb/samsung_usb_phy.h>
+
+extern int s5p_usb_phy_init(struct platform_device *pdev, int type);
+extern int s5p_usb_phy_exit(struct platform_device *pdev, int type);
+
+#endif /* __PLAT_SAMSUNG_USB_PHY_H */

arch/arm/plat-samsung/include/plat/wakeup-mask.h

@@ -0,0 +1,44 @@
+/* arch/arm/plat-samsung/include/plat/wakeup-mask.h
+ *
+ * Copyright 2010 Ben Dooks <ben-linux@fluff.org>
+ *
+ * Support for wakeup mask interrupts on newer SoCs
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+*/
+
+#ifndef __PLAT_WAKEUP_MASK_H
+#define __PLAT_WAKEUP_MASK_H __file__
+
+/* if no irq yet defined, but still want to mask */
+#define NO_WAKEUP_IRQ (0x90000000)
+
+/**
+ * struct samsung_wakeup_mask - wakeup mask information
+ * @irq: The interrupt associated with this wakeup.
+ * @bit: The bit, as a (1 << bitno) controlling this source.
+ */ 
+struct samsung_wakeup_mask {
+	unsigned int	irq;
+	u32		bit;
+};
+
+/**
+ * samsung_sync_wakemask - sync wakeup mask information for pm
+ * @reg: The register that is used.
+ * @masks: The list of masks to use.
+ * @nr_masks: The number of entries pointed to buy @masks.
+ *
+ * Synchronise the wakeup mask information at suspend time from the list
+ * of interrupts and control bits in @masks. We do this at suspend time
+ * as overriding the relevant irq chips is harder and the register is only
+ * required to be correct before we enter sleep.
+ */
+extern void samsung_sync_wakemask(void __iomem *reg,
+				  struct samsung_wakeup_mask *masks,
+				  int nr_masks);
+
+#endif /* __PLAT_WAKEUP_MASK_H */

arch/arm/plat-samsung/init.c

@@ -0,0 +1,178 @@
+/* linux/arch/arm/plat-s3c/init.c
+ *
+ * Copyright (c) 2008 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *	http://armlinux.simtec.co.uk/
+ *
+ * S3C series CPU initialisation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+/*
+ * NOTE: Code in this file is not used on S3C64xx when booting with
+ * Device Tree support.
+ */
+
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/interrupt.h>
+#include <linux/ioport.h>
+#include <linux/serial_core.h>
+#include <linux/serial_s3c.h>
+#include <linux/platform_device.h>
+#include <linux/of.h>
+
+#include <asm/mach/arch.h>
+#include <asm/mach/map.h>
+
+#include <plat/cpu.h>
+#include <plat/devs.h>
+
+static struct cpu_table *cpu;
+
+static struct cpu_table * __init s3c_lookup_cpu(unsigned long idcode,
+						struct cpu_table *tab,
+						unsigned int count)
+{
+	for (; count != 0; count--, tab++) {
+		if ((idcode & tab->idmask) == (tab->idcode & tab->idmask))
+			return tab;
+	}
+
+	return NULL;
+}
+
+void __init s3c_init_cpu(unsigned long idcode,
+			 struct cpu_table *cputab, unsigned int cputab_size)
+{
+	cpu = s3c_lookup_cpu(idcode, cputab, cputab_size);
+
+	if (cpu == NULL) {
+		printk(KERN_ERR "Unknown CPU type 0x%08lx\n", idcode);
+		panic("Unknown S3C24XX CPU");
+	}
+
+	printk("CPU %s (id 0x%08lx)\n", cpu->name, idcode);
+
+	if (cpu->init == NULL) {
+		printk(KERN_ERR "CPU %s support not enabled\n", cpu->name);
+		panic("Unsupported Samsung CPU");
+	}
+
+	if (cpu->map_io)
+		cpu->map_io();
+}
+
+/* s3c24xx_init_clocks
+ *
+ * Initialise the clock subsystem and associated information from the
+ * given master crystal value.
+ *
+ * xtal  = 0 -> use default PLL crystal value (normally 12MHz)
+ *      != 0 -> PLL crystal value in Hz
+*/
+
+void __init s3c24xx_init_clocks(int xtal)
+{
+	if (xtal == 0)
+		xtal = 12*1000*1000;
+
+	if (cpu == NULL)
+		panic("s3c24xx_init_clocks: no cpu setup?\n");
+
+	if (cpu->init_clocks == NULL)
+		panic("s3c24xx_init_clocks: cpu has no clock init\n");
+	else
+		(cpu->init_clocks)(xtal);
+}
+
+/* uart management */
+#if IS_ENABLED(CONFIG_SAMSUNG_ATAGS)
+static int nr_uarts __initdata = 0;
+
+#ifdef CONFIG_SERIAL_SAMSUNG_UARTS
+static struct s3c2410_uartcfg uart_cfgs[CONFIG_SERIAL_SAMSUNG_UARTS];
+#endif
+
+/* s3c24xx_init_uartdevs
+ *
+ * copy the specified platform data and configuration into our central
+ * set of devices, before the data is thrown away after the init process.
+ *
+ * This also fills in the array passed to the serial driver for the
+ * early initialisation of the console.
+*/
+
+void __init s3c24xx_init_uartdevs(char *name,
+				  struct s3c24xx_uart_resources *res,
+				  struct s3c2410_uartcfg *cfg, int no)
+{
+#ifdef CONFIG_SERIAL_SAMSUNG_UARTS
+	struct platform_device *platdev;
+	struct s3c2410_uartcfg *cfgptr = uart_cfgs;
+	struct s3c24xx_uart_resources *resp;
+	int uart;
+
+	memcpy(cfgptr, cfg, sizeof(struct s3c2410_uartcfg) * no);
+
+	for (uart = 0; uart < no; uart++, cfg++, cfgptr++) {
+		platdev = s3c24xx_uart_src[cfgptr->hwport];
+
+		resp = res + cfgptr->hwport;
+
+		s3c24xx_uart_devs[uart] = platdev;
+
+		platdev->name = name;
+		platdev->resource = resp->resources;
+		platdev->num_resources = resp->nr_resources;
+
+		platdev->dev.platform_data = cfgptr;
+	}
+
+	nr_uarts = no;
+#endif
+}
+
+void __init s3c24xx_init_uarts(struct s3c2410_uartcfg *cfg, int no)
+{
+	if (cpu == NULL)
+		return;
+
+	if (cpu->init_uarts == NULL && IS_ENABLED(CONFIG_SAMSUNG_ATAGS)) {
+		printk(KERN_ERR "s3c24xx_init_uarts: cpu has no uart init\n");
+	} else
+		(cpu->init_uarts)(cfg, no);
+}
+#endif
+
+static int __init s3c_arch_init(void)
+{
+	int ret;
+
+	/* init is only needed for ATAGS based platforms */
+	if (!IS_ENABLED(CONFIG_ATAGS) ||
+	    (!soc_is_s3c24xx() && !soc_is_s3c64xx()))
+		return 0;
+
+	// do the correct init for cpu
+
+	if (cpu == NULL) {
+		/* Not needed when booting with device tree. */
+		if (of_have_populated_dt())
+			return 0;
+		panic("s3c_arch_init: NULL cpu\n");
+	}
+
+	ret = (cpu->init)();
+	if (ret != 0)
+		return ret;
+#if IS_ENABLED(CONFIG_SAMSUNG_ATAGS)
+	ret = platform_add_devices(s3c24xx_uart_devs, nr_uarts);
+#endif
+	return ret;
+}
+
+arch_initcall(s3c_arch_init);

