Merge branch 'next' into for-linus

Prepare first set of updates for 3.10 merge window.

112 files changed, 1313 insertions, 1733 deletions

diff --git a/arch/sparc/Kconfig b/arch/sparc/Kconfig
index b6b442b0d793..289127d5241c 100644
--- a/arch/sparc/Kconfig
+++ b/arch/sparc/Kconfig
@@ -20,9 +20,9 @@ config SPARC
 	select HAVE_ARCH_TRACEHOOK
 	select SYSCTL_EXCEPTION_TRACE
 	select ARCH_WANT_OPTIONAL_GPIOLIB
+	select ARCH_HAS_ATOMIC64_DEC_IF_POSITIVE
 	select RTC_CLASS
 	select RTC_DRV_M48T59
-	select HAVE_IRQ_WORK
 	select HAVE_DMA_ATTRS
 	select HAVE_DMA_API_DEBUG
 	select HAVE_ARCH_JUMP_LABEL
@@ -40,12 +40,15 @@ config SPARC
 	select GENERIC_STRNCPY_FROM_USER
 	select GENERIC_STRNLEN_USER
 	select MODULES_USE_ELF_RELA
+	select ODD_RT_SIGACTION
+	select OLD_SIGSUSPEND
 
 config SPARC32
 	def_bool !64BIT
 	select GENERIC_ATOMIC64
 	select CLZ_TAB
 	select HAVE_UID16
+	select OLD_SIGACTION
 
 config SPARC64
 	def_bool 64BIT
@@ -60,6 +63,7 @@ config SPARC64
 	select HAVE_MEMBLOCK
 	select HAVE_MEMBLOCK_NODE_MAP
 	select HAVE_SYSCALL_WRAPPERS
+	select HAVE_ARCH_TRANSPARENT_HUGEPAGE
 	select HAVE_DYNAMIC_FTRACE
 	select HAVE_FTRACE_MCOUNT_RECORD
 	select HAVE_SYSCALL_TRACEPOINTS
@@ -142,9 +146,6 @@ config GENERIC_GPIO
 	help
 	  Generic GPIO API support
 
-config ARCH_NO_VIRT_TO_BUS
-	def_bool y
-
 config ARCH_SUPPORTS_DEBUG_PAGEALLOC
 	def_bool y if SPARC64
 
@@ -542,6 +543,7 @@ config COMPAT
 	select COMPAT_BINFMT_ELF
 	select HAVE_UID16
 	select ARCH_WANT_OLD_COMPAT_IPC
+	select COMPAT_OLD_SIGACTION
 
 config SYSVIPC_COMPAT
 	bool
diff --git a/arch/sparc/boot/piggyback.c b/arch/sparc/boot/piggyback.c
index c0a798fcf030..bb7c95161d71 100644
--- a/arch/sparc/boot/piggyback.c
+++ b/arch/sparc/boot/piggyback.c
@@ -81,18 +81,18 @@ static void usage(void)
 
 static int start_line(const char *line)
 {
-	if (strcmp(line + 8, " T _start\n") == 0)
+	if (strcmp(line + 10, " _start\n") == 0)
 		return 1;
-	else if (strcmp(line + 16, " T _start\n") == 0)
+	else if (strcmp(line + 18, " _start\n") == 0)
 		return 1;
 	return 0;
 }
 
 static int end_line(const char *line)
 {
-	if (strcmp(line + 8, " A _end\n") == 0)
+	if (strcmp(line + 10, " _end\n") == 0)
 		return 1;
-	else if (strcmp (line + 16, " A _end\n") == 0)
+	else if (strcmp (line + 18, " _end\n") == 0)
 		return 1;
 	return 0;
 }
@@ -100,8 +100,8 @@ static int end_line(const char *line)
 /*
  * Find address for start and end in System.map.
  * The file looks like this:
- * f0004000 T _start
- * f0379f79 A _end
+ * f0004000 ... _start
+ * f0379f79 ... _end
  * 1234567890123456
  * ^coloumn 1
  * There is support for 64 bit addresses too.
diff --git a/arch/sparc/crypto/Makefile b/arch/sparc/crypto/Makefile
index 6ae1ad5e502b..5d469d81761f 100644
--- a/arch/sparc/crypto/Makefile
+++ b/arch/sparc/crypto/Makefile
@@ -13,13 +13,13 @@ obj-$(CONFIG_CRYPTO_DES_SPARC64) += camellia-sparc64.o
 
 obj-$(CONFIG_CRYPTO_CRC32C_SPARC64) += crc32c-sparc64.o
 
-sha1-sparc64-y := sha1_asm.o sha1_glue.o crop_devid.o
-sha256-sparc64-y := sha256_asm.o sha256_glue.o crop_devid.o
-sha512-sparc64-y := sha512_asm.o sha512_glue.o crop_devid.o
-md5-sparc64-y := md5_asm.o md5_glue.o crop_devid.o
+sha1-sparc64-y := sha1_asm.o sha1_glue.o
+sha256-sparc64-y := sha256_asm.o sha256_glue.o
+sha512-sparc64-y := sha512_asm.o sha512_glue.o
+md5-sparc64-y := md5_asm.o md5_glue.o
 
-aes-sparc64-y := aes_asm.o aes_glue.o crop_devid.o
-des-sparc64-y := des_asm.o des_glue.o crop_devid.o
-camellia-sparc64-y := camellia_asm.o camellia_glue.o crop_devid.o
+aes-sparc64-y := aes_asm.o aes_glue.o
+des-sparc64-y := des_asm.o des_glue.o
+camellia-sparc64-y := camellia_asm.o camellia_glue.o
 
-crc32c-sparc64-y := crc32c_asm.o crc32c_glue.o crop_devid.o
+crc32c-sparc64-y := crc32c_asm.o crc32c_glue.o
diff --git a/arch/sparc/crypto/aes_asm.S b/arch/sparc/crypto/aes_asm.S
index 23f6cbb910d3..1cda8aa7cb85 100644
--- a/arch/sparc/crypto/aes_asm.S
+++ b/arch/sparc/crypto/aes_asm.S
@@ -1024,7 +1024,11 @@ ENTRY(aes_sparc64_ecb_encrypt_256)
 	 add		%o2, 0x20, %o2
 	brlz,pt		%o3, 11f
 	 nop
-10:	ldx		[%o1 + 0x00], %g3
+10:	ldd		[%o0 + 0xd0], %f56
+	ldd		[%o0 + 0xd8], %f58
+	ldd		[%o0 + 0xe0], %f60
+	ldd		[%o0 + 0xe8], %f62
+	ldx		[%o1 + 0x00], %g3
 	ldx		[%o1 + 0x08], %g7
 	xor		%g1, %g3, %g3
 	xor		%g2, %g7, %g7
@@ -1128,9 +1132,9 @@ ENTRY(aes_sparc64_ecb_decrypt_256)
 	/* %o0=&key[key_len], %o1=input, %o2=output, %o3=len */
 	ldx		[%o0 - 0x10], %g1
 	subcc		%o3, 0x10, %o3
+	ldx		[%o0 - 0x08], %g2
 	be		10f
-	 ldx		[%o0 - 0x08], %g2
-	sub		%o0, 0xf0, %o0
+	 sub		%o0, 0xf0, %o0
 1:	ldx		[%o1 + 0x00], %g3
 	ldx		[%o1 + 0x08], %g7
 	ldx		[%o1 + 0x10], %o4
@@ -1154,7 +1158,11 @@ ENTRY(aes_sparc64_ecb_decrypt_256)
 	 add		%o2, 0x20, %o2
 	brlz,pt		%o3, 11f
 	 nop
-10:	ldx		[%o1 + 0x00], %g3
+10:	ldd		[%o0 + 0x18], %f56
+	ldd		[%o0 + 0x10], %f58
+	ldd		[%o0 + 0x08], %f60
+	ldd		[%o0 + 0x00], %f62
+	ldx		[%o1 + 0x00], %g3
 	ldx		[%o1 + 0x08], %g7
 	xor		%g1, %g3, %g3
 	xor		%g2, %g7, %g7
@@ -1511,11 +1519,11 @@ ENTRY(aes_sparc64_ctr_crypt_256)
 	 add		%o2, 0x20, %o2
 	brlz,pt		%o3, 11f
 	 nop
-	ldd		[%o0 + 0xd0], %f56
+10:	ldd		[%o0 + 0xd0], %f56
 	ldd		[%o0 + 0xd8], %f58
 	ldd		[%o0 + 0xe0], %f60
 	ldd		[%o0 + 0xe8], %f62
-10:	xor		%g1, %g3, %o5
+	xor		%g1, %g3, %o5
 	MOVXTOD_O5_F0
 	xor		%g2, %g7, %o5
 	MOVXTOD_O5_F2
diff --git a/arch/sparc/crypto/aes_glue.c b/arch/sparc/crypto/aes_glue.c
index 8f1c9980f637..503e6d96ad4e 100644
--- a/arch/sparc/crypto/aes_glue.c
+++ b/arch/sparc/crypto/aes_glue.c
@@ -222,6 +222,7 @@ static int ecb_encrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	ctx->ops->load_encrypt_keys(&ctx->key[0]);
 	while ((nbytes = walk.nbytes)) {
@@ -251,6 +252,7 @@ static int ecb_decrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	ctx->ops->load_decrypt_keys(&ctx->key[0]);
 	key_end = &ctx->key[ctx->expanded_key_length / sizeof(u64)];
@@ -280,6 +282,7 @@ static int cbc_encrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	ctx->ops->load_encrypt_keys(&ctx->key[0]);
 	while ((nbytes = walk.nbytes)) {
@@ -309,6 +312,7 @@ static int cbc_decrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	ctx->ops->load_decrypt_keys(&ctx->key[0]);
 	key_end = &ctx->key[ctx->expanded_key_length / sizeof(u64)];
@@ -329,6 +333,22 @@ static int cbc_decrypt(struct blkcipher_desc *desc,
 	return err;
 }
 
+static void ctr_crypt_final(struct crypto_sparc64_aes_ctx *ctx,
+			    struct blkcipher_walk *walk)
+{
+	u8 *ctrblk = walk->iv;
+	u64 keystream[AES_BLOCK_SIZE / sizeof(u64)];
+	u8 *src = walk->src.virt.addr;
+	u8 *dst = walk->dst.virt.addr;
+	unsigned int nbytes = walk->nbytes;
+
+	ctx->ops->ecb_encrypt(&ctx->key[0], (const u64 *)ctrblk,
+			      keystream, AES_BLOCK_SIZE);
+	crypto_xor((u8 *) keystream, src, nbytes);
+	memcpy(dst, keystream, nbytes);
+	crypto_inc(ctrblk, AES_BLOCK_SIZE);
+}
+
 static int ctr_crypt(struct blkcipher_desc *desc,
 		     struct scatterlist *dst, struct scatterlist *src,
 		     unsigned int nbytes)
@@ -338,10 +358,11 @@ static int ctr_crypt(struct blkcipher_desc *desc,
 	int err;
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
-	err = blkcipher_walk_virt(desc, &walk);
+	err = blkcipher_walk_virt_block(desc, &walk, AES_BLOCK_SIZE);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	ctx->ops->load_encrypt_keys(&ctx->key[0]);
-	while ((nbytes = walk.nbytes)) {
+	while ((nbytes = walk.nbytes) >= AES_BLOCK_SIZE) {
 		unsigned int block_len = nbytes & AES_BLOCK_MASK;
 
 		if (likely(block_len)) {
@@ -353,6 +374,10 @@ static int ctr_crypt(struct blkcipher_desc *desc,
 		nbytes &= AES_BLOCK_SIZE - 1;
 		err = blkcipher_walk_done(desc, &walk, nbytes);
 	}
+	if (walk.nbytes) {
+		ctr_crypt_final(ctx, &walk);
+		err = blkcipher_walk_done(desc, &walk, 0);
+	}
 	fprs_write(0);
 	return err;
 }
@@ -418,7 +443,7 @@ static struct crypto_alg algs[] = { {
 	.cra_driver_name	= "ctr-aes-sparc64",
 	.cra_priority		= SPARC_CR_OPCODE_PRIORITY,
 	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
-	.cra_blocksize		= AES_BLOCK_SIZE,
+	.cra_blocksize		= 1,
 	.cra_ctxsize		= sizeof(struct crypto_sparc64_aes_ctx),
 	.cra_alignmask		= 7,
 	.cra_type		= &crypto_blkcipher_type,
@@ -475,3 +500,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("AES Secure Hash Algorithm, sparc64 aes opcode accelerated");
 
 MODULE_ALIAS("aes");
+
+#include "crop_devid.c"
diff --git a/arch/sparc/crypto/camellia_glue.c b/arch/sparc/crypto/camellia_glue.c
index 42905c084299..888f6260b4ec 100644
--- a/arch/sparc/crypto/camellia_glue.c
+++ b/arch/sparc/crypto/camellia_glue.c
@@ -98,6 +98,7 @@ static int __ecb_crypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	if (encrypt)
 		key = &ctx->encrypt_key[0];
@@ -160,6 +161,7 @@ static int cbc_encrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	key = &ctx->encrypt_key[0];
 	camellia_sparc64_load_keys(key, ctx->key_len);
@@ -198,6 +200,7 @@ static int cbc_decrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	key = &ctx->decrypt_key[0];
 	camellia_sparc64_load_keys(key, ctx->key_len);
@@ -320,3 +323,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("Camellia Cipher Algorithm, sparc64 camellia opcode accelerated");
 
 MODULE_ALIAS("aes");
+
+#include "crop_devid.c"
diff --git a/arch/sparc/crypto/crc32c_glue.c b/arch/sparc/crypto/crc32c_glue.c
index 0bd89cea8d8e..5162fad912ce 100644
--- a/arch/sparc/crypto/crc32c_glue.c
+++ b/arch/sparc/crypto/crc32c_glue.c
@@ -177,3 +177,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("CRC32c (Castagnoli), sparc64 crc32c opcode accelerated");
 
 MODULE_ALIAS("crc32c");
+
+#include "crop_devid.c"
diff --git a/arch/sparc/crypto/des_asm.S b/arch/sparc/crypto/des_asm.S
index 30b6e90b28b2..b5c8fc269b5f 100644
--- a/arch/sparc/crypto/des_asm.S
+++ b/arch/sparc/crypto/des_asm.S
@@ -376,6 +376,7 @@ ENTRY(des3_ede_sparc64_ecb_crypt)
 1:	ldd	[%o1 + 0x00], %f60
 	DES3_LOOP_BODY(60)
 	std	%f60, [%o2 + 0x00]
+	add	%o1, 0x08, %o1
 	subcc	%o3, 0x08, %o3
 	bne,pt	%icc, 1b
 	 add	%o2, 0x08, %o2
diff --git a/arch/sparc/crypto/des_glue.c b/arch/sparc/crypto/des_glue.c
index c4940c2d3073..3065bc61f9d3 100644
--- a/arch/sparc/crypto/des_glue.c
+++ b/arch/sparc/crypto/des_glue.c
@@ -100,6 +100,7 @@ static int __ecb_crypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	if (encrypt)
 		des_sparc64_load_keys(&ctx->encrypt_expkey[0]);
@@ -147,6 +148,7 @@ static int cbc_encrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	des_sparc64_load_keys(&ctx->encrypt_expkey[0]);
 	while ((nbytes = walk.nbytes)) {
@@ -177,6 +179,7 @@ static int cbc_decrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	des_sparc64_load_keys(&ctx->decrypt_expkey[0]);
 	while ((nbytes = walk.nbytes)) {
@@ -266,6 +269,7 @@ static int __ecb3_crypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	if (encrypt)
 		K = &ctx->encrypt_expkey[0];
@@ -317,6 +321,7 @@ static int cbc3_encrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	K = &ctx->encrypt_expkey[0];
 	des3_ede_sparc64_load_keys(K);
@@ -352,6 +357,7 @@ static int cbc3_decrypt(struct blkcipher_desc *desc,
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
 
 	K = &ctx->decrypt_expkey[0];
 	des3_ede_sparc64_load_keys(K);
@@ -527,3 +533,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("DES & Triple DES EDE Cipher Algorithms, sparc64 des opcode accelerated");
 
 MODULE_ALIAS("des");
+
+#include "crop_devid.c"
diff --git a/arch/sparc/crypto/md5_glue.c b/arch/sparc/crypto/md5_glue.c
index 603d723038ce..09a9ea1dfb69 100644
--- a/arch/sparc/crypto/md5_glue.c
+++ b/arch/sparc/crypto/md5_glue.c
@@ -186,3 +186,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("MD5 Secure Hash Algorithm, sparc64 md5 opcode accelerated");
 
 MODULE_ALIAS("md5");
+
+#include "crop_devid.c"
diff --git a/arch/sparc/crypto/sha1_glue.c b/arch/sparc/crypto/sha1_glue.c
index 2bbb20bee9f1..6cd5f29e1e0d 100644
--- a/arch/sparc/crypto/sha1_glue.c
+++ b/arch/sparc/crypto/sha1_glue.c
@@ -181,3 +181,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("SHA1 Secure Hash Algorithm, sparc64 sha1 opcode accelerated");
 
 MODULE_ALIAS("sha1");
+
+#include "crop_devid.c"
diff --git a/arch/sparc/crypto/sha256_glue.c b/arch/sparc/crypto/sha256_glue.c
index 591e656bd891..04f555ab2680 100644
--- a/arch/sparc/crypto/sha256_glue.c
+++ b/arch/sparc/crypto/sha256_glue.c
@@ -239,3 +239,5 @@ MODULE_DESCRIPTION("SHA-224 and SHA-256 Secure Hash Algorithm, sparc64 sha256 op
 
 MODULE_ALIAS("sha224");
 MODULE_ALIAS("sha256");
+
+#include "crop_devid.c"
diff --git a/arch/sparc/crypto/sha512_glue.c b/arch/sparc/crypto/sha512_glue.c
index 486f0a2b7001..f04d1994d19a 100644
--- a/arch/sparc/crypto/sha512_glue.c
+++ b/arch/sparc/crypto/sha512_glue.c
@@ -224,3 +224,5 @@ MODULE_DESCRIPTION("SHA-384 and SHA-512 Secure Hash Algorithm, sparc64 sha512 op
 
 MODULE_ALIAS("sha384");
 MODULE_ALIAS("sha512");
+
+#include "crop_devid.c"
diff --git a/arch/sparc/include/asm/Kbuild b/arch/sparc/include/asm/Kbuild
index 645a58da0e86..e26d430ce2fd 100644
--- a/arch/sparc/include/asm/Kbuild
+++ b/arch/sparc/include/asm/Kbuild
@@ -8,4 +8,5 @@ generic-y += local64.h
 generic-y += irq_regs.h
 generic-y += local.h
 generic-y += module.h
+generic-y += trace_clock.h
 generic-y += word-at-a-time.h
diff --git a/arch/sparc/include/asm/atomic_64.h b/arch/sparc/include/asm/atomic_64.h
index ce35a1cf1a20..be56a244c9cf 100644
--- a/arch/sparc/include/asm/atomic_64.h
+++ b/arch/sparc/include/asm/atomic_64.h
@@ -1,7 +1,7 @@
 /* atomic.h: Thankfully the V9 is at least reasonable for this
  *           stuff.
  *
- * Copyright (C) 1996, 1997, 2000 David S. Miller (davem@redhat.com)
+ * Copyright (C) 1996, 1997, 2000, 2012 David S. Miller (davem@redhat.com)
  */
 
 #ifndef __ARCH_SPARC64_ATOMIC__
@@ -106,6 +106,8 @@ static inline long atomic64_add_unless(atomic64_t *v, long a, long u)
 
 #define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1, 0)
 
+extern long atomic64_dec_if_positive(atomic64_t *v);
+
 /* Atomic operations are already serializing */
 #define smp_mb__before_atomic_dec()	barrier()
 #define smp_mb__after_atomic_dec()	barrier()
diff --git a/arch/sparc/include/asm/backoff.h b/arch/sparc/include/asm/backoff.h
index db3af0d30fb1..4e02086b839c 100644
--- a/arch/sparc/include/asm/backoff.h
+++ b/arch/sparc/include/asm/backoff.h
@@ -1,6 +1,46 @@
 #ifndef _SPARC64_BACKOFF_H
 #define _SPARC64_BACKOFF_H
 
+/* The macros in this file implement an exponential backoff facility
+ * for atomic operations.
+ *
+ * When multiple threads compete on an atomic operation, it is
+ * possible for one thread to be continually denied a successful
+ * completion of the compare-and-swap instruction.  Heavily
+ * threaded cpu implementations like Niagara can compound this
+ * problem even further.
+ *
+ * When an atomic operation fails and needs to be retried, we spin a
+ * certain number of times.  At each subsequent failure of the same
+ * operation we double the spin count, realizing an exponential
+ * backoff.
+ *
+ * When we spin, we try to use an operation that will cause the
+ * current cpu strand to block, and therefore make the core fully
+ * available to any other other runnable strands.  There are two
+ * options, based upon cpu capabilities.
+ *
+ * On all cpus prior to SPARC-T4 we do three dummy reads of the
+ * condition code register.  Each read blocks the strand for something
+ * between 40 and 50 cpu cycles.
+ *
+ * For SPARC-T4 and later we have a special "pause" instruction
+ * available.  This is implemented using writes to register %asr27.
+ * The cpu will block the number of cycles written into the register,
+ * unless a disrupting trap happens first.  SPARC-T4 specifically
+ * implements pause with a granularity of 8 cycles.  Each strand has
+ * an internal pause counter which decrements every 8 cycles.  So the
+ * chip shifts the %asr27 value down by 3 bits, and writes the result
+ * into the pause counter.  If a value smaller than 8 is written, the
+ * chip blocks for 1 cycle.
+ *
+ * To achieve the same amount of backoff as the three %ccr reads give
+ * on earlier chips, we shift the backoff value up by 7 bits.  (Three
+ * %ccr reads block for about 128 cycles, 1 << 7 == 128) We write the
+ * whole amount we want to block into the pause register, rather than
+ * loop writing 128 each time.
+ */
+
 #define BACKOFF_LIMIT	(4 * 1024)
 
 #ifdef CONFIG_SMP
@@ -11,16 +51,25 @@
 #define BACKOFF_LABEL(spin_label, continue_label) \
 	spin_label
 
-#define BACKOFF_SPIN(reg, tmp, label)	\
-	mov	reg, tmp; \
-88:	brnz,pt	tmp, 88b; \
-	 sub	tmp, 1, tmp; \
-	set	BACKOFF_LIMIT, tmp; \
-	cmp	reg, tmp; \
-	bg,pn	%xcc, label; \
-	 nop; \
-	ba,pt	%xcc, label; \
-	 sllx	reg, 1, reg;
+#define BACKOFF_SPIN(reg, tmp, label)		\
+	mov		reg, tmp;		\
+88:	rd		%ccr, %g0;		\
+	rd		%ccr, %g0;		\
+	rd		%ccr, %g0;		\
+	.section	.pause_3insn_patch,"ax";\
+	.word		88b;			\
+	sllx		tmp, 7, tmp;		\
+	wr		tmp, 0, %asr27;		\
+	clr		tmp;			\
+	.previous;				\
+	brnz,pt		tmp, 88b;		\
+	 sub		tmp, 1, tmp;		\
+	set		BACKOFF_LIMIT, tmp;	\
+	cmp		reg, tmp;		\
+	bg,pn		%xcc, label;		\
+	 nop;					\
+	ba,pt		%xcc, label;		\
+	 sllx		reg, 1, reg;
 
 #else
 
diff --git a/arch/sparc/include/asm/compat.h b/arch/sparc/include/asm/compat.h
index cef99fbc0a21..830502fe62b4 100644
--- a/arch/sparc/include/asm/compat.h
+++ b/arch/sparc/include/asm/compat.h
@@ -232,9 +232,10 @@ static inline void __user *arch_compat_alloc_user_space(long len)
 	struct pt_regs *regs = current_thread_info()->kregs;
 	unsigned long usp = regs->u_regs[UREG_I6];
 
-	if (!(test_thread_flag(TIF_32BIT)))
+	if (test_thread_64bit_stack(usp))
 		usp += STACK_BIAS;
-	else
+
+	if (test_thread_flag(TIF_32BIT))
 		usp &= 0xffffffffUL;
 
 	usp -= len;
diff --git a/arch/sparc/include/asm/compat_signal.h b/arch/sparc/include/asm/compat_signal.h
index b759eab9b51c..9ed1f128b4d1 100644
--- a/arch/sparc/include/asm/compat_signal.h
+++ b/arch/sparc/include/asm/compat_signal.h
@@ -18,12 +18,6 @@ struct __old_sigaction32 {
 	unsigned int    	sa_flags;
 	unsigned		sa_restorer;     /* not used by Linux/SPARC yet */
 };
-
-typedef struct sigaltstack32 {
-	u32			ss_sp;
-	int			ss_flags;
-	compat_size_t		ss_size;
-} stack_t32;
 #endif
 
 #endif /* !(_COMPAT_SIGNAL_H) */
diff --git a/arch/sparc/include/asm/dma-mapping.h b/arch/sparc/include/asm/dma-mapping.h
index 8493fd3c7ba5..05fe53f5346e 100644
--- a/arch/sparc/include/asm/dma-mapping.h
+++ b/arch/sparc/include/asm/dma-mapping.h
@@ -59,6 +59,7 @@ static inline void dma_free_attrs(struct device *dev, size_t size,
 
 static inline int dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
 {
+	debug_dma_mapping_error(dev, dma_addr);
 	return (dma_addr == DMA_ERROR_CODE);
 }
 
diff --git a/arch/sparc/include/asm/elf_32.h b/arch/sparc/include/asm/elf_32.h
index ac74a2c98e6d..a24e41fcdde1 100644
--- a/arch/sparc/include/asm/elf_32.h
+++ b/arch/sparc/include/asm/elf_32.h
@@ -128,7 +128,4 @@ typedef struct {
 
 #define ELF_PLATFORM	(NULL)
 
-#define SET_PERSONALITY(ex) \
-	set_personality(PER_LINUX | (current->personality & (~PER_MASK)))
-
 #endif /* !(__ASMSPARC_ELF_H) */
diff --git a/arch/sparc/include/asm/hugetlb.h b/arch/sparc/include/asm/hugetlb.h
index 8c5eed6d267f..7eb57d245044 100644
--- a/arch/sparc/include/asm/hugetlb.h
+++ b/arch/sparc/include/asm/hugetlb.h
@@ -12,7 +12,6 @@ pte_t huge_ptep_get_and_clear(struct mm_struct *mm, unsigned long addr,
 
 static inline void hugetlb_prefault_arch_hook(struct mm_struct *mm)
 {
-	hugetlb_setup(mm);
 }
 
 static inline int is_hugepage_only_range(struct mm_struct *mm,
@@ -61,14 +60,20 @@ static inline pte_t huge_pte_wrprotect(pte_t pte)
 static inline void huge_ptep_set_wrprotect(struct mm_struct *mm,
 					   unsigned long addr, pte_t *ptep)
 {
-	ptep_set_wrprotect(mm, addr, ptep);
+	pte_t old_pte = *ptep;
+	set_huge_pte_at(mm, addr, ptep, pte_wrprotect(old_pte));
 }
 
 static inline int huge_ptep_set_access_flags(struct vm_area_struct *vma,
 					     unsigned long addr, pte_t *ptep,
 					     pte_t pte, int dirty)
 {
-	return ptep_set_access_flags(vma, addr, ptep, pte, dirty);
+	int changed = !pte_same(*ptep, pte);
+	if (changed) {
+		set_huge_pte_at(vma->vm_mm, addr, ptep, pte);
+		flush_tlb_page(vma, addr);
+	}
+	return changed;
 }
 
 static inline pte_t huge_ptep_get(pte_t *ptep)
diff --git a/arch/sparc/include/asm/page_64.h b/arch/sparc/include/asm/page_64.h
index 4b39f74d6ca0..e15538899f3d 100644
--- a/arch/sparc/include/asm/page_64.h
+++ b/arch/sparc/include/asm/page_64.h
@@ -27,8 +27,8 @@
 #ifndef __ASSEMBLY__
 
 #if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
-struct mm_struct;
-extern void hugetlb_setup(struct mm_struct *mm);
+struct pt_regs;
+extern void hugetlb_setup(struct pt_regs *regs);
 #endif
 
 #define WANT_PAGE_VIRTUAL
diff --git a/arch/sparc/include/asm/parport.h b/arch/sparc/include/asm/parport.h
index cb33608cc68f..c55291e5b83e 100644
--- a/arch/sparc/include/asm/parport.h
+++ b/arch/sparc/include/asm/parport.h
@@ -103,7 +103,7 @@ static inline unsigned int get_dma_residue(unsigned int dmanr)
 	return ebus_dma_residue(&sparc_ebus_dmas[dmanr].info);
 }
 
-static int __devinit ecpp_probe(struct platform_device *op)
+static int ecpp_probe(struct platform_device *op)
 {
 	unsigned long base = op->resource[0].start;
 	unsigned long config = op->resource[1].start;
@@ -192,7 +192,7 @@ out_err:
 	return err;
 }
 
-static int __devexit ecpp_remove(struct platform_device *op)
+static int ecpp_remove(struct platform_device *op)
 {
 	struct parport *p = dev_get_drvdata(&op->dev);
 	int slot = p->dma;
@@ -242,7 +242,7 @@ static struct platform_driver ecpp_driver = {
 		.of_match_table = ecpp_match,
 	},
 	.probe			= ecpp_probe,
-	.remove			= __devexit_p(ecpp_remove),
+	.remove			= ecpp_remove,
 };
 
 static int parport_pc_find_nonpci_ports(int autoirq, int autodma)
diff --git a/arch/sparc/include/asm/pgtable_64.h b/arch/sparc/include/asm/pgtable_64.h
index 95515f1e7cef..08fcce90316b 100644
--- a/arch/sparc/include/asm/pgtable_64.h
+++ b/arch/sparc/include/asm/pgtable_64.h
@@ -71,7 +71,6 @@
 #define PMD_PADDR	_AC(0xfffffffe,UL)
 #define PMD_PADDR_SHIFT	_AC(11,UL)
 
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 #define PMD_ISHUGE	_AC(0x00000001,UL)
 
