[PATCH] powerpc: Implement support for setting little-endian mode via prctl

This adds the PowerPC part of the code to allow processes to change
their endian mode via prctl.

This also extends the alignment exception handler to be able to fix up
alignment exceptions that occur in little-endian mode, both for
"PowerPC" little-endian and true little-endian.

We always enter signal handlers in big-endian mode -- the support for
little-endian mode does not amount to the creation of a little-endian
user/kernel ABI.  If the signal handler returns, the endian mode is
restored to what it was when the signal was delivered.

We have two new kernel CPU feature bits, one for PPC little-endian and
one for true little-endian.  Most of the classic 32-bit processors
support PPC little-endian, and this is reflected in the CPU feature
table.  There are two corresponding feature bits reported to userland
in the AT_HWCAP aux vector entry.

This is based on an earlier patch by Anton Blanchard.

Signed-off-by: Paul Mackerras <paulus@samba.org>

6 files changed, 231 insertions, 115 deletions

diff --git a/arch/powerpc/kernel/align.c b/arch/powerpc/kernel/align.c
index faaec9c6f78f..4734b5de599d 100644
--- a/arch/powerpc/kernel/align.c
+++ b/arch/powerpc/kernel/align.c
@@ -35,17 +35,19 @@ struct aligninfo {
 
 #define INVALID	{ 0, 0 }
 
-#define LD	1	/* load */
-#define ST	2	/* store */
-#define	SE	4	/* sign-extend value */
-#define F	8	/* to/from fp regs */
-#define U	0x10	/* update index register */
-#define M	0x20	/* multiple load/store */
-#define SW	0x40	/* byte swap int or ... */
-#define S	0x40	/* ... single-precision fp */
-#define SX	0x40	/* byte count in XER */
+/* Bits in the flags field */
+#define LD	0	/* load */
+#define ST	1	/* store */
+#define	SE	2	/* sign-extend value */
+#define F	4	/* to/from fp regs */
+#define U	8	/* update index register */
+#define M	0x10	/* multiple load/store */
+#define SW	0x20	/* byte swap */
+#define S	0x40	/* single-precision fp or... */
+#define SX	0x40	/* ... byte count in XER */
 #define HARD	0x80	/* string, stwcx. */
 
+/* DSISR bits reported for a DCBZ instruction: */
 #define DCBZ	0x5f	/* 8xx/82xx dcbz faults when cache not enabled */
 
 #define SWAP(a, b)	(t = (a), (a) = (b), (b) = t)
@@ -256,12 +258,16 @@ static int emulate_dcbz(struct pt_regs *regs, unsigned char __user *addr)
 #define REG_BYTE(rp, i)		*((u8 *)(rp) + (i))
 #endif
 
+#define SWIZ_PTR(p)		((unsigned char __user *)((p) ^ swiz))
+
 static int emulate_multiple(struct pt_regs *regs, unsigned char __user *addr,
 			    unsigned int reg, unsigned int nb,
-			    unsigned int flags, unsigned int instr)
+			    unsigned int flags, unsigned int instr,
+			    unsigned long swiz)
 {
 	unsigned long *rptr;
-	unsigned int nb0, i;
+	unsigned int nb0, i, bswiz;
+	unsigned long p;
 
 	/*
 	 * We do not try to emulate 8 bytes multiple as they aren't really
@@ -280,9 +286,12 @@ static int emulate_multiple(struct pt_regs *regs, unsigned char __user *addr,
 			if (nb == 0)
 				return 1;
 		} else {
-			if (__get_user(instr,
-				       (unsigned int __user *)regs->nip))
+			unsigned long pc = regs->nip ^ (swiz & 4);
+
+			if (__get_user(instr, (unsigned int __user *)pc))
 				return -EFAULT;
+			if (swiz == 0 && (flags & SW))
+				instr = cpu_to_le32(instr);
 			nb = (instr >> 11) & 0x1f;
 			if (nb == 0)
 				nb = 32;
@@ -300,7 +309,10 @@ static int emulate_multiple(struct pt_regs *regs, unsigned char __user *addr,
 		return -EFAULT;	/* bad address */
 
 	rptr = &regs->gpr[reg];
-	if (flags & LD) {
+	p = (unsigned long) addr;
+	bswiz = (flags & SW)? 3: 0;
+
+	if (!(flags & ST)) {
 		/*
 		 * This zeroes the top 4 bytes of the affected registers
 		 * in 64-bit mode, and also zeroes out any remaining
@@ -311,26 +323,28 @@ static int emulate_multiple(struct pt_regs *regs, unsigned char __user *addr,
 			memset(&regs->gpr[0], 0,
 			       ((nb0 + 3) / 4) * sizeof(unsigned long));
 