arch/arm/plat-samsung/platformdata.c

@@ -0,0 +1,60 @@
+/* linux/arch/arm/plat-samsung/platformdata.c
+ *
+ * Copyright 2010 Ben Dooks <ben-linux <at> fluff.org>
+ *
+ * Helper for platform data setting
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/string.h>
+#include <linux/platform_device.h>
+
+#include <plat/devs.h>
+#include <plat/sdhci.h>
+
+void __init *s3c_set_platdata(void *pd, size_t pdsize,
+			      struct platform_device *pdev)
+{
+	void *npd;
+
+	if (!pd) {
+		/* too early to use dev_name(), may not be registered */
+		printk(KERN_ERR "%s: no platform data supplied\n", pdev->name);
+		return NULL;
+	}
+
+	npd = kmemdup(pd, pdsize, GFP_KERNEL);
+	if (!npd) {
+		printk(KERN_ERR "%s: cannot clone platform data\n", pdev->name);
+		return NULL;
+	}
+
+	pdev->dev.platform_data = npd;
+	return npd;
+}
+
+void s3c_sdhci_set_platdata(struct s3c_sdhci_platdata *pd,
+			     struct s3c_sdhci_platdata *set)
+{
+	set->cd_type = pd->cd_type;
+	set->ext_cd_init = pd->ext_cd_init;
+	set->ext_cd_cleanup = pd->ext_cd_cleanup;
+	set->ext_cd_gpio = pd->ext_cd_gpio;
+	set->ext_cd_gpio_invert = pd->ext_cd_gpio_invert;
+
+	if (pd->max_width)
+		set->max_width = pd->max_width;
+	if (pd->cfg_gpio)
+		set->cfg_gpio = pd->cfg_gpio;
+	if (pd->host_caps)
+		set->host_caps |= pd->host_caps;
+	if (pd->host_caps2)
+		set->host_caps2 |= pd->host_caps2;
+	if (pd->pm_caps)
+		set->pm_caps |= pd->pm_caps;
+}