 /* This is the PMD layout when PMD_ISHUGE is set.  With 4MB huge
@@ -86,7 +85,6 @@
 #define PMD_HUGE_ACCESSED	_AC(0x00000080,UL)
 #define PMD_HUGE_EXEC		_AC(0x00000040,UL)
 #define PMD_HUGE_SPLITTING	_AC(0x00000020,UL)
-#endif
 
 /* PGDs point to PMD tables which are 8K aligned.  */
 #define PGD_PADDR	_AC(0xfffffffc,UL)
@@ -617,11 +615,23 @@ static inline unsigned long pte_present(pte_t pte)
 	return val;
 }
 
+#define pte_accessible pte_accessible
+static inline unsigned long pte_accessible(pte_t a)
+{
+	return pte_val(a) & _PAGE_VALID;
+}
+
 static inline unsigned long pte_special(pte_t pte)
 {
 	return pte_val(pte) & _PAGE_SPECIAL;
 }
 
+static inline int pmd_large(pmd_t pmd)
+{
+	return (pmd_val(pmd) & (PMD_ISHUGE | PMD_HUGE_PRESENT)) ==
+		(PMD_ISHUGE | PMD_HUGE_PRESENT);
+}
+
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
 static inline int pmd_young(pmd_t pmd)
 {
@@ -640,12 +650,6 @@ static inline unsigned long pmd_pfn(pmd_t pmd)
 	return val >> (PAGE_SHIFT - PMD_PADDR_SHIFT);
 }
 
-static inline int pmd_large(pmd_t pmd)
-{
-	return (pmd_val(pmd) & (PMD_ISHUGE | PMD_HUGE_PRESENT)) ==
-		(PMD_ISHUGE | PMD_HUGE_PRESENT);
-}
-
 static inline int pmd_trans_splitting(pmd_t pmd)
 {
 	return (pmd_val(pmd) & (PMD_ISHUGE|PMD_HUGE_SPLITTING)) ==
@@ -802,7 +806,7 @@ static inline void __set_pte_at(struct mm_struct *mm, unsigned long addr,
 	 * SUN4V NOTE: _PAGE_VALID is the same value in both the SUN4U
 	 *             and SUN4V pte layout, so this inline test is fine.
 	 */
-	if (likely(mm != &init_mm) && (pte_val(orig) & _PAGE_VALID))
+	if (likely(mm != &init_mm) && pte_accessible(orig))
 		tlb_batch_add(mm, addr, ptep, orig, fullmm);
 }
 
diff --git a/arch/sparc/include/asm/processor_32.h b/arch/sparc/include/asm/processor_32.h
index f74ac9ee33a8..2c7baa4c4505 100644
--- a/arch/sparc/include/asm/processor_32.h
+++ b/arch/sparc/include/asm/processor_32.h
@@ -106,7 +106,6 @@ static inline void start_thread(struct pt_regs * regs, unsigned long pc,
 
 /* Free all resources held by a thread. */
 #define release_thread(tsk)		do { } while(0)
-extern pid_t kernel_thread(int (*fn)(void *), void * arg, unsigned long flags);
 
 extern unsigned long get_wchan(struct task_struct *);
 
@@ -119,6 +118,7 @@ extern unsigned long get_wchan(struct task_struct *);
 extern struct task_struct *last_task_used_math;
 
 #define cpu_relax()	barrier()
+extern void (*sparc_idle)(void);
 
 #endif
 
diff --git a/arch/sparc/include/asm/processor_64.h b/arch/sparc/include/asm/processor_64.h
index 4e5a483122a0..cce72ce4c334 100644
--- a/arch/sparc/include/asm/processor_64.h
+++ b/arch/sparc/include/asm/processor_64.h
@@ -94,6 +94,7 @@ struct thread_struct {
 #ifndef __ASSEMBLY__
 
 #include <linux/types.h>
+#include <asm/fpumacro.h>
 
 /* Return saved PC of a blocked thread. */
 struct task_struct;
@@ -143,6 +144,10 @@ do { \
 	: \
 	: "r" (regs), "r" (sp - sizeof(struct reg_window) - STACK_BIAS), \
 	  "i" ((const unsigned long)(&((struct pt_regs *)0)->u_regs[0]))); \
+	fprs_write(0);	\
+	current_thread_info()->xfsr[0] = 0;	\
+	current_thread_info()->fpsaved[0] = 0;	\
+	regs->tstate &= ~TSTATE_PEF;	\
 } while (0)
 
 #define start_thread32(regs, pc, sp) \
@@ -183,20 +188,37 @@ do { \
 	: \
 	: "r" (regs), "r" (sp - sizeof(struct reg_window32)), \
 	  "i" ((const unsigned long)(&((struct pt_regs *)0)->u_regs[0]))); \
+	fprs_write(0);	\
+	current_thread_info()->xfsr[0] = 0;	\
+	current_thread_info()->fpsaved[0] = 0;	\
+	regs->tstate &= ~TSTATE_PEF;	\
 } while (0)
 
 /* Free all resources held by a thread. */
 #define release_thread(tsk)		do { } while (0)
 
-extern pid_t kernel_thread(int (*fn)(void *), void * arg, unsigned long flags);
-
 extern unsigned long get_wchan(struct task_struct *task);
 
 #define task_pt_regs(tsk) (task_thread_info(tsk)->kregs)
 #define KSTK_EIP(tsk)  (task_pt_regs(tsk)->tpc)
 #define KSTK_ESP(tsk)  (task_pt_regs(tsk)->u_regs[UREG_FP])
 
-#define cpu_relax()	barrier()
+/* Please see the commentary in asm/backoff.h for a description of
+ * what these instructions are doing and how they have been choosen.
+ * To make a long story short, we are trying to yield the current cpu
+ * strand during busy loops.
+ */
+#define cpu_relax()	asm volatile("\n99:\n\t"			\
+				     "rd	%%ccr, %%g0\n\t"	\
+				     "rd	%%ccr, %%g0\n\t"	\
+				     "rd	%%ccr, %%g0\n\t"	\
+				     ".section	.pause_3insn_patch,\"ax\"\n\t"\
+				     ".word	99b\n\t"		\
+				     "wr	%%g0, 128, %%asr27\n\t"	\
+				     "nop\n\t"				\
+				     "nop\n\t"				\
+				     ".previous"			\
+				     ::: "memory")
 
 /* Prefetch support.  This is tuned for UltraSPARC-III and later.
  * UltraSPARC-I will treat these as nops, and UltraSPARC-II has
diff --git a/arch/sparc/include/asm/prom.h b/arch/sparc/include/asm/prom.h
index c28765110706..67c62578d170 100644
--- a/arch/sparc/include/asm/prom.h
+++ b/arch/sparc/include/asm/prom.h
@@ -63,5 +63,13 @@ extern char *of_console_options;
 extern void irq_trans_init(struct device_node *dp);
 extern char *build_path_component(struct device_node *dp);
 
+/* SPARC has local implementations */
+extern int of_address_to_resource(struct device_node *dev, int index,
+				  struct resource *r);
+#define of_address_to_resource of_address_to_resource
+
+void __iomem *of_iomap(struct device_node *node, int index);
+#define of_iomap of_iomap
+
 #endif /* __KERNEL__ */
 #endif /* _SPARC_PROM_H */
diff --git a/arch/sparc/include/asm/ptrace.h b/arch/sparc/include/asm/ptrace.h
index da43bdc62294..bdfafd7af46f 100644
--- a/arch/sparc/include/asm/ptrace.h
+++ b/arch/sparc/include/asm/ptrace.h
@@ -32,6 +32,9 @@ static inline bool pt_regs_clear_syscall(struct pt_regs *regs)
 #define arch_ptrace_stop(exit_code, info) \
 	synchronize_user_stack()
 
+#define current_pt_regs() \
+	((struct pt_regs *)((unsigned long)current_thread_info() + THREAD_SIZE) - 1)
+
 struct global_reg_snapshot {
 	unsigned long		tstate;
 	unsigned long		tpc;
@@ -55,9 +58,7 @@ union global_cpu_snapshot {
 
 extern union global_cpu_snapshot global_cpu_snapshot[NR_CPUS];
 
-#define force_successful_syscall_return()	    \
-do {	current_thread_info()->syscall_noerror = 1; \
-} while (0)
+#define force_successful_syscall_return() set_thread_noerror(1)
 #define user_mode(regs) (!((regs)->tstate & TSTATE_PRIV))
 #define instruction_pointer(regs) ((regs)->tpc)
 #define instruction_pointer_set(regs, val) ((regs)->tpc = (val))
@@ -100,6 +101,9 @@ static inline bool pt_regs_clear_syscall(struct pt_regs *regs)
 #define arch_ptrace_stop(exit_code, info) \
 	synchronize_user_stack()
 
+#define current_pt_regs() \
+	((struct pt_regs *)((unsigned long)current_thread_info() + THREAD_SIZE) - 1)
+
 #define user_mode(regs) (!((regs)->psr & PSR_PS))
 #define instruction_pointer(regs) ((regs)->pc)
 #define user_stack_pointer(regs) ((regs)->u_regs[UREG_FP])
diff --git a/arch/sparc/include/asm/signal.h b/arch/sparc/include/asm/signal.h
index d243c2ae02d2..c33ce3f2ba84 100644
--- a/arch/sparc/include/asm/signal.h
+++ b/arch/sparc/include/asm/signal.h
@@ -21,12 +21,8 @@
  */
 #define SA_STATIC_ALLOC         0x8000
 
-struct k_sigaction {
-	struct			__new_sigaction sa;
-	void			__user *ka_restorer;
-};
-
-#define ptrace_signal_deliver(regs, cookie) do { } while (0)
+#define __ARCH_HAS_KA_RESTORER
+#define __ARCH_HAS_SA_RESTORER
 
 #endif /* !(__ASSEMBLY__) */
 #endif /* !(__SPARC_SIGNAL_H) */
diff --git a/arch/sparc/include/asm/switch_to_64.h b/arch/sparc/include/asm/switch_to_64.h
index 7923c4a2be38..cad36f56fa03 100644
--- a/arch/sparc/include/asm/switch_to_64.h
+++ b/arch/sparc/include/asm/switch_to_64.h
@@ -23,7 +23,7 @@ do {	flush_tlb_pending();						\
 	/* If you are tempted to conditionalize the following */	\
 	/* so that ASI is only written if it changes, think again. */	\
 	__asm__ __volatile__("wr %%g0, %0, %%asi"			\
-	: : "r" (__thread_flag_byte_ptr(task_thread_info(next))[TI_FLAG_BYTE_CURRENT_DS]));\
+	: : "r" (task_thread_info(next)->current_ds));\
 	trap_block[current_thread_info()->cpu].thread =			\
 		task_thread_info(next);					\
 	__asm__ __volatile__(						\
diff --git a/arch/sparc/include/asm/syscalls.h b/arch/sparc/include/asm/syscalls.h
index 45a43f637a14..bf8972adea17 100644
--- a/arch/sparc/include/asm/syscalls.h
+++ b/arch/sparc/include/asm/syscalls.h
@@ -8,6 +8,4 @@ extern asmlinkage long sparc_do_fork(unsigned long clone_flags,
 				     struct pt_regs *regs,
 				     unsigned long stack_size);
 
-extern asmlinkage int sparc_execve(struct pt_regs *regs);
-
 #endif /* _SPARC64_SYSCALLS_H */
diff --git a/arch/sparc/include/asm/thread_info_64.h b/arch/sparc/include/asm/thread_info_64.h
index 4e2276631081..269bd92313df 100644
--- a/arch/sparc/include/asm/thread_info_64.h
+++ b/arch/sparc/include/asm/thread_info_64.h
@@ -14,12 +14,12 @@
 #define TI_FLAG_FAULT_CODE_SHIFT	56
 #define TI_FLAG_BYTE_WSTATE		1
 #define TI_FLAG_WSTATE_SHIFT		48
-#define TI_FLAG_BYTE_CWP		2
-#define TI_FLAG_CWP_SHIFT		40
-#define TI_FLAG_BYTE_CURRENT_DS		3
-#define TI_FLAG_CURRENT_DS_SHIFT	32
-#define TI_FLAG_BYTE_FPDEPTH		4
-#define TI_FLAG_FPDEPTH_SHIFT		24
+#define TI_FLAG_BYTE_NOERROR		2
+#define TI_FLAG_BYTE_NOERROR_SHIFT	40
+#define TI_FLAG_BYTE_FPDEPTH		3
+#define TI_FLAG_FPDEPTH_SHIFT		32
+#define TI_FLAG_BYTE_CWP		4
+#define TI_FLAG_CWP_SHIFT		24
 #define TI_FLAG_BYTE_WSAVED		5
 #define TI_FLAG_WSAVED_SHIFT		16
 
@@ -47,7 +47,7 @@ struct thread_info {
 	struct exec_domain	*exec_domain;
 	int			preempt_count;	/* 0 => preemptable, <0 => BUG */
 	__u8			new_child;
-	__u8			syscall_noerror;
+	__u8			current_ds;
 	__u16			cpu;
 
 	unsigned long		*utraps;
@@ -74,9 +74,9 @@ struct thread_info {
 #define TI_FAULT_CODE	(TI_FLAGS + TI_FLAG_BYTE_FAULT_CODE)
 #define TI_WSTATE	(TI_FLAGS + TI_FLAG_BYTE_WSTATE)
 #define TI_CWP		(TI_FLAGS + TI_FLAG_BYTE_CWP)
-#define TI_CURRENT_DS	(TI_FLAGS + TI_FLAG_BYTE_CURRENT_DS)
 #define TI_FPDEPTH	(TI_FLAGS + TI_FLAG_BYTE_FPDEPTH)
 #define TI_WSAVED	(TI_FLAGS + TI_FLAG_BYTE_WSAVED)
+#define TI_SYS_NOERROR	(TI_FLAGS + TI_FLAG_BYTE_NOERROR)
 #define TI_FPSAVED	0x00000010
 #define TI_KSP		0x00000018
 #define TI_FAULT_ADDR	0x00000020
@@ -84,7 +84,7 @@ struct thread_info {
 #define TI_EXEC_DOMAIN	0x00000030
 #define TI_PRE_COUNT	0x00000038
 #define TI_NEW_CHILD	0x0000003c
-#define TI_SYS_NOERROR	0x0000003d
+#define TI_CURRENT_DS	0x0000003d
 #define TI_CPU		0x0000003e
 #define TI_UTRAPS	0x00000040
 #define TI_REG_WINDOW	0x00000048
@@ -121,7 +121,7 @@ struct thread_info {
 #define INIT_THREAD_INFO(tsk)				\
 {							\
 	.task		=	&tsk,			\
-	.flags		= ((unsigned long)ASI_P) << TI_FLAG_CURRENT_DS_SHIFT,	\
+	.current_ds	=	ASI_P,			\
 	.exec_domain	=	&default_exec_domain,	\
 	.preempt_count	=	INIT_PREEMPT_COUNT,	\
 	.restart_block	= {				\
@@ -153,13 +153,12 @@ register struct thread_info *current_thread_info_reg asm("g6");
 #define set_thread_wstate(val)		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_WSTATE] = (val))
 #define get_thread_cwp()		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_CWP])
 #define set_thread_cwp(val)		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_CWP] = (val))
-#define get_thread_current_ds()		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_CURRENT_DS])
-#define set_thread_current_ds(val)	(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_CURRENT_DS] = (val))
+#define get_thread_noerror()		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_NOERROR])
+#define set_thread_noerror(val)		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_NOERROR] = (val))
 #define get_thread_fpdepth()		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_FPDEPTH])
 #define set_thread_fpdepth(val)		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_FPDEPTH] = (val))
 #define get_thread_wsaved()		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_WSAVED])
 #define set_thread_wsaved(val)		(__cur_thread_flag_byte_ptr[TI_FLAG_BYTE_WSAVED] = (val))
-
 #endif /* !(__ASSEMBLY__) */
 
 /*
@@ -259,6 +258,11 @@ static inline bool test_and_clear_restore_sigmask(void)
 
 #define tsk_is_polling(t) test_tsk_thread_flag(t, TIF_POLLING_NRFLAG)
 
+#define thread32_stack_is_64bit(__SP) (((__SP) & 0x1) != 0)
+#define test_thread_64bit_stack(__SP) \
+	((test_thread_flag(TIF_32BIT) && !thread32_stack_is_64bit(__SP)) ? \
+	 false : true)
+
 #endif	/* !__ASSEMBLY__ */
 
 #endif /* __KERNEL__ */
diff --git a/arch/sparc/include/asm/timer_64.h b/arch/sparc/include/asm/timer_64.h
index ef3c3682debf..01197d8215c4 100644
--- a/arch/sparc/include/asm/timer_64.h
+++ b/arch/sparc/include/asm/timer_64.h
@@ -24,7 +24,7 @@ struct sparc64_tick_ops {
 extern struct sparc64_tick_ops *tick_ops;
 
 extern unsigned long sparc64_get_clock_tick(unsigned int cpu);
-extern void __devinit setup_sparc64_timer(void);
+extern void setup_sparc64_timer(void);
 extern void __init time_init(void);
 
 #endif /* _SPARC64_TIMER_H */
diff --git a/arch/sparc/include/asm/tsb.h b/arch/sparc/include/asm/tsb.h
index b4c258de4443..e696432b950d 100644
--- a/arch/sparc/include/asm/tsb.h
+++ b/arch/sparc/include/asm/tsb.h
@@ -157,17 +157,26 @@ extern struct tsb_phys_patch_entry __tsb_phys_patch, __tsb_phys_patch_end;
 	andn		REG2, 0x7, REG2; \
 	add		REG1, REG2, REG1;
 
-	/* This macro exists only to make the PMD translator below easier
-	 * to read.  It hides the ELF section switch for the sun4v code
-	 * patching.
+	/* These macros exists only to make the PMD translator below
+	 * easier to read.  It hides the ELF section switch for the
+	 * sun4v code patching.
 	 */
-#define OR_PTE_BIT(REG, NAME)				\
+#define OR_PTE_BIT_1INSN(REG, NAME)			\
 661:	or		REG, _PAGE_##NAME##_4U, REG;	\
 	.section	.sun4v_1insn_patch, "ax";	\
 	.word		661b;				\
 	or		REG, _PAGE_##NAME##_4V, REG;	\
 	.previous;
 
+#define OR_PTE_BIT_2INSN(REG, TMP, NAME)		\
+661:	sethi		%hi(_PAGE_##NAME##_4U), TMP;	\
+	or		REG, TMP, REG;			\
+	.section	.sun4v_2insn_patch, "ax";	\
+	.word		661b;				\
+	mov		-1, TMP;			\
+	or		REG, _PAGE_##NAME##_4V, REG;	\
+	.previous;
+
 	/* Load into REG the PTE value for VALID, CACHE, and SZHUGE.  */
 #define BUILD_PTE_VALID_SZHUGE_CACHE(REG)				   \
 661:	sethi		%uhi(_PAGE_VALID|_PAGE_SZHUGE_4U), REG;		   \
@@ -214,12 +223,13 @@ extern struct tsb_phys_patch_entry __tsb_phys_patch, __tsb_phys_patch_end;
 	 andn		REG1, PMD_HUGE_PROTBITS, REG2;			      \
 	sllx		REG2, PMD_PADDR_SHIFT, REG2;			      \
 	/* REG2 now holds PFN << PAGE_SHIFT */				      \
-	andcc		REG1, PMD_HUGE_EXEC, %g0;			      \
-	bne,a,pt	%xcc, 1f;					      \
-	 OR_PTE_BIT(REG2, EXEC);					      \
-1:	andcc		REG1, PMD_HUGE_WRITE, %g0;			      \
+	andcc		REG1, PMD_HUGE_WRITE, %g0;			      \
 	bne,a,pt	%xcc, 1f;					      \
-	 OR_PTE_BIT(REG2, W);						      \
+	 OR_PTE_BIT_1INSN(REG2, W);					      \
+1:	andcc		REG1, PMD_HUGE_EXEC, %g0;			      \
+	be,pt		%xcc, 1f;					      \
+	 nop;								      \
+	OR_PTE_BIT_2INSN(REG2, REG1, EXEC);				      \
 	/* REG1 can now be clobbered, build final PTE */		      \
 1:	BUILD_PTE_VALID_SZHUGE_CACHE(REG1);				      \
 	ba,pt		%xcc, PTE_LABEL;				      \
diff --git a/arch/sparc/include/asm/ttable.h b/arch/sparc/include/asm/ttable.h
index 48f2807d3265..71b5a67522ab 100644
--- a/arch/sparc/include/asm/ttable.h
+++ b/arch/sparc/include/asm/ttable.h
@@ -372,7 +372,9 @@ etrap_spill_fixup_64bit:				\
 
 /* Normal 32bit spill */
 #define SPILL_2_GENERIC(ASI)				\
-	srl	%sp, 0, %sp;				\
+	and	%sp, 1, %g3;				\
+	brnz,pn	%g3, (. - (128 + 4));			\
+	 srl	%sp, 0, %sp;				\
 	stwa	%l0, [%sp + %g0] ASI;			\
 	mov	0x04, %g3;				\
 	stwa	%l1, [%sp + %g3] ASI;			\
@@ -398,14 +400,16 @@ etrap_spill_fixup_64bit:				\
 	stwa	%i6, [%g1 + %g0] ASI;			\
 	stwa	%i7, [%g1 + %g3] ASI;			\
 	saved;						\
-        retry; nop; nop;				\
+        retry;						\
 	b,a,pt	%xcc, spill_fixup_dax;			\
 	b,a,pt	%xcc, spill_fixup_mna;			\
 	b,a,pt	%xcc, spill_fixup;
 
 #define SPILL_2_GENERIC_ETRAP		\
 etrap_user_spill_32bit:			\
-	srl	%sp, 0, %sp;		\
+	and	%sp, 1, %g3;		\
+	brnz,pn	%g3, etrap_user_spill_64bit;	\
+	 srl	%sp, 0, %sp;		\
 	stwa	%l0, [%sp + 0x00] %asi;	\
 	stwa	%l1, [%sp + 0x04] %asi;	\
 	stwa	%l2, [%sp + 0x08] %asi;	\
@@ -427,7 +431,7 @@ etrap_user_spill_32bit:			\
 	ba,pt	%xcc, etrap_save;	\
 	 wrpr	%g1, %cwp;		\
 	nop; nop; nop; nop;		\
-	nop; nop; nop; nop;		\
+	nop; nop;			\
 	ba,a,pt	%xcc, etrap_spill_fixup_32bit; \
 	ba,a,pt	%xcc, etrap_spill_fixup_32bit; \
 	ba,a,pt	%xcc, etrap_spill_fixup_32bit;
@@ -592,7 +596,9 @@ user_rtt_fill_64bit:					\
 
 /* Normal 32bit fill */
 #define FILL_2_GENERIC(ASI)				\
-	srl	%sp, 0, %sp;				\
+	and	%sp, 1, %g3;				\
+	brnz,pn	%g3, (. - (128 + 4));			\
+	 srl	%sp, 0, %sp;				\
 	lduwa	[%sp + %g0] ASI, %l0;			\
 	mov	0x04, %g2;				\
 	mov	0x08, %g3;				\
@@ -616,14 +622,16 @@ user_rtt_fill_64bit:					\
 	lduwa	[%g1 + %g3] ASI, %i6;			\
 	lduwa	[%g1 + %g5] ASI, %i7;			\
 	restored;					\
-	retry; nop; nop; nop; nop;			\
+	retry; nop; nop;				\
 	b,a,pt	%xcc, fill_fixup_dax;			\
 	b,a,pt	%xcc, fill_fixup_mna;			\
 	b,a,pt	%xcc, fill_fixup;
 
 #define FILL_2_GENERIC_RTRAP				\
 user_rtt_fill_32bit:					\
-	srl	%sp, 0, %sp;				\
+	and	%sp, 1, %g3;				\
+	brnz,pn	%g3, user_rtt_fill_64bit;		\
+	 srl	%sp, 0, %sp;				\
 	lduwa	[%sp + 0x00] %asi, %l0;			\
 	lduwa	[%sp + 0x04] %asi, %l1;			\
 	lduwa	[%sp + 0x08] %asi, %l2;			\
@@ -643,7 +651,7 @@ user_rtt_fill_32bit:					\
 	ba,pt	%xcc, user_rtt_pre_restore;		\
 	 restored;					\
 	nop; nop; nop; nop; nop;			\
-	nop; nop; nop; nop; nop;			\
+	nop; nop; nop;					\
 	ba,a,pt	%xcc, user_rtt_fill_fixup;		\
 	ba,a,pt	%xcc, user_rtt_fill_fixup;		\
 	ba,a,pt	%xcc, user_rtt_fill_fixup;
diff --git a/arch/sparc/include/asm/uaccess_64.h b/arch/sparc/include/asm/uaccess_64.h
index 73083e1d38d9..e562d3caee57 100644
--- a/arch/sparc/include/asm/uaccess_64.h
+++ b/arch/sparc/include/asm/uaccess_64.h
@@ -38,14 +38,14 @@
 #define VERIFY_READ	0
 #define VERIFY_WRITE	1
 
-#define get_fs() ((mm_segment_t) { get_thread_current_ds() })
+#define get_fs() ((mm_segment_t){(current_thread_info()->current_ds)})
 #define get_ds() (KERNEL_DS)
 
 #define segment_eq(a,b)  ((a).seg == (b).seg)
 
 #define set_fs(val)								\
 do {										\
-	set_thread_current_ds((val).seg);					\
+	current_thread_info()->current_ds =(val).seg;				\
 	__asm__ __volatile__ ("wr %%g0, %0, %%asi" : : "r" ((val).seg));	\
 } while(0)
 
diff --git a/arch/sparc/include/asm/unistd.h b/arch/sparc/include/asm/unistd.h
index 0ecea6ed943e..5356810bd7e7 100644
--- a/arch/sparc/include/asm/unistd.h
+++ b/arch/sparc/include/asm/unistd.h
@@ -38,12 +38,10 @@
 #define __ARCH_WANT_SYS_OLDUMOUNT
 #define __ARCH_WANT_SYS_SIGPENDING
 #define __ARCH_WANT_SYS_SIGPROCMASK
-#define __ARCH_WANT_SYS_RT_SIGSUSPEND
 #ifdef __32bit_syscall_numbers__
 #define __ARCH_WANT_SYS_IPC
 #else
 #define __ARCH_WANT_COMPAT_SYS_TIME
-#define __ARCH_WANT_COMPAT_SYS_RT_SIGSUSPEND
 #define __ARCH_WANT_COMPAT_SYS_SENDFILE
 #endif
 
diff --git a/arch/sparc/include/uapi/asm/ioctls.h b/arch/sparc/include/uapi/asm/ioctls.h
index 9155f7041d44..897d1723fa14 100644
--- a/arch/sparc/include/uapi/asm/ioctls.h
+++ b/arch/sparc/include/uapi/asm/ioctls.h
@@ -21,6 +21,9 @@
 #define TCSETSF2	_IOW('T', 15, struct termios2)
 #define TIOCGDEV	_IOR('T',0x32, unsigned int) /* Get primary device node of /dev/console */
 #define TIOCVHANGUP	_IO('T', 0x37)
+#define TIOCGPKT	_IOR('T', 0x38, int) /* Get packet mode state */
+#define TIOCGPTLCK	_IOR('T', 0x39, int) /* Get Pty lock state */
+#define TIOCGEXCL	_IOR('T', 0x40, int) /* Get exclusive mode state */
 
 /* Note that all the ioctls that are not available in Linux have a 
  * double underscore on the front to: a) avoid some programs to
diff --git a/arch/sparc/include/uapi/asm/signal.h b/arch/sparc/include/uapi/asm/signal.h
index 1a041892538f..f387400fcfdf 100644
--- a/arch/sparc/include/uapi/asm/signal.h
+++ b/arch/sparc/include/uapi/asm/signal.h
@@ -147,18 +147,13 @@ struct sigstack {
 #define SIG_UNBLOCK        0x02	/* for unblocking signals */
 #define SIG_SETMASK        0x04	/* for setting the signal mask */
 
-/*
- * sigaltstack controls
- */
-#define SS_ONSTACK	1
-#define SS_DISABLE	2
-
 #define MINSIGSTKSZ	4096
 #define SIGSTKSZ	16384
 
 
 #include <asm-generic/signal-defs.h>
 
+#ifndef __KERNEL__
 struct __new_sigaction {
 	__sighandler_t		sa_handler;
 	unsigned long		sa_flags;
@@ -172,6 +167,7 @@ struct __old_sigaction {
 	unsigned long		sa_flags;
 	void			(*sa_restorer)(void);  /* not used by Linux/SPARC yet */
 };
+#endif
 
 typedef struct sigaltstack {
 	void			__user *ss_sp;
diff --git a/arch/sparc/include/uapi/asm/socket.h b/arch/sparc/include/uapi/asm/socket.h
index bea1568ae4af..cbbad74b2e06 100644
--- a/arch/sparc/include/uapi/asm/socket.h
+++ b/arch/sparc/include/uapi/asm/socket.h
@@ -15,7 +15,7 @@
 #define SO_PEERCRED	0x0040
 #define SO_LINGER	0x0080
 #define SO_OOBINLINE	0x0100
-/* To add :#define SO_REUSEPORT 0x0200 */
+#define SO_REUSEPORT	0x0200
 #define SO_BSDCOMPAT    0x0400
 #define SO_RCVLOWAT     0x0800
 #define SO_SNDLOWAT     0x1000
@@ -41,6 +41,7 @@
 
 #define SO_ATTACH_FILTER	0x001a
 #define SO_DETACH_FILTER        0x001b
+#define SO_GET_FILTER		SO_ATTACH_FILTER
 
 #define SO_PEERNAME		0x001c
 #define SO_TIMESTAMP		0x001d
@@ -65,6 +66,7 @@
 /* Instruct lower device to use last 4-bytes of skb data as FCS */
 #define SO_NOFCS		0x0027
 
+#define SO_LOCK_FILTER		0x0028
 
 /* Security levels - as per NRL IPv6 - don't actually do anything */
 #define SO_SECURITY_AUTHENTICATION		0x5001
diff --git a/arch/sparc/include/uapi/asm/unistd.h b/arch/sparc/include/uapi/asm/unistd.h
index 8974ef7ae920..62ced589bcf7 100644
--- a/arch/sparc/include/uapi/asm/unistd.h
+++ b/arch/sparc/include/uapi/asm/unistd.h
@@ -405,8 +405,14 @@
 #define __NR_setns		337
 #define __NR_process_vm_readv	338
 #define __NR_process_vm_writev	339
+#define __NR_kern_features	340
+#define __NR_kcmp		341
+#define __NR_finit_module	342
 