-		for (i = 0; i < nb; ++i)
-			if (__get_user(REG_BYTE(rptr, i), addr + i))
+		for (i = 0; i < nb; ++i, ++p)
+			if (__get_user(REG_BYTE(rptr, i ^ bswiz), SWIZ_PTR(p)))
 				return -EFAULT;
 		if (nb0 > 0) {
 			rptr = &regs->gpr[0];
 			addr += nb;
-			for (i = 0; i < nb0; ++i)
-				if (__get_user(REG_BYTE(rptr, i), addr + i))
+			for (i = 0; i < nb0; ++i, ++p)
+				if (__get_user(REG_BYTE(rptr, i ^ bswiz),
+					       SWIZ_PTR(p)))
 					return -EFAULT;
 		}
 
 	} else {
-		for (i = 0; i < nb; ++i)
-			if (__put_user(REG_BYTE(rptr, i), addr + i))
+		for (i = 0; i < nb; ++i, ++p)
+			if (__put_user(REG_BYTE(rptr, i ^ bswiz), SWIZ_PTR(p)))
 				return -EFAULT;
 		if (nb0 > 0) {
 			rptr = &regs->gpr[0];
 			addr += nb;
-			for (i = 0; i < nb0; ++i)
-				if (__put_user(REG_BYTE(rptr, i), addr + i))
+			for (i = 0; i < nb0; ++i, ++p)
+				if (__put_user(REG_BYTE(rptr, i ^ bswiz),
+					       SWIZ_PTR(p)))
 					return -EFAULT;
 		}
 	}
@@ -352,7 +366,7 @@ int fix_alignment(struct pt_regs *regs)
 	unsigned int reg, areg;
 	unsigned int dsisr;
 	unsigned char __user *addr;
-	unsigned char __user *p;
+	unsigned long p, swiz;
 	int ret, t;
 	union {
 		u64 ll;
@@ -380,11 +394,15 @@ int fix_alignment(struct pt_regs *regs)
 	 * let's make one up from the instruction
 	 */
 	if (cpu_has_feature(CPU_FTR_NODSISRALIGN)) {
-		unsigned int real_instr;
-		if (unlikely(__get_user(real_instr,
-					(unsigned int __user *)regs->nip)))
+		unsigned long pc = regs->nip;
+
+		if (cpu_has_feature(CPU_FTR_PPC_LE) && (regs->msr & MSR_LE))
+			pc ^= 4;
+		if (unlikely(__get_user(instr, (unsigned int __user *)pc)))
 			return -EFAULT;
-		dsisr = make_dsisr(real_instr);
+		if (cpu_has_feature(CPU_FTR_REAL_LE) && (regs->msr & MSR_LE))
+			instr = cpu_to_le32(instr);
+		dsisr = make_dsisr(instr);
 	}
 
 	/* extract the operation and registers from the dsisr */
@@ -397,6 +415,24 @@ int fix_alignment(struct pt_regs *regs)
 	nb = aligninfo[instr].len;
 	flags = aligninfo[instr].flags;
 
+	/* Byteswap little endian loads and stores */
+	swiz = 0;
+	if (regs->msr & MSR_LE) {
+		flags ^= SW;
+		/*
+		 * So-called "PowerPC little endian" mode works by
+		 * swizzling addresses rather than by actually doing
+		 * any byte-swapping.  To emulate this, we XOR each
+		 * byte address with 7.  We also byte-swap, because
+		 * the processor's address swizzling depends on the
+		 * operand size (it xors the address with 7 for bytes,
+		 * 6 for halfwords, 4 for words, 0 for doublewords) but
+		 * we will xor with 7 and load/store each byte separately.
+		 */
+		if (cpu_has_feature(CPU_FTR_PPC_LE))
+			swiz = 7;
+	}
+
 	/* DAR has the operand effective address */
 	addr = (unsigned char __user *)regs->dar;
 