arch/arm/plat-samsung/pm-check.c

@@ -0,0 +1,237 @@
+/* linux/arch/arm/plat-s3c/pm-check.c
+ *  originally in linux/arch/arm/plat-s3c24xx/pm.c
+ *
+ * Copyright (c) 2004-2008 Simtec Electronics
+ *	http://armlinux.simtec.co.uk
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * S3C Power Mangament - suspend/resume memory corruption check.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/kernel.h>
+#include <linux/suspend.h>
+#include <linux/init.h>
+#include <linux/crc32.h>
+#include <linux/ioport.h>
+#include <linux/slab.h>
+
+#include <plat/pm-common.h>
+
+#if CONFIG_SAMSUNG_PM_CHECK_CHUNKSIZE < 1
+#error CONFIG_SAMSUNG_PM_CHECK_CHUNKSIZE must be a positive non-zero value
+#endif
+
+/* suspend checking code...
+ *
+ * this next area does a set of crc checks over all the installed
+ * memory, so the system can verify if the resume was ok.
+ *
+ * CONFIG_SAMSUNG_PM_CHECK_CHUNKSIZE defines the block-size for the CRC,
+ * increasing it will mean that the area corrupted will be less easy to spot,
+ * and reducing the size will cause the CRC save area to grow
+*/
+
+#define CHECK_CHUNKSIZE (CONFIG_SAMSUNG_PM_CHECK_CHUNKSIZE * 1024)
+
+static u32 crc_size;	/* size needed for the crc block */
+static u32 *crcs;	/* allocated over suspend/resume */
+
+typedef u32 *(run_fn_t)(struct resource *ptr, u32 *arg);
+
+/* s3c_pm_run_res
+ *
+ * go through the given resource list, and look for system ram
+*/
+
+static void s3c_pm_run_res(struct resource *ptr, run_fn_t fn, u32 *arg)
+{
+	while (ptr != NULL) {
+		if (ptr->child != NULL)
+			s3c_pm_run_res(ptr->child, fn, arg);
+
+		if ((ptr->flags & IORESOURCE_SYSTEM_RAM)
+				== IORESOURCE_SYSTEM_RAM) {
+			S3C_PMDBG("Found system RAM at %08lx..%08lx\n",
+				  (unsigned long)ptr->start,
+				  (unsigned long)ptr->end);
+			arg = (fn)(ptr, arg);
+		}
+
+		ptr = ptr->sibling;
+	}
+}
+
+static void s3c_pm_run_sysram(run_fn_t fn, u32 *arg)
+{
+	s3c_pm_run_res(&iomem_resource, fn, arg);
+}
+
+static u32 *s3c_pm_countram(struct resource *res, u32 *val)
+{
+	u32 size = (u32)resource_size(res);
+
+	size += CHECK_CHUNKSIZE-1;
+	size /= CHECK_CHUNKSIZE;
+
+	S3C_PMDBG("Area %08lx..%08lx, %d blocks\n",
+		  (unsigned long)res->start, (unsigned long)res->end, size);
+
+	*val += size * sizeof(u32);
+	return val;
+}
+
+/* s3c_pm_prepare_check
+ *
+ * prepare the necessary information for creating the CRCs. This
+ * must be done before the final save, as it will require memory
+ * allocating, and thus touching bits of the kernel we do not
+ * know about.
+*/
+
+void s3c_pm_check_prepare(void)
+{
+	crc_size = 0;
+
+	s3c_pm_run_sysram(s3c_pm_countram, &crc_size);
+
+	S3C_PMDBG("s3c_pm_prepare_check: %u checks needed\n", crc_size);
+
+	crcs = kmalloc(crc_size+4, GFP_KERNEL);
+	if (crcs == NULL)
+		printk(KERN_ERR "Cannot allocated CRC save area\n");
+}
+
+static u32 *s3c_pm_makecheck(struct resource *res, u32 *val)
+{
+	unsigned long addr, left;
+
+	for (addr = res->start; addr < res->end;
+	     addr += CHECK_CHUNKSIZE) {
+		left = res->end - addr;
+
+		if (left > CHECK_CHUNKSIZE)
+			left = CHECK_CHUNKSIZE;
+
+		*val = crc32_le(~0, phys_to_virt(addr), left);
+		val++;
+	}
+
+	return val;
+}
+
+/* s3c_pm_check_store
+ *
+ * compute the CRC values for the memory blocks before the final
+ * sleep.
+*/
+
+void s3c_pm_check_store(void)
+{
+	if (crcs != NULL)
+		s3c_pm_run_sysram(s3c_pm_makecheck, crcs);
+}
+
+/* in_region
+ *
+ * return TRUE if the area defined by ptr..ptr+size contains the
+ * what..what+whatsz
+*/
+
+static inline int in_region(void *ptr, int size, void *what, size_t whatsz)
+{
+	if ((what+whatsz) < ptr)
+		return 0;
+
+	if (what > (ptr+size))
+		return 0;
+
+	return 1;
+}
+
+/**
+ * s3c_pm_runcheck() - helper to check a resource on restore.
+ * @res: The resource to check
+ * @vak: Pointer to list of CRC32 values to check.
+ *
+ * Called from the s3c_pm_check_restore() via s3c_pm_run_sysram(), this
+ * function runs the given memory resource checking it against the stored
+ * CRC to ensure that memory is restored. The function tries to skip as
+ * many of the areas used during the suspend process.
+ */
+static u32 *s3c_pm_runcheck(struct resource *res, u32 *val)
+{
+	unsigned long addr;
+	unsigned long left;
+	void *stkpage;
+	void *ptr;
+	u32 calc;
+
+	stkpage = (void *)((u32)&calc & ~PAGE_MASK);
+
+	for (addr = res->start; addr < res->end;
+	     addr += CHECK_CHUNKSIZE) {
+		left = res->end - addr;
+
+		if (left > CHECK_CHUNKSIZE)
+			left = CHECK_CHUNKSIZE;
+
+		ptr = phys_to_virt(addr);
+
+		if (in_region(ptr, left, stkpage, 4096)) {
+			S3C_PMDBG("skipping %08lx, has stack in\n", addr);
+			goto skip_check;
+		}
+
+		if (in_region(ptr, left, crcs, crc_size)) {
+			S3C_PMDBG("skipping %08lx, has crc block in\n", addr);
+			goto skip_check;
+		}
+
+		/* calculate and check the checksum */
+
+		calc = crc32_le(~0, ptr, left);
+		if (calc != *val) {
+			printk(KERN_ERR "Restore CRC error at "
+			       "%08lx (%08x vs %08x)\n", addr, calc, *val);
+
+			S3C_PMDBG("Restore CRC error at %08lx (%08x vs %08x)\n",
+			    addr, calc, *val);
+		}
+
+	skip_check:
+		val++;
+	}
+
+	return val;
+}
+
+/**
+ * s3c_pm_check_restore() - memory check called on resume
+ *
+ * check the CRCs after the restore event and free the memory used
+ * to hold them
+*/
+void s3c_pm_check_restore(void)
+{
+	if (crcs != NULL)
+		s3c_pm_run_sysram(s3c_pm_runcheck, crcs);
+}
+
+/**
+ * s3c_pm_check_cleanup() - free memory resources
+ *
+ * Free the resources that where allocated by the suspend
+ * memory check code. We do this separately from the
+ * s3c_pm_check_restore() function as we cannot call any
+ * functions that might sleep during that resume.
+ */
+void s3c_pm_check_cleanup(void)
+{
+	kfree(crcs);
+	crcs = NULL;
+}
+