-#define NR_syscalls		340
+#define NR_syscalls		343
+
+/* Bitmask values returned from kern_features system call.  */
+#define KERN_FEATURE_MIXED_MODE_STACK	0x00000001
 
 #ifdef __32bit_syscall_numbers__
 /* Sparc 32-bit only has the "setresuid32", "getresuid32" variants,
diff --git a/arch/sparc/kernel/apc.c b/arch/sparc/kernel/apc.c
index 812e10bbb0b3..eefda32b595e 100644
--- a/arch/sparc/kernel/apc.c
+++ b/arch/sparc/kernel/apc.c
@@ -20,6 +20,7 @@
 #include <asm/uaccess.h>
 #include <asm/auxio.h>
 #include <asm/apc.h>
+#include <asm/processor.h>
 
 /* Debugging
  * 
@@ -31,7 +32,7 @@
 #define APC_DEVNAME "apc"
 
 static u8 __iomem *regs;
-static int apc_no_idle __devinitdata = 0;
+static int apc_no_idle = 0;
 
 #define apc_readb(offs)		(sbus_readb(regs+offs))
 #define apc_writeb(val, offs) 	(sbus_writeb(val, regs+offs))
@@ -138,7 +139,7 @@ static const struct file_operations apc_fops = {
 
 static struct miscdevice apc_miscdev = { APC_MINOR, APC_DEVNAME, &apc_fops };
 
-static int __devinit apc_probe(struct platform_device *op)
+static int apc_probe(struct platform_device *op)
 {
 	int err;
 
@@ -158,7 +159,7 @@ static int __devinit apc_probe(struct platform_device *op)
 
 	/* Assign power management IDLE handler */
 	if (!apc_no_idle)
-		pm_idle = apc_swift_idle;	
+		sparc_idle = apc_swift_idle;
 
 	printk(KERN_INFO "%s: power management initialized%s\n", 
 	       APC_DEVNAME, apc_no_idle ? " (CPU idle disabled)" : "");
diff --git a/arch/sparc/kernel/auxio_64.c b/arch/sparc/kernel/auxio_64.c
index 773091ac71a3..86e55778e4af 100644
--- a/arch/sparc/kernel/auxio_64.c
+++ b/arch/sparc/kernel/auxio_64.c
@@ -102,7 +102,7 @@ static const struct of_device_id auxio_match[] = {
 
 MODULE_DEVICE_TABLE(of, auxio_match);
 
-static int __devinit auxio_probe(struct platform_device *dev)
+static int auxio_probe(struct platform_device *dev)
 {
 	struct device_node *dp = dev->dev.of_node;
 	unsigned long size;
diff --git a/arch/sparc/kernel/central.c b/arch/sparc/kernel/central.c
index 9708851a8b9f..052b5a44318f 100644
--- a/arch/sparc/kernel/central.c
+++ b/arch/sparc/kernel/central.c
@@ -33,7 +33,7 @@ struct fhc {
 	struct platform_device	leds_pdev;
 };
 
-static int __devinit clock_board_calc_nslots(struct clock_board *p)
+static int clock_board_calc_nslots(struct clock_board *p)
 {
 	u8 reg = upa_readb(p->clock_regs + CLOCK_STAT1) & 0xc0;
 
@@ -60,7 +60,7 @@ static int __devinit clock_board_calc_nslots(struct clock_board *p)
 	}
 }
 
-static int __devinit clock_board_probe(struct platform_device *op)
+static int clock_board_probe(struct platform_device *op)
 {
 	struct clock_board *p = kzalloc(sizeof(*p), GFP_KERNEL);
 	int err = -ENOMEM;
@@ -157,7 +157,7 @@ static struct platform_driver clock_board_driver = {
 	},
 };
 
-static int __devinit fhc_probe(struct platform_device *op)
+static int fhc_probe(struct platform_device *op)
 {
 	struct fhc *p = kzalloc(sizeof(*p), GFP_KERNEL);
 	int err = -ENOMEM;
diff --git a/arch/sparc/kernel/chmc.c b/arch/sparc/kernel/chmc.c
index 5f450260981d..dbb210d74e21 100644
--- a/arch/sparc/kernel/chmc.c
+++ b/arch/sparc/kernel/chmc.c
@@ -336,9 +336,9 @@ static int jbusmc_print_dimm(int syndrome_code,
 	return 0;
 }
 
-static u64 __devinit jbusmc_dimm_group_size(u64 base,
-					    const struct linux_prom64_registers *mem_regs,
-					    int num_mem_regs)
+static u64 jbusmc_dimm_group_size(u64 base,
+				  const struct linux_prom64_registers *mem_regs,
+				  int num_mem_regs)
 {
 	u64 max = base + (8UL * 1024 * 1024 * 1024);
 	u64 max_seen = base;
@@ -363,10 +363,10 @@ static u64 __devinit jbusmc_dimm_group_size(u64 base,
 	return max_seen - base;
 }
 
-static void __devinit jbusmc_construct_one_dimm_group(struct jbusmc *p,
-						      unsigned long index,
-						      const struct linux_prom64_registers *mem_regs,
-						      int num_mem_regs)
+static void jbusmc_construct_one_dimm_group(struct jbusmc *p,
+					    unsigned long index,
+					    const struct linux_prom64_registers *mem_regs,
+					    int num_mem_regs)
 {
 	struct jbusmc_dimm_group *dp = &p->dimm_groups[index];
 
@@ -378,9 +378,9 @@ static void __devinit jbusmc_construct_one_dimm_group(struct jbusmc *p,
 	dp->size = jbusmc_dimm_group_size(dp->base_addr, mem_regs, num_mem_regs);
 }
 
-static void __devinit jbusmc_construct_dimm_groups(struct jbusmc *p,
-						   const struct linux_prom64_registers *mem_regs,
-						   int num_mem_regs)
+static void jbusmc_construct_dimm_groups(struct jbusmc *p,
+					 const struct linux_prom64_registers *mem_regs,
+					 int num_mem_regs)
 {
 	if (p->mc_reg_1 & JB_MC_REG1_DIMM1_BANK0) {
 		jbusmc_construct_one_dimm_group(p, 0, mem_regs, num_mem_regs);
@@ -392,7 +392,7 @@ static void __devinit jbusmc_construct_dimm_groups(struct jbusmc *p,
 	}
 }
 
-static int __devinit jbusmc_probe(struct platform_device *op)
+static int jbusmc_probe(struct platform_device *op)
 {
 	const struct linux_prom64_registers *mem_regs;
 	struct device_node *mem_node;
@@ -689,7 +689,7 @@ static void chmc_fetch_decode_regs(struct chmc *p)
 				      chmc_read_mcreg(p, CHMCTRL_DECODE4));
 }
 
-static int __devinit chmc_probe(struct platform_device *op)
+static int chmc_probe(struct platform_device *op)
 {
 	struct device_node *dp = op->dev.of_node;
 	unsigned long ver;
@@ -763,7 +763,7 @@ out_free:
 	goto out;
 }
 
-static int __devinit us3mc_probe(struct platform_device *op)
+static int us3mc_probe(struct platform_device *op)
 {
 	if (mc_type == MC_TYPE_SAFARI)
 		return chmc_probe(op);
@@ -772,21 +772,21 @@ static int __devinit us3mc_probe(struct platform_device *op)
 	return -ENODEV;
 }
 
-static void __devexit chmc_destroy(struct platform_device *op, struct chmc *p)
+static void chmc_destroy(struct platform_device *op, struct chmc *p)
 {
 	list_del(&p->list);
 	of_iounmap(&op->resource[0], p->regs, 0x48);
 	kfree(p);
 }
 
-static void __devexit jbusmc_destroy(struct platform_device *op, struct jbusmc *p)
+static void jbusmc_destroy(struct platform_device *op, struct jbusmc *p)
 {
 	mc_list_del(&p->list);
 	of_iounmap(&op->resource[0], p->regs, JBUSMC_REGS_SIZE);
 	kfree(p);
 }
 
-static int __devexit us3mc_remove(struct platform_device *op)
+static int us3mc_remove(struct platform_device *op)
 {
 	void *p = dev_get_drvdata(&op->dev);
 
@@ -814,7 +814,7 @@ static struct platform_driver us3mc_driver = {
 		.of_match_table = us3mc_match,
 	},
 	.probe		= us3mc_probe,
-	.remove		= __devexit_p(us3mc_remove),
+	.remove		= us3mc_remove,
 };
 
 static inline bool us3mc_platform(void)
diff --git a/arch/sparc/kernel/ds.c b/arch/sparc/kernel/ds.c
index f09257c86107..75bb608c423e 100644
--- a/arch/sparc/kernel/ds.c
+++ b/arch/sparc/kernel/ds.c
@@ -29,7 +29,7 @@
 #define DRV_MODULE_VERSION	"1.0"
 #define DRV_MODULE_RELDATE	"Jul 11, 2007"
 
-static char version[] __devinitdata =
+static char version[] =
 	DRV_MODULE_NAME ".c:v" DRV_MODULE_VERSION " (" DRV_MODULE_RELDATE ")\n";
 MODULE_AUTHOR("David S. Miller (davem@davemloft.net)");
 MODULE_DESCRIPTION("Sun LDOM domain services driver");
@@ -1146,8 +1146,7 @@ static void ds_event(void *arg, int event)
 	spin_unlock_irqrestore(&ds_lock, flags);
 }
 
-static int __devinit ds_probe(struct vio_dev *vdev,
-			      const struct vio_device_id *id)
+static int ds_probe(struct vio_dev *vdev, const struct vio_device_id *id)
 {
 	static int ds_version_printed;
 	struct ldc_channel_config ds_cfg = {
diff --git a/arch/sparc/kernel/entry.S b/arch/sparc/kernel/entry.S
index dcaa1cf0de40..e2a030045089 100644
--- a/arch/sparc/kernel/entry.S
+++ b/arch/sparc/kernel/entry.S
@@ -806,23 +806,10 @@ sys_nis_syscall:
 	call	c_sys_nis_syscall
 	 mov	%l5, %o7
 
-	.align	4
-	.globl	sys_execve
-sys_execve:
-	mov	%o7, %l5
-	add	%sp, STACKFRAME_SZ, %o0		! pt_regs *regs arg
-	call	sparc_execve
-	 mov	%l5, %o7
-
-	.globl	sunos_execv
 sunos_execv:
-	st	%g0, [%sp + STACKFRAME_SZ + PT_I2]
-
-	call	sparc_execve
-	 add	%sp, STACKFRAME_SZ, %o0
-
-	b	ret_sys_call
-	 ld	[%sp + STACKFRAME_SZ + PT_I0], %o0
+	.globl	sunos_execv
+	b	sys_execve
+	 clr	%i2
 
 	.align	4
 	.globl	sys_sparc_pipe
@@ -833,14 +820,6 @@ sys_sparc_pipe:
 	 mov	%l5, %o7
 
 	.align	4
-	.globl	sys_sigaltstack
-sys_sigaltstack:
-	mov	%o7, %l5
-	mov	%fp, %o2
-	call	do_sigaltstack
-	 mov	%l5, %o7
-
-	.align	4
 	.globl	sys_sigstack
 sys_sigstack:
 	mov	%o7, %l5
@@ -959,17 +938,9 @@ flush_patch_four:
         .align  4
 linux_sparc_ni_syscall:
 	sethi   %hi(sys_ni_syscall), %l7
-	b       syscall_is_too_hard
+	b       do_syscall
 	 or     %l7, %lo(sys_ni_syscall), %l7
 
-linux_fast_syscall:
-	andn	%l7, 3, %l7
-	mov	%i0, %o0
-	mov	%i1, %o1
-	mov 	%i2, %o2
-	jmpl	%l7 + %g0, %g0
-	 mov	%i3, %o3
-
 linux_syscall_trace:
 	add	%sp, STACKFRAME_SZ, %o0
 	call	syscall_trace
@@ -991,6 +962,23 @@ ret_from_fork:
 	b	ret_sys_call
 	 ld	[%sp + STACKFRAME_SZ + PT_I0], %o0
 
+	.globl	ret_from_kernel_thread
+ret_from_kernel_thread:
+	call	schedule_tail
+	 ld	[%g3 + TI_TASK], %o0
+	ld	[%sp + STACKFRAME_SZ + PT_G1], %l0
+	call	%l0
+	 ld	[%sp + STACKFRAME_SZ + PT_G2], %o0
+	rd	%psr, %l1
+	ld	[%sp + STACKFRAME_SZ + PT_PSR], %l0
+	andn	%l0, PSR_CWP, %l0
+	nop
+	and	%l1, PSR_CWP, %l1
+	or	%l0, %l1, %l0
+	st	%l0, [%sp + STACKFRAME_SZ + PT_PSR]
+	b	ret_sys_call
+	 mov	0, %o0
+
 	/* Linux native system calls enter here... */
 	.align	4
 	.globl	linux_sparc_syscall
@@ -1002,11 +990,8 @@ linux_sparc_syscall:
 	bgeu	linux_sparc_ni_syscall
 	 sll	%g1, 2, %l4
 	ld	[%l7 + %l4], %l7
-	andcc	%l7, 1, %g0
-	bne	linux_fast_syscall
-	 /* Just do first insn from SAVE_ALL in the delay slot */
 
-syscall_is_too_hard:
+do_syscall:
 	SAVE_ALL_HEAD
 	 rd	%wim, %l3
 
diff --git a/arch/sparc/kernel/entry.h b/arch/sparc/kernel/entry.h
index 0c218e4c0881..cc3c5cb47cda 100644
--- a/arch/sparc/kernel/entry.h
+++ b/arch/sparc/kernel/entry.h
@@ -59,6 +59,13 @@ struct popc_6insn_patch_entry {
 extern struct popc_6insn_patch_entry __popc_6insn_patch,
 	__popc_6insn_patch_end;
 
+struct pause_patch_entry {
+	unsigned int	addr;
+	unsigned int	insns[3];
+};
+extern struct pause_patch_entry __pause_3insn_patch,
+	__pause_3insn_patch_end;
+
 extern void __init per_cpu_patch(void);
 extern void sun4v_patch_1insn_range(struct sun4v_1insn_patch_entry *,
 				    struct sun4v_1insn_patch_entry *);
diff --git a/arch/sparc/kernel/etrap_64.S b/arch/sparc/kernel/etrap_64.S
index 786b185e6e3f..1276ca2567ba 100644
--- a/arch/sparc/kernel/etrap_64.S
+++ b/arch/sparc/kernel/etrap_64.S
@@ -92,8 +92,10 @@ etrap_save:	save	%g2, -STACK_BIAS, %sp
 		rdpr	%wstate, %g2
 		wrpr	%g0, 0, %canrestore
 		sll	%g2, 3, %g2
+
+		/* Set TI_SYS_FPDEPTH to 1 and clear TI_SYS_NOERROR.  */
 		mov	1, %l5
-		stb	%l5, [%l6 + TI_FPDEPTH]
+		sth	%l5, [%l6 + TI_SYS_NOERROR]
 
 		wrpr	%g3, 0, %otherwin
 		wrpr	%g2, 0, %wstate
@@ -152,7 +154,9 @@ etrap_save:	save	%g2, -STACK_BIAS, %sp
 		add	%l6, TI_FPSAVED + 1, %l4
 		srl	%l5, 1, %l3
 		add	%l5, 2, %l5
-		stb	%l5, [%l6 + TI_FPDEPTH]
+
+		/* Set TI_SYS_FPDEPTH to %l5 and clear TI_SYS_NOERROR.  */
+		sth	%l5, [%l6 + TI_SYS_NOERROR]
 		ba,pt	%xcc, 2b
 		 stb	%g0, [%l4 + %l3]
 		nop
diff --git a/arch/sparc/kernel/kernel.h b/arch/sparc/kernel/kernel.h
index 291bb5de9ce0..a702d9ab019c 100644
--- a/arch/sparc/kernel/kernel.h
+++ b/arch/sparc/kernel/kernel.h
@@ -48,6 +48,10 @@ extern void sun4m_init_IRQ(void);
 extern void sun4m_unmask_profile_irq(void);
 extern void sun4m_clear_profile_irq(int cpu);
 
+/* sun4m_smp.c */
+void sun4m_cpu_pre_starting(void *arg);
+void sun4m_cpu_pre_online(void *arg);
+
 /* sun4d_irq.c */
 extern spinlock_t sun4d_imsk_lock;
 
@@ -60,6 +64,14 @@ extern int show_sun4d_interrupts(struct seq_file *, void *);
 extern void sun4d_distribute_irqs(void);
 extern void sun4d_free_irq(unsigned int irq, void *dev_id);
 
+/* sun4d_smp.c */
+void sun4d_cpu_pre_starting(void *arg);
+void sun4d_cpu_pre_online(void *arg);
+
+/* leon_smp.c */
+void leon_cpu_pre_starting(void *arg);
+void leon_cpu_pre_online(void *arg);
+
 /* head_32.S */
 extern unsigned int t_nmi[];
 extern unsigned int linux_trap_ipi15_sun4d[];
diff --git a/arch/sparc/kernel/kgdb_32.c b/arch/sparc/kernel/kgdb_32.c
index 2e424a576a36..dcf210811af4 100644
--- a/arch/sparc/kernel/kgdb_32.c
+++ b/arch/sparc/kernel/kgdb_32.c
@@ -5,6 +5,7 @@
 
 #include <linux/kgdb.h>
 #include <linux/kdebug.h>
+#include <linux/sched.h>
 
 #include <asm/kdebug.h>
 #include <asm/ptrace.h>
diff --git a/arch/sparc/kernel/kprobes.c b/arch/sparc/kernel/kprobes.c
index a39d1ba5a119..e72212148d2a 100644
--- a/arch/sparc/kernel/kprobes.c
+++ b/arch/sparc/kernel/kprobes.c
@@ -511,7 +511,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
 {
 	struct kretprobe_instance *ri = NULL;
 	struct hlist_head *head, empty_rp;
-	struct hlist_node *node, *tmp;
+	struct hlist_node *tmp;
 	unsigned long flags, orig_ret_address = 0;
 	unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline;
 
@@ -531,7 +531,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
 	 *       real return address, and all the rest will point to
 	 *       kretprobe_trampoline
 	 */
-	hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
+	hlist_for_each_entry_safe(ri, tmp, head, hlist) {
 		if (ri->task != current)
 			/* another task is sharing our hash bucket */
 			continue;
@@ -559,7 +559,7 @@ int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
 	kretprobe_hash_unlock(current, &flags);
 	preempt_enable_no_resched();
 
-	hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) {
+	hlist_for_each_entry_safe(ri, tmp, &empty_rp, hlist) {
 		hlist_del(&ri->hlist);
 		kfree(ri);
 	}
diff --git a/arch/sparc/kernel/ldc.c b/arch/sparc/kernel/ldc.c
index 81d92fc9983b..54df554b82d9 100644
--- a/arch/sparc/kernel/ldc.c
+++ b/arch/sparc/kernel/ldc.c
@@ -27,7 +27,7 @@
 #define DRV_MODULE_VERSION	"1.1"
 #define DRV_MODULE_RELDATE	"July 22, 2008"
 
-static char version[] __devinitdata =
+static char version[] =
 	DRV_MODULE_NAME ".c:v" DRV_MODULE_VERSION " (" DRV_MODULE_RELDATE ")\n";
 #define LDC_PACKET_SIZE		64
 
@@ -953,9 +953,8 @@ static HLIST_HEAD(ldc_channel_list);
 static int __ldc_channel_exists(unsigned long id)
 {
 	struct ldc_channel *lp;
-	struct hlist_node *n;
 
-	hlist_for_each_entry(lp, n, &ldc_channel_list, list) {
+	hlist_for_each_entry(lp, &ldc_channel_list, list) {
 		if (lp->id == id)
 			return 1;
 	}
diff --git a/arch/sparc/kernel/leon_kernel.c b/arch/sparc/kernel/leon_kernel.c
index f8b6eee40bde..87f60ee65433 100644
--- a/arch/sparc/kernel/leon_kernel.c
+++ b/arch/sparc/kernel/leon_kernel.c
@@ -56,11 +56,13 @@ static inline unsigned int leon_eirq_get(int cpu)
 static void leon_handle_ext_irq(unsigned int irq, struct irq_desc *desc)
 {
 	unsigned int eirq;
+	struct irq_bucket *p;
 	int cpu = sparc_leon3_cpuid();
 
 	eirq = leon_eirq_get(cpu);
-	if ((eirq & 0x10) && irq_map[eirq]->irq) /* bit4 tells if IRQ happened */
-		generic_handle_irq(irq_map[eirq]->irq);
+	p = irq_map[eirq];
+	if ((eirq & 0x10) && p && p->irq) /* bit4 tells if IRQ happened */
+		generic_handle_irq(p->irq);
 }
 
 /* The extended IRQ controller has been found, this function registers it */
diff --git a/arch/sparc/kernel/leon_pci.c b/arch/sparc/kernel/leon_pci.c
index fc0521161568..852dc8430528 100644
--- a/arch/sparc/kernel/leon_pci.c
+++ b/arch/sparc/kernel/leon_pci.c
@@ -43,7 +43,7 @@ void leon_pci_init(struct platform_device *ofdev, struct leon_pci_info *info)
 	}
 }
 
-void __devinit pcibios_fixup_bus(struct pci_bus *pbus)
+void pcibios_fixup_bus(struct pci_bus *pbus)
 {
 	struct pci_dev *dev;
 	int i, has_io, has_mem;
diff --git a/arch/sparc/kernel/leon_pci_grpci2.c b/arch/sparc/kernel/leon_pci_grpci2.c
index b1bc38853a3d..fc4320886a3a 100644
--- a/arch/sparc/kernel/leon_pci_grpci2.c
+++ b/arch/sparc/kernel/leon_pci_grpci2.c
@@ -668,7 +668,7 @@ static irqreturn_t grpci2_err_interrupt(int irq, void *arg)
 	return IRQ_HANDLED;
 }
 
-static int __devinit grpci2_of_probe(struct platform_device *ofdev)
+static int grpci2_of_probe(struct platform_device *ofdev)
 {
 	struct grpci2_regs *regs;
 	struct grpci2_priv *priv;
diff --git a/arch/sparc/kernel/leon_pmc.c b/arch/sparc/kernel/leon_pmc.c
index 4e174321097d..708bca435219 100644
--- a/arch/sparc/kernel/leon_pmc.c
+++ b/arch/sparc/kernel/leon_pmc.c
@@ -9,6 +9,7 @@
 #include <asm/leon_amba.h>
 #include <asm/cpu_type.h>
 #include <asm/leon.h>
+#include <asm/processor.h>
 
 /* List of Systems that need fixup instructions around power-down instruction */
 unsigned int pmc_leon_fixup_ids[] = {
@@ -69,9 +70,9 @@ static int __init leon_pmc_install(void)
 	if (sparc_cpu_model == sparc_leon) {
 		/* Assign power management IDLE handler */
 		if (pmc_leon_need_fixup())
-			pm_idle = pmc_leon_idle_fixup;
+			sparc_idle = pmc_leon_idle_fixup;
 		else
-			pm_idle = pmc_leon_idle;
+			sparc_idle = pmc_leon_idle;
 
 		printk(KERN_INFO "leon: power management initialized\n");
 	}
diff --git a/arch/sparc/kernel/leon_smp.c b/arch/sparc/kernel/leon_smp.c
index 0f3fb6d9c8ef..9b40c9c12a0c 100644
--- a/arch/sparc/kernel/leon_smp.c
+++ b/arch/sparc/kernel/leon_smp.c
@@ -69,31 +69,19 @@ static inline unsigned long do_swap(volatile unsigned long *ptr,
 	return val;
 }
 
-void __cpuinit leon_callin(void)
+void __cpuinit leon_cpu_pre_starting(void *arg)
 {
-	int cpuid = hard_smp_processor_id();
-
-	local_ops->cache_all();
-	local_ops->tlb_all();
 	leon_configure_cache_smp();
+}
 
-	notify_cpu_starting(cpuid);
-
-	/* Get our local ticker going. */
-	register_percpu_ce(cpuid);
-
-	calibrate_delay();
-	smp_store_cpu_info(cpuid);
-
-	local_ops->cache_all();
-	local_ops->tlb_all();
+void __cpuinit leon_cpu_pre_online(void *arg)
+{
+	int cpuid = hard_smp_processor_id();
 
-	/*
-	 * Unblock the master CPU _only_ when the scheduler state
-	 * of all secondary CPUs will be up-to-date, so after
-	 * the SMP initialization the master will be just allowed
-	 * to call the scheduler code.
-	 * Allow master to continue.
+	/* Allow master to continue. The master will then give us the
+	 * go-ahead by setting the smp_commenced_mask and will wait without
+	 * timeouts until our setup is completed fully (signified by
+	 * our bit being set in the cpu_online_mask).
 	 */
 	do_swap(&cpu_callin_map[cpuid], 1);
 
@@ -110,9 +98,6 @@ void __cpuinit leon_callin(void)
 
 	while (!cpumask_test_cpu(cpuid, &smp_commenced_mask))
 		mb();
-
-	local_irq_enable();
-	set_cpu_online(cpuid, true);
 }
 
 /*
diff --git a/arch/sparc/kernel/module.c b/arch/sparc/kernel/module.c
index f1ddc0d23679..4435488ebe25 100644
--- a/arch/sparc/kernel/module.c
+++ b/arch/sparc/kernel/module.c
@@ -43,10 +43,6 @@ void *module_alloc(unsigned long size)
 {
 	void *ret;
 
-	/* We handle the zero case fine, unlike vmalloc */
-	if (size == 0)
-		return NULL;
-
 	ret = module_map(size);
 	if (ret)
 		memset(ret, 0, size);
diff --git a/arch/sparc/kernel/pci.c b/arch/sparc/kernel/pci.c
index 75b31bcdeadf..baf4366e2d6a 100644
--- a/arch/sparc/kernel/pci.c
+++ b/arch/sparc/kernel/pci.c
@@ -356,7 +356,7 @@ static struct pci_dev *of_create_pci_dev(struct pci_pbm_info *pbm,
 	return dev;
 }
 
-static void __devinit apb_calc_first_last(u8 map, u32 *first_p, u32 *last_p)
+static void apb_calc_first_last(u8 map, u32 *first_p, u32 *last_p)
 {
 	u32 idx, first, last;
 
@@ -378,9 +378,9 @@ static void __devinit apb_calc_first_last(u8 map, u32 *first_p, u32 *last_p)
 /* Cook up fake bus resources for SUNW,simba PCI bridges which lack
  * a proper 'ranges' property.
  */
-static void __devinit apb_fake_ranges(struct pci_dev *dev,
-				      struct pci_bus *bus,
-				      struct pci_pbm_info *pbm)
+static void apb_fake_ranges(struct pci_dev *dev,
+			    struct pci_bus *bus,
+			    struct pci_pbm_info *pbm)
 {
 	struct pci_bus_region region;
 	struct resource *res;
@@ -404,15 +404,15 @@ static void __devinit apb_fake_ranges(struct pci_dev *dev,
 	pcibios_bus_to_resource(dev, res, &region);
 }
 
-static void __devinit pci_of_scan_bus(struct pci_pbm_info *pbm,
-				      struct device_node *node,
-				      struct pci_bus *bus);
+static void pci_of_scan_bus(struct pci_pbm_info *pbm,
+			    struct device_node *node,
+			    struct pci_bus *bus);
 
 #define GET_64BIT(prop, i)	((((u64) (prop)[(i)]) << 32) | (prop)[(i)+1])
 
-static void __devinit of_scan_pci_bridge(struct pci_pbm_info *pbm,
-					 struct device_node *node,
-					 struct pci_dev *dev)
+static void of_scan_pci_bridge(struct pci_pbm_info *pbm,
+			       struct device_node *node,
+			       struct pci_dev *dev)
 {
 	struct pci_bus *bus;
 	const u32 *busrange, *ranges;
@@ -503,9 +503,9 @@ after_ranges:
 	pci_of_scan_bus(pbm, node, bus);
 }
 
-static void __devinit pci_of_scan_bus(struct pci_pbm_info *pbm,
-				      struct device_node *node,
-				      struct pci_bus *bus)
+static void pci_of_scan_bus(struct pci_pbm_info *pbm,
+			    struct device_node *node,
+			    struct pci_bus *bus)
 {
 	struct device_node *child;
 	const u32 *reg;
@@ -564,7 +564,7 @@ show_pciobppath_attr(struct device * dev, struct device_attribute * attr, char *
 
 static DEVICE_ATTR(obppath, S_IRUSR | S_IRGRP | S_IROTH, show_pciobppath_attr, NULL);
 
-static void __devinit pci_bus_register_of_sysfs(struct pci_bus *bus)
+static void pci_bus_register_of_sysfs(struct pci_bus *bus)
 {
 	struct pci_dev *dev;
 	struct pci_bus *child_bus;
@@ -585,8 +585,8 @@ static void __devinit pci_bus_register_of_sysfs(struct pci_bus *bus)
 		pci_bus_register_of_sysfs(child_bus);
 }
 
-struct pci_bus * __devinit pci_scan_one_pbm(struct pci_pbm_info *pbm,
-					    struct device *parent)
+struct pci_bus *pci_scan_one_pbm(struct pci_pbm_info *pbm,
+				 struct device *parent)
 {
 	LIST_HEAD(resources);
 	struct device_node *node = pbm->op->dev.of_node;
@@ -618,7 +618,7 @@ struct pci_bus * __devinit pci_scan_one_pbm(struct pci_pbm_info *pbm,
 	return bus;
 }
 
-void __devinit pcibios_fixup_bus(struct pci_bus *pbus)
+void pcibios_fixup_bus(struct pci_bus *pbus)
 {
 }
 
@@ -949,8 +949,7 @@ static int __init pcibios_init(void)
 subsys_initcall(pcibios_init);
 
 #ifdef CONFIG_SYSFS
-static void __devinit pci_bus_slot_names(struct device_node *node,
-					 struct pci_bus *bus)
+static void pci_bus_slot_names(struct device_node *node, struct pci_bus *bus)
 {
 	const struct pci_slot_names {
 		u32	slot_mask;
diff --git a/arch/sparc/kernel/pci_fire.c b/arch/sparc/kernel/pci_fire.c
index 188f935276fd..e60fc6a67e9b 100644
--- a/arch/sparc/kernel/pci_fire.c
+++ b/arch/sparc/kernel/pci_fire.c
@@ -408,8 +408,8 @@ static void pci_fire_hw_init(struct pci_pbm_info *pbm)
 	upa_writeq(~(u64)0, pbm->pbm_regs + FIRE_PEC_IENAB);
 }
 