@@ -412,7 +448,8 @@ int fix_alignment(struct pt_regs *regs)
 	 * function
 	 */
 	if (flags & M)
-		return emulate_multiple(regs, addr, reg, nb, flags, instr);
+		return emulate_multiple(regs, addr, reg, nb,
+					flags, instr, swiz);
 
 	/* Verify the address of the operand */
 	if (unlikely(user_mode(regs) &&
@@ -431,51 +468,71 @@ int fix_alignment(struct pt_regs *regs)
 	/* If we are loading, get the data from user space, else
 	 * get it from register values
 	 */
-	if (flags & LD) {
+	if (!(flags & ST)) {
 		data.ll = 0;
 		ret = 0;
-		p = addr;
+		p = (unsigned long) addr;
 		switch (nb) {
 		case 8:
-			ret |= __get_user(data.v[0], p++);
-			ret |= __get_user(data.v[1], p++);
-			ret |= __get_user(data.v[2], p++);
-			ret |= __get_user(data.v[3], p++);
+			ret |= __get_user(data.v[0], SWIZ_PTR(p++));
+			ret |= __get_user(data.v[1], SWIZ_PTR(p++));
+			ret |= __get_user(data.v[2], SWIZ_PTR(p++));
+			ret |= __get_user(data.v[3], SWIZ_PTR(p++));
 		case 4:
-			ret |= __get_user(data.v[4], p++);
-			ret |= __get_user(data.v[5], p++);
+			ret |= __get_user(data.v[4], SWIZ_PTR(p++));
+			ret |= __get_user(data.v[5], SWIZ_PTR(p++));
 		case 2:
-			ret |= __get_user(data.v[6], p++);
-			ret |= __get_user(data.v[7], p++);
+			ret |= __get_user(data.v[6], SWIZ_PTR(p++));
+			ret |= __get_user(data.v[7], SWIZ_PTR(p++));
 			if (unlikely(ret))
 				return -EFAULT;
 		}
-	} else if (flags & F)
+	} else if (flags & F) {
 		data.dd = current->thread.fpr[reg];
-	else
+		if (flags & S) {
+			/* Single-precision FP store requires conversion... */
+#ifdef CONFIG_PPC_FPU
+			preempt_disable();
+			enable_kernel_fp();
+			cvt_df(&data.dd, (float *)&data.v[4], &current->thread);
+			preempt_enable();
+#else
+			return 0;
+#endif
+		}
+	} else
 		data.ll = regs->gpr[reg];
 
-	/* Perform other misc operations like sign extension, byteswap,
+	if (flags & SW) {
+		switch (nb) {
+		case 8:
+			SWAP(data.v[0], data.v[7]);
+			SWAP(data.v[1], data.v[6]);
+			SWAP(data.v[2], data.v[5]);
+			SWAP(data.v[3], data.v[4]);
+			break;
+		case 4:
+			SWAP(data.v[4], data.v[7]);
+			SWAP(data.v[5], data.v[6]);
+			break;
+		case 2:
+			SWAP(data.v[6], data.v[7]);
+			break;
+		}
+	}
+
+	/* Perform other misc operations like sign extension
 	 * or floating point single precision conversion
 	 */
-	switch (flags & ~U) {
+	switch (flags & ~(U|SW)) {
 	case LD+SE:	/* sign extend */
 		if ( nb == 2 )
 			data.ll = data.x16.low16;
 		else	/* nb must be 4 */
 			data.ll = data.x32.low32;
 		break;
-	case LD+S:	/* byte-swap */
-	case ST+S:
-		if (nb == 2) {
-			SWAP(data.v[6], data.v[7]);
-		} else {
-			SWAP(data.v[4], data.v[7]);
-			SWAP(data.v[5], data.v[6]);
-		}
-		break;
 
-	/* Single-precision FP load and store require conversions... */
+	/* Single-precision FP load requires conversion... */
 	case LD+F+S:
 #ifdef CONFIG_PPC_FPU
 		preempt_disable();
@@ -486,34 +543,24 @@ int fix_alignment(struct pt_regs *regs)
 		return 0;
 #endif
 		break;
-	case ST+F+S:
-#ifdef CONFIG_PPC_FPU
-		preempt_disable();
-		enable_kernel_fp();
-		cvt_df(&data.dd, (float *)&data.v[4], &current->thread);
-		preempt_enable();
-#else
-		return 0;
-#endif
-		break;
 	}
 