arch/arm/plat-samsung/pm-common.c

@@ -0,0 +1,75 @@
+/*
+ * Copyright (C) 2013 Samsung Electronics Co., Ltd.
+ *	Tomasz Figa <t.figa@samsung.com>
+ * Copyright (C) 2008 Openmoko, Inc.
+ * Copyright (C) 2004-2008 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *	http://armlinux.simtec.co.uk/
+ *
+ * Samsung common power management helper functions.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/io.h>
+#include <linux/kernel.h>
+
+#include <plat/pm-common.h>
+
+/* helper functions to save and restore register state */
+
+/**
+ * s3c_pm_do_save() - save a set of registers for restoration on resume.
+ * @ptr: Pointer to an array of registers.
+ * @count: Size of the ptr array.
+ *
+ * Run through the list of registers given, saving their contents in the
+ * array for later restoration when we wakeup.
+ */
+void s3c_pm_do_save(struct sleep_save *ptr, int count)
+{
+	for (; count > 0; count--, ptr++) {
+		ptr->val = readl_relaxed(ptr->reg);
+		S3C_PMDBG("saved %p value %08lx\n", ptr->reg, ptr->val);
+	}
+}
+
+/**
+ * s3c_pm_do_restore() - restore register values from the save list.
+ * @ptr: Pointer to an array of registers.
+ * @count: Size of the ptr array.
+ *
+ * Restore the register values saved from s3c_pm_do_save().
+ *
+ * Note, we do not use S3C_PMDBG() in here, as the system may not have
+ * restore the UARTs state yet
+*/
+
+void s3c_pm_do_restore(const struct sleep_save *ptr, int count)
+{
+	for (; count > 0; count--, ptr++) {
+		pr_debug("restore %p (restore %08lx, was %08x)\n",
+				ptr->reg, ptr->val, readl_relaxed(ptr->reg));
+
+		writel_relaxed(ptr->val, ptr->reg);
+	}
+}
+
+/**
+ * s3c_pm_do_restore_core() - early restore register values from save list.
+ *
+ * This is similar to s3c_pm_do_restore() except we try and minimise the
+ * side effects of the function in case registers that hardware might need
+ * to work has been restored.
+ *
+ * WARNING: Do not put any debug in here that may effect memory or use
+ * peripherals, as things may be changing!
+*/
+
+void s3c_pm_do_restore_core(const struct sleep_save *ptr, int count)
+{
+	for (; count > 0; count--, ptr++)
+		writel_relaxed(ptr->val, ptr->reg);
+}

arch/arm/plat-samsung/pm-debug.c

@@ -0,0 +1,99 @@
+/*
+ * Copyright (C) 2013 Samsung Electronics Co., Ltd.
+ *	Tomasz Figa <t.figa@samsung.com>
+ * Copyright (C) 2008 Openmoko, Inc.
+ * Copyright (C) 2004-2008 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *	http://armlinux.simtec.co.uk/
+ *
+ * Samsung common power management (suspend to RAM) debug support
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/serial_core.h>
+#include <linux/serial_s3c.h>
+#include <linux/io.h>
+
+#include <asm/mach/map.h>
+
+#include <plat/cpu.h>
+#include <plat/pm-common.h>
+
+#ifdef CONFIG_SAMSUNG_ATAGS
+#include <plat/pm.h>
+#include <mach/pm-core.h>
+#else
+static inline void s3c_pm_debug_init_uart(void) {}
+static inline void s3c_pm_arch_update_uart(void __iomem *regs,
+					   struct pm_uart_save *save) {}
+#endif
+
+static struct pm_uart_save uart_save;
+
+extern void printascii(const char *);
+
+void s3c_pm_dbg(const char *fmt, ...)
+{
+	va_list va;
+	char buff[256];
+
+	va_start(va, fmt);
+	vsnprintf(buff, sizeof(buff), fmt, va);
+	va_end(va);
+
+	printascii(buff);
+}
+
+void s3c_pm_debug_init(void)
+{
+	/* restart uart clocks so we can use them to output */
+	s3c_pm_debug_init_uart();
+}
+
+static inline void __iomem *s3c_pm_uart_base(void)
+{
+	unsigned long paddr;
+	unsigned long vaddr;
+
+	debug_ll_addr(&paddr, &vaddr);
+
+	return (void __iomem *)vaddr;
+}
+
+void s3c_pm_save_uarts(void)
+{
+	void __iomem *regs = s3c_pm_uart_base();
+	struct pm_uart_save *save = &uart_save;
+
+	save->ulcon = __raw_readl(regs + S3C2410_ULCON);
+	save->ucon = __raw_readl(regs + S3C2410_UCON);
+	save->ufcon = __raw_readl(regs + S3C2410_UFCON);
+	save->umcon = __raw_readl(regs + S3C2410_UMCON);
+	save->ubrdiv = __raw_readl(regs + S3C2410_UBRDIV);
+
+	if (!soc_is_s3c2410())
+		save->udivslot = __raw_readl(regs + S3C2443_DIVSLOT);
+
+	S3C_PMDBG("UART[%p]: ULCON=%04x, UCON=%04x, UFCON=%04x, UBRDIV=%04x\n",
+		  regs, save->ulcon, save->ucon, save->ufcon, save->ubrdiv);
+}
+
+void s3c_pm_restore_uarts(void)
+{
+	void __iomem *regs = s3c_pm_uart_base();
+	struct pm_uart_save *save = &uart_save;
+
+	s3c_pm_arch_update_uart(regs, save);
+
+	__raw_writel(save->ulcon, regs + S3C2410_ULCON);
+	__raw_writel(save->ucon,  regs + S3C2410_UCON);
+	__raw_writel(save->ufcon, regs + S3C2410_UFCON);
+	__raw_writel(save->umcon, regs + S3C2410_UMCON);
+	__raw_writel(save->ubrdiv, regs + S3C2410_UBRDIV);
+
+	if (!soc_is_s3c2410())
+		__raw_writel(save->udivslot, regs + S3C2443_DIVSLOT);
+}