-static int __devinit pci_fire_pbm_init(struct pci_pbm_info *pbm,
-				       struct platform_device *op, u32 portid)
+static int pci_fire_pbm_init(struct pci_pbm_info *pbm,
+			     struct platform_device *op, u32 portid)
 {
 	const struct linux_prom64_registers *regs;
 	struct device_node *dp = op->dev.of_node;
@@ -454,7 +454,7 @@ static int __devinit pci_fire_pbm_init(struct pci_pbm_info *pbm,
 	return 0;
 }
 
-static int __devinit fire_probe(struct platform_device *op)
+static int fire_probe(struct platform_device *op)
 {
 	struct device_node *dp = op->dev.of_node;
 	struct pci_pbm_info *pbm;
diff --git a/arch/sparc/kernel/pci_impl.h b/arch/sparc/kernel/pci_impl.h
index 918a2031c8bb..5f688531f48c 100644
--- a/arch/sparc/kernel/pci_impl.h
+++ b/arch/sparc/kernel/pci_impl.h
@@ -88,7 +88,7 @@ struct pci_pbm_info {
 	int				chip_revision;
 
 	/* Name used for top-level resources. */
-	char				*name;
+	const char			*name;
 
 	/* OBP specific information. */
 	struct platform_device		*op;
diff --git a/arch/sparc/kernel/pci_psycho.c b/arch/sparc/kernel/pci_psycho.c
index f4d29e15ce71..c647634ead2b 100644
--- a/arch/sparc/kernel/pci_psycho.c
+++ b/arch/sparc/kernel/pci_psycho.c
@@ -366,8 +366,8 @@ static void pbm_config_busmastering(struct pci_pbm_info *pbm)
 	pci_config_write8(addr, 64);
 }
 
-static void __devinit psycho_scan_bus(struct pci_pbm_info *pbm,
-				      struct device *parent)
+static void psycho_scan_bus(struct pci_pbm_info *pbm,
+			    struct device *parent)
 {
 	pbm_config_busmastering(pbm);
 	pbm->is_66mhz_capable = 0;
@@ -483,15 +483,15 @@ static void psycho_pbm_strbuf_init(struct pci_pbm_info *pbm,
 #define PSYCHO_MEMSPACE_B	0x180000000UL
 #define PSYCHO_MEMSPACE_SIZE	0x07fffffffUL
 
-static void __devinit psycho_pbm_init(struct pci_pbm_info *pbm,
-				      struct platform_device *op, int is_pbm_a)
+static void psycho_pbm_init(struct pci_pbm_info *pbm,
+			    struct platform_device *op, int is_pbm_a)
 {
 	psycho_pbm_init_common(pbm, op, "PSYCHO", PBM_CHIP_TYPE_PSYCHO);
 	psycho_pbm_strbuf_init(pbm, is_pbm_a);
 	psycho_scan_bus(pbm, &op->dev);
 }
 
-static struct pci_pbm_info * __devinit psycho_find_sibling(u32 upa_portid)
+static struct pci_pbm_info *psycho_find_sibling(u32 upa_portid)
 {
 	struct pci_pbm_info *pbm;
 
@@ -504,7 +504,7 @@ static struct pci_pbm_info * __devinit psycho_find_sibling(u32 upa_portid)
 
 #define PSYCHO_CONFIGSPACE	0x001000000UL
 
-static int __devinit psycho_probe(struct platform_device *op)
+static int psycho_probe(struct platform_device *op)
 {
 	const struct linux_prom64_registers *pr_regs;
 	struct device_node *dp = op->dev.of_node;
diff --git a/arch/sparc/kernel/pci_sabre.c b/arch/sparc/kernel/pci_sabre.c
index 3efaa4644d60..6f00d27e8dac 100644
--- a/arch/sparc/kernel/pci_sabre.c
+++ b/arch/sparc/kernel/pci_sabre.c
@@ -403,8 +403,7 @@ static void apb_init(struct pci_bus *sabre_bus)
 	}
 }
 
-static void __devinit sabre_scan_bus(struct pci_pbm_info *pbm,
-				     struct device *parent)
+static void sabre_scan_bus(struct pci_pbm_info *pbm, struct device *parent)
 {
 	static int once;
 
@@ -443,8 +442,8 @@ static void __devinit sabre_scan_bus(struct pci_pbm_info *pbm,
 	sabre_register_error_handlers(pbm);
 }
 
-static void __devinit sabre_pbm_init(struct pci_pbm_info *pbm,
-				     struct platform_device *op)
+static void sabre_pbm_init(struct pci_pbm_info *pbm,
+			   struct platform_device *op)
 {
 	psycho_pbm_init_common(pbm, op, "SABRE", PBM_CHIP_TYPE_SABRE);
 	pbm->pci_afsr = pbm->controller_regs + SABRE_PIOAFSR;
@@ -454,7 +453,7 @@ static void __devinit sabre_pbm_init(struct pci_pbm_info *pbm,
 }
 
 static const struct of_device_id sabre_match[];
-static int __devinit sabre_probe(struct platform_device *op)
+static int sabre_probe(struct platform_device *op)
 {
 	const struct of_device_id *match;
 	const struct linux_prom64_registers *pr_regs;
diff --git a/arch/sparc/kernel/pci_schizo.c b/arch/sparc/kernel/pci_schizo.c
index 13d4aa20b5a5..8f76f23dac38 100644
--- a/arch/sparc/kernel/pci_schizo.c
+++ b/arch/sparc/kernel/pci_schizo.c
@@ -1064,8 +1064,7 @@ static void pbm_config_busmastering(struct pci_pbm_info *pbm)
 	pci_config_write8(addr, 64);
 }
 
-static void __devinit schizo_scan_bus(struct pci_pbm_info *pbm,
-				      struct device *parent)
+static void schizo_scan_bus(struct pci_pbm_info *pbm, struct device *parent)
 {
 	pbm_config_busmastering(pbm);
 	pbm->is_66mhz_capable =
@@ -1307,9 +1306,9 @@ static void schizo_pbm_hw_init(struct pci_pbm_info *pbm)
 	}
 }
 
-static int __devinit schizo_pbm_init(struct pci_pbm_info *pbm,
-				     struct platform_device *op, u32 portid,
-				     int chip_type)
+static int schizo_pbm_init(struct pci_pbm_info *pbm,
+			   struct platform_device *op, u32 portid,
+			   int chip_type)
 {
 	const struct linux_prom64_registers *regs;
 	struct device_node *dp = op->dev.of_node;
@@ -1400,8 +1399,7 @@ static inline int portid_compare(u32 x, u32 y, int chip_type)
 	return (x == y);
 }
 
-static struct pci_pbm_info * __devinit schizo_find_sibling(u32 portid,
-							   int chip_type)
+static struct pci_pbm_info *schizo_find_sibling(u32 portid, int chip_type)
 {
 	struct pci_pbm_info *pbm;
 
@@ -1412,7 +1410,7 @@ static struct pci_pbm_info * __devinit schizo_find_sibling(u32 portid,
 	return NULL;
 }
 
-static int __devinit __schizo_init(struct platform_device *op, unsigned long chip_type)
+static int __schizo_init(struct platform_device *op, unsigned long chip_type)
 {
 	struct device_node *dp = op->dev.of_node;
 	struct pci_pbm_info *pbm;
@@ -1460,7 +1458,7 @@ out_err:
 }
 
 static const struct of_device_id schizo_match[];
-static int __devinit schizo_probe(struct platform_device *op)
+static int schizo_probe(struct platform_device *op)
 {
 	const struct of_device_id *match;
 
diff --git a/arch/sparc/kernel/pci_sun4v.c b/arch/sparc/kernel/pci_sun4v.c
index 051b69caeffd..d07f6b29aed8 100644
--- a/arch/sparc/kernel/pci_sun4v.c
+++ b/arch/sparc/kernel/pci_sun4v.c
@@ -536,8 +536,7 @@ static struct dma_map_ops sun4v_dma_ops = {
 	.unmap_sg			= dma_4v_unmap_sg,
 };
 
-static void __devinit pci_sun4v_scan_bus(struct pci_pbm_info *pbm,
-					 struct device *parent)
+static void pci_sun4v_scan_bus(struct pci_pbm_info *pbm, struct device *parent)
 {
 	struct property *prop;
 	struct device_node *dp;
@@ -550,8 +549,8 @@ static void __devinit pci_sun4v_scan_bus(struct pci_pbm_info *pbm,
 	/* XXX register error interrupt handlers XXX */
 }
 
-static unsigned long __devinit probe_existing_entries(struct pci_pbm_info *pbm,
-						      struct iommu *iommu)
+static unsigned long probe_existing_entries(struct pci_pbm_info *pbm,
+					    struct iommu *iommu)
 {
 	struct iommu_arena *arena = &iommu->arena;
 	unsigned long i, cnt = 0;
@@ -578,7 +577,7 @@ static unsigned long __devinit probe_existing_entries(struct pci_pbm_info *pbm,
 	return cnt;
 }
 
-static int __devinit pci_sun4v_iommu_init(struct pci_pbm_info *pbm)
+static int pci_sun4v_iommu_init(struct pci_pbm_info *pbm)
 {
 	static const u32 vdma_default[] = { 0x80000000, 0x80000000 };
 	struct iommu *iommu = pbm->iommu;
@@ -879,8 +878,8 @@ static void pci_sun4v_msi_init(struct pci_pbm_info *pbm)
 }
 #endif /* !(CONFIG_PCI_MSI) */
 
-static int __devinit pci_sun4v_pbm_init(struct pci_pbm_info *pbm,
-					struct platform_device *op, u32 devhandle)
+static int pci_sun4v_pbm_init(struct pci_pbm_info *pbm,
+			      struct platform_device *op, u32 devhandle)
 {
 	struct device_node *dp = op->dev.of_node;
 	int err;
@@ -919,7 +918,7 @@ static int __devinit pci_sun4v_pbm_init(struct pci_pbm_info *pbm,
 	return 0;
 }
 
-static int __devinit pci_sun4v_probe(struct platform_device *op)
+static int pci_sun4v_probe(struct platform_device *op)
 {
 	const struct linux_prom64_registers *regs;
 	static int hvapi_negotiated = 0;
diff --git a/arch/sparc/kernel/pcic.c b/arch/sparc/kernel/pcic.c
index 521fdf1b20e5..09f4fdd8d808 100644
--- a/arch/sparc/kernel/pcic.c
+++ b/arch/sparc/kernel/pcic.c
@@ -439,8 +439,7 @@ int pcic_present(void)
 	return pcic0_up;
 }
 
-static int __devinit pdev_to_pnode(struct linux_pbm_info *pbm,
-				    struct pci_dev *pdev)
+static int pdev_to_pnode(struct linux_pbm_info *pbm, struct pci_dev *pdev)
 {
 	struct linux_prom_pci_registers regs[PROMREG_MAX];
 	int err;
@@ -595,7 +594,7 @@ pcic_fill_irq(struct linux_pcic *pcic, struct pci_dev *dev, int node)
 /*
  * Normally called from {do_}pci_scan_bus...
  */
-void __devinit pcibios_fixup_bus(struct pci_bus *bus)
+void pcibios_fixup_bus(struct pci_bus *bus)
 {
 	struct pci_dev *dev;
 	int i, has_io, has_mem;
diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c
index 885a8af74064..b5c38faa4ead 100644
--- a/arch/sparc/kernel/perf_event.c
+++ b/arch/sparc/kernel/perf_event.c
@@ -1762,15 +1762,25 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry,
 
 	ufp = regs->u_regs[UREG_I6] & 0xffffffffUL;
 	do {
-		struct sparc_stackf32 *usf, sf;
 		unsigned long pc;
 
-		usf = (struct sparc_stackf32 *) ufp;
-		if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
-			break;
+		if (thread32_stack_is_64bit(ufp)) {
+			struct sparc_stackf *usf, sf;
 
-		pc = sf.callers_pc;
-		ufp = (unsigned long)sf.fp;
+			ufp += STACK_BIAS;
+			usf = (struct sparc_stackf *) ufp;
+			if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
+				break;
+			pc = sf.callers_pc & 0xffffffff;
+			ufp = ((unsigned long) sf.fp) & 0xffffffff;
+		} else {
+			struct sparc_stackf32 *usf, sf;
+			usf = (struct sparc_stackf32 *) ufp;
+			if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
+				break;
+			pc = sf.callers_pc;
+			ufp = (unsigned long)sf.fp;
+		}
 		perf_callchain_store(entry, pc);
 	} while (entry->nr < PERF_MAX_STACK_DEPTH);
 }
diff --git a/arch/sparc/kernel/pmc.c b/arch/sparc/kernel/pmc.c
index 0e3202239ff5..8b7297faca79 100644
--- a/arch/sparc/kernel/pmc.c
+++ b/arch/sparc/kernel/pmc.c
@@ -17,6 +17,7 @@
 #include <asm/oplib.h>
 #include <asm/uaccess.h>
 #include <asm/auxio.h>
+#include <asm/processor.h>
 
 /* Debug
  *
@@ -52,7 +53,7 @@ static void pmc_swift_idle(void)
 #endif
 }
 
-static int __devinit pmc_probe(struct platform_device *op)
+static int pmc_probe(struct platform_device *op)
 {
 	regs = of_ioremap(&op->resource[0], 0,
 			  resource_size(&op->resource[0]), PMC_OBPNAME);
@@ -63,7 +64,7 @@ static int __devinit pmc_probe(struct platform_device *op)
 
 #ifndef PMC_NO_IDLE
 	/* Assign power management IDLE handler */
-	pm_idle = pmc_swift_idle;
+	sparc_idle = pmc_swift_idle;
 #endif
 
 	printk(KERN_INFO "%s: power management initialized\n", PMC_DEVNAME);
diff --git a/arch/sparc/kernel/power.c b/arch/sparc/kernel/power.c
index 0d39075063b2..4cb23c41553f 100644
--- a/arch/sparc/kernel/power.c
+++ b/arch/sparc/kernel/power.c
@@ -23,7 +23,7 @@ static irqreturn_t power_handler(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
-static int __devinit has_button_interrupt(unsigned int irq, struct device_node *dp)
+static int has_button_interrupt(unsigned int irq, struct device_node *dp)
 {
 	if (irq == 0xffffffff)
 		return 0;
@@ -33,7 +33,7 @@ static int __devinit has_button_interrupt(unsigned int irq, struct device_node *
 	return 1;
 }
 
-static int __devinit power_probe(struct platform_device *op)
+static int power_probe(struct platform_device *op)
 {
 	struct resource *res = &op->resource[0];
 	unsigned int irq = op->archdata.irqs[0];
diff --git a/arch/sparc/kernel/process_32.c b/arch/sparc/kernel/process_32.c
index 487bffb36f5e..62eede13831a 100644
--- a/arch/sparc/kernel/process_32.c
+++ b/arch/sparc/kernel/process_32.c
@@ -43,8 +43,7 @@
  * Power management idle function 
  * Set in pm platform drivers (apc.c and pmc.c)
  */
-void (*pm_idle)(void);
-EXPORT_SYMBOL(pm_idle);
+void (*sparc_idle)(void);
 
 /* 
  * Power-off handler instantiation for pm.h compliance
@@ -75,8 +74,8 @@ void cpu_idle(void)
 	/* endless idle loop with no priority at all */
 	for (;;) {
 		while (!need_resched()) {
-			if (pm_idle)
-				(*pm_idle)();
+			if (sparc_idle)
+				(*sparc_idle)();
 			else
 				cpu_relax();
 		}
@@ -286,8 +285,7 @@ asmlinkage int sparc_do_fork(unsigned long clone_flags,
 	parent_tid_ptr = regs->u_regs[UREG_I2];
 	child_tid_ptr = regs->u_regs[UREG_I4];
 
-	ret = do_fork(clone_flags, stack_start,
-		      regs, stack_size,
+	ret = do_fork(clone_flags, stack_start, stack_size,
 		      (int __user *) parent_tid_ptr,
 		      (int __user *) child_tid_ptr);
 
@@ -316,13 +314,13 @@ asmlinkage int sparc_do_fork(unsigned long clone_flags,
  * XXX See comment above sys_vfork in sparc64. todo.
  */
 extern void ret_from_fork(void);
+extern void ret_from_kernel_thread(void);
 
 int copy_thread(unsigned long clone_flags, unsigned long sp,
-		unsigned long unused,
-		struct task_struct *p, struct pt_regs *regs)
+		unsigned long arg, struct task_struct *p)
 {
 	struct thread_info *ti = task_thread_info(p);
-	struct pt_regs *childregs;
+	struct pt_regs *childregs, *regs = current_pt_regs();
 	char *new_stack;
 
 #ifndef CONFIG_SMP
@@ -336,16 +334,13 @@ int copy_thread(unsigned long clone_flags, unsigned long sp,
 	}
 
 	/*
-	 *  p->thread_info         new_stack   childregs
-	 *  !                      !           !             {if(PSR_PS) }
-	 *  V                      V (stk.fr.) V  (pt_regs)  { (stk.fr.) }
-	 *  +----- - - - - - ------+===========+============={+==========}+
+	 *  p->thread_info         new_stack   childregs stack bottom
+	 *  !                      !           !             !
+	 *  V                      V (stk.fr.) V  (pt_regs)  V
+	 *  +----- - - - - - ------+===========+=============+
 	 */
 	new_stack = task_stack_page(p) + THREAD_SIZE;
-	if (regs->psr & PSR_PS)
-		new_stack -= STACKFRAME_SZ;
 	new_stack -= STACKFRAME_SZ + TRACEREG_SZ;
-	memcpy(new_stack, (char *)regs - STACKFRAME_SZ, STACKFRAME_SZ + TRACEREG_SZ);
 	childregs = (struct pt_regs *) (new_stack + STACKFRAME_SZ);
 
 	/*
@@ -356,55 +351,58 @@ int copy_thread(unsigned long clone_flags, unsigned long sp,
 	 * Thus, kpsr|=PSR_PIL.
 	 */
 	ti->ksp = (unsigned long) new_stack;
+	p->thread.kregs = childregs;
+
+	if (unlikely(p->flags & PF_KTHREAD)) {
+		extern int nwindows;
+		unsigned long psr;
+		memset(new_stack, 0, STACKFRAME_SZ + TRACEREG_SZ);
+		p->thread.flags |= SPARC_FLAG_KTHREAD;
+		p->thread.current_ds = KERNEL_DS;
+		ti->kpc = (((unsigned long) ret_from_kernel_thread) - 0x8);
+		childregs->u_regs[UREG_G1] = sp; /* function */
+		childregs->u_regs[UREG_G2] = arg;
+		psr = childregs->psr = get_psr();
+		ti->kpsr = psr | PSR_PIL;
+		ti->kwim = 1 << (((psr & PSR_CWP) + 1) % nwindows);
+		return 0;
+	}
+	memcpy(new_stack, (char *)regs - STACKFRAME_SZ, STACKFRAME_SZ + TRACEREG_SZ);
+	childregs->u_regs[UREG_FP] = sp;
+	p->thread.flags &= ~SPARC_FLAG_KTHREAD;
+	p->thread.current_ds = USER_DS;
 	ti->kpc = (((unsigned long) ret_from_fork) - 0x8);
 	ti->kpsr = current->thread.fork_kpsr | PSR_PIL;
 	ti->kwim = current->thread.fork_kwim;
 
-	if(regs->psr & PSR_PS) {
-		extern struct pt_regs fake_swapper_regs;
+	if (sp != regs->u_regs[UREG_FP]) {
+		struct sparc_stackf __user *childstack;
+		struct sparc_stackf __user *parentstack;
 
-		p->thread.kregs = &fake_swapper_regs;
-		new_stack += STACKFRAME_SZ + TRACEREG_SZ;
-		childregs->u_regs[UREG_FP] = (unsigned long) new_stack;
-		p->thread.flags |= SPARC_FLAG_KTHREAD;
-		p->thread.current_ds = KERNEL_DS;
-		memcpy(new_stack, (void *)regs->u_regs[UREG_FP], STACKFRAME_SZ);
-		childregs->u_regs[UREG_G6] = (unsigned long) ti;
-	} else {
-		p->thread.kregs = childregs;
-		childregs->u_regs[UREG_FP] = sp;
-		p->thread.flags &= ~SPARC_FLAG_KTHREAD;
-		p->thread.current_ds = USER_DS;
-
-		if (sp != regs->u_regs[UREG_FP]) {
-			struct sparc_stackf __user *childstack;
-			struct sparc_stackf __user *parentstack;
-
-			/*
-			 * This is a clone() call with supplied user stack.
-			 * Set some valid stack frames to give to the child.
-			 */
-			childstack = (struct sparc_stackf __user *)
-				(sp & ~0xfUL);
-			parentstack = (struct sparc_stackf __user *)
-				regs->u_regs[UREG_FP];
+		/*
+		 * This is a clone() call with supplied user stack.
+		 * Set some valid stack frames to give to the child.
+		 */
+		childstack = (struct sparc_stackf __user *)
+			(sp & ~0xfUL);
+		parentstack = (struct sparc_stackf __user *)
+			regs->u_regs[UREG_FP];
 
 #if 0
-			printk("clone: parent stack:\n");
-			show_stackframe(parentstack);
+		printk("clone: parent stack:\n");
+		show_stackframe(parentstack);
 #endif
 
-			childstack = clone_stackframe(childstack, parentstack);
-			if (!childstack)
-				return -EFAULT;
+		childstack = clone_stackframe(childstack, parentstack);
+		if (!childstack)
+			return -EFAULT;
 
 #if 0
-			printk("clone: child stack:\n");
-			show_stackframe(childstack);
+		printk("clone: child stack:\n");
+		show_stackframe(childstack);
 #endif
 
-			childregs->u_regs[UREG_FP] = (unsigned long)childstack;
-		}
+		childregs->u_regs[UREG_FP] = (unsigned long)childstack;
 	}
 
 #ifdef CONFIG_SMP
@@ -475,69 +473,6 @@ int dump_fpu (struct pt_regs * regs, elf_fpregset_t * fpregs)
 	return 1;
 }
 
-/*
- * sparc_execve() executes a new program after the asm stub has set
- * things up for us.  This should basically do what I want it to.
- */
-asmlinkage int sparc_execve(struct pt_regs *regs)
-{
-	int error, base = 0;
-	struct filename *filename;
-
-	/* Check for indirect call. */
-	if(regs->u_regs[UREG_G1] == 0)
-		base = 1;
-
-	filename = getname((char __user *)regs->u_regs[base + UREG_I0]);
-	error = PTR_ERR(filename);
-	if(IS_ERR(filename))
-		goto out;
-	error = do_execve(filename->name,
-			  (const char __user *const  __user *)
-			  regs->u_regs[base + UREG_I1],
-			  (const char __user *const  __user *)
-			  regs->u_regs[base + UREG_I2],
-			  regs);
-	putname(filename);
-out:
-	return error;
-}
-
-/*
- * This is the mechanism for creating a new kernel thread.
- *
- * NOTE! Only a kernel-only process(ie the swapper or direct descendants
- * who haven't done an "execve()") should use this: it will work within
- * a system call from a "real" process, but the process memory space will
- * not be freed until both the parent and the child have exited.
- */
-pid_t kernel_thread(int (*fn)(void *), void * arg, unsigned long flags)
-{
-	long retval;
-
-	__asm__ __volatile__("mov %4, %%g2\n\t"    /* Set aside fn ptr... */
-			     "mov %5, %%g3\n\t"    /* and arg. */
-			     "mov %1, %%g1\n\t"
-			     "mov %2, %%o0\n\t"    /* Clone flags. */
-			     "mov 0, %%o1\n\t"     /* usp arg == 0 */
-			     "t 0x10\n\t"          /* Linux/Sparc clone(). */
-			     "cmp %%o1, 0\n\t"
-			     "be 1f\n\t"           /* The parent, just return. */
-			     " nop\n\t"            /* Delay slot. */
-			     "jmpl %%g2, %%o7\n\t" /* Call the function. */
-			     " mov %%g3, %%o0\n\t" /* Get back the arg in delay. */
-			     "mov %3, %%g1\n\t"
-			     "t 0x10\n\t"          /* Linux/Sparc exit(). */
-			     /* Notreached by child. */
-			     "1: mov %%o0, %0\n\t" :
-			     "=r" (retval) :
-			     "i" (__NR_clone), "r" (flags | CLONE_VM | CLONE_UNTRACED),
-			     "i" (__NR_exit),  "r" (fn), "r" (arg) :
-			     "g1", "g2", "g3", "o0", "o1", "memory", "cc");
-	return retval;
-}
-EXPORT_SYMBOL(kernel_thread);
-
 unsigned long get_wchan(struct task_struct *task)
 {
 	unsigned long pc, fp, bias = 0;
diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c
index d778248ef3f8..cdb80b2adbe0 100644
--- a/arch/sparc/kernel/process_64.c
+++ b/arch/sparc/kernel/process_64.c
@@ -452,13 +452,16 @@ void flush_thread(void)
 /* It's a bit more tricky when 64-bit tasks are involved... */
 static unsigned long clone_stackframe(unsigned long csp, unsigned long psp)
 {
+	bool stack_64bit = test_thread_64bit_stack(psp);
 	unsigned long fp, distance, rval;
 
-	if (!(test_thread_flag(TIF_32BIT))) {
+	if (stack_64bit) {
 		csp += STACK_BIAS;
 		psp += STACK_BIAS;
 		__get_user(fp, &(((struct reg_window __user *)psp)->ins[6]));
 		fp += STACK_BIAS;
+		if (test_thread_flag(TIF_32BIT))
+			fp &= 0xffffffff;
 	} else
 		__get_user(fp, &(((struct reg_window32 __user *)psp)->ins[6]));
 
@@ -472,7 +475,7 @@ static unsigned long clone_stackframe(unsigned long csp, unsigned long psp)
 	rval = (csp - distance);
 	if (copy_in_user((void __user *) rval, (void __user *) psp, distance))
 		rval = 0;
-	else if (test_thread_flag(TIF_32BIT)) {
+	else if (!stack_64bit) {
 		if (put_user(((u32)csp),
 			     &(((struct reg_window32 __user *)rval)->ins[6])))
 			rval = 0;
@@ -507,18 +510,18 @@ void synchronize_user_stack(void)
 
 	flush_user_windows();
 	if ((window = get_thread_wsaved()) != 0) {
-		int winsize = sizeof(struct reg_window);
-		int bias = 0;
-
-		if (test_thread_flag(TIF_32BIT))
-			winsize = sizeof(struct reg_window32);
-		else
-			bias = STACK_BIAS;
-
 		window -= 1;
 		do {
-			unsigned long sp = (t->rwbuf_stkptrs[window] + bias);
 			struct reg_window *rwin = &t->reg_window[window];
+			int winsize = sizeof(struct reg_window);
+			unsigned long sp;
+
+			sp = t->rwbuf_stkptrs[window];
+
+			if (test_thread_64bit_stack(sp))
+				sp += STACK_BIAS;
+			else
+				winsize = sizeof(struct reg_window32);
 
 			if (!copy_to_user((char __user *)sp, rwin, winsize)) {
 				shift_window_buffer(window, get_thread_wsaved() - 1, t);
@@ -544,13 +547,6 @@ void fault_in_user_windows(void)
 {
 	struct thread_info *t = current_thread_info();
 	unsigned long window;
-	int winsize = sizeof(struct reg_window);
-	int bias = 0;
-
-	if (test_thread_flag(TIF_32BIT))
-		winsize = sizeof(struct reg_window32);
-	else
-		bias = STACK_BIAS;
 
 	flush_user_windows();
 	window = get_thread_wsaved();
@@ -558,8 +554,16 @@ void fault_in_user_windows(void)
 	if (likely(window != 0)) {
 		window -= 1;
 		do {
-			unsigned long sp = (t->rwbuf_stkptrs[window] + bias);
 			struct reg_window *rwin = &t->reg_window[window];
+			int winsize = sizeof(struct reg_window);
+			unsigned long sp;
+
+			sp = t->rwbuf_stkptrs[window];
+
+			if (test_thread_64bit_stack(sp))
+				sp += STACK_BIAS;
+			else
+				winsize = sizeof(struct reg_window32);
 
 			if (unlikely(sp & 0x7UL))
 				stack_unaligned(sp);
@@ -597,8 +601,7 @@ asmlinkage long sparc_do_fork(unsigned long clone_flags,
 		child_tid_ptr = (int __user *) regs->u_regs[UREG_I4];
 	}
 
-	ret = do_fork(clone_flags, stack_start,
-		      regs, stack_size,
+	ret = do_fork(clone_flags, stack_start, stack_size,
 		      parent_tid_ptr, child_tid_ptr);
 
 	/* If we get an error and potentially restart the system
@@ -618,64 +621,55 @@ asmlinkage long sparc_do_fork(unsigned long clone_flags,
  * Child  -->  %o0 == parents pid, %o1 == 1
  */
 int copy_thread(unsigned long clone_flags, unsigned long sp,
-		unsigned long unused,
-		struct task_struct *p, struct pt_regs *regs)
+		unsigned long arg, struct task_struct *p)
 {
 	struct thread_info *t = task_thread_info(p);
+	struct pt_regs *regs = current_pt_regs();
 	struct sparc_stackf *parent_sf;
 	unsigned long child_stack_sz;
 	char *child_trap_frame;
-	int kernel_thread;
-
-	kernel_thread = (regs->tstate & TSTATE_PRIV) ? 1 : 0;
-	parent_sf = ((struct sparc_stackf *) regs) - 1;
 
 	/* Calculate offset to stack_frame & pt_regs */
-	child_stack_sz = ((STACKFRAME_SZ + TRACEREG_SZ) +
-			  (kernel_thread ? STACKFRAME_SZ : 0));
+	child_stack_sz = (STACKFRAME_SZ + TRACEREG_SZ);
 	child_trap_frame = (task_stack_page(p) +
 			    (THREAD_SIZE - child_stack_sz));
-	memcpy(child_trap_frame, parent_sf, child_stack_sz);
 