 	/* Store result to memory or update registers */
 	if (flags & ST) {
 		ret = 0;
-		p = addr;
+		p = (unsigned long) addr;
 		switch (nb) {
 		case 8:
-			ret |= __put_user(data.v[0], p++);
-			ret |= __put_user(data.v[1], p++);
-			ret |= __put_user(data.v[2], p++);
-			ret |= __put_user(data.v[3], p++);
+			ret |= __put_user(data.v[0], SWIZ_PTR(p++));
+			ret |= __put_user(data.v[1], SWIZ_PTR(p++));
+			ret |= __put_user(data.v[2], SWIZ_PTR(p++));
+			ret |= __put_user(data.v[3], SWIZ_PTR(p++));
 		case 4:
-			ret |= __put_user(data.v[4], p++);
-			ret |= __put_user(data.v[5], p++);
+			ret |= __put_user(data.v[4], SWIZ_PTR(p++));
+			ret |= __put_user(data.v[5], SWIZ_PTR(p++));
 		case 2:
-			ret |= __put_user(data.v[6], p++);
-			ret |= __put_user(data.v[7], p++);
+			ret |= __put_user(data.v[6], SWIZ_PTR(p++));
+			ret |= __put_user(data.v[7], SWIZ_PTR(p++));
 		}
 		if (unlikely(ret))
 			return -EFAULT;
diff --git a/arch/powerpc/kernel/cputable.c b/arch/powerpc/kernel/cputable.c
index 83f9ab139d4c..dfe2fcfb20a0 100644
--- a/arch/powerpc/kernel/cputable.c
+++ b/arch/powerpc/kernel/cputable.c
@@ -54,7 +54,8 @@ extern void __setup_cpu_ppc970(unsigned long offset, struct cpu_spec* spec);
 #define COMMON_USER_POWER5_PLUS	(COMMON_USER_PPC64 | PPC_FEATURE_POWER5_PLUS|\
 				 PPC_FEATURE_SMT | PPC_FEATURE_ICACHE_SNOOP)
 #define COMMON_USER_POWER6	(COMMON_USER_PPC64 | PPC_FEATURE_ARCH_2_05 |\
-				 PPC_FEATURE_SMT | PPC_FEATURE_ICACHE_SNOOP)
+				 PPC_FEATURE_SMT | PPC_FEATURE_ICACHE_SNOOP | \
+				 PPC_FEATURE_TRUE_LE)
 #define COMMON_USER_BOOKE	(PPC_FEATURE_32 | PPC_FEATURE_HAS_MMU | \
 				 PPC_FEATURE_BOOKE)
 