arch/arm/plat-samsung/pm-gpio.c

@@ -0,0 +1,386 @@
+
+/* linux/arch/arm/plat-s3c/pm-gpio.c
+ *
+ * Copyright 2008 Openmoko, Inc.
+ * Copyright 2008 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *	http://armlinux.simtec.co.uk/
+ *
+ * S3C series GPIO PM code
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/kernel.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/gpio.h>
+
+#include <mach/gpio-samsung.h>
+
+#include <plat/gpio-core.h>
+#include <plat/pm.h>
+
+/* PM GPIO helpers */
+
+#define OFFS_CON	(0x00)
+#define OFFS_DAT	(0x04)
+#define OFFS_UP		(0x08)
+
+static void samsung_gpio_pm_1bit_save(struct samsung_gpio_chip *chip)
+{
+	chip->pm_save[0] = __raw_readl(chip->base + OFFS_CON);
+	chip->pm_save[1] = __raw_readl(chip->base + OFFS_DAT);
+}
+
+static void samsung_gpio_pm_1bit_resume(struct samsung_gpio_chip *chip)
+{
+	void __iomem *base = chip->base;
+	u32 old_gpcon = __raw_readl(base + OFFS_CON);
+	u32 old_gpdat = __raw_readl(base + OFFS_DAT);
+	u32 gps_gpcon = chip->pm_save[0];
+	u32 gps_gpdat = chip->pm_save[1];
+	u32 gpcon;
+
+	/* GPACON only has one bit per control / data and no PULLUPs.
+	 * GPACON[x] = 0 => Output, 1 => SFN */
+
+	/* first set all SFN bits to SFN */
+
+	gpcon = old_gpcon | gps_gpcon;
+	__raw_writel(gpcon, base + OFFS_CON);
+
+	/* now set all the other bits */
+
+	__raw_writel(gps_gpdat, base + OFFS_DAT);
+	__raw_writel(gps_gpcon, base + OFFS_CON);
+
+	S3C_PMDBG("%s: CON %08x => %08x, DAT %08x => %08x\n",
+		  chip->chip.label, old_gpcon, gps_gpcon, old_gpdat, gps_gpdat);
+}
+
+struct samsung_gpio_pm samsung_gpio_pm_1bit = {
+	.save	= samsung_gpio_pm_1bit_save,
+	.resume = samsung_gpio_pm_1bit_resume,
+};
+
+static void samsung_gpio_pm_2bit_save(struct samsung_gpio_chip *chip)
+{
+	chip->pm_save[0] = __raw_readl(chip->base + OFFS_CON);
+	chip->pm_save[1] = __raw_readl(chip->base + OFFS_DAT);
+	chip->pm_save[2] = __raw_readl(chip->base + OFFS_UP);
+}
+
+/* Test whether the given masked+shifted bits of an GPIO configuration
+ * are one of the SFN (special function) modes. */
+
+static inline int is_sfn(unsigned long con)
+{
+	return con >= 2;
+}
+
+/* Test if the given masked+shifted GPIO configuration is an input */
+
+static inline int is_in(unsigned long con)
+{
+	return con == 0;
+}
+
+/* Test if the given masked+shifted GPIO configuration is an output */
+
+static inline int is_out(unsigned long con)
+{
+	return con == 1;
+}
+
+/**
+ * samsung_gpio_pm_2bit_resume() - restore the given GPIO bank
+ * @chip: The chip information to resume.
+ *
+ * Restore one of the GPIO banks that was saved during suspend. This is
+ * not as simple as once thought, due to the possibility of glitches
+ * from the order that the CON and DAT registers are set in.
+ *
+ * The three states the pin can be are {IN,OUT,SFN} which gives us 9
+ * combinations of changes to check. Three of these, if the pin stays
+ * in the same configuration can be discounted. This leaves us with
+ * the following:
+ *
+ * { IN => OUT }  Change DAT first
+ * { IN => SFN }  Change CON first
+ * { OUT => SFN } Change CON first, so new data will not glitch
+ * { OUT => IN }  Change CON first, so new data will not glitch
+ * { SFN => IN }  Change CON first
+ * { SFN => OUT } Change DAT first, so new data will not glitch [1]
+ *
+ * We do not currently deal with the UP registers as these control
+ * weak resistors, so a small delay in change should not need to bring
+ * these into the calculations.
+ *
+ * [1] this assumes that writing to a pin DAT whilst in SFN will set the
+ *     state for when it is next output.
+ */
+static void samsung_gpio_pm_2bit_resume(struct samsung_gpio_chip *chip)
+{
+	void __iomem *base = chip->base;
+	u32 old_gpcon = __raw_readl(base + OFFS_CON);
+	u32 old_gpdat = __raw_readl(base + OFFS_DAT);
+	u32 gps_gpcon = chip->pm_save[0];
+	u32 gps_gpdat = chip->pm_save[1];
+	u32 gpcon, old, new, mask;
+	u32 change_mask = 0x0;
+	int nr;
+
+	/* restore GPIO pull-up settings */
+	__raw_writel(chip->pm_save[2], base + OFFS_UP);
+
+	/* Create a change_mask of all the items that need to have
+	 * their CON value changed before their DAT value, so that
+	 * we minimise the work between the two settings.
+	 */
+
+	for (nr = 0, mask = 0x03; nr < 32; nr += 2, mask <<= 2) {
+		old = (old_gpcon & mask) >> nr;
+		new = (gps_gpcon & mask) >> nr;
+
+		/* If there is no change, then skip */
+
+		if (old == new)
+			continue;
+
+		/* If both are special function, then skip */
+
+		if (is_sfn(old) && is_sfn(new))
+			continue;
+
+		/* Change is IN => OUT, do not change now */
+
+		if (is_in(old) && is_out(new))
+			continue;
+
+		/* Change is SFN => OUT, do not change now */
+
+		if (is_sfn(old) && is_out(new))
+			continue;
+
+		/* We should now be at the case of IN=>SFN,
+		 * OUT=>SFN, OUT=>IN, SFN=>IN. */
+
+		change_mask |= mask;
+	}
+
+
+	/* Write the new CON settings */
+
+	gpcon = old_gpcon & ~change_mask;
+	gpcon |= gps_gpcon & change_mask;
+
+	__raw_writel(gpcon, base + OFFS_CON);
+
+	/* Now change any items that require DAT,CON */
+
+	__raw_writel(gps_gpdat, base + OFFS_DAT);