-	t->flags = (t->flags & ~((0xffUL << TI_FLAG_CWP_SHIFT) |
-				 (0xffUL << TI_FLAG_CURRENT_DS_SHIFT))) |
-		(((regs->tstate + 1) & TSTATE_CWP) << TI_FLAG_CWP_SHIFT);
 	t->new_child = 1;
 	t->ksp = ((unsigned long) child_trap_frame) - STACK_BIAS;
 	t->kregs = (struct pt_regs *) (child_trap_frame +
 				       sizeof(struct sparc_stackf));
 	t->fpsaved[0] = 0;
 
-	if (kernel_thread) {
-		struct sparc_stackf *child_sf = (struct sparc_stackf *)
-			(child_trap_frame + (STACKFRAME_SZ + TRACEREG_SZ));
-
-		/* Zero terminate the stack backtrace.  */
-		child_sf->fp = NULL;
-		t->kregs->u_regs[UREG_FP] =
-		  ((unsigned long) child_sf) - STACK_BIAS;
+	if (unlikely(p->flags & PF_KTHREAD)) {
+		memset(child_trap_frame, 0, child_stack_sz);
+		__thread_flag_byte_ptr(t)[TI_FLAG_BYTE_CWP] = 
+			(current_pt_regs()->tstate + 1) & TSTATE_CWP;
+		t->current_ds = ASI_P;
+		t->kregs->u_regs[UREG_G1] = sp; /* function */
+		t->kregs->u_regs[UREG_G2] = arg;
+		return 0;
+	}
 
-		t->flags |= ((long)ASI_P << TI_FLAG_CURRENT_DS_SHIFT);
-		t->kregs->u_regs[UREG_G6] = (unsigned long) t;
-		t->kregs->u_regs[UREG_G4] = (unsigned long) t->task;
-	} else {
-		if (t->flags & _TIF_32BIT) {
-			sp &= 0x00000000ffffffffUL;
-			regs->u_regs[UREG_FP] &= 0x00000000ffffffffUL;
-		}
-		t->kregs->u_regs[UREG_FP] = sp;
-		t->flags |= ((long)ASI_AIUS << TI_FLAG_CURRENT_DS_SHIFT);
-		if (sp != regs->u_regs[UREG_FP]) {
-			unsigned long csp;
-
-			csp = clone_stackframe(sp, regs->u_regs[UREG_FP]);
-			if (!csp)
-				return -EFAULT;
-			t->kregs->u_regs[UREG_FP] = csp;
-		}
-		if (t->utraps)
-			t->utraps[0]++;
+	parent_sf = ((struct sparc_stackf *) regs) - 1;
+	memcpy(child_trap_frame, parent_sf, child_stack_sz);
+	if (t->flags & _TIF_32BIT) {
+		sp &= 0x00000000ffffffffUL;
+		regs->u_regs[UREG_FP] &= 0x00000000ffffffffUL;
 	}
+	t->kregs->u_regs[UREG_FP] = sp;
+	__thread_flag_byte_ptr(t)[TI_FLAG_BYTE_CWP] = 
+		(regs->tstate + 1) & TSTATE_CWP;
+	t->current_ds = ASI_AIUS;
+	if (sp != regs->u_regs[UREG_FP]) {
+		unsigned long csp;
+
+		csp = clone_stackframe(sp, regs->u_regs[UREG_FP]);
+		if (!csp)
+			return -EFAULT;
+		t->kregs->u_regs[UREG_FP] = csp;
+	}
+	if (t->utraps)
+		t->utraps[0]++;
 
 	/* Set the return value for the child. */
 	t->kregs->u_regs[UREG_I0] = current->pid;
@@ -690,45 +684,6 @@ int copy_thread(unsigned long clone_flags, unsigned long sp,
 	return 0;
 }
 
-/*
- * This is the mechanism for creating a new kernel thread.
- *
- * NOTE! Only a kernel-only process(ie the swapper or direct descendants
- * who haven't done an "execve()") should use this: it will work within
- * a system call from a "real" process, but the process memory space will
- * not be freed until both the parent and the child have exited.
- */
-pid_t kernel_thread(int (*fn)(void *), void * arg, unsigned long flags)
-{
-	long retval;
-
-	/* If the parent runs before fn(arg) is called by the child,
-	 * the input registers of this function can be clobbered.
-	 * So we stash 'fn' and 'arg' into global registers which
-	 * will not be modified by the parent.
-	 */
-	__asm__ __volatile__("mov %4, %%g2\n\t"	   /* Save FN into global */
-			     "mov %5, %%g3\n\t"	   /* Save ARG into global */
-			     "mov %1, %%g1\n\t"	   /* Clone syscall nr. */
-			     "mov %2, %%o0\n\t"	   /* Clone flags. */
-			     "mov 0, %%o1\n\t"	   /* usp arg == 0 */
-			     "t 0x6d\n\t"	   /* Linux/Sparc clone(). */
-			     "brz,a,pn %%o1, 1f\n\t" /* Parent, just return. */
-			     " mov %%o0, %0\n\t"
-			     "jmpl %%g2, %%o7\n\t"   /* Call the function. */
-			     " mov %%g3, %%o0\n\t"   /* Set arg in delay. */
-			     "mov %3, %%g1\n\t"
-			     "t 0x6d\n\t"	   /* Linux/Sparc exit(). */
-			     /* Notreached by child. */
-			     "1:" :
-			     "=r" (retval) :
-			     "i" (__NR_clone), "r" (flags | CLONE_VM | CLONE_UNTRACED),
-			     "i" (__NR_exit),  "r" (fn), "r" (arg) :
-			     "g1", "g2", "g3", "o0", "o1", "memory", "cc");
-	return retval;
-}
-EXPORT_SYMBOL(kernel_thread);
-
 typedef struct {
 	union {
 		unsigned int	pr_regs[32];
@@ -795,41 +750,6 @@ int dump_fpu (struct pt_regs * regs, elf_fpregset_t * fpregs)
 }
 EXPORT_SYMBOL(dump_fpu);
 
-/*
- * sparc_execve() executes a new program after the asm stub has set
- * things up for us.  This should basically do what I want it to.
- */
-asmlinkage int sparc_execve(struct pt_regs *regs)
-{
-	int error, base = 0;
-	struct filename *filename;
-
-	/* User register window flush is done by entry.S */
-
-	/* Check for indirect call. */
-	if (regs->u_regs[UREG_G1] == 0)
-		base = 1;
-
-	filename = getname((char __user *)regs->u_regs[base + UREG_I0]);
-	error = PTR_ERR(filename);
-	if (IS_ERR(filename))
-		goto out;
-	error = do_execve(filename->name,
-			  (const char __user *const __user *)
-			  regs->u_regs[base + UREG_I1],
-			  (const char __user *const __user *)
-			  regs->u_regs[base + UREG_I2], regs);
-	putname(filename);
-	if (!error) {
-		fprs_write(0);
-		current_thread_info()->xfsr[0] = 0;
-		current_thread_info()->fpsaved[0] = 0;
-		regs->tstate &= ~TSTATE_PEF;
-	}
-out:
-	return error;
-}
-
 unsigned long get_wchan(struct task_struct *task)
 {
 	unsigned long pc, fp, bias = 0;
diff --git a/arch/sparc/kernel/prom_common.c b/arch/sparc/kernel/prom_common.c
index 1303021748c8..9f20566b0773 100644
--- a/arch/sparc/kernel/prom_common.c
+++ b/arch/sparc/kernel/prom_common.c
@@ -64,7 +64,7 @@ int of_set_property(struct device_node *dp, const char *name, void *val, int len
 	err = -ENODEV;
 
 	mutex_lock(&of_set_property_mutex);
-	write_lock(&devtree_lock);
+	raw_spin_lock(&devtree_lock);
 	prevp = &dp->properties;
 	while (*prevp) {
 		struct property *prop = *prevp;
@@ -91,7 +91,7 @@ int of_set_property(struct device_node *dp, const char *name, void *val, int len
 		}
 		prevp = &(*prevp)->next;
 	}
-	write_unlock(&devtree_lock);
+	raw_spin_unlock(&devtree_lock);
 	mutex_unlock(&of_set_property_mutex);
 
 	/* XXX Upate procfs if necessary... */
diff --git a/arch/sparc/kernel/ptrace_64.c b/arch/sparc/kernel/ptrace_64.c
index 484dabac7045..7ff45e4ba681 100644
--- a/arch/sparc/kernel/ptrace_64.c
+++ b/arch/sparc/kernel/ptrace_64.c
@@ -151,7 +151,7 @@ static int regwindow64_get(struct task_struct *target,
 {
 	unsigned long rw_addr = regs->u_regs[UREG_I6];
 
-	if (test_tsk_thread_flag(current, TIF_32BIT)) {
+	if (!test_thread_64bit_stack(rw_addr)) {
 		struct reg_window32 win32;
 		int i;
 
@@ -176,7 +176,7 @@ static int regwindow64_set(struct task_struct *target,
 {
 	unsigned long rw_addr = regs->u_regs[UREG_I6];
 
-	if (test_tsk_thread_flag(current, TIF_32BIT)) {
+	if (!test_thread_64bit_stack(rw_addr)) {
 		struct reg_window32 win32;
 		int i;
 
diff --git a/arch/sparc/kernel/sbus.c b/arch/sparc/kernel/sbus.c
index 1271b3a27d4e..be5bdf93c767 100644
--- a/arch/sparc/kernel/sbus.c
+++ b/arch/sparc/kernel/sbus.c
@@ -554,10 +554,8 @@ static void __init sbus_iommu_init(struct platform_device *op)
 	regs = pr->phys_addr;
 
 	iommu = kzalloc(sizeof(*iommu), GFP_ATOMIC);
-	if (!iommu)
-		goto fatal_memory_error;
 	strbuf = kzalloc(sizeof(*strbuf), GFP_ATOMIC);
-	if (!strbuf)
+	if (!iommu || !strbuf)
 		goto fatal_memory_error;
 
 	op->dev.archdata.iommu = iommu;
@@ -656,6 +654,8 @@ static void __init sbus_iommu_init(struct platform_device *op)
 	return;
 
 fatal_memory_error:
+	kfree(iommu);
+	kfree(strbuf);
 	prom_printf("sbus_iommu_init: Fatal memory allocation error.\n");
 }
 
diff --git a/arch/sparc/kernel/setup_64.c b/arch/sparc/kernel/setup_64.c
index 0800e71d8a88..88a127b9c69e 100644
--- a/arch/sparc/kernel/setup_64.c
+++ b/arch/sparc/kernel/setup_64.c
@@ -115,7 +115,7 @@ static void __init process_switch(char c)
 			break;
 		}
 		cheetah_pcache_forced_on = 1;
-		add_taint(TAINT_MACHINE_CHECK);
+		add_taint(TAINT_MACHINE_CHECK, LOCKDEP_NOW_UNRELIABLE);
 		cheetah_enable_pcache();
 		break;
 
@@ -316,6 +316,25 @@ static void __init popc_patch(void)
 	}
 }
 
+static void __init pause_patch(void)
+{
+	struct pause_patch_entry *p;
+
+	p = &__pause_3insn_patch;
+	while (p < &__pause_3insn_patch_end) {
+		unsigned long i, addr = p->addr;
+
+		for (i = 0; i < 3; i++) {
+			*(unsigned int *) (addr +  (i * 4)) = p->insns[i];
+			wmb();
+			__asm__ __volatile__("flush	%0"
+					     : : "r" (addr +  (i * 4)));
+		}
+
+		p++;
+	}
+}
+
 #ifdef CONFIG_SMP
 void __init boot_cpu_id_too_large(int cpu)
 {
@@ -528,6 +547,8 @@ static void __init init_sparc64_elf_hwcap(void)
 
 	if (sparc64_elf_hwcap & AV_SPARC_POPC)
 		popc_patch();
+	if (sparc64_elf_hwcap & AV_SPARC_PAUSE)
+		pause_patch();
 }
 
 void __init setup_arch(char **cmdline_p)
diff --git a/arch/sparc/kernel/signal32.c b/arch/sparc/kernel/signal32.c
index 53e48f721ce3..b524f91dd0e5 100644
--- a/arch/sparc/kernel/signal32.c
+++ b/arch/sparc/kernel/signal32.c
@@ -61,7 +61,7 @@ struct rt_signal_frame32 {
 	compat_sigset_t		mask;
 	/* __siginfo_fpu_t * */ u32 fpu_save;
 	unsigned int		insns[2];
-	stack_t32		stack;
+	compat_stack_t		stack;
 	unsigned int		extra_size; /* Should be sizeof(siginfo_extra_v8plus_t) */
 	/* Only valid if (regs.psr & (PSR_VERS|PSR_IMPL)) == PSR_V8PLUS */
 	siginfo_extra_v8plus_t	v8plus;
@@ -230,13 +230,11 @@ segv:
 asmlinkage void do_rt_sigreturn32(struct pt_regs *regs)
 {
 	struct rt_signal_frame32 __user *sf;
-	unsigned int psr, pc, npc, u_ss_sp;
+	unsigned int psr, pc, npc;
 	compat_uptr_t fpu_save;
 	compat_uptr_t rwin_save;
-	mm_segment_t old_fs;
 	sigset_t set;
 	compat_sigset_t seta;
-	stack_t st;
 	int err, i;
 	
 	/* Always make any pending restarted system calls return -EINTR */
@@ -295,20 +293,10 @@ asmlinkage void do_rt_sigreturn32(struct pt_regs *regs)
 	if (!err && fpu_save)
 		err |= restore_fpu_state(regs, compat_ptr(fpu_save));
 	err |= copy_from_user(&seta, &sf->mask, sizeof(compat_sigset_t));
-	err |= __get_user(u_ss_sp, &sf->stack.ss_sp);
-	st.ss_sp = compat_ptr(u_ss_sp);
-	err |= __get_user(st.ss_flags, &sf->stack.ss_flags);
-	err |= __get_user(st.ss_size, &sf->stack.ss_size);
+	err |= compat_restore_altstack(&sf->stack);
 	if (err)
 		goto segv;
 		
-	/* It is more difficult to avoid calling this function than to
-	   call it and ignore errors.  */
-	old_fs = get_fs();
-	set_fs(KERNEL_DS);
-	do_sigaltstack((stack_t __user *) &st, NULL, (unsigned long)sf);
-	set_fs(old_fs);
-	
 	err |= __get_user(rwin_save, &sf->rwin_save);
 	if (!err && rwin_save) {
 		if (restore_rwin_state(compat_ptr(rwin_save)))
@@ -335,7 +323,7 @@ static int invalid_frame_pointer(void __user *fp, int fplen)
 	return 0;
 }
 
-static void __user *get_sigframe(struct sigaction *sa, struct pt_regs *regs, unsigned long framesize)
+static void __user *get_sigframe(struct ksignal *ksig, struct pt_regs *regs, unsigned long framesize)
 {
 	unsigned long sp;
 	
@@ -350,12 +338,7 @@ static void __user *get_sigframe(struct sigaction *sa, struct pt_regs *regs, uns
 		return (void __user *) -1L;
 
 	/* This is the X/Open sanctioned signal stack switching.  */
-	if (sa->sa_flags & SA_ONSTACK) {
-		if (sas_ss_flags(sp) == 0)
-			sp = current->sas_ss_sp + current->sas_ss_size;
-	}
-
-	sp -= framesize;
+	sp = sigsp(sp, ksig) - framesize;
 
 	/* Always align the stack frame.  This handles two cases.  First,
 	 * sigaltstack need not be mindful of platform specific stack
@@ -426,8 +409,8 @@ out_irqs_on:
 
 }
 
-static int setup_frame32(struct k_sigaction *ka, struct pt_regs *regs,
-			 int signo, sigset_t *oldset)
+static int setup_frame32(struct ksignal *ksig, struct pt_regs *regs,
+			 sigset_t *oldset)
 {
 	struct signal_frame32 __user *sf;
 	int i, err, wsaved;
@@ -449,10 +432,12 @@ static int setup_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 		sigframe_size += sizeof(__siginfo_rwin_t);
 
 	sf = (struct signal_frame32 __user *)
-		get_sigframe(&ka->sa, regs, sigframe_size);
+		get_sigframe(ksig, regs, sigframe_size);
 	
-	if (invalid_frame_pointer(sf, sigframe_size))
-		goto sigill;
+	if (invalid_frame_pointer(sf, sigframe_size)) {
+		do_exit(SIGILL);
+		return -EINVAL;
+	}
 
 	tail = (sf + 1);
 
@@ -526,16 +511,16 @@ static int setup_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 		err |= __put_user(rp->ins[7], &sf->ss.callers_pc);
 	}	
 	if (err)
-		goto sigsegv;
+		return err;
 
 	/* 3. signal handler back-trampoline and parameters */
 	regs->u_regs[UREG_FP] = (unsigned long) sf;
-	regs->u_regs[UREG_I0] = signo;
+	regs->u_regs[UREG_I0] = ksig->sig;
 	regs->u_regs[UREG_I1] = (unsigned long) &sf->info;
 	regs->u_regs[UREG_I2] = (unsigned long) &sf->info;
 
 	/* 4. signal handler */
-	regs->tpc = (unsigned long) ka->sa.sa_handler;
+	regs->tpc = (unsigned long) ksig->ka.sa.sa_handler;
 	regs->tnpc = (regs->tpc + 4);
 	if (test_thread_flag(TIF_32BIT)) {
 		regs->tpc &= 0xffffffff;
@@ -543,8 +528,8 @@ static int setup_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 	}
 
 	/* 5. return to kernel instructions */
-	if (ka->ka_restorer) {
-		regs->u_regs[UREG_I7] = (unsigned long)ka->ka_restorer;
+	if (ksig->ka.ka_restorer) {
+		regs->u_regs[UREG_I7] = (unsigned long)ksig->ka.ka_restorer;
 	} else {
 		unsigned long address = ((unsigned long)&(sf->insns[0]));
 
@@ -553,23 +538,14 @@ static int setup_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 		err  = __put_user(0x821020d8, &sf->insns[0]); /*mov __NR_sigreturn, %g1*/
 		err |= __put_user(0x91d02010, &sf->insns[1]); /*t 0x10*/
 		if (err)
-			goto sigsegv;
+			return err;
 		flush_signal_insns(address);
 	}
 	return 0;
-
-sigill:
-	do_exit(SIGILL);
-	return -EINVAL;
-
-sigsegv:
-	force_sigsegv(signo, current);
-	return -EFAULT;
 }
 
-static int setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs,
-			    unsigned long signr, sigset_t *oldset,
-			    siginfo_t *info)
+static int setup_rt_frame32(struct ksignal *ksig, struct pt_regs *regs,
+			    sigset_t *oldset)
 {
 	struct rt_signal_frame32 __user *sf;
 	int i, err, wsaved;
@@ -591,10 +567,12 @@ static int setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 		sigframe_size += sizeof(__siginfo_rwin_t);
 
 	sf = (struct rt_signal_frame32 __user *)
-		get_sigframe(&ka->sa, regs, sigframe_size);
+		get_sigframe(ksig, regs, sigframe_size);
 	
-	if (invalid_frame_pointer(sf, sigframe_size))
-		goto sigill;
+	if (invalid_frame_pointer(sf, sigframe_size)) {
+		do_exit(SIGILL);
+		return -EINVAL;
+	}
 
 	tail = (sf + 1);
 
@@ -639,12 +617,10 @@ static int setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 	}
 
 	/* Update the siginfo structure.  */
-	err |= copy_siginfo_to_user32(&sf->info, info);
+	err |= copy_siginfo_to_user32(&sf->info, &ksig->info);
 	
 	/* Setup sigaltstack */
-	err |= __put_user(current->sas_ss_sp, &sf->stack.ss_sp);
-	err |= __put_user(sas_ss_flags(regs->u_regs[UREG_FP]), &sf->stack.ss_flags);
-	err |= __put_user(current->sas_ss_size, &sf->stack.ss_size);
+	err |= __compat_save_altstack(&sf->stack, regs->u_regs[UREG_FP]);
 
 	switch (_NSIG_WORDS) {
 	case 4: seta.sig[7] = (oldset->sig[3] >> 32);
@@ -674,16 +650,16 @@ static int setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 		err |= __put_user(rp->ins[7], &sf->ss.callers_pc);
 	}
 	if (err)
-		goto sigsegv;
+		return err;
 	
 	/* 3. signal handler back-trampoline and parameters */
 	regs->u_regs[UREG_FP] = (unsigned long) sf;
-	regs->u_regs[UREG_I0] = signr;
+	regs->u_regs[UREG_I0] = ksig->sig;
 	regs->u_regs[UREG_I1] = (unsigned long) &sf->info;
 	regs->u_regs[UREG_I2] = (unsigned long) &sf->regs;
 
 	/* 4. signal handler */
-	regs->tpc = (unsigned long) ka->sa.sa_handler;
+	regs->tpc = (unsigned long) ksig->ka.sa.sa_handler;
 	regs->tnpc = (regs->tpc + 4);
 	if (test_thread_flag(TIF_32BIT)) {
 		regs->tpc &= 0xffffffff;
@@ -691,8 +667,8 @@ static int setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 	}
 
 	/* 5. return to kernel instructions */
-	if (ka->ka_restorer)
-		regs->u_regs[UREG_I7] = (unsigned long)ka->ka_restorer;
+	if (ksig->ka.ka_restorer)
+		regs->u_regs[UREG_I7] = (unsigned long)ksig->ka.ka_restorer;
 	else {
 		unsigned long address = ((unsigned long)&(sf->insns[0]));
 
@@ -704,36 +680,25 @@ static int setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 		/* t 0x10 */
 		err |= __put_user(0x91d02010, &sf->insns[1]);
 		if (err)
-			goto sigsegv;
+			return err;
 
 		flush_signal_insns(address);
 	}
 	return 0;
-
-sigill:
-	do_exit(SIGILL);
-	return -EINVAL;
-
-sigsegv:
-	force_sigsegv(signr, current);
-	return -EFAULT;
 }
 
-static inline void handle_signal32(unsigned long signr, struct k_sigaction *ka,
-				  siginfo_t *info,
-				  sigset_t *oldset, struct pt_regs *regs)
+static inline void handle_signal32(struct ksignal *ksig, 
+				  struct pt_regs *regs)
 {
+	sigset_t *oldset = sigmask_to_save();
 	int err;
 
-	if (ka->sa.sa_flags & SA_SIGINFO)
-		err = setup_rt_frame32(ka, regs, signr, oldset, info);
+	if (ksig->ka.sa.sa_flags & SA_SIGINFO)
+		err = setup_rt_frame32(ksig, regs, oldset);
 	else
-		err = setup_frame32(ka, regs, signr, oldset);
-
-	if (err)
-		return;
+		err = setup_frame32(ksig, regs, oldset);
 
-	signal_delivered(signr, info, ka, regs, 0);
+	signal_setup_done(err, ksig, 0);
 }
 
 static inline void syscall_restart32(unsigned long orig_i0, struct pt_regs *regs,
@@ -761,52 +726,43 @@ static inline void syscall_restart32(unsigned long orig_i0, struct pt_regs *regs
  * want to handle. Thus you cannot kill init even with a SIGKILL even by
  * mistake.
  */
-void do_signal32(sigset_t *oldset, struct pt_regs * regs)
+void do_signal32(struct pt_regs * regs)
 {
-	struct k_sigaction ka;
-	unsigned long orig_i0;
-	int restart_syscall;
-	siginfo_t info;
-	int signr;
-	
-	signr = get_signal_to_deliver(&info, &ka, regs, NULL);
+	struct ksignal ksig;
+	unsigned long orig_i0 = 0;
+	int restart_syscall = 0;
+	bool has_handler = get_signal(&ksig);
 
-	restart_syscall = 0;
-	orig_i0 = 0;
 	if (pt_regs_is_syscall(regs) &&
 	    (regs->tstate & (TSTATE_XCARRY | TSTATE_ICARRY))) {
 		restart_syscall = 1;
 		orig_i0 = regs->u_regs[UREG_G6];
 	}
 
-	if (signr > 0) {
+	if (has_handler) {
 		if (restart_syscall)
-			syscall_restart32(orig_i0, regs, &ka.sa);
-		handle_signal32(signr, &ka, &info, oldset, regs);
-		return;
-	}
-	if (restart_syscall &&
-	    (regs->u_regs[UREG_I0] == ERESTARTNOHAND ||
-	     regs->u_regs[UREG_I0] == ERESTARTSYS ||
-	     regs->u_regs[UREG_I0] == ERESTARTNOINTR)) {
-		/* replay the system call when we are done */
-		regs->u_regs[UREG_I0] = orig_i0;
-		regs->tpc -= 4;
-		regs->tnpc -= 4;
-		pt_regs_clear_syscall(regs);
-	}
-	if (restart_syscall &&
-	    regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) {
-		regs->u_regs[UREG_G1] = __NR_restart_syscall;
-		regs->tpc -= 4;
-		regs->tnpc -= 4;
-		pt_regs_clear_syscall(regs);
+			syscall_restart32(orig_i0, regs, &ksig.ka.sa);
+		handle_signal32(&ksig, regs);
+	} else {
+		if (restart_syscall) {
+			switch (regs->u_regs[UREG_I0]) {
+			case ERESTARTNOHAND:
+	     		case ERESTARTSYS:
+			case ERESTARTNOINTR:
+				/* replay the system call when we are done */
+				regs->u_regs[UREG_I0] = orig_i0;
+				regs->tpc -= 4;
+				regs->tnpc -= 4;
+				pt_regs_clear_syscall(regs);
+			case ERESTART_RESTARTBLOCK:
+				regs->u_regs[UREG_G1] = __NR_restart_syscall;
+				regs->tpc -= 4;
+				regs->tnpc -= 4;
+				pt_regs_clear_syscall(regs);
+			}
+		}
+		restore_saved_sigmask();
 	}
-
-	/* If there's no signal to deliver, we just put the saved sigmask
-	 * back
-	 */
-	restore_saved_sigmask();
 }
 
 struct sigstack32 {
@@ -856,29 +812,3 @@ asmlinkage int do_sys32_sigstack(u32 u_ssptr, u32 u_ossptr, unsigned long sp)
 out:
 	return ret;
 }
-
-asmlinkage long do_sys32_sigaltstack(u32 ussa, u32 uossa, unsigned long sp)
-{
-	stack_t uss, uoss;
-	u32 u_ss_sp = 0;
-	int ret;
-	mm_segment_t old_fs;
-	stack_t32 __user *uss32 = compat_ptr(ussa);
-	stack_t32 __user *uoss32 = compat_ptr(uossa);
-	
-	if (ussa && (get_user(u_ss_sp, &uss32->ss_sp) ||
-		    __get_user(uss.ss_flags, &uss32->ss_flags) ||
-		    __get_user(uss.ss_size, &uss32->ss_size)))
-		return -EFAULT;
-	uss.ss_sp = compat_ptr(u_ss_sp);
-	old_fs = get_fs();
-	set_fs(KERNEL_DS);
-	ret = do_sigaltstack(ussa ? (stack_t __user *) &uss : NULL,
-			     uossa ? (stack_t __user *) &uoss : NULL, sp);
-	set_fs(old_fs);
-	if (!ret && uossa && (put_user(ptr_to_compat(uoss.ss_sp), &uoss32->ss_sp) ||
-		    __put_user(uoss.ss_flags, &uoss32->ss_flags) ||
-		    __put_user(uoss.ss_size, &uoss32->ss_size)))
-		return -EFAULT;
-	return ret;
-}
diff --git a/arch/sparc/kernel/signal_32.c b/arch/sparc/kernel/signal_32.c
index 68f9c8650af4..7d5d8e1f8415 100644
--- a/arch/sparc/kernel/signal_32.c
+++ b/arch/sparc/kernel/signal_32.c
@@ -59,18 +59,6 @@ struct rt_signal_frame {
 #define SF_ALIGNEDSZ  (((sizeof(struct signal_frame) + 7) & (~7)))
 #define RT_ALIGNEDSZ  (((sizeof(struct rt_signal_frame) + 7) & (~7)))
 
-static int _sigpause_common(old_sigset_t set)
-{
-	sigset_t blocked;
-	siginitset(&blocked, set);
-	return sigsuspend(&blocked);
-}
-
-asmlinkage int sys_sigsuspend(old_sigset_t set)
-{
-	return _sigpause_common(set);
-}
-
 asmlinkage void do_sigreturn(struct pt_regs *regs)
 {
 	struct signal_frame __user *sf;
@@ -141,9 +129,7 @@ asmlinkage void do_rt_sigreturn(struct pt_regs *regs)
 	unsigned int psr, pc, npc;
 	__siginfo_fpu_t __user *fpu_save;
 	__siginfo_rwin_t __user *rwin_save;
-	mm_segment_t old_fs;
 	sigset_t set;
-	stack_t st;
 	int err;
 
 	synchronize_user_stack();
@@ -171,8 +157,7 @@ asmlinkage void do_rt_sigreturn(struct pt_regs *regs)
 	if (!err && fpu_save)
 		err |= restore_fpu_state(regs, fpu_save);
 	err |= __copy_from_user(&set, &sf->mask, sizeof(sigset_t));
-	
-	err |= __copy_from_user(&st, &sf->stack, sizeof(stack_t));
+	err |= restore_altstack(&sf->stack);
 	
 	if (err)
 		goto segv;
@@ -180,14 +165,6 @@ asmlinkage void do_rt_sigreturn(struct pt_regs *regs)
 	regs->pc = pc;
 	regs->npc = npc;
 	
-	/* It is more difficult to avoid calling this function than to
-	 * call it and ignore errors.
-	 */
-	old_fs = get_fs();
-	set_fs(KERNEL_DS);
-	do_sigaltstack((const stack_t __user *) &st, NULL, (unsigned long)sf);
-	set_fs(old_fs);
-
 	err |= __get_user(rwin_save, &sf->rwin_save);
 	if (!err && rwin_save) {
 		if (restore_rwin_state(rwin_save))
@@ -209,7 +186,7 @@ static inline int invalid_frame_pointer(void __user *fp, int fplen)
 	return 0;
 }
 
-static inline void __user *get_sigframe(struct sigaction *sa, struct pt_regs *regs, unsigned long framesize)
+static inline void __user *get_sigframe(struct ksignal *ksig, struct pt_regs *regs, unsigned long framesize)
 {
 	unsigned long sp = regs->u_regs[UREG_FP];
 
@@ -221,12 +198,7 @@ static inline void __user *get_sigframe(struct sigaction *sa, struct pt_regs *re
 		return (void __user *) -1L;
 