@@ -74,7 +75,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00400000,
 		.cpu_name		= "POWER3 (630)",
 		.cpu_features		= CPU_FTRS_POWER3,
-		.cpu_user_features	= COMMON_USER_PPC64,
+		.cpu_user_features	= COMMON_USER_PPC64|PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 128,
 		.dcache_bsize		= 128,
 		.num_pmcs		= 8,
@@ -87,7 +88,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00410000,
 		.cpu_name		= "POWER3 (630+)",
 		.cpu_features		= CPU_FTRS_POWER3,
-		.cpu_user_features	= COMMON_USER_PPC64,
+		.cpu_user_features	= COMMON_USER_PPC64|PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 128,
 		.dcache_bsize		= 128,
 		.num_pmcs		= 8,
@@ -318,7 +319,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00030000,
 		.cpu_name		= "603",
 		.cpu_features		= CPU_FTRS_603,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.cpu_setup		= __setup_cpu_603,
@@ -329,7 +330,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00060000,
 		.cpu_name		= "603e",
 		.cpu_features		= CPU_FTRS_603,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.cpu_setup		= __setup_cpu_603,
@@ -340,7 +341,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00070000,
 		.cpu_name		= "603ev",
 		.cpu_features		= CPU_FTRS_603,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.cpu_setup		= __setup_cpu_603,
@@ -351,7 +352,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00040000,
 		.cpu_name		= "604",
 		.cpu_features		= CPU_FTRS_604,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 2,
@@ -363,7 +364,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00090000,
 		.cpu_name		= "604e",
 		.cpu_features		= CPU_FTRS_604,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -375,7 +376,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00090000,
 		.cpu_name		= "604r",
 		.cpu_features		= CPU_FTRS_604,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -387,7 +388,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x000a0000,
 		.cpu_name		= "604ev",
 		.cpu_features		= CPU_FTRS_604,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -399,7 +400,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00084202,
 		.cpu_name		= "740/750",
 		.cpu_features		= CPU_FTRS_740_NOTAU,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -411,7 +412,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00080100,
 		.cpu_name		= "750CX",
 		.cpu_features		= CPU_FTRS_750,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -423,7 +424,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00082200,
 		.cpu_name		= "750CX",
 		.cpu_features		= CPU_FTRS_750,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -435,7 +436,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00082210,
 		.cpu_name		= "750CXe",
 		.cpu_features		= CPU_FTRS_750,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -447,7 +448,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00083214,
 		.cpu_name		= "750CXe",
 		.cpu_features		= CPU_FTRS_750,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -459,7 +460,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00083000,
 		.cpu_name		= "745/755",
 		.cpu_features		= CPU_FTRS_750,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -471,7 +472,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x70000100,
 		.cpu_name		= "750FX",
 		.cpu_features		= CPU_FTRS_750FX1,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -483,7 +484,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x70000200,
 		.cpu_name		= "750FX",
 		.cpu_features		= CPU_FTRS_750FX2,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -495,7 +496,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x70000000,
 		.cpu_name		= "750FX",
 		.cpu_features		= CPU_FTRS_750FX,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -507,7 +508,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x70020000,
 		.cpu_name		= "750GX",
 		.cpu_features		= CPU_FTRS_750GX,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -519,7 +520,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x00080000,
 		.cpu_name		= "740/750",
 		.cpu_features		= CPU_FTRS_740,
-		.cpu_user_features	= COMMON_USER,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -531,7 +532,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x000c1101,
 		.cpu_name		= "7400 (1.1)",
 		.cpu_features		= CPU_FTRS_7400_NOTAU,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -543,7 +545,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x000c0000,
 		.cpu_name		= "7400",
 		.cpu_features		= CPU_FTRS_7400,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -555,7 +558,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x800c0000,
 		.cpu_name		= "7410",
 		.cpu_features		= CPU_FTRS_7400,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 4,
@@ -567,7 +571,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80000200,
 		.cpu_name		= "7450",
 		.cpu_features		= CPU_FTRS_7450_20,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -581,7 +586,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80000201,
 		.cpu_name		= "7450",
 		.cpu_features		= CPU_FTRS_7450_21,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -595,7 +601,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80000000,
 		.cpu_name		= "7450",
 		.cpu_features		= CPU_FTRS_7450_23,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -609,7 +616,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80010100,
 		.cpu_name		= "7455",
 		.cpu_features		= CPU_FTRS_7455_1,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -623,7 +631,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80010200,
 		.cpu_name		= "7455",
 		.cpu_features		= CPU_FTRS_7455_20,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -637,7 +646,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80010000,
 		.cpu_name		= "7455",
 		.cpu_features		= CPU_FTRS_7455,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -651,7 +661,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80020100,
 		.cpu_name		= "7447/7457",
 		.cpu_features		= CPU_FTRS_7447_10,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -665,7 +676,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80020101,
 		.cpu_name		= "7447/7457",
 		.cpu_features		= CPU_FTRS_7447_10,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -679,7 +691,7 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80020000,
 		.cpu_name		= "7447/7457",
 		.cpu_features		= CPU_FTRS_7447,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -693,7 +705,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80030000,
 		.cpu_name		= "7447A",
 		.cpu_features		= CPU_FTRS_7447A,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
@@ -707,7 +720,8 @@ struct cpu_spec	cpu_specs[] = {
 		.pvr_value		= 0x80040000,
 		.cpu_name		= "7448",
 		.cpu_features		= CPU_FTRS_7447A,
-		.cpu_user_features	= COMMON_USER | PPC_FEATURE_HAS_ALTIVEC_COMP,
+		.cpu_user_features	= COMMON_USER |
+			PPC_FEATURE_HAS_ALTIVEC_COMP | PPC_FEATURE_PPC_LE,
 		.icache_bsize		= 32,
 		.dcache_bsize		= 32,
 		.num_pmcs		= 6,
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 2dd47d2dd998..2d35d83961b2 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -708,6 +708,50 @@ int get_fpexc_mode(struct task_struct *tsk, unsigned long adr)
 	return put_user(val, (unsigned int __user *) adr);
 }
 