+	__raw_writel(gps_gpcon, base + OFFS_CON);
+
+	S3C_PMDBG("%s: CON %08x => %08x, DAT %08x => %08x\n",
+		  chip->chip.label, old_gpcon, gps_gpcon, old_gpdat, gps_gpdat);
+}
+
+struct samsung_gpio_pm samsung_gpio_pm_2bit = {
+	.save	= samsung_gpio_pm_2bit_save,
+	.resume = samsung_gpio_pm_2bit_resume,
+};
+
+#if defined(CONFIG_ARCH_S3C64XX)
+static void samsung_gpio_pm_4bit_save(struct samsung_gpio_chip *chip)
+{
+	chip->pm_save[1] = __raw_readl(chip->base + OFFS_CON);
+	chip->pm_save[2] = __raw_readl(chip->base + OFFS_DAT);
+	chip->pm_save[3] = __raw_readl(chip->base + OFFS_UP);
+
+	if (chip->chip.ngpio > 8)
+		chip->pm_save[0] = __raw_readl(chip->base - 4);
+}
+
+static u32 samsung_gpio_pm_4bit_mask(u32 old_gpcon, u32 gps_gpcon)
+{
+	u32 old, new, mask;
+	u32 change_mask = 0x0;
+	int nr;
+
+	for (nr = 0, mask = 0x0f; nr < 16; nr += 4, mask <<= 4) {
+		old = (old_gpcon & mask) >> nr;
+		new = (gps_gpcon & mask) >> nr;
+
+		/* If there is no change, then skip */
+
+		if (old == new)
+			continue;
+
+		/* If both are special function, then skip */
+
+		if (is_sfn(old) && is_sfn(new))
+			continue;
+
+		/* Change is IN => OUT, do not change now */
+
+		if (is_in(old) && is_out(new))
+			continue;
+
+		/* Change is SFN => OUT, do not change now */
+
+		if (is_sfn(old) && is_out(new))
+			continue;
+
+		/* We should now be at the case of IN=>SFN,
+		 * OUT=>SFN, OUT=>IN, SFN=>IN. */
+
+		change_mask |= mask;
+	}
+
+	return change_mask;
+}
+
+static void samsung_gpio_pm_4bit_con(struct samsung_gpio_chip *chip, int index)
+{
+	void __iomem *con = chip->base + (index * 4);
+	u32 old_gpcon = __raw_readl(con);
+	u32 gps_gpcon = chip->pm_save[index + 1];
+	u32 gpcon, mask;
+
+	mask = samsung_gpio_pm_4bit_mask(old_gpcon, gps_gpcon);
+
+	gpcon = old_gpcon & ~mask;
+	gpcon |= gps_gpcon & mask;
+
+	__raw_writel(gpcon, con);
+}
+
+static void samsung_gpio_pm_4bit_resume(struct samsung_gpio_chip *chip)
+{
+	void __iomem *base = chip->base;
+	u32 old_gpcon[2];
+	u32 old_gpdat = __raw_readl(base + OFFS_DAT);
+	u32 gps_gpdat = chip->pm_save[2];
+
+	/* First, modify the CON settings */
+
+	old_gpcon[0] = 0;
+	old_gpcon[1] = __raw_readl(base + OFFS_CON);
+
+	samsung_gpio_pm_4bit_con(chip, 0);
+	if (chip->chip.ngpio > 8) {
+		old_gpcon[0] = __raw_readl(base - 4);
+		samsung_gpio_pm_4bit_con(chip, -1);
+	}
+
+	/* Now change the configurations that require DAT,CON */
+
+	__raw_writel(chip->pm_save[2], base + OFFS_DAT);
+	__raw_writel(chip->pm_save[1], base + OFFS_CON);
+	if (chip->chip.ngpio > 8)
+		__raw_writel(chip->pm_save[0], base - 4);
+
+	__raw_writel(chip->pm_save[2], base + OFFS_DAT);
+	__raw_writel(chip->pm_save[3], base + OFFS_UP);
+
+	if (chip->chip.ngpio > 8) {
+		S3C_PMDBG("%s: CON4 %08x,%08x => %08x,%08x, DAT %08x => %08x\n",
+			  chip->chip.label, old_gpcon[0], old_gpcon[1],
+			  __raw_readl(base - 4),
+			  __raw_readl(base + OFFS_CON),
+			  old_gpdat, gps_gpdat);
+	} else
+		S3C_PMDBG("%s: CON4 %08x => %08x, DAT %08x => %08x\n",
+			  chip->chip.label, old_gpcon[1],
+			  __raw_readl(base + OFFS_CON),
+			  old_gpdat, gps_gpdat);
+}
+
+struct samsung_gpio_pm samsung_gpio_pm_4bit = {
+	.save	= samsung_gpio_pm_4bit_save,
+	.resume = samsung_gpio_pm_4bit_resume,
+};
+#endif /* CONFIG_ARCH_S3C64XX */
+
+/**
+ * samsung_pm_save_gpio() - save gpio chip data for suspend
+ * @ourchip: The chip for suspend.
+ */
+static void samsung_pm_save_gpio(struct samsung_gpio_chip *ourchip)
+{
+	struct samsung_gpio_pm *pm = ourchip->pm;
+
+	if (pm == NULL || pm->save == NULL)
+		S3C_PMDBG("%s: no pm for %s\n", __func__, ourchip->chip.label);
+	else
+		pm->save(ourchip);
+}
+
+/**
+ * samsung_pm_save_gpios() - Save the state of the GPIO banks.
+ *
+ * For all the GPIO banks, save the state of each one ready for going
+ * into a suspend mode.
+ */
+void samsung_pm_save_gpios(void)
+{
+	struct samsung_gpio_chip *ourchip;
+	unsigned int gpio_nr;
+
+	for (gpio_nr = 0; gpio_nr < S3C_GPIO_END;) {
+		ourchip = samsung_gpiolib_getchip(gpio_nr);
+		if (!ourchip) {
+			gpio_nr++;
+			continue;
+		}
+
+		samsung_pm_save_gpio(ourchip);
+
+		S3C_PMDBG("%s: save %08x,%08x,%08x,%08x\n",
+			  ourchip->chip.label,
+			  ourchip->pm_save[0],
+			  ourchip->pm_save[1],
+			  ourchip->pm_save[2],
+			  ourchip->pm_save[3]);
+
+		gpio_nr += ourchip->chip.ngpio;
+		gpio_nr += CONFIG_S3C_GPIO_SPACE;
+	}
+}
+
+/**
+ * samsung_pm_resume_gpio() - restore gpio chip data after suspend
+ * @ourchip: The suspended chip.
+ */
+static void samsung_pm_resume_gpio(struct samsung_gpio_chip *ourchip)
+{
+	struct samsung_gpio_pm *pm = ourchip->pm;
+
+	if (pm == NULL || pm->resume == NULL)
+		S3C_PMDBG("%s: no pm for %s\n", __func__, ourchip->chip.label);
+	else
+		pm->resume(ourchip);
+}
+
+void samsung_pm_restore_gpios(void)
+{
+	struct samsung_gpio_chip *ourchip;
+	unsigned int gpio_nr;
+
+	for (gpio_nr = 0; gpio_nr < S3C_GPIO_END;) {
+		ourchip = samsung_gpiolib_getchip(gpio_nr);
+		if (!ourchip) {
+			gpio_nr++;
+			continue;
+		}
+
+		samsung_pm_resume_gpio(ourchip);
+
+		gpio_nr += ourchip->chip.ngpio;
+		gpio_nr += CONFIG_S3C_GPIO_SPACE;
+	}
+}