 	/* This is the X/Open sanctioned signal stack switching.  */
-	if (sa->sa_flags & SA_ONSTACK) {
-		if (sas_ss_flags(sp) == 0)
-			sp = current->sas_ss_sp + current->sas_ss_size;
-	}
-
-	sp -= framesize;
+	sp = sigsp(sp, ksig) - framesize;
 
 	/* Always align the stack frame.  This handles two cases.  First,
 	 * sigaltstack need not be mindful of platform specific stack
@@ -239,8 +211,8 @@ static inline void __user *get_sigframe(struct sigaction *sa, struct pt_regs *re
 	return (void __user *) sp;
 }
 
-static int setup_frame(struct k_sigaction *ka, struct pt_regs *regs,
-		       int signo, sigset_t *oldset)
+static int setup_frame(struct ksignal *ksig, struct pt_regs *regs,
+		       sigset_t *oldset)
 {
 	struct signal_frame __user *sf;
 	int sigframe_size, err, wsaved;
@@ -258,10 +230,12 @@ static int setup_frame(struct k_sigaction *ka, struct pt_regs *regs,
 		sigframe_size += sizeof(__siginfo_rwin_t);
 
 	sf = (struct signal_frame __user *)
-		get_sigframe(&ka->sa, regs, sigframe_size);
+		get_sigframe(ksig, regs, sigframe_size);
 
-	if (invalid_frame_pointer(sf, sigframe_size))
-		goto sigill_and_return;
+	if (invalid_frame_pointer(sf, sigframe_size)) {
+		do_exit(SIGILL);
+		return -EINVAL;
+	}
 
 	tail = sf + 1;
 
@@ -300,21 +274,21 @@ static int setup_frame(struct k_sigaction *ka, struct pt_regs *regs,
 		err |= __copy_to_user(sf, rp, sizeof(struct reg_window32));
 	}
 	if (err)
-		goto sigsegv;
+		return err;
 	
 	/* 3. signal handler back-trampoline and parameters */
 	regs->u_regs[UREG_FP] = (unsigned long) sf;
-	regs->u_regs[UREG_I0] = signo;
+	regs->u_regs[UREG_I0] = ksig->sig;
 	regs->u_regs[UREG_I1] = (unsigned long) &sf->info;
 	regs->u_regs[UREG_I2] = (unsigned long) &sf->info;
 
 	/* 4. signal handler */
-	regs->pc = (unsigned long) ka->sa.sa_handler;
+	regs->pc = (unsigned long) ksig->ka.sa.sa_handler;
 	regs->npc = (regs->pc + 4);
 
 	/* 5. return to kernel instructions */
-	if (ka->ka_restorer)
-		regs->u_regs[UREG_I7] = (unsigned long)ka->ka_restorer;
+	if (ksig->ka.ka_restorer)
+		regs->u_regs[UREG_I7] = (unsigned long)ksig->ka.ka_restorer;
 	else {
 		regs->u_regs[UREG_I7] = (unsigned long)(&(sf->insns[0]) - 2);
 
@@ -324,24 +298,16 @@ static int setup_frame(struct k_sigaction *ka, struct pt_regs *regs,
 		/* t 0x10 */
 		err |= __put_user(0x91d02010, &sf->insns[1]);
 		if (err)
-			goto sigsegv;
+			return err;
 
 		/* Flush instruction space. */
 		flush_sig_insns(current->mm, (unsigned long) &(sf->insns[0]));
 	}
 	return 0;
-
-sigill_and_return:
-	do_exit(SIGILL);
-	return -EINVAL;
-
-sigsegv:
-	force_sigsegv(signo, current);
-	return -EFAULT;
 }
 
-static int setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
-			  int signo, sigset_t *oldset, siginfo_t *info)
+static int setup_rt_frame(struct ksignal *ksig, struct pt_regs *regs,
+			  sigset_t *oldset)
 {
 	struct rt_signal_frame __user *sf;
 	int sigframe_size, wsaved;
@@ -357,9 +323,11 @@ static int setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
 	if (wsaved)
 		sigframe_size += sizeof(__siginfo_rwin_t);
 	sf = (struct rt_signal_frame __user *)
-		get_sigframe(&ka->sa, regs, sigframe_size);
-	if (invalid_frame_pointer(sf, sigframe_size))
-		goto sigill;
+		get_sigframe(ksig, regs, sigframe_size);
+	if (invalid_frame_pointer(sf, sigframe_size)) {
+		do_exit(SIGILL);
+		return -EINVAL;
+	}
 
 	tail = sf + 1;
 	err  = __put_user(regs->pc, &sf->regs.pc);
@@ -391,9 +359,7 @@ static int setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
 	err |= __copy_to_user(&sf->mask, &oldset->sig[0], sizeof(sigset_t));
 	
 	/* Setup sigaltstack */
-	err |= __put_user(current->sas_ss_sp, &sf->stack.ss_sp);
-	err |= __put_user(sas_ss_flags(regs->u_regs[UREG_FP]), &sf->stack.ss_flags);
-	err |= __put_user(current->sas_ss_size, &sf->stack.ss_size);
+	err |= __save_altstack(&sf->stack, regs->u_regs[UREG_FP]);
 	
 	if (!wsaved) {
 		err |= __copy_to_user(sf, (char *) regs->u_regs[UREG_FP],
@@ -405,21 +371,21 @@ static int setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
 		err |= __copy_to_user(sf, rp, sizeof(struct reg_window32));
 	}
 
-	err |= copy_siginfo_to_user(&sf->info, info);
+	err |= copy_siginfo_to_user(&sf->info, &ksig->info);
 
 	if (err)
-		goto sigsegv;
+		return err;
 
 	regs->u_regs[UREG_FP] = (unsigned long) sf;
-	regs->u_regs[UREG_I0] = signo;
+	regs->u_regs[UREG_I0] = ksig->sig;
 	regs->u_regs[UREG_I1] = (unsigned long) &sf->info;
 	regs->u_regs[UREG_I2] = (unsigned long) &sf->regs;
 
-	regs->pc = (unsigned long) ka->sa.sa_handler;
+	regs->pc = (unsigned long) ksig->ka.sa.sa_handler;
 	regs->npc = (regs->pc + 4);
 
-	if (ka->ka_restorer)
-		regs->u_regs[UREG_I7] = (unsigned long)ka->ka_restorer;
+	if (ksig->ka.ka_restorer)
+		regs->u_regs[UREG_I7] = (unsigned long)ksig->ka.ka_restorer;
 	else {
 		regs->u_regs[UREG_I7] = (unsigned long)(&(sf->insns[0]) - 2);
 
@@ -429,38 +395,25 @@ static int setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
 		/* t 0x10 */
 		err |= __put_user(0x91d02010, &sf->insns[1]);
 		if (err)
-			goto sigsegv;
+			return err;
 
 		/* Flush instruction space. */
 		flush_sig_insns(current->mm, (unsigned long) &(sf->insns[0]));
 	}
 	return 0;
-
-sigill:
-	do_exit(SIGILL);
-	return -EINVAL;
-
-sigsegv:
-	force_sigsegv(signo, current);
-	return -EFAULT;
 }
 
 static inline void
-handle_signal(unsigned long signr, struct k_sigaction *ka,
-	      siginfo_t *info, struct pt_regs *regs)
+handle_signal(struct ksignal *ksig, struct pt_regs *regs)
 {
 	sigset_t *oldset = sigmask_to_save();
 	int err;
 
-	if (ka->sa.sa_flags & SA_SIGINFO)
-		err = setup_rt_frame(ka, regs, signr, oldset, info);
+	if (ksig->ka.sa.sa_flags & SA_SIGINFO)
+		err = setup_rt_frame(ksig, regs, oldset);
 	else
-		err = setup_frame(ka, regs, signr, oldset);
-
-	if (err)
-		return;
-
-	signal_delivered(signr, info, ka, regs, 0);
+		err = setup_frame(ksig, regs, oldset);
+	signal_setup_done(err, ksig, 0);
 }
 
 static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
@@ -490,10 +443,9 @@ static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
  */
 static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 {
-	struct k_sigaction ka;
+	struct ksignal ksig;
 	int restart_syscall;
-	siginfo_t info;
-	int signr;
+	bool has_handler;
 
 	/* It's a lot of work and synchronization to add a new ptrace
 	 * register for GDB to save and restore in order to get
@@ -516,7 +468,7 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 	if (pt_regs_is_syscall(regs) && (regs->psr & PSR_C))
 		regs->u_regs[UREG_G6] = orig_i0;
 
-	signr = get_signal_to_deliver(&info, &ka, regs, NULL);
+	has_handler = get_signal(&ksig);
 
 	/* If the debugger messes with the program counter, it clears
 	 * the software "in syscall" bit, directing us to not perform
@@ -528,35 +480,30 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 		orig_i0 = regs->u_regs[UREG_G6];
 	}
 
-
-	if (signr > 0) {
+	if (has_handler) {
 		if (restart_syscall)
-			syscall_restart(orig_i0, regs, &ka.sa);
-		handle_signal(signr, &ka, &info, regs);
-		return;
-	}
-	if (restart_syscall &&
-	    (regs->u_regs[UREG_I0] == ERESTARTNOHAND ||
-	     regs->u_regs[UREG_I0] == ERESTARTSYS ||
-	     regs->u_regs[UREG_I0] == ERESTARTNOINTR)) {
-		/* replay the system call when we are done */
-		regs->u_regs[UREG_I0] = orig_i0;
-		regs->pc -= 4;
-		regs->npc -= 4;
-		pt_regs_clear_syscall(regs);
-	}
-	if (restart_syscall &&
-	    regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) {
-		regs->u_regs[UREG_G1] = __NR_restart_syscall;
-		regs->pc -= 4;
-		regs->npc -= 4;
-		pt_regs_clear_syscall(regs);
+			syscall_restart(orig_i0, regs, &ksig.ka.sa);
+		handle_signal(&ksig, regs);
+	} else {
+		if (restart_syscall) {
+			switch (regs->u_regs[UREG_I0]) {
+			case ERESTARTNOHAND:
+	     		case ERESTARTSYS:
+			case ERESTARTNOINTR:
+				/* replay the system call when we are done */
+				regs->u_regs[UREG_I0] = orig_i0;
+				regs->pc -= 4;
+				regs->npc -= 4;
+				pt_regs_clear_syscall(regs);
+			case ERESTART_RESTARTBLOCK:
+				regs->u_regs[UREG_G1] = __NR_restart_syscall;
+				regs->pc -= 4;
+				regs->npc -= 4;
+				pt_regs_clear_syscall(regs);
+			}
+		}
+		restore_saved_sigmask();
 	}
-
-	/* if there's no signal to deliver, we just put the saved sigmask
-	 * back
-	 */
-	restore_saved_sigmask();
 }
 
 void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0,
diff --git a/arch/sparc/kernel/signal_64.c b/arch/sparc/kernel/signal_64.c
index 867de2f8189c..35923e8abd82 100644
--- a/arch/sparc/kernel/signal_64.c
+++ b/arch/sparc/kernel/signal_64.c
@@ -236,23 +236,6 @@ struct rt_signal_frame {
 	__siginfo_rwin_t	*rwin_save;
 };
 
-static long _sigpause_common(old_sigset_t set)
-{
-	sigset_t blocked;
-	siginitset(&blocked, set);
-	return sigsuspend(&blocked);
-}
-
-asmlinkage long sys_sigpause(unsigned int set)
-{
-	return _sigpause_common(set);
-}
-
-asmlinkage long sys_sigsuspend(old_sigset_t set)
-{
-	return _sigpause_common(set);
-}
-
 void do_rt_sigreturn(struct pt_regs *regs)
 {
 	struct rt_signal_frame __user *sf;
@@ -295,8 +278,7 @@ void do_rt_sigreturn(struct pt_regs *regs)
 		err |= restore_fpu_state(regs, fpu_save);
 
 	err |= __copy_from_user(&set, &sf->mask, sizeof(sigset_t));
-	err |= do_sigaltstack(&sf->stack, NULL, (unsigned long)sf);
-
+	err |= restore_altstack(&sf->stack);
 	if (err)
 		goto segv;
 
@@ -326,7 +308,7 @@ static int invalid_frame_pointer(void __user *fp)
 	return 0;
 }
 
-static inline void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, unsigned long framesize)
+static inline void __user *get_sigframe(struct ksignal *ksig, struct pt_regs *regs, unsigned long framesize)
 {
 	unsigned long sp = regs->u_regs[UREG_FP] + STACK_BIAS;
 
@@ -338,12 +320,7 @@ static inline void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *
 		return (void __user *) -1L;
 
 	/* This is the X/Open sanctioned signal stack switching.  */
-	if (ka->sa.sa_flags & SA_ONSTACK) {
-		if (sas_ss_flags(sp) == 0)
-			sp = current->sas_ss_sp + current->sas_ss_size;
-	}
-
-	sp -= framesize;
+	sp = sigsp(sp, ksig) - framesize;
 
 	/* Always align the stack frame.  This handles two cases.  First,
 	 * sigaltstack need not be mindful of platform specific stack
@@ -357,8 +334,7 @@ static inline void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *
 }
 
 static inline int
-setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
-	       int signo, sigset_t *oldset, siginfo_t *info)
+setup_rt_frame(struct ksignal *ksig, struct pt_regs *regs)
 {
 	struct rt_signal_frame __user *sf;
 	int wsaved, err, sf_size;
@@ -376,10 +352,12 @@ setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
 	if (wsaved)
 		sf_size += sizeof(__siginfo_rwin_t);
 	sf = (struct rt_signal_frame __user *)
-		get_sigframe(ka, regs, sf_size);
+		get_sigframe(ksig, regs, sf_size);
 
-	if (invalid_frame_pointer (sf))
-		goto sigill;
+	if (invalid_frame_pointer (sf)) {
+		do_exit(SIGILL);	/* won't return, actually */
+		return -EINVAL;
+	}
 
 	tail = (sf + 1);
 
@@ -405,11 +383,9 @@ setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
 	}
 	
 	/* Setup sigaltstack */
-	err |= __put_user(current->sas_ss_sp, &sf->stack.ss_sp);
-	err |= __put_user(sas_ss_flags(regs->u_regs[UREG_FP]), &sf->stack.ss_flags);
-	err |= __put_user(current->sas_ss_size, &sf->stack.ss_size);
+	err |= __save_altstack(&sf->stack, regs->u_regs[UREG_FP]);
 
-	err |= copy_to_user(&sf->mask, oldset, sizeof(sigset_t));
+	err |= copy_to_user(&sf->mask, sigmask_to_save(), sizeof(sigset_t));
 
 	if (!wsaved) {
 		err |= copy_in_user((u64 __user *)sf,
@@ -422,18 +398,18 @@ setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
 		rp = &current_thread_info()->reg_window[wsaved - 1];
 		err |= copy_to_user(sf, rp, sizeof(struct reg_window));
 	}
-	if (info)
-		err |= copy_siginfo_to_user(&sf->info, info);
+	if (ksig->ka.sa.sa_flags & SA_SIGINFO)
+		err |= copy_siginfo_to_user(&sf->info, &ksig->info);
 	else {
-		err |= __put_user(signo, &sf->info.si_signo);
+		err |= __put_user(ksig->sig, &sf->info.si_signo);
 		err |= __put_user(SI_NOINFO, &sf->info.si_code);
 	}
 	if (err)
-		goto sigsegv;
+		return err;
 	
 	/* 3. signal handler back-trampoline and parameters */
 	regs->u_regs[UREG_FP] = ((unsigned long) sf) - STACK_BIAS;
-	regs->u_regs[UREG_I0] = signo;
+	regs->u_regs[UREG_I0] = ksig->sig;
 	regs->u_regs[UREG_I1] = (unsigned long) &sf->info;
 
 	/* The sigcontext is passed in this way because of how it
@@ -443,37 +419,15 @@ setup_rt_frame(struct k_sigaction *ka, struct pt_regs *regs,
 	regs->u_regs[UREG_I2] = (unsigned long) &sf->info;
 
 	/* 5. signal handler */
-	regs->tpc = (unsigned long) ka->sa.sa_handler;
+	regs->tpc = (unsigned long) ksig->ka.sa.sa_handler;
 	regs->tnpc = (regs->tpc + 4);
 	if (test_thread_flag(TIF_32BIT)) {
 		regs->tpc &= 0xffffffff;
 		regs->tnpc &= 0xffffffff;
 	}
 	/* 4. return to kernel instructions */
-	regs->u_regs[UREG_I7] = (unsigned long)ka->ka_restorer;
+	regs->u_regs[UREG_I7] = (unsigned long)ksig->ka.ka_restorer;
 	return 0;
-
-sigill:
-	do_exit(SIGILL);
-	return -EINVAL;
-
-sigsegv:
-	force_sigsegv(signo, current);
-	return -EFAULT;
-}
-
-static inline void handle_signal(unsigned long signr, struct k_sigaction *ka,
-				siginfo_t *info,
-				sigset_t *oldset, struct pt_regs *regs)
-{
-	int err;
-
-	err = setup_rt_frame(ka, regs, signr, oldset,
-			     (ka->sa.sa_flags & SA_SIGINFO) ? info : NULL);
-	if (err)
-		return;
-
-	signal_delivered(signr, info, ka, regs, 0);
 }
 
 static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
@@ -503,11 +457,9 @@ static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,
  */
 static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 {
-	struct k_sigaction ka;
+	struct ksignal ksig;
 	int restart_syscall;
-	sigset_t *oldset = sigmask_to_save();
-	siginfo_t info;
-	int signr;
+	bool has_handler;
 	
 	/* It's a lot of work and synchronization to add a new ptrace
 	 * register for GDB to save and restore in order to get
@@ -533,13 +485,13 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 
 #ifdef CONFIG_COMPAT
 	if (test_thread_flag(TIF_32BIT)) {
-		extern void do_signal32(sigset_t *, struct pt_regs *);
-		do_signal32(oldset, regs);
+		extern void do_signal32(struct pt_regs *);
+		do_signal32(regs);
 		return;
 	}
 #endif	
 
-	signr = get_signal_to_deliver(&info, &ka, regs, NULL);
+	has_handler = get_signal(&ksig);
 
 	restart_syscall = 0;
 	if (pt_regs_is_syscall(regs) &&
@@ -548,34 +500,30 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 		orig_i0 = regs->u_regs[UREG_G6];
 	}
 
-	if (signr > 0) {
+	if (has_handler) {
 		if (restart_syscall)
-			syscall_restart(orig_i0, regs, &ka.sa);
-		handle_signal(signr, &ka, &info, oldset, regs);
-		return;
-	}
-	if (restart_syscall &&
-	    (regs->u_regs[UREG_I0] == ERESTARTNOHAND ||
-	     regs->u_regs[UREG_I0] == ERESTARTSYS ||
-	     regs->u_regs[UREG_I0] == ERESTARTNOINTR)) {
-		/* replay the system call when we are done */
-		regs->u_regs[UREG_I0] = orig_i0;
-		regs->tpc -= 4;
-		regs->tnpc -= 4;
-		pt_regs_clear_syscall(regs);
-	}
-	if (restart_syscall &&
-	    regs->u_regs[UREG_I0] == ERESTART_RESTARTBLOCK) {
-		regs->u_regs[UREG_G1] = __NR_restart_syscall;
-		regs->tpc -= 4;
-		regs->tnpc -= 4;
-		pt_regs_clear_syscall(regs);
+			syscall_restart(orig_i0, regs, &ksig.ka.sa);
+		signal_setup_done(setup_rt_frame(&ksig, regs), &ksig, 0);
+	} else {
+		if (restart_syscall) {
+			switch (regs->u_regs[UREG_I0]) {
+			case ERESTARTNOHAND:
+	     		case ERESTARTSYS:
+			case ERESTARTNOINTR:
+				/* replay the system call when we are done */
+				regs->u_regs[UREG_I0] = orig_i0;
+				regs->tpc -= 4;
+				regs->tnpc -= 4;
+				pt_regs_clear_syscall(regs);
+			case ERESTART_RESTARTBLOCK:
+				regs->u_regs[UREG_G1] = __NR_restart_syscall;
+				regs->tpc -= 4;
+				regs->tnpc -= 4;
+				pt_regs_clear_syscall(regs);
+			}
+		}
+		restore_saved_sigmask();
 	}
-
-	/* If there's no signal to deliver, we just put the saved sigmask
-	 * back
-	 */
-	restore_saved_sigmask();
 }
 
 void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long thread_info_flags)
diff --git a/arch/sparc/kernel/smp_32.c b/arch/sparc/kernel/smp_32.c
index 79db45e5134a..9e7e6d718367 100644
--- a/arch/sparc/kernel/smp_32.c
+++ b/arch/sparc/kernel/smp_32.c
@@ -20,6 +20,7 @@
 #include <linux/seq_file.h>
 #include <linux/cache.h>
 #include <linux/delay.h>
+#include <linux/cpu.h>
 
 #include <asm/ptrace.h>
 #include <linux/atomic.h>
@@ -32,8 +33,10 @@
 #include <asm/cacheflush.h>
 #include <asm/tlbflush.h>
 #include <asm/cpudata.h>
+#include <asm/timer.h>
 #include <asm/leon.h>
 
+#include "kernel.h"
 #include "irq.h"
 
 volatile unsigned long cpu_callin_map[NR_CPUS] __cpuinitdata = {0,};
@@ -294,6 +297,89 @@ int __cpuinit __cpu_up(unsigned int cpu, struct task_struct *tidle)
 	return ret;
 }
 
+void __cpuinit arch_cpu_pre_starting(void *arg)
+{
+	local_ops->cache_all();
+	local_ops->tlb_all();
+
+	switch(sparc_cpu_model) {
+	case sun4m:
+		sun4m_cpu_pre_starting(arg);
+		break;
+	case sun4d:
+		sun4d_cpu_pre_starting(arg);
+		break;
+	case sparc_leon:
+		leon_cpu_pre_starting(arg);
+		break;
+	default:
+		BUG();
+	}
+}
+
+void __cpuinit arch_cpu_pre_online(void *arg)
+{
+	unsigned int cpuid = hard_smp_processor_id();
+
+	register_percpu_ce(cpuid);
+
+	calibrate_delay();
+	smp_store_cpu_info(cpuid);
+
+	local_ops->cache_all();
+	local_ops->tlb_all();
+
+	switch(sparc_cpu_model) {
+	case sun4m:
+		sun4m_cpu_pre_online(arg);
+		break;
+	case sun4d:
+		sun4d_cpu_pre_online(arg);
+		break;
+	case sparc_leon:
+		leon_cpu_pre_online(arg);
+		break;
+	default:
+		BUG();
+	}
+}
+
+void __cpuinit sparc_start_secondary(void *arg)
+{
+	unsigned int cpu;
+
+	/*
+	 * SMP booting is extremely fragile in some architectures. So run
+	 * the cpu initialization code first before anything else.
+	 */
+	arch_cpu_pre_starting(arg);
+
+	preempt_disable();
+	cpu = smp_processor_id();
+
+	/* Invoke the CPU_STARTING notifier callbacks */
+	notify_cpu_starting(cpu);
+
+	arch_cpu_pre_online(arg);
+
+	/* Set the CPU in the cpu_online_mask */
+	set_cpu_online(cpu, true);
+
+	/* Enable local interrupts now */
+	local_irq_enable();
+
+	wmb();
+	cpu_idle();
+
+	/* We should never reach here! */
+	BUG();
+}
+
+void __cpuinit smp_callin(void)
+{
+	sparc_start_secondary(NULL);
+}
+
 void smp_bogo(struct seq_file *m)
 {
 	int i;
diff --git a/arch/sparc/kernel/smp_64.c b/arch/sparc/kernel/smp_64.c
index d94b878577b7..537eb66abd06 100644
--- a/arch/sparc/kernel/smp_64.c
+++ b/arch/sparc/kernel/smp_64.c
@@ -1180,7 +1180,7 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 {
 }
 
-void __devinit smp_prepare_boot_cpu(void)
+void smp_prepare_boot_cpu(void)
 {
 }
 
@@ -1194,7 +1194,7 @@ void __init smp_setup_processor_id(void)
 		xcall_deliver_impl = hypervisor_xcall_deliver;
 }
 
-void __devinit smp_fill_in_sib_core_maps(void)
+void smp_fill_in_sib_core_maps(void)
 {
 	unsigned int i;
 
diff --git a/arch/sparc/kernel/sun4d_smp.c b/arch/sparc/kernel/sun4d_smp.c
index ddaea31de586..c9eb82f23d92 100644
--- a/arch/sparc/kernel/sun4d_smp.c
+++ b/arch/sparc/kernel/sun4d_smp.c
@@ -50,10 +50,9 @@ static inline void show_leds(int cpuid)
 			      "i" (ASI_M_CTL));
 }
 
-void __cpuinit smp4d_callin(void)
+void __cpuinit sun4d_cpu_pre_starting(void *arg)
 {
 	int cpuid = hard_smp_processor_id();
-	unsigned long flags;
 
 	/* Show we are alive */
 	cpu_leds[cpuid] = 0x6;
@@ -61,26 +60,20 @@ void __cpuinit smp4d_callin(void)
 
 	/* Enable level15 interrupt, disable level14 interrupt for now */
 	cc_set_imsk((cc_get_imsk() & ~0x8000) | 0x4000);
+}
 
-	local_ops->cache_all();
-	local_ops->tlb_all();
+void __cpuinit sun4d_cpu_pre_online(void *arg)
+{
+	unsigned long flags;
+	int cpuid;
 
-	notify_cpu_starting(cpuid);
-	/*
-	 * Unblock the master CPU _only_ when the scheduler state
+	cpuid = hard_smp_processor_id();
+
+	/* Unblock the master CPU _only_ when the scheduler state
 	 * of all secondary CPUs will be up-to-date, so after
 	 * the SMP initialization the master will be just allowed
 	 * to call the scheduler code.
 	 */
-	/* Get our local ticker going. */
-	register_percpu_ce(cpuid);
-
-	calibrate_delay();
-	smp_store_cpu_info(cpuid);
-	local_ops->cache_all();
-	local_ops->tlb_all();
-
-	/* Allow master to continue. */
 	sun4d_swap((unsigned long *)&cpu_callin_map[cpuid], 1);
 	local_ops->cache_all();
 	local_ops->tlb_all();
@@ -106,16 +99,12 @@ void __cpuinit smp4d_callin(void)
 	local_ops->cache_all();
 	local_ops->tlb_all();
 
-	local_irq_enable();	/* We don't allow PIL 14 yet */
-
 	while (!cpumask_test_cpu(cpuid, &smp_commenced_mask))
 		barrier();
 
 	spin_lock_irqsave(&sun4d_imsk_lock, flags);
 	cc_set_imsk(cc_get_imsk() & ~0x4000); /* Allow PIL 14 as well */
 	spin_unlock_irqrestore(&sun4d_imsk_lock, flags);
-	set_cpu_online(cpuid, true);
-
 }
 
 /*
diff --git a/arch/sparc/kernel/sun4m_smp.c b/arch/sparc/kernel/sun4m_smp.c
index 128af7304288..8a65f158153d 100644
--- a/arch/sparc/kernel/sun4m_smp.c
+++ b/arch/sparc/kernel/sun4m_smp.c
@@ -34,30 +34,19 @@ swap_ulong(volatile unsigned long *ptr, unsigned long val)
 	return val;
 }
 
-void __cpuinit smp4m_callin(void)
+void __cpuinit sun4m_cpu_pre_starting(void *arg)
 {
-	int cpuid = hard_smp_processor_id();
-
-	local_ops->cache_all();
-	local_ops->tlb_all();
-
-	notify_cpu_starting(cpuid);
-
-	register_percpu_ce(cpuid);
-
-	calibrate_delay();
-	smp_store_cpu_info(cpuid);
+}
 
-	local_ops->cache_all();
-	local_ops->tlb_all();
+void __cpuinit sun4m_cpu_pre_online(void *arg)
+{
+	int cpuid = hard_smp_processor_id();
 
-	/*
-	 * Unblock the master CPU _only_ when the scheduler state
-	 * of all secondary CPUs will be up-to-date, so after
-	 * the SMP initialization the master will be just allowed
-	 * to call the scheduler code.
+	/* Allow master to continue. The master will then give us the
+	 * go-ahead by setting the smp_commenced_mask and will wait without
+	 * timeouts until our setup is completed fully (signified by
+	 * our bit being set in the cpu_online_mask).
 	 */
-	/* Allow master to continue. */
 	swap_ulong(&cpu_callin_map[cpuid], 1);
 
 	/* XXX: What's up with all the flushes? */
@@ -75,10 +64,6 @@ void __cpuinit smp4m_callin(void)
 
 	while (!cpumask_test_cpu(cpuid, &smp_commenced_mask))
 		mb();
-
-	local_irq_enable();
-
-	set_cpu_online(cpuid, true);
 }
 
 /*
diff --git a/arch/sparc/kernel/sys32.S b/arch/sparc/kernel/sys32.S
index 44025f4ba41f..240a3cecc11e 100644
--- a/arch/sparc/kernel/sys32.S
+++ b/arch/sparc/kernel/sys32.S
@@ -36,108 +36,22 @@ STUB:	sra	REG1, 0, REG1; \
 	jmpl	%g1 + %lo(SYSCALL), %g0; \
 	sra	REG3, 0, REG3
 