+int set_endian(struct task_struct *tsk, unsigned int val)
+{
+	struct pt_regs *regs = tsk->thread.regs;
+
+	if ((val == PR_ENDIAN_LITTLE && !cpu_has_feature(CPU_FTR_REAL_LE)) ||
+	    (val == PR_ENDIAN_PPC_LITTLE && !cpu_has_feature(CPU_FTR_PPC_LE)))
+		return -EINVAL;
+
+	if (regs == NULL)
+		return -EINVAL;
+
+	if (val == PR_ENDIAN_BIG)
+		regs->msr &= ~MSR_LE;
+	else if (val == PR_ENDIAN_LITTLE || val == PR_ENDIAN_PPC_LITTLE)
+		regs->msr |= MSR_LE;
+	else
+		return -EINVAL;
+
+	return 0;
+}
+
+int get_endian(struct task_struct *tsk, unsigned long adr)
+{
+	struct pt_regs *regs = tsk->thread.regs;
+	unsigned int val;
+
+	if (!cpu_has_feature(CPU_FTR_PPC_LE) &&
+	    !cpu_has_feature(CPU_FTR_REAL_LE))
+		return -EINVAL;
+
+	if (regs == NULL)
+		return -EINVAL;
+
+	if (regs->msr & MSR_LE) {
+		if (cpu_has_feature(CPU_FTR_REAL_LE))
+			val = PR_ENDIAN_LITTLE;
+		else
+			val = PR_ENDIAN_PPC_LITTLE;
+	} else
+		val = PR_ENDIAN_BIG;
+
+	return put_user(val, (unsigned int __user *)adr);
+}
+
 #define TRUNC_PTR(x)	((typeof(x))(((unsigned long)(x)) & 0xffffffff))
 
 int sys_clone(unsigned long clone_flags, unsigned long usp,
diff --git a/arch/powerpc/kernel/signal_32.c b/arch/powerpc/kernel/signal_32.c
index 22f078984845..237faeec2ec2 100644
--- a/arch/powerpc/kernel/signal_32.c
+++ b/arch/powerpc/kernel/signal_32.c
@@ -419,9 +419,7 @@ static long restore_user_regs(struct pt_regs *regs,
 {
 	long err;
 	unsigned int save_r2 = 0;
-#if defined(CONFIG_ALTIVEC) || defined(CONFIG_SPE)
 	unsigned long msr;
-#endif
 
 	/*
 	 * restore general registers but not including MSR or SOFTE. Also
@@ -430,11 +428,16 @@ static long restore_user_regs(struct pt_regs *regs,
 	if (!sig)
 		save_r2 = (unsigned int)regs->gpr[2];
 	err = restore_general_regs(regs, sr);
+	err |= __get_user(msr, &sr->mc_gregs[PT_MSR]);
 	if (!sig)
 		regs->gpr[2] = (unsigned long) save_r2;
 	if (err)
 		return 1;
 
+	/* if doing signal return, restore the previous little-endian mode */
+	if (sig)
+		regs->msr = (regs->msr & ~MSR_LE) | (msr & MSR_LE);
+
 	/*
 	 * Do this before updating the thread state in
 	 * current->thread.fpr/vr/evr.  That way, if we get preempted
@@ -455,7 +458,7 @@ static long restore_user_regs(struct pt_regs *regs,
 	/* force the process to reload the altivec registers from
 	   current->thread when it next does altivec instructions */
 	regs->msr &= ~MSR_VEC;
-	if (!__get_user(msr, &sr->mc_gregs[PT_MSR]) && (msr & MSR_VEC) != 0) {
+	if (msr & MSR_VEC) {
 		/* restore altivec registers from the stack */
 		if (__copy_from_user(current->thread.vr, &sr->mc_vregs,
 				     sizeof(sr->mc_vregs)))
@@ -472,7 +475,7 @@ static long restore_user_regs(struct pt_regs *regs,
 	/* force the process to reload the spe registers from
 	   current->thread when it next does spe instructions */
 	regs->msr &= ~MSR_SPE;
-	if (!__get_user(msr, &sr->mc_gregs[PT_MSR]) && (msr & MSR_SPE) != 0) {
+	if (msr & MSR_SPE) {
 		/* restore spe registers from the stack */
 		if (__copy_from_user(current->thread.evr, &sr->mc_vregs,
 				     ELF_NEVRREG * sizeof(u32)))
@@ -777,6 +780,8 @@ static int handle_rt_signal(unsigned long sig, struct k_sigaction *ka,
 	regs->gpr[5] = (unsigned long) &rt_sf->uc;
 	regs->gpr[6] = (unsigned long) rt_sf;
 	regs->nip = (unsigned long) ka->sa.sa_handler;
+	/* enter the signal handler in big-endian mode */
+	regs->msr &= ~MSR_LE;
 	regs->trap = 0;
 	return 1;
 