arch/arm/plat-samsung/pm.c

@@ -0,0 +1,204 @@
+/* linux/arch/arm/plat-s3c/pm.c
+ *
+ * Copyright 2008 Openmoko, Inc.
+ * Copyright 2004-2008 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *	http://armlinux.simtec.co.uk/
+ *
+ * S3C common power management (suspend to ram) support.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/init.h>
+#include <linux/suspend.h>
+#include <linux/errno.h>
+#include <linux/delay.h>
+#include <linux/of.h>
+#include <linux/serial_s3c.h>
+#include <linux/io.h>
+
+#include <asm/cacheflush.h>
+#include <asm/suspend.h>
+
+#include <mach/map.h>
+#include <mach/regs-clock.h>
+#include <mach/regs-irq.h>
+#include <mach/irqs.h>
+
+#include <asm/irq.h>
+
+#include <plat/pm.h>
+#include <mach/pm-core.h>
+
+/* for external use */
+
+unsigned long s3c_pm_flags;
+
+/* The IRQ ext-int code goes here, it is too small to currently bother
+ * with its own file. */
+
+unsigned long s3c_irqwake_intmask	= 0xffffffffL;
+unsigned long s3c_irqwake_eintmask	= 0xffffffffL;
+
+int s3c_irqext_wake(struct irq_data *data, unsigned int state)
+{
+	unsigned long bit = 1L << IRQ_EINT_BIT(data->irq);
+
+	if (!(s3c_irqwake_eintallow & bit))
+		return -ENOENT;
+
+	printk(KERN_INFO "wake %s for irq %d\n",
+	       state ? "enabled" : "disabled", data->irq);
+
+	if (!state)
+		s3c_irqwake_eintmask |= bit;
+	else
+		s3c_irqwake_eintmask &= ~bit;
+
+	return 0;
+}
+
+void (*pm_cpu_prep)(void);
+int (*pm_cpu_sleep)(unsigned long);
+
+#define any_allowed(mask, allow) (((mask) & (allow)) != (allow))
+
+/* s3c_pm_enter
+ *
+ * central control for sleep/resume process
+*/
+
+static int s3c_pm_enter(suspend_state_t state)
+{
+	int ret;
+	/* ensure the debug is initialised (if enabled) */
+
+	s3c_pm_debug_init();
+
+	S3C_PMDBG("%s(%d)\n", __func__, state);
+
+	if (pm_cpu_prep == NULL || pm_cpu_sleep == NULL) {
+		printk(KERN_ERR "%s: error: no cpu sleep function\n", __func__);
+		return -EINVAL;
+	}
+
+	/* check if we have anything to wake-up with... bad things seem
+	 * to happen if you suspend with no wakeup (system will often
+	 * require a full power-cycle)
+	*/
+
+	if (!of_have_populated_dt() &&
+	    !any_allowed(s3c_irqwake_intmask, s3c_irqwake_intallow) &&
+	    !any_allowed(s3c_irqwake_eintmask, s3c_irqwake_eintallow)) {
+		printk(KERN_ERR "%s: No wake-up sources!\n", __func__);
+		printk(KERN_ERR "%s: Aborting sleep\n", __func__);
+		return -EINVAL;
+	}
+
+	/* save all necessary core registers not covered by the drivers */
+
+	if (!of_have_populated_dt()) {
+		samsung_pm_save_gpios();
+		samsung_pm_saved_gpios();
+	}
+
+	s3c_pm_save_uarts();
+	s3c_pm_save_core();
+
+	/* set the irq configuration for wake */
+
+	s3c_pm_configure_extint();
+
+	S3C_PMDBG("sleep: irq wakeup masks: %08lx,%08lx\n",
+	    s3c_irqwake_intmask, s3c_irqwake_eintmask);
+
+	s3c_pm_arch_prepare_irqs();
+
+	/* call cpu specific preparation */
+
+	pm_cpu_prep();
+
+	/* flush cache back to ram */
+
+	flush_cache_all();
+
+	s3c_pm_check_store();
+
+	/* send the cpu to sleep... */
+
+	s3c_pm_arch_stop_clocks();
+
+	/* this will also act as our return point from when
+	 * we resume as it saves its own register state and restores it
+	 * during the resume.  */
+
+	ret = cpu_suspend(0, pm_cpu_sleep);
+	if (ret)
+		return ret;
+
+	/* restore the system state */
+
+	s3c_pm_restore_core();
+	s3c_pm_restore_uarts();
+
+	if (!of_have_populated_dt()) {
+		samsung_pm_restore_gpios();
+		s3c_pm_restored_gpios();
+	}
+
+	s3c_pm_debug_init();
+
+	/* check what irq (if any) restored the system */
+
+	s3c_pm_arch_show_resume_irqs();
+
+	S3C_PMDBG("%s: post sleep, preparing to return\n", __func__);
+
+	/* LEDs should now be 1110 */
+	s3c_pm_debug_smdkled(1 << 1, 0);
+
+	s3c_pm_check_restore();
+
+	/* ok, let's return from sleep */
+
+	S3C_PMDBG("S3C PM Resume (post-restore)\n");
+	return 0;
+}
+
+static int s3c_pm_prepare(void)
+{
+	/* prepare check area if configured */
+
+	s3c_pm_check_prepare();
+	return 0;
+}
+
+static void s3c_pm_finish(void)
+{
+	s3c_pm_check_cleanup();
+}
+
+static const struct platform_suspend_ops s3c_pm_ops = {
+	.enter		= s3c_pm_enter,
+	.prepare	= s3c_pm_prepare,
+	.finish		= s3c_pm_finish,
+	.valid		= suspend_valid_only_mem,
+};
+
+/* s3c_pm_init
+ *
+ * Attach the power management functions. This should be called
+ * from the board specific initialisation if the board supports
+ * it.
+*/
+
+int __init s3c_pm_init(void)
+{
+	printk("S3C Power Management, Copyright 2004 Simtec Electronics\n");
+
+	suspend_set_ops(&s3c_pm_ops);
+	return 0;
+}