-#define SIGN4(STUB,SYSCALL,REG1,REG2,REG3,REG4) \
-	.align	32; \
-	.globl	STUB; \
-STUB:	sra	REG1, 0, REG1; \
-	sethi	%hi(SYSCALL), %g1; \
-	sra	REG2, 0, REG2; \
-	sra	REG3, 0, REG3; \
-	jmpl	%g1 + %lo(SYSCALL), %g0; \
-	sra	REG4, 0, REG4
-
-SIGN1(sys32_exit, sparc_exit, %o0)
-SIGN1(sys32_exit_group, sys_exit_group, %o0)
-SIGN1(sys32_wait4, compat_sys_wait4, %o2)
-SIGN1(sys32_creat, sys_creat, %o1)
-SIGN1(sys32_mknod, sys_mknod, %o1)
-SIGN1(sys32_umount, sys_umount, %o1)
-SIGN1(sys32_signal, sys_signal, %o0)
-SIGN1(sys32_access, sys_access, %o1)
-SIGN1(sys32_msync, sys_msync, %o2)
-SIGN2(sys32_reboot, sys_reboot, %o0, %o1)
-SIGN1(sys32_setitimer, compat_sys_setitimer, %o0)
-SIGN1(sys32_getitimer, compat_sys_getitimer, %o0)
-SIGN1(sys32_sethostname, sys_sethostname, %o1)
-SIGN1(sys32_swapon, sys_swapon, %o1)
-SIGN1(sys32_sigaction, compat_sys_sigaction, %o0)
-SIGN1(sys32_rt_sigaction, compat_sys_rt_sigaction, %o0)
-SIGN1(sys32_sigprocmask, compat_sys_sigprocmask, %o0)
-SIGN1(sys32_rt_sigprocmask, compat_sys_rt_sigprocmask, %o0)
-SIGN2(sys32_rt_sigqueueinfo, compat_sys_rt_sigqueueinfo, %o0, %o1)
 SIGN1(sys32_getrusage, compat_sys_getrusage, %o0)
-SIGN1(sys32_setxattr, sys_setxattr, %o4)
-SIGN1(sys32_lsetxattr, sys_lsetxattr, %o4)
-SIGN1(sys32_fsetxattr, sys_fsetxattr, %o4)
-SIGN1(sys32_fgetxattr, sys_fgetxattr, %o0)
-SIGN1(sys32_flistxattr, sys_flistxattr, %o0)
-SIGN1(sys32_fremovexattr, sys_fremovexattr, %o0)
-SIGN2(sys32_tkill, sys_tkill, %o0, %o1)
-SIGN1(sys32_epoll_create, sys_epoll_create, %o0)
-SIGN3(sys32_epoll_ctl, sys_epoll_ctl, %o0, %o1, %o2)
-SIGN3(sys32_epoll_wait, sys_epoll_wait, %o0, %o2, %o3)
 SIGN1(sys32_readahead, compat_sys_readahead, %o0)
 SIGN2(sys32_fadvise64, compat_sys_fadvise64, %o0, %o4)
 SIGN2(sys32_fadvise64_64, compat_sys_fadvise64_64, %o0, %o5)
-SIGN2(sys32_bdflush, sys_bdflush, %o0, %o1)
-SIGN1(sys32_mlockall, sys_mlockall, %o0)
 SIGN1(sys32_clock_nanosleep, compat_sys_clock_nanosleep, %o1)
 SIGN1(sys32_timer_settime, compat_sys_timer_settime, %o1)
 SIGN1(sys32_io_submit, compat_sys_io_submit, %o1)
 SIGN1(sys32_mq_open, compat_sys_mq_open, %o1)
 SIGN1(sys32_select, compat_sys_select, %o0)
-SIGN1(sys32_mkdir, sys_mkdir, %o1)
 SIGN3(sys32_futex, compat_sys_futex, %o1, %o2, %o5)
-SIGN1(sys32_sysfs, compat_sys_sysfs, %o0)
 SIGN2(sys32_sendfile, compat_sys_sendfile, %o0, %o1)
-SIGN2(sys32_sendfile64, sys_sendfile, %o0, %o1)
-SIGN1(sys32_prctl, sys_prctl, %o0)
-SIGN1(sys32_sched_rr_get_interval, compat_sys_sched_rr_get_interval, %o0)
-SIGN2(sys32_waitpid, sys_waitpid, %o0, %o2)
-SIGN1(sys32_getgroups, sys_getgroups, %o0)
-SIGN1(sys32_getpgid, sys_getpgid, %o0)
-SIGN2(sys32_getpriority, sys_getpriority, %o0, %o1)
-SIGN1(sys32_getsid, sys_getsid, %o0)
-SIGN2(sys32_kill, sys_kill, %o0, %o1)
-SIGN1(sys32_nice, sys_nice, %o0)
-SIGN1(sys32_lseek, sys_lseek, %o1)
-SIGN2(sys32_open, sparc32_open, %o1, %o2)
-SIGN1(sys32_readlink, sys_readlink, %o2)
-SIGN1(sys32_sched_get_priority_max, sys_sched_get_priority_max, %o0)
-SIGN1(sys32_sched_get_priority_min, sys_sched_get_priority_min, %o0)
-SIGN1(sys32_sched_getparam, sys_sched_getparam, %o0)
-SIGN1(sys32_sched_getscheduler, sys_sched_getscheduler, %o0)
-SIGN1(sys32_sched_setparam, sys_sched_setparam, %o0)
-SIGN2(sys32_sched_setscheduler, sys_sched_setscheduler, %o0, %o1)
-SIGN1(sys32_getdomainname, sys_getdomainname, %o1)
-SIGN1(sys32_setdomainname, sys_setdomainname, %o1)
-SIGN1(sys32_setgroups, sys_setgroups, %o0)
-SIGN2(sys32_setpgid, sys_setpgid, %o0, %o1)
-SIGN3(sys32_setpriority, sys_setpriority, %o0, %o1, %o2)
-SIGN1(sys32_ssetmask, sys_ssetmask, %o0)
-SIGN2(sys32_syslog, sys_syslog, %o0, %o2)
-SIGN1(sys32_umask, sys_umask, %o0)
-SIGN3(sys32_tgkill, sys_tgkill, %o0, %o1, %o2)
-SIGN1(sys32_sendto, sys_sendto, %o0)
 SIGN1(sys32_recvfrom, compat_sys_recvfrom, %o0)
-SIGN3(sys32_socket, sys_socket, %o0, %o1, %o2)
-SIGN2(sys32_connect, sys_connect, %o0, %o2)
-SIGN2(sys32_bind, sys_bind, %o0, %o2)
-SIGN2(sys32_listen, sys_listen, %o0, %o1)
 SIGN1(sys32_recvmsg, compat_sys_recvmsg, %o0)
 SIGN1(sys32_sendmsg, compat_sys_sendmsg, %o0)
-SIGN2(sys32_shutdown, sys_shutdown, %o0, %o1)
-SIGN3(sys32_socketpair, sys_socketpair, %o0, %o1, %o2)
-SIGN1(sys32_getpeername, sys_getpeername, %o0)
-SIGN1(sys32_getsockname, sys_getsockname, %o0)
-SIGN2(sys32_ioprio_get, sys_ioprio_get, %o0, %o1)
-SIGN3(sys32_ioprio_set, sys_ioprio_set, %o0, %o1, %o2)
-SIGN2(sys32_splice, sys_splice, %o0, %o2)
 SIGN2(sys32_sync_file_range, compat_sync_file_range, %o0, %o5)
-SIGN2(sys32_tee, sys_tee, %o0, %o1)
 SIGN1(sys32_vmsplice, compat_sys_vmsplice, %o0)
-SIGN1(sys32_truncate, sys_truncate, %o1)
-SIGN1(sys32_ftruncate, sys_ftruncate, %o1)
 
 	.globl		sys32_mmap2
 sys32_mmap2:
diff --git a/arch/sparc/kernel/sys_sparc32.c b/arch/sparc/kernel/sys_sparc32.c
index c3239811a1b5..f38f2280fade 100644
--- a/arch/sparc/kernel/sys_sparc32.c
+++ b/arch/sparc/kernel/sys_sparc32.c
@@ -206,147 +206,19 @@ asmlinkage long compat_sys_fstatat64(unsigned int dfd,
 	return cp_compat_stat64(&stat, statbuf);
 }
 
-asmlinkage long compat_sys_sysfs(int option, u32 arg1, u32 arg2)
+COMPAT_SYSCALL_DEFINE3(sparc_sigaction, int, sig,
+			struct compat_old_sigaction __user *,act,
+			struct compat_old_sigaction __user *,oact)
 {
-	return sys_sysfs(option, arg1, arg2);
-}
-
-asmlinkage long compat_sys_sched_rr_get_interval(compat_pid_t pid, struct compat_timespec __user *interval)
-{
-	struct timespec t;
-	int ret;
-	mm_segment_t old_fs = get_fs ();
-	
-	set_fs (KERNEL_DS);
-	ret = sys_sched_rr_get_interval(pid, (struct timespec __user *) &t);
-	set_fs (old_fs);
-	if (put_compat_timespec(&t, interval))
-		return -EFAULT;
-	return ret;
-}
-
-asmlinkage long compat_sys_rt_sigprocmask(int how,
-					  compat_sigset_t __user *set,
-					  compat_sigset_t __user *oset,
-					  compat_size_t sigsetsize)
-{
-	sigset_t s;
-	compat_sigset_t s32;
-	int ret;
-	mm_segment_t old_fs = get_fs();
-	
-	if (set) {
-		if (copy_from_user (&s32, set, sizeof(compat_sigset_t)))
-			return -EFAULT;
-		switch (_NSIG_WORDS) {
-		case 4: s.sig[3] = s32.sig[6] | (((long)s32.sig[7]) << 32);
-		case 3: s.sig[2] = s32.sig[4] | (((long)s32.sig[5]) << 32);
-		case 2: s.sig[1] = s32.sig[2] | (((long)s32.sig[3]) << 32);
-		case 1: s.sig[0] = s32.sig[0] | (((long)s32.sig[1]) << 32);
-		}
-	}
-	set_fs (KERNEL_DS);
-	ret = sys_rt_sigprocmask(how,
-				 set ? (sigset_t __user *) &s : NULL,
-				 oset ? (sigset_t __user *) &s : NULL,
-				 sigsetsize);
-	set_fs (old_fs);
-	if (ret) return ret;
-	if (oset) {
-		switch (_NSIG_WORDS) {
-		case 4: s32.sig[7] = (s.sig[3] >> 32); s32.sig[6] = s.sig[3];
-		case 3: s32.sig[5] = (s.sig[2] >> 32); s32.sig[4] = s.sig[2];
-		case 2: s32.sig[3] = (s.sig[1] >> 32); s32.sig[2] = s.sig[1];
-		case 1: s32.sig[1] = (s.sig[0] >> 32); s32.sig[0] = s.sig[0];
-		}
-		if (copy_to_user (oset, &s32, sizeof(compat_sigset_t)))
-			return -EFAULT;
-	}
-	return 0;
-}
-
-asmlinkage long sys32_rt_sigpending(compat_sigset_t __user *set,
-				    compat_size_t sigsetsize)
-{
-	sigset_t s;
-	compat_sigset_t s32;
-	int ret;
-	mm_segment_t old_fs = get_fs();
-		
-	set_fs (KERNEL_DS);
-	ret = sys_rt_sigpending((sigset_t __user *) &s, sigsetsize);
-	set_fs (old_fs);
-	if (!ret) {
-		switch (_NSIG_WORDS) {
-		case 4: s32.sig[7] = (s.sig[3] >> 32); s32.sig[6] = s.sig[3];
-		case 3: s32.sig[5] = (s.sig[2] >> 32); s32.sig[4] = s.sig[2];
-		case 2: s32.sig[3] = (s.sig[1] >> 32); s32.sig[2] = s.sig[1];
-		case 1: s32.sig[1] = (s.sig[0] >> 32); s32.sig[0] = s.sig[0];
-		}
-		if (copy_to_user (set, &s32, sizeof(compat_sigset_t)))
-			return -EFAULT;
-	}
-	return ret;
-}
-
-asmlinkage long compat_sys_rt_sigqueueinfo(int pid, int sig,
-					   struct compat_siginfo __user *uinfo)
-{
-	siginfo_t info;
-	int ret;
-	mm_segment_t old_fs = get_fs();
-	
-	if (copy_siginfo_from_user32(&info, uinfo))
-		return -EFAULT;
-
-	set_fs (KERNEL_DS);
-	ret = sys_rt_sigqueueinfo(pid, sig, (siginfo_t __user *) &info);
-	set_fs (old_fs);
-	return ret;
-}
-
-asmlinkage long compat_sys_sigaction(int sig, struct old_sigaction32 __user *act,
-				     struct old_sigaction32 __user *oact)
-{
-        struct k_sigaction new_ka, old_ka;
-        int ret;
-
 	WARN_ON_ONCE(sig >= 0);
-	sig = -sig;
-
-        if (act) {
-		compat_old_sigset_t mask;
-		u32 u_handler, u_restorer;
-		
-		ret = get_user(u_handler, &act->sa_handler);
-		new_ka.sa.sa_handler =  compat_ptr(u_handler);
-		ret |= __get_user(u_restorer, &act->sa_restorer);
-		new_ka.sa.sa_restorer = compat_ptr(u_restorer);
-		ret |= __get_user(new_ka.sa.sa_flags, &act->sa_flags);
-		ret |= __get_user(mask, &act->sa_mask);
-		if (ret)
-			return ret;
-		new_ka.ka_restorer = NULL;
-		siginitset(&new_ka.sa.sa_mask, mask);
-        }
-
-        ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL);
-
-	if (!ret && oact) {
-		ret = put_user(ptr_to_compat(old_ka.sa.sa_handler), &oact->sa_handler);
-		ret |= __put_user(ptr_to_compat(old_ka.sa.sa_restorer), &oact->sa_restorer);
-		ret |= __put_user(old_ka.sa.sa_flags, &oact->sa_flags);
-		ret |= __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask);
-        }
-
-	return ret;
+	return compat_sys_sigaction(-sig, act, oact);
 }
 
-asmlinkage long compat_sys_rt_sigaction(int sig,
-					struct sigaction32 __user *act,
-					struct sigaction32 __user *oact,
-					void __user *restorer,
-					compat_size_t sigsetsize)
+COMPAT_SYSCALL_DEFINE5(rt_sigaction, int, sig,
+			struct compat_sigaction __user *,act,
+			struct compat_sigaction __user *,oact,
+			void __user *,restorer,
+			compat_size_t,sigsetsize)
 {
         struct k_sigaction new_ka, old_ka;
         int ret;
@@ -363,12 +235,7 @@ asmlinkage long compat_sys_rt_sigaction(int sig,
 		ret = get_user(u_handler, &act->sa_handler);
 		new_ka.sa.sa_handler =  compat_ptr(u_handler);
 		ret |= __copy_from_user(&set32, &act->sa_mask, sizeof(compat_sigset_t));
-		switch (_NSIG_WORDS) {
-		case 4: new_ka.sa.sa_mask.sig[3] = set32.sig[6] | (((long)set32.sig[7]) << 32);
-		case 3: new_ka.sa.sa_mask.sig[2] = set32.sig[4] | (((long)set32.sig[5]) << 32);
-		case 2: new_ka.sa.sa_mask.sig[1] = set32.sig[2] | (((long)set32.sig[3]) << 32);
-		case 1: new_ka.sa.sa_mask.sig[0] = set32.sig[0] | (((long)set32.sig[1]) << 32);
-		}
+		sigset_from_compat(&new_ka.sa.sa_mask, &set32);
 		ret |= __get_user(new_ka.sa.sa_flags, &act->sa_flags);
 		ret |= __get_user(u_restorer, &act->sa_restorer);
 		new_ka.sa.sa_restorer = compat_ptr(u_restorer);
@@ -379,12 +246,7 @@ asmlinkage long compat_sys_rt_sigaction(int sig,
 	ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL);
 
 	if (!ret && oact) {
-		switch (_NSIG_WORDS) {
-		case 4: set32.sig[7] = (old_ka.sa.sa_mask.sig[3] >> 32); set32.sig[6] = old_ka.sa.sa_mask.sig[3];
-		case 3: set32.sig[5] = (old_ka.sa.sa_mask.sig[2] >> 32); set32.sig[4] = old_ka.sa.sa_mask.sig[2];
-		case 2: set32.sig[3] = (old_ka.sa.sa_mask.sig[1] >> 32); set32.sig[2] = old_ka.sa.sa_mask.sig[1];
-		case 1: set32.sig[1] = (old_ka.sa.sa_mask.sig[0] >> 32); set32.sig[0] = old_ka.sa.sa_mask.sig[0];
-		}
+		sigset_to_compat(&set32, &old_ka.sa.sa_mask);
 		ret = put_user(ptr_to_compat(old_ka.sa.sa_handler), &oact->sa_handler);
 		ret |= __copy_to_user(&oact->sa_mask, &set32, sizeof(compat_sigset_t));
 		ret |= __put_user(old_ka.sa.sa_flags, &oact->sa_flags);
@@ -396,71 +258,6 @@ asmlinkage long compat_sys_rt_sigaction(int sig,
         return ret;
 }
 
-/*
- * sparc32_execve() executes a new program after the asm stub has set
- * things up for us.  This should basically do what I want it to.
- */
-asmlinkage long sparc32_execve(struct pt_regs *regs)
-{
-	int error, base = 0;
-	struct filename *filename;
-
-	/* User register window flush is done by entry.S */
-
-	/* Check for indirect call. */
-	if ((u32)regs->u_regs[UREG_G1] == 0)
-		base = 1;
-
-	filename = getname(compat_ptr(regs->u_regs[base + UREG_I0]));
-	error = PTR_ERR(filename);
-	if (IS_ERR(filename))
-		goto out;
-
-	error = compat_do_execve(filename->name,
-				 compat_ptr(regs->u_regs[base + UREG_I1]),
-				 compat_ptr(regs->u_regs[base + UREG_I2]), regs);
-
-	putname(filename);
-
-	if (!error) {
-		fprs_write(0);
-		current_thread_info()->xfsr[0] = 0;
-		current_thread_info()->fpsaved[0] = 0;
-		regs->tstate &= ~TSTATE_PEF;
-	}
-out:
-	return error;
-}
-
-#ifdef CONFIG_MODULES
-
-asmlinkage long sys32_init_module(void __user *umod, u32 len,
-				  const char __user *uargs)
-{
-	return sys_init_module(umod, len, uargs);
-}
-
-asmlinkage long sys32_delete_module(const char __user *name_user,
-				    unsigned int flags)
-{
-	return sys_delete_module(name_user, flags);
-}
-
-#else /* CONFIG_MODULES */
-
-asmlinkage long sys32_init_module(const char __user *name_user,
-				  struct module __user *mod_user)
-{
-	return -ENOSYS;
-}
-
-asmlinkage long sys32_delete_module(const char __user *name_user)
-{
-	return -ENOSYS;
-}
-
-#endif  /* CONFIG_MODULES */
-
 asmlinkage compat_ssize_t sys32_pread64(unsigned int fd,
 					char __user *ubuf,
 					compat_size_t count,
@@ -506,16 +303,6 @@ long compat_sys_fadvise64_64(int fd,
 				advice);
 }
 
-/* This is just a version for 32-bit applications which does
- * not force O_LARGEFILE on.
- */
-
-asmlinkage long sparc32_open(const char __user *filename,
-			     int flags, int mode)
-{
-	return do_sys_open(AT_FDCWD, filename, flags, mode);
-}
-
 long sys32_lookup_dcookie(unsigned long cookie_high,
 			  unsigned long cookie_low,
 			  char __user *buf, size_t len)
diff --git a/arch/sparc/kernel/sys_sparc_32.c b/arch/sparc/kernel/sys_sparc_32.c
index 0c9b31b22e07..3a8d1844402e 100644
--- a/arch/sparc/kernel/sys_sparc_32.c
+++ b/arch/sparc/kernel/sys_sparc_32.c
@@ -34,11 +34,9 @@ asmlinkage unsigned long sys_getpagesize(void)
 	return PAGE_SIZE; /* Possibly older binaries want 8192 on sun4's? */
 }
 
-#define COLOUR_ALIGN(addr)      (((addr)+SHMLBA-1)&~(SHMLBA-1))
-
 unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, unsigned long len, unsigned long pgoff, unsigned long flags)
 {
-	struct vm_area_struct * vmm;
+	struct vm_unmapped_area_info info;
 
 	if (flags & MAP_FIXED) {
 		/* We do not accept a shared mapping if it would violate
@@ -56,21 +54,14 @@ unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, unsi
 	if (!addr)
 		addr = TASK_UNMAPPED_BASE;
 
-	if (flags & MAP_SHARED)
-		addr = COLOUR_ALIGN(addr);
-	else
-		addr = PAGE_ALIGN(addr);
-
-	for (vmm = find_vma(current->mm, addr); ; vmm = vmm->vm_next) {
-		/* At this point:  (!vmm || addr < vmm->vm_end). */
-		if (TASK_SIZE - PAGE_SIZE - len < addr)
-			return -ENOMEM;
-		if (!vmm || addr + len <= vmm->vm_start)
-			return addr;
-		addr = vmm->vm_end;
-		if (flags & MAP_SHARED)
-			addr = COLOUR_ALIGN(addr);
-	}
+	info.flags = 0;
+	info.length = len;
+	info.low_limit = addr;
+	info.high_limit = TASK_SIZE;
+	info.align_mask = (flags & MAP_SHARED) ?
+		(PAGE_MASK & (SHMLBA - 1)) : 0;
+	info.align_offset = pgoff << PAGE_SHIFT;
+	return vm_unmapped_area(&info);
 }
 
 /*
@@ -169,49 +160,19 @@ sparc_breakpoint (struct pt_regs *regs)
 #endif
 }
 
-asmlinkage int
-sparc_sigaction (int sig, const struct old_sigaction __user *act,
-		 struct old_sigaction __user *oact)
+SYSCALL_DEFINE3(sparc_sigaction, int, sig,
+		struct old_sigaction __user *,act,
+		struct old_sigaction __user *,oact)
 {
-	struct k_sigaction new_ka, old_ka;
-	int ret;
-
 	WARN_ON_ONCE(sig >= 0);
-	sig = -sig;
-
-	if (act) {
-		unsigned long mask;
-
-		if (!access_ok(VERIFY_READ, act, sizeof(*act)) ||
-		    __get_user(new_ka.sa.sa_handler, &act->sa_handler) ||
-		    __get_user(new_ka.sa.sa_restorer, &act->sa_restorer) ||
-		    __get_user(new_ka.sa.sa_flags, &act->sa_flags) ||
-		    __get_user(mask, &act->sa_mask))
-			return -EFAULT;
-		siginitset(&new_ka.sa.sa_mask, mask);
-		new_ka.ka_restorer = NULL;
-	}
-
-	ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL);
-
-	if (!ret && oact) {
-		if (!access_ok(VERIFY_WRITE, oact, sizeof(*oact)) ||
-		    __put_user(old_ka.sa.sa_handler, &oact->sa_handler) ||
-		    __put_user(old_ka.sa.sa_restorer, &oact->sa_restorer) ||
-		    __put_user(old_ka.sa.sa_flags, &oact->sa_flags) ||
-		    __put_user(old_ka.sa.sa_mask.sig[0], &oact->sa_mask))
-			return -EFAULT;
-	}
-
-	return ret;
+	return sys_sigaction(-sig, act, oact);
 }
 
-asmlinkage long
-sys_rt_sigaction(int sig,
-		 const struct sigaction __user *act,
-		 struct sigaction __user *oact,
-		 void __user *restorer,
-		 size_t sigsetsize)
+SYSCALL_DEFINE5(rt_sigaction, int, sig,
+		 const struct sigaction __user *, act,
+		 struct sigaction __user *, oact,
+		 void __user *, restorer,
+		 size_t, sigsetsize)
 {
 	struct k_sigaction new_ka, old_ka;
 	int ret;
@@ -258,27 +219,3 @@ out:
 	up_read(&uts_sem);
 	return err;
 }
-
-/*
- * Do a system call from kernel instead of calling sys_execve so we
- * end up with proper pt_regs.
- */
-int kernel_execve(const char *filename,
-		  const char *const argv[],
-		  const char *const envp[])
-{
-	long __res;
-	register long __g1 __asm__ ("g1") = __NR_execve;
-	register long __o0 __asm__ ("o0") = (long)(filename);
-	register long __o1 __asm__ ("o1") = (long)(argv);
-	register long __o2 __asm__ ("o2") = (long)(envp);
-	asm volatile ("t 0x10\n\t"
-		      "bcc 1f\n\t"
-		      "mov %%o0, %0\n\t"
-		      "sub %%g0, %%o0, %0\n\t"
-		      "1:\n\t"
-		      : "=r" (__res), "=&r" (__o0)
-		      : "1" (__o0), "r" (__o1), "r" (__o2), "r" (__g1)
-		      : "cc");
-	return __res;
-}
diff --git a/arch/sparc/kernel/sys_sparc_64.c b/arch/sparc/kernel/sys_sparc_64.c
index 11c6c9603e71..708bc29d36a8 100644
--- a/arch/sparc/kernel/sys_sparc_64.c
+++ b/arch/sparc/kernel/sys_sparc_64.c
@@ -75,7 +75,7 @@ static inline int invalid_64bit_range(unsigned long addr, unsigned long len)
  *    the spitfire/niagara VA-hole.
  */
 
-static inline unsigned long COLOUR_ALIGN(unsigned long addr,
+static inline unsigned long COLOR_ALIGN(unsigned long addr,
 					 unsigned long pgoff)
 {
 	unsigned long base = (addr+SHMLBA-1)&~(SHMLBA-1);
@@ -84,24 +84,13 @@ static inline unsigned long COLOUR_ALIGN(unsigned long addr,
 	return base + off;
 }
 
-static inline unsigned long COLOUR_ALIGN_DOWN(unsigned long addr,
-					      unsigned long pgoff)
-{
-	unsigned long base = addr & ~(SHMLBA-1);
-	unsigned long off = (pgoff<<PAGE_SHIFT) & (SHMLBA-1);
-
-	if (base + off <= addr)
-		return base + off;
-	return base - off;
-}
-
 unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, unsigned long len, unsigned long pgoff, unsigned long flags)
 {
 	struct mm_struct *mm = current->mm;
 	struct vm_area_struct * vma;
 	unsigned long task_size = TASK_SIZE;
-	unsigned long start_addr;
 	int do_color_align;
+	struct vm_unmapped_area_info info;
 
 	if (flags & MAP_FIXED) {
 		/* We do not accept a shared mapping if it would violate
@@ -124,7 +113,7 @@ unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, unsi
 
 	if (addr) {
 		if (do_color_align)
-			addr = COLOUR_ALIGN(addr, pgoff);
+			addr = COLOR_ALIGN(addr, pgoff);
 		else
 			addr = PAGE_ALIGN(addr);
 
@@ -134,50 +123,22 @@ unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, unsi
 			return addr;
 	}
 
-	if (len > mm->cached_hole_size) {
-	        start_addr = addr = mm->free_area_cache;
-	} else {
-	        start_addr = addr = TASK_UNMAPPED_BASE;
-	        mm->cached_hole_size = 0;
+	info.flags = 0;
+	info.length = len;
+	info.low_limit = TASK_UNMAPPED_BASE;
+	info.high_limit = min(task_size, VA_EXCLUDE_START);
+	info.align_mask = do_color_align ? (PAGE_MASK & (SHMLBA - 1)) : 0;
+	info.align_offset = pgoff << PAGE_SHIFT;
+	addr = vm_unmapped_area(&info);
+
+	if ((addr & ~PAGE_MASK) && task_size > VA_EXCLUDE_END) {
+		VM_BUG_ON(addr != -ENOMEM);
+		info.low_limit = VA_EXCLUDE_END;
+		info.high_limit = task_size;
+		addr = vm_unmapped_area(&info);
 	}
 
-	task_size -= len;
-
-full_search:
-	if (do_color_align)
-		addr = COLOUR_ALIGN(addr, pgoff);
-	else
-		addr = PAGE_ALIGN(addr);
-
-	for (vma = find_vma(mm, addr); ; vma = vma->vm_next) {
-		/* At this point:  (!vma || addr < vma->vm_end). */
-		if (addr < VA_EXCLUDE_START &&
-		    (addr + len) >= VA_EXCLUDE_START) {
-			addr = VA_EXCLUDE_END;
-			vma = find_vma(mm, VA_EXCLUDE_END);
-		}
-		if (unlikely(task_size < addr)) {
-			if (start_addr != TASK_UNMAPPED_BASE) {
-				start_addr = addr = TASK_UNMAPPED_BASE;
-				mm->cached_hole_size = 0;
-				goto full_search;
-			}
-			return -ENOMEM;
-		}
-		if (likely(!vma || addr + len <= vma->vm_start)) {
-			/*
-			 * Remember the place where we stopped the search:
-			 */
-			mm->free_area_cache = addr + len;
-			return addr;
-		}
-		if (addr + mm->cached_hole_size < vma->vm_start)
-		        mm->cached_hole_size = vma->vm_start - addr;
-
-		addr = vma->vm_end;
-		if (do_color_align)
-			addr = COLOUR_ALIGN(addr, pgoff);
-	}
+	return addr;
 }
 
 unsigned long
@@ -190,6 +151,7 @@ arch_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
 	unsigned long task_size = STACK_TOP32;
 	unsigned long addr = addr0;
 	int do_color_align;
+	struct vm_unmapped_area_info info;
 
 	/* This should only ever run for 32-bit processes.  */
 	BUG_ON(!test_thread_flag(TIF_32BIT));
@@ -214,7 +176,7 @@ arch_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
 	/* requesting a specific address */
 	if (addr) {
 		if (do_color_align)
-			addr = COLOUR_ALIGN(addr, pgoff);
+			addr = COLOR_ALIGN(addr, pgoff);
 		else
 			addr = PAGE_ALIGN(addr);
 
@@ -224,73 +186,27 @@ arch_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
 			return addr;
 	}
 
-	/* check if free_area_cache is useful for us */
-	if (len <= mm->cached_hole_size) {
- 	        mm->cached_hole_size = 0;
- 		mm->free_area_cache = mm->mmap_base;
- 	}
+	info.flags = VM_UNMAPPED_AREA_TOPDOWN;
+	info.length = len;
+	info.low_limit = PAGE_SIZE;
+	info.high_limit = mm->mmap_base;
+	info.align_mask = do_color_align ? (PAGE_MASK & (SHMLBA - 1)) : 0;
+	info.align_offset = pgoff << PAGE_SHIFT;
+	addr = vm_unmapped_area(&info);
 