@@ -1047,6 +1052,8 @@ static int handle_signal(unsigned long sig, struct k_sigaction *ka,
 	regs->gpr[3] = sig;
 	regs->gpr[4] = (unsigned long) sc;
 	regs->nip = (unsigned long) ka->sa.sa_handler;
+	/* enter the signal handler in big-endian mode */
+	regs->msr &= ~MSR_LE;
 	regs->trap = 0;
 
 	return 1;
diff --git a/arch/powerpc/kernel/signal_64.c b/arch/powerpc/kernel/signal_64.c
index 23ba69c26913..66a5fbe31989 100644
--- a/arch/powerpc/kernel/signal_64.c
+++ b/arch/powerpc/kernel/signal_64.c
@@ -141,9 +141,7 @@ static long restore_sigcontext(struct pt_regs *regs, sigset_t *set, int sig,
 	unsigned long err = 0;
 	unsigned long save_r13 = 0;
 	elf_greg_t *gregs = (elf_greg_t *)regs;
-#ifdef CONFIG_ALTIVEC
 	unsigned long msr;
-#endif
 	int i;
 
 	/* If this is not a signal return, we preserve the TLS in r13 */
@@ -154,7 +152,12 @@ static long restore_sigcontext(struct pt_regs *regs, sigset_t *set, int sig,
 	err |= __copy_from_user(regs, &sc->gp_regs,
 				PT_MSR*sizeof(unsigned long));
 
-	/* skip MSR and SOFTE */
+	/* get MSR separately, transfer the LE bit if doing signal return */
+	err |= __get_user(msr, &sc->gp_regs[PT_MSR]);
+	if (sig)
+		regs->msr = (regs->msr & ~MSR_LE) | (msr & MSR_LE);
+
+	/* skip SOFTE */
 	for (i = PT_MSR+1; i <= PT_RESULT; i++) {
 		if (i == PT_SOFTE)
 			continue;
@@ -179,7 +182,6 @@ static long restore_sigcontext(struct pt_regs *regs, sigset_t *set, int sig,
 
 #ifdef CONFIG_ALTIVEC
 	err |= __get_user(v_regs, &sc->v_regs);
-	err |= __get_user(msr, &sc->gp_regs[PT_MSR]);
 	if (err)
 		return err;
 	/* Copy 33 vec registers (vr0..31 and vscr) from the stack */
@@ -410,6 +412,8 @@ static int setup_rt_frame(int signr, struct k_sigaction *ka, siginfo_t *info,
 
 	/* Set up "regs" so we "return" to the signal handler. */
 	err |= get_user(regs->nip, &funct_desc_ptr->entry);
+	/* enter the signal handler in big-endian mode */
+	regs->msr &= ~MSR_LE;
 	regs->gpr[1] = newsp;
 	err |= get_user(regs->gpr[2], &funct_desc_ptr->toc);
 	regs->gpr[3] = signr;
diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
index 064a52564692..03def5715494 100644
--- a/arch/powerpc/kernel/traps.c
+++ b/arch/powerpc/kernel/traps.c
@@ -658,7 +658,7 @@ static int emulate_instruction(struct pt_regs *regs)
 	u32 instword;
 	u32 rd;
 
-	if (!user_mode(regs))
+	if (!user_mode(regs) || (regs->msr & MSR_LE))
 		return -EINVAL;
 	CHECK_FULL_REGS(regs);