arch/arm/plat-samsung/wakeup-mask.c

@@ -0,0 +1,47 @@
+/* arch/arm/plat-samsung/wakeup-mask.c
+ *
+ * Copyright 2010 Ben Dooks <ben-linux@fluff.org>
+ *
+ * Support for wakeup mask interrupts on newer SoCs
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/kernel.h>
+#include <linux/spinlock.h>
+#include <linux/device.h>
+#include <linux/types.h>
+#include <linux/irq.h>
+#include <linux/io.h>
+
+#include <plat/wakeup-mask.h>
+#include <plat/pm.h>
+
+void samsung_sync_wakemask(void __iomem *reg,
+			   struct samsung_wakeup_mask *mask, int nr_mask)
+{
+	struct irq_data *data;
+	u32 val;
+
+	val = __raw_readl(reg);
+
+	for (; nr_mask > 0; nr_mask--, mask++) {
+		if (mask->irq == NO_WAKEUP_IRQ) {
+			val |= mask->bit;
+			continue;
+		}
+
+		data = irq_get_irq_data(mask->irq);
+
+		/* bit of a liberty to read this directly from irq_data. */
+		if (irqd_is_wakeup_set(data))
+			val &= ~mask->bit;
+		else
+			val |= mask->bit;
+	}
+
+	printk(KERN_INFO "wakemask %08x => %08x\n", __raw_readl(reg), val);
+	__raw_writel(val, reg);
+}

arch/arm/plat-samsung/watchdog-reset.c

@@ -0,0 +1,97 @@
+/* arch/arm/plat-samsung/watchdog-reset.c
+ *
+ * Copyright (c) 2008 Simtec Electronics
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * Copyright (c) 2013 Tomasz Figa <tomasz.figa@gmail.com>
+ *
+ * Watchdog reset support for Samsung SoCs.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+*/
+
+#include <linux/clk.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/delay.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+
+#define S3C2410_WTCON			0x00
+#define S3C2410_WTDAT			0x04
+#define S3C2410_WTCNT			0x08
+
+#define S3C2410_WTCON_ENABLE		(1 << 5)
+#define S3C2410_WTCON_DIV16		(0 << 3)
+#define S3C2410_WTCON_RSTEN		(1 << 0)
+#define S3C2410_WTCON_PRESCALE(x)	((x) << 8)
+
+static void __iomem *wdt_base;
+static struct clk *wdt_clock;
+
+void samsung_wdt_reset(void)
+{
+	if (!wdt_base) {
+		pr_err("%s: wdt reset not initialized\n", __func__);
+		/* delay to allow the serial port to show the message */
+		mdelay(50);
+		return;
+	}
+
+	if (!IS_ERR(wdt_clock))
+		clk_prepare_enable(wdt_clock);
+
+	/* disable watchdog, to be safe  */
+	__raw_writel(0, wdt_base + S3C2410_WTCON);
+
+	/* put initial values into count and data */
+	__raw_writel(0x80, wdt_base + S3C2410_WTCNT);
+	__raw_writel(0x80, wdt_base + S3C2410_WTDAT);
+
+	/* set the watchdog to go and reset... */
+	__raw_writel(S3C2410_WTCON_ENABLE | S3C2410_WTCON_DIV16 |
+			S3C2410_WTCON_RSTEN | S3C2410_WTCON_PRESCALE(0x20),
+			wdt_base + S3C2410_WTCON);
+
+	/* wait for reset to assert... */
+	mdelay(500);
+
+	pr_err("Watchdog reset failed to assert reset\n");
+
+	/* delay to allow the serial port to show the message */
+	mdelay(50);
+}
+
+#ifdef CONFIG_OF
+static const struct of_device_id s3c2410_wdt_match[] = {
+	{ .compatible = "samsung,s3c2410-wdt" },
+	{},
+};
+
+void __init samsung_wdt_reset_of_init(void)
+{
+	struct device_node *np;
+
+	np = of_find_matching_node(NULL, s3c2410_wdt_match);
+	if (!np) {
+		pr_err("%s: failed to find watchdog node\n", __func__);
+		return;
+	}
+
+	wdt_base = of_iomap(np, 0);
+	if (!wdt_base) {
+		pr_err("%s: failed to map watchdog registers\n", __func__);
+		return;
+	}
+
+	wdt_clock = of_clk_get(np, 0);
+}
+#endif
+
+void __init samsung_wdt_reset_init(void __iomem *base)
+{
+	wdt_base = base;
+	wdt_clock = clk_get(NULL, "watchdog");
+}