-	/* either no address requested or can't fit in requested address hole */
-	addr = mm->free_area_cache;
-	if (do_color_align) {
-		unsigned long base = COLOUR_ALIGN_DOWN(addr-len, pgoff);
-
-		addr = base + len;
-	}
-
-	/* make sure it can fit in the remaining address space */
-	if (likely(addr > len)) {
-		vma = find_vma(mm, addr-len);
-		if (!vma || addr <= vma->vm_start) {
-			/* remember the address as a hint for next time */
-			return (mm->free_area_cache = addr-len);
-		}
-	}
-
-	if (unlikely(mm->mmap_base < len))
-		goto bottomup;
-
-	addr = mm->mmap_base-len;
-	if (do_color_align)
-		addr = COLOUR_ALIGN_DOWN(addr, pgoff);
-
-	do {
-		/*
-		 * Lookup failure means no vma is above this address,
-		 * else if new region fits below vma->vm_start,
-		 * return with success:
-		 */
-		vma = find_vma(mm, addr);
-		if (likely(!vma || addr+len <= vma->vm_start)) {
-			/* remember the address as a hint for next time */
-			return (mm->free_area_cache = addr);
-		}
-
- 		/* remember the largest hole we saw so far */
- 		if (addr + mm->cached_hole_size < vma->vm_start)
- 		        mm->cached_hole_size = vma->vm_start - addr;
-
-		/* try just below the current vma->vm_start */
-		addr = vma->vm_start-len;
-		if (do_color_align)
-			addr = COLOUR_ALIGN_DOWN(addr, pgoff);
-	} while (likely(len < vma->vm_start));
-
-bottomup:
 	/*
 	 * A failed mmap() very likely causes application failure,
 	 * so fall back to the bottom-up function here. This scenario
 	 * can happen with large stack limits and large mmap()
 	 * allocations.
 	 */
-	mm->cached_hole_size = ~0UL;
-  	mm->free_area_cache = TASK_UNMAPPED_BASE;
-	addr = arch_get_unmapped_area(filp, addr0, len, pgoff, flags);
-	/*
-	 * Restore the topdown base:
-	 */
-	mm->free_area_cache = mm->mmap_base;
-	mm->cached_hole_size = ~0UL;
+	if (addr & ~PAGE_MASK) {
+		VM_BUG_ON(addr != -ENOMEM);
+		info.flags = 0;
+		info.low_limit = TASK_UNMAPPED_BASE;
+		info.high_limit = STACK_TOP32;
+		addr = vm_unmapped_area(&info);
+	}
 
 	return addr;
 }
@@ -730,24 +646,7 @@ SYSCALL_DEFINE5(rt_sigaction, int, sig, const struct sigaction __user *, act,
 	return ret;
 }
 
-/*
- * Do a system call from kernel instead of calling sys_execve so we
- * end up with proper pt_regs.
- */
-int kernel_execve(const char *filename,
-		  const char *const argv[],
-		  const char *const envp[])
+asmlinkage long sys_kern_features(void)
 {
-	long __res;
-	register long __g1 __asm__ ("g1") = __NR_execve;
-	register long __o0 __asm__ ("o0") = (long)(filename);
-	register long __o1 __asm__ ("o1") = (long)(argv);
-	register long __o2 __asm__ ("o2") = (long)(envp);
-	asm volatile ("t 0x6d\n\t"
-		      "sub %%g0, %%o0, %0\n\t"
-		      "movcc %%xcc, %%o0, %0\n\t"
-		      : "=r" (__res), "=&r" (__o0)
-		      : "1" (__o0), "r" (__o1), "r" (__o2), "r" (__g1)
-		      : "cc");
-	return __res;
+	return KERN_FEATURE_MIXED_MODE_STACK;
 }
diff --git a/arch/sparc/kernel/syscalls.S b/arch/sparc/kernel/syscalls.S
index 7f5f65d0b3fd..22a1098961f5 100644
--- a/arch/sparc/kernel/syscalls.S
+++ b/arch/sparc/kernel/syscalls.S
@@ -1,23 +1,19 @@
 	/* SunOS's execv() call only specifies the argv argument, the
 	 * environment settings are the same as the calling processes.
 	 */
-sys_execve:
-	sethi	%hi(sparc_execve), %g1
-	ba,pt	%xcc, execve_merge
-	 or	%g1, %lo(sparc_execve), %g1
+sys64_execve:
+	set	sys_execve, %g1
+	jmpl	%g1, %g0
+	 flushw
 
 #ifdef CONFIG_COMPAT
 sunos_execv:
-	stx	%g0, [%sp + PTREGS_OFF + PT_V9_I2]
+	mov	%g0, %o2
 sys32_execve:
-	sethi	%hi(sparc32_execve), %g1
-	or	%g1, %lo(sparc32_execve), %g1
-#endif
-
-execve_merge:
-	flushw
+	set	compat_sys_execve, %g1
 	jmpl	%g1, %g0
-	 add	%sp, PTREGS_OFF, %o0
+	 flushw
+#endif
 
 	.align	32
 sys_sparc_pipe:
@@ -29,16 +25,10 @@ sys_nis_syscall:
 sys_memory_ordering:
 	ba,pt	%xcc, sparc_memory_ordering
 	 add	%sp, PTREGS_OFF, %o1
-sys_sigaltstack:
-	ba,pt	%xcc, do_sigaltstack
-	 add	%i6, STACK_BIAS, %o2
 #ifdef CONFIG_COMPAT
 sys32_sigstack:
 	ba,pt	%xcc, do_sys32_sigstack
 	 mov	%i6, %o2
-sys32_sigaltstack:
-	ba,pt	%xcc, do_sys32_sigaltstack
-	 mov	%i6, %o2
 #endif
 	.align	32
 #ifdef CONFIG_COMPAT
@@ -112,16 +102,31 @@ sys_clone:
 ret_from_syscall:
 	/* Clear current_thread_info()->new_child. */
 	stb	%g0, [%g6 + TI_NEW_CHILD]
-	ldx	[%g6 + TI_FLAGS], %l0
 	call	schedule_tail
 	 mov	%g7, %o0
+	ldx	[%sp + PTREGS_OFF + PT_V9_I0], %o0
+	brnz,pt	%o0, ret_sys_call
+	 ldx	[%g6 + TI_FLAGS], %l0
+	ldx	[%sp + PTREGS_OFF + PT_V9_G1], %l1
+	call	%l1
+	 ldx	[%sp + PTREGS_OFF + PT_V9_G2], %o0
 	ba,pt	%xcc, ret_sys_call
-	 ldx	[%sp + PTREGS_OFF + PT_V9_I0], %o0
+	 mov	0, %o0
+
+	.globl	sparc_exit_group
+	.type	sparc_exit_group,#function
+sparc_exit_group:
+	sethi	%hi(sys_exit_group), %g7
+	ba,pt	%xcc, 1f
+	 or	%g7, %lo(sys_exit_group), %g7
+	.size	sparc_exit_group,.-sparc_exit_group
 
 	.globl	sparc_exit
 	.type	sparc_exit,#function
 sparc_exit:
-	rdpr	%pstate, %g2
+	sethi	%hi(sys_exit), %g7
+	or	%g7, %lo(sys_exit), %g7
+1:	rdpr	%pstate, %g2
 	wrpr	%g2, PSTATE_IE, %pstate
 	rdpr	%otherwin, %g1
 	rdpr	%cansave, %g3
@@ -129,7 +134,7 @@ sparc_exit:
 	wrpr	%g3, 0x0, %cansave
 	wrpr	%g0, 0x0, %otherwin
 	wrpr	%g2, 0x0, %pstate
-	ba,pt	%xcc, sys_exit
+	jmpl	%g7, %g0
 	 stb	%g0, [%g6 + TI_WSAVED]
 	.size	sparc_exit,.-sparc_exit
 
@@ -222,7 +227,6 @@ ret_sys_call:
 	ldx	[%sp + PTREGS_OFF + PT_V9_TNPC], %l1 ! pc = npc
 
 2:
-	stb	%g0, [%g6 + TI_SYS_NOERROR]
 	/* System call success, clear Carry condition code. */
 	andn	%g3, %g2, %g3
 3:
diff --git a/arch/sparc/kernel/systbls.h b/arch/sparc/kernel/systbls.h
index 118759cd7342..26e6dd72e92a 100644
--- a/arch/sparc/kernel/systbls.h
+++ b/arch/sparc/kernel/systbls.h
@@ -3,8 +3,8 @@
 
 #include <linux/kernel.h>
 #include <linux/types.h>
+#include <linux/signal.h>
 #include <asm/utrap.h>
-#include <asm/signal.h>
 
 extern asmlinkage unsigned long sys_getpagesize(void);
 extern asmlinkage long sparc_pipe(struct pt_regs *regs);
@@ -36,8 +36,6 @@ extern asmlinkage long sys_rt_sigaction(int sig,
 
 extern asmlinkage void sparc64_set_context(struct pt_regs *regs);
 extern asmlinkage void sparc64_get_context(struct pt_regs *regs);
-extern asmlinkage long sys_sigpause(unsigned int set);
-extern asmlinkage long sys_sigsuspend(old_sigset_t set);
 extern void do_rt_sigreturn(struct pt_regs *regs);
 
 #endif /* _SYSTBLS_H */
diff --git a/arch/sparc/kernel/systbls_32.S b/arch/sparc/kernel/systbls_32.S
index 63402f9e9f51..7b87171ecf1e 100644
--- a/arch/sparc/kernel/systbls_32.S
+++ b/arch/sparc/kernel/systbls_32.S
@@ -55,7 +55,7 @@ sys_call_table:
 /*180*/	.long sys_flistxattr, sys_removexattr, sys_lremovexattr, sys_sigpending, sys_ni_syscall
 /*185*/	.long sys_setpgid, sys_fremovexattr, sys_tkill, sys_exit_group, sys_newuname
 /*190*/	.long sys_init_module, sys_personality, sparc_remap_file_pages, sys_epoll_create, sys_epoll_ctl
-/*195*/	.long sys_epoll_wait, sys_ioprio_set, sys_getppid, sparc_sigaction, sys_sgetmask
+/*195*/	.long sys_epoll_wait, sys_ioprio_set, sys_getppid, sys_sparc_sigaction, sys_sgetmask
 /*200*/	.long sys_ssetmask, sys_sigsuspend, sys_newlstat, sys_uselib, sys_old_readdir
 /*205*/	.long sys_readahead, sys_socketcall, sys_syslog, sys_lookup_dcookie, sys_fadvise64
 /*210*/	.long sys_fadvise64_64, sys_tgkill, sys_waitpid, sys_swapoff, sys_sysinfo
@@ -85,3 +85,4 @@ sys_call_table:
 /*325*/	.long sys_pwritev, sys_rt_tgsigqueueinfo, sys_perf_event_open, sys_recvmmsg, sys_fanotify_init
 /*330*/	.long sys_fanotify_mark, sys_prlimit64, sys_name_to_handle_at, sys_open_by_handle_at, sys_clock_adjtime
 /*335*/	.long sys_syncfs, sys_sendmmsg, sys_setns, sys_process_vm_readv, sys_process_vm_writev
+/*340*/	.long sys_ni_syscall, sys_kcmp, sys_finit_module
diff --git a/arch/sparc/kernel/systbls_64.S b/arch/sparc/kernel/systbls_64.S
index 3a58e0d66f51..088134834dab 100644
--- a/arch/sparc/kernel/systbls_64.S
+++ b/arch/sparc/kernel/systbls_64.S
@@ -18,63 +18,63 @@
 
 	.globl sys_call_table32
 sys_call_table32:
-/*0*/	.word sys_restart_syscall, sys32_exit, sys_fork, sys_read, sys_write
-/*5*/	.word sys32_open, sys_close, sys32_wait4, sys32_creat, sys_link
-/*10*/  .word sys_unlink, sunos_execv, sys_chdir, sys_chown16, sys32_mknod
-/*15*/	.word sys_chmod, sys_lchown16, sys_brk, sys_nis_syscall, sys32_lseek
+/*0*/	.word sys_restart_syscall, sparc_exit, sys_fork, sys_read, sys_write
+/*5*/	.word compat_sys_open, sys_close, compat_sys_wait4, sys_creat, sys_link
+/*10*/  .word sys_unlink, sunos_execv, sys_chdir, sys_chown16, sys_mknod
+/*15*/	.word sys_chmod, sys_lchown16, sys_brk, sys_nis_syscall, compat_sys_lseek
 /*20*/	.word sys_getpid, sys_capget, sys_capset, sys_setuid16, sys_getuid16
-/*25*/	.word sys32_vmsplice, compat_sys_ptrace, sys_alarm, sys32_sigaltstack, sys_pause
-/*30*/	.word compat_sys_utime, sys_lchown, sys_fchown, sys32_access, sys32_nice
-	.word sys_chown, sys_sync, sys32_kill, compat_sys_newstat, sys32_sendfile
+/*25*/	.word sys32_vmsplice, compat_sys_ptrace, sys_alarm, compat_sys_sigaltstack, sys_pause
+/*30*/	.word compat_sys_utime, sys_lchown, sys_fchown, sys_access, sys_nice
+	.word sys_chown, sys_sync, sys_kill, compat_sys_newstat, sys32_sendfile
 /*40*/	.word compat_sys_newlstat, sys_dup, sys_sparc_pipe, compat_sys_times, sys_getuid
-	.word sys32_umount, sys_setgid16, sys_getgid16, sys32_signal, sys_geteuid16
+	.word sys_umount, sys_setgid16, sys_getgid16, sys_signal, sys_geteuid16
 /*50*/	.word sys_getegid16, sys_acct, sys_nis_syscall, sys_getgid, compat_sys_ioctl
-	.word sys32_reboot, sys32_mmap2, sys_symlink, sys32_readlink, sys32_execve
-/*60*/	.word sys32_umask, sys_chroot, compat_sys_newfstat, compat_sys_fstat64, sys_getpagesize
-	.word sys32_msync, sys_vfork, sys32_pread64, sys32_pwrite64, sys_geteuid
+	.word sys_reboot, sys32_mmap2, sys_symlink, sys_readlink, sys32_execve
+/*60*/	.word sys_umask, sys_chroot, compat_sys_newfstat, compat_sys_fstat64, sys_getpagesize
+	.word sys_msync, sys_vfork, sys32_pread64, sys32_pwrite64, sys_geteuid
 /*70*/	.word sys_getegid, sys_mmap, sys_setreuid, sys_munmap, sys_mprotect
 	.word sys_madvise, sys_vhangup, sys32_truncate64, sys_mincore, sys_getgroups16
-/*80*/	.word sys_setgroups16, sys_getpgrp, sys32_setgroups, sys32_setitimer, sys32_ftruncate64
-	.word sys32_swapon, sys32_getitimer, sys_setuid, sys32_sethostname, sys_setgid
+/*80*/	.word sys_setgroups16, sys_getpgrp, sys_setgroups, compat_sys_setitimer, sys32_ftruncate64
+	.word sys_swapon, compat_sys_getitimer, sys_setuid, sys_sethostname, sys_setgid
 /*90*/	.word sys_dup2, sys_setfsuid, compat_sys_fcntl, sys32_select, sys_setfsgid
-	.word sys_fsync, sys32_setpriority, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*100*/ .word sys32_getpriority, sys32_rt_sigreturn, sys32_rt_sigaction, sys32_rt_sigprocmask, sys32_rt_sigpending
-	.word compat_sys_rt_sigtimedwait, sys32_rt_sigqueueinfo, compat_sys_rt_sigsuspend, sys_setresuid, sys_getresuid
+	.word sys_fsync, sys_setpriority, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*100*/ .word sys_getpriority, sys32_rt_sigreturn, compat_sys_rt_sigaction, compat_sys_rt_sigprocmask, compat_sys_rt_sigpending
+	.word compat_sys_rt_sigtimedwait, compat_sys_rt_sigqueueinfo, compat_sys_rt_sigsuspend, sys_setresuid, sys_getresuid
 /*110*/	.word sys_setresgid, sys_getresgid, sys_setregid, sys_nis_syscall, sys_nis_syscall
-	.word sys32_getgroups, compat_sys_gettimeofday, sys32_getrusage, sys_nis_syscall, sys_getcwd
+	.word sys_getgroups, compat_sys_gettimeofday, sys32_getrusage, sys_nis_syscall, sys_getcwd
 /*120*/	.word compat_sys_readv, compat_sys_writev, compat_sys_settimeofday, sys_fchown16, sys_fchmod
-	.word sys_nis_syscall, sys_setreuid16, sys_setregid16, sys_rename, sys32_truncate
-/*130*/	.word sys32_ftruncate, sys_flock, compat_sys_lstat64, sys_nis_syscall, sys_nis_syscall
-	.word sys_nis_syscall, sys32_mkdir, sys_rmdir, compat_sys_utimes, compat_sys_stat64
-/*140*/	.word sys32_sendfile64, sys_nis_syscall, sys32_futex, sys_gettid, compat_sys_getrlimit
-	.word compat_sys_setrlimit, sys_pivot_root, sys32_prctl, sys_pciconfig_read, sys_pciconfig_write
+	.word sys_nis_syscall, sys_setreuid16, sys_setregid16, sys_rename, compat_sys_truncate
+/*130*/	.word compat_sys_ftruncate, sys_flock, compat_sys_lstat64, sys_nis_syscall, sys_nis_syscall
+	.word sys_nis_syscall, sys_mkdir, sys_rmdir, compat_sys_utimes, compat_sys_stat64
+/*140*/	.word sys_sendfile64, sys_nis_syscall, sys32_futex, sys_gettid, compat_sys_getrlimit
+	.word compat_sys_setrlimit, sys_pivot_root, sys_prctl, sys_pciconfig_read, sys_pciconfig_write
 /*150*/	.word sys_nis_syscall, sys_inotify_init, sys_inotify_add_watch, sys_poll, sys_getdents64
 	.word compat_sys_fcntl64, sys_inotify_rm_watch, compat_sys_statfs, compat_sys_fstatfs, sys_oldumount
-/*160*/	.word compat_sys_sched_setaffinity, compat_sys_sched_getaffinity, sys32_getdomainname, sys32_setdomainname, sys_nis_syscall
-	.word sys_quotactl, sys_set_tid_address, compat_sys_mount, compat_sys_ustat, sys32_setxattr
-/*170*/	.word sys32_lsetxattr, sys32_fsetxattr, sys_getxattr, sys_lgetxattr, compat_sys_getdents
-	.word sys_setsid, sys_fchdir, sys32_fgetxattr, sys_listxattr, sys_llistxattr
-/*180*/	.word sys32_flistxattr, sys_removexattr, sys_lremovexattr, compat_sys_sigpending, sys_ni_syscall
-	.word sys32_setpgid, sys32_fremovexattr, sys32_tkill, sys32_exit_group, sys_newuname
-/*190*/	.word sys32_init_module, sys_sparc64_personality, sys_remap_file_pages, sys32_epoll_create, sys32_epoll_ctl
-	.word sys32_epoll_wait, sys32_ioprio_set, sys_getppid, sys32_sigaction, sys_sgetmask
-/*200*/	.word sys32_ssetmask, sys_sigsuspend, compat_sys_newlstat, sys_uselib, compat_sys_old_readdir
-	.word sys32_readahead, sys32_socketcall, sys32_syslog, sys32_lookup_dcookie, sys32_fadvise64
-/*210*/	.word sys32_fadvise64_64, sys32_tgkill, sys32_waitpid, sys_swapoff, compat_sys_sysinfo
-	.word compat_sys_ipc, sys32_sigreturn, sys_clone, sys32_ioprio_get, compat_sys_adjtimex
-/*220*/	.word sys32_sigprocmask, sys_ni_syscall, sys32_delete_module, sys_ni_syscall, sys32_getpgid
-	.word sys32_bdflush, sys32_sysfs, sys_nis_syscall, sys_setfsuid16, sys_setfsgid16
-/*230*/	.word sys32_select, compat_sys_time, sys32_splice, compat_sys_stime, compat_sys_statfs64
-	.word compat_sys_fstatfs64, sys_llseek, sys_mlock, sys_munlock, sys32_mlockall
-/*240*/	.word sys_munlockall, sys32_sched_setparam, sys32_sched_getparam, sys32_sched_setscheduler, sys32_sched_getscheduler
-	.word sys_sched_yield, sys32_sched_get_priority_max, sys32_sched_get_priority_min, sys32_sched_rr_get_interval, compat_sys_nanosleep
-/*250*/	.word sys_mremap, compat_sys_sysctl, sys32_getsid, sys_fdatasync, sys_nis_syscall
+/*160*/	.word compat_sys_sched_setaffinity, compat_sys_sched_getaffinity, sys_getdomainname, sys_setdomainname, sys_nis_syscall
+	.word sys_quotactl, sys_set_tid_address, compat_sys_mount, compat_sys_ustat, sys_setxattr
+/*170*/	.word sys_lsetxattr, sys_fsetxattr, sys_getxattr, sys_lgetxattr, compat_sys_getdents
+	.word sys_setsid, sys_fchdir, sys_fgetxattr, sys_listxattr, sys_llistxattr
+/*180*/	.word sys_flistxattr, sys_removexattr, sys_lremovexattr, compat_sys_sigpending, sys_ni_syscall
+	.word sys_setpgid, sys_fremovexattr, sys_tkill, sparc_exit_group, sys_newuname
+/*190*/	.word sys_init_module, sys_sparc64_personality, sys_remap_file_pages, sys_epoll_create, sys_epoll_ctl
+	.word sys_epoll_wait, sys_ioprio_set, sys_getppid, compat_sys_sparc_sigaction, sys_sgetmask
+/*200*/	.word sys_ssetmask, sys_sigsuspend, compat_sys_newlstat, sys_uselib, compat_sys_old_readdir
+	.word sys32_readahead, sys32_socketcall, sys_syslog, sys32_lookup_dcookie, sys32_fadvise64
+/*210*/	.word sys32_fadvise64_64, sys_tgkill, sys_waitpid, sys_swapoff, compat_sys_sysinfo
+	.word compat_sys_ipc, sys32_sigreturn, sys_clone, sys_ioprio_get, compat_sys_adjtimex
+/*220*/	.word compat_sys_sigprocmask, sys_ni_syscall, sys_delete_module, sys_ni_syscall, sys_getpgid
+	.word sys_bdflush, sys_sysfs, sys_nis_syscall, sys_setfsuid16, sys_setfsgid16
+/*230*/	.word sys32_select, compat_sys_time, sys_splice, compat_sys_stime, compat_sys_statfs64
+	.word compat_sys_fstatfs64, sys_llseek, sys_mlock, sys_munlock, sys_mlockall
+/*240*/	.word sys_munlockall, sys_sched_setparam, sys_sched_getparam, sys_sched_setscheduler, sys_sched_getscheduler
+	.word sys_sched_yield, sys_sched_get_priority_max, sys_sched_get_priority_min, compat_sys_sched_rr_get_interval, compat_sys_nanosleep
+/*250*/	.word sys_mremap, compat_sys_sysctl, sys_getsid, sys_fdatasync, sys_nis_syscall
 	.word sys32_sync_file_range, compat_sys_clock_settime, compat_sys_clock_gettime, compat_sys_clock_getres, sys32_clock_nanosleep
 /*260*/	.word compat_sys_sched_getaffinity, compat_sys_sched_setaffinity, sys32_timer_settime, compat_sys_timer_gettime, sys_timer_getoverrun
 	.word sys_timer_delete, compat_sys_timer_create, sys_ni_syscall, compat_sys_io_setup, sys_io_destroy
 /*270*/	.word sys32_io_submit, sys_io_cancel, compat_sys_io_getevents, sys32_mq_open, sys_mq_unlink
 	.word compat_sys_mq_timedsend, compat_sys_mq_timedreceive, compat_sys_mq_notify, compat_sys_mq_getsetattr, compat_sys_waitid
-/*280*/	.word sys32_tee, sys_add_key, sys_request_key, compat_sys_keyctl, compat_sys_openat
+/*280*/	.word sys_tee, sys_add_key, sys_request_key, compat_sys_keyctl, compat_sys_openat
 	.word sys_mkdirat, sys_mknodat, sys_fchownat, compat_sys_futimesat, compat_sys_fstatat64
 /*290*/	.word sys_unlinkat, sys_renameat, sys_linkat, sys_symlinkat, sys_readlinkat
 	.word sys_fchmodat, sys_faccessat, compat_sys_pselect6, compat_sys_ppoll, sys_unshare
@@ -86,6 +86,7 @@ sys_call_table32:
 	.word compat_sys_pwritev, compat_sys_rt_tgsigqueueinfo, sys_perf_event_open, compat_sys_recvmmsg, sys_fanotify_init
 /*330*/	.word sys32_fanotify_mark, sys_prlimit64, sys_name_to_handle_at, compat_sys_open_by_handle_at, compat_sys_clock_adjtime
 	.word sys_syncfs, compat_sys_sendmmsg, sys_setns, compat_sys_process_vm_readv, compat_sys_process_vm_writev
+/*340*/	.word sys_kern_features, sys_kcmp, sys_finit_module
 
 #endif /* CONFIG_COMPAT */
 
@@ -106,7 +107,7 @@ sys_call_table:
 /*40*/	.word sys_newlstat, sys_dup, sys_sparc_pipe, sys_times, sys_nis_syscall
 	.word sys_umount, sys_setgid, sys_getgid, sys_signal, sys_geteuid
 /*50*/	.word sys_getegid, sys_acct, sys_memory_ordering, sys_nis_syscall, sys_ioctl
-	.word sys_reboot, sys_nis_syscall, sys_symlink, sys_readlink, sys_execve
+	.word sys_reboot, sys_nis_syscall, sys_symlink, sys_readlink, sys64_execve
 /*60*/	.word sys_umask, sys_chroot, sys_newfstat, sys_fstat64, sys_getpagesize
 	.word sys_msync, sys_vfork, sys_pread64, sys_pwrite64, sys_nis_syscall
 /*70*/	.word sys_nis_syscall, sys_mmap, sys_nis_syscall, sys_64_munmap, sys_mprotect
@@ -132,7 +133,7 @@ sys_call_table:
 /*170*/	.word sys_lsetxattr, sys_fsetxattr, sys_getxattr, sys_lgetxattr, sys_getdents
 	.word sys_setsid, sys_fchdir, sys_fgetxattr, sys_listxattr, sys_llistxattr
 /*180*/	.word sys_flistxattr, sys_removexattr, sys_lremovexattr, sys_nis_syscall, sys_ni_syscall
-	.word sys_setpgid, sys_fremovexattr, sys_tkill, sys_exit_group, sys_newuname
+	.word sys_setpgid, sys_fremovexattr, sys_tkill, sparc_exit_group, sys_newuname
 /*190*/	.word sys_init_module, sys_sparc64_personality, sys_remap_file_pages, sys_epoll_create, sys_epoll_ctl
 	.word sys_epoll_wait, sys_ioprio_set, sys_getppid, sys_nis_syscall, sys_sgetmask
 /*200*/	.word sys_ssetmask, sys_nis_syscall, sys_newlstat, sys_uselib, sys_nis_syscall
@@ -163,3 +164,4 @@ sys_call_table:
 	.word sys_pwritev, sys_rt_tgsigqueueinfo, sys_perf_event_open, sys_recvmmsg, sys_fanotify_init
 /*330*/	.word sys_fanotify_mark, sys_prlimit64, sys_name_to_handle_at, sys_open_by_handle_at, sys_clock_adjtime
 	.word sys_syncfs, sys_sendmmsg, sys_setns, sys_process_vm_readv, sys_process_vm_writev
+/*340*/	.word sys_kern_features, sys_kcmp, sys_finit_module
diff --git a/arch/sparc/kernel/time_32.c b/arch/sparc/kernel/time_32.c
index 953641549e82..c4c27b0f9063 100644
--- a/arch/sparc/kernel/time_32.c
+++ b/arch/sparc/kernel/time_32.c
@@ -278,7 +278,7 @@ static struct platform_device m48t59_rtc = {
 	},
 };
 
-static int __devinit clock_probe(struct platform_device *op)
+static int clock_probe(struct platform_device *op)
 {
 	struct device_node *dp = op->dev.of_node;
 	const char *model = of_get_property(dp, "model", NULL);
diff --git a/arch/sparc/kernel/time_64.c b/arch/sparc/kernel/time_64.c
index e861072b9c52..c3d82b5f54ca 100644
--- a/arch/sparc/kernel/time_64.c
+++ b/arch/sparc/kernel/time_64.c
@@ -419,7 +419,7 @@ static struct platform_device rtc_cmos_device = {
 	.num_resources	= 1,
 };
 
-static int __devinit rtc_probe(struct platform_device *op)
+static int rtc_probe(struct platform_device *op)
 {
 	struct resource *r;
 
@@ -477,7 +477,7 @@ static struct platform_device rtc_bq4802_device = {
 	.num_resources	= 1,
 };
 
-static int __devinit bq4802_probe(struct platform_device *op)
+static int bq4802_probe(struct platform_device *op)
 {
 
 	printk(KERN_INFO "%s: BQ4802 regs at 0x%llx\n",
@@ -534,7 +534,7 @@ static struct platform_device m48t59_rtc = {
 	},
 };
 
-static int __devinit mostek_probe(struct platform_device *op)
+static int mostek_probe(struct platform_device *op)
 {
 	struct device_node *dp = op->dev.of_node;
 
@@ -746,7 +746,7 @@ void __irq_entry timer_interrupt(int irq, struct pt_regs *regs)
 	set_irq_regs(old_regs);
 }
 
-void __devinit setup_sparc64_timer(void)
+void setup_sparc64_timer(void)
 {
 	struct clock_event_device *sevt;
 	unsigned long pstate;
@@ -844,7 +844,7 @@ unsigned long long sched_clock(void)
 		>> SPARC64_NSEC_PER_CYC_SHIFT;
 }
 
-int __devinit read_current_timer(unsigned long *timer_val)
+int read_current_timer(unsigned long *timer_val)
 {
 	*timer_val = tick_ops->get_tick();
 	return 0;
diff --git a/arch/sparc/kernel/trampoline_32.S b/arch/sparc/kernel/trampoline_32.S
index af27acab4486..6cdb08cdabf0 100644
--- a/arch/sparc/kernel/trampoline_32.S
+++ b/arch/sparc/kernel/trampoline_32.S
@@ -79,18 +79,15 @@ cpu3_startup:
 	 nop
 
 	/* Start this processor. */
-	call	smp4m_callin
+	call	smp_callin
 	 nop
 
-	b,a	smp_do_cpu_idle
+	b,a	smp_panic
 
 	.text
 	.align	4
 
-smp_do_cpu_idle:
-	call	cpu_idle
-	 mov	0, %o0
-
+smp_panic:
 	call	cpu_panic
 	 nop
 
@@ -144,10 +141,10 @@ sun4d_cpu_startup:
 	 nop
 
 	/* Start this processor. */
-	call	smp4d_callin
+	call	smp_callin
 	 nop
 
-	b,a	smp_do_cpu_idle
+	b,a	smp_panic
 
 	__CPUINIT
 	.align	4
@@ -201,7 +198,7 @@ leon_smp_cpu_startup:
 	 nop
 
 	/* Start this processor. */
-	call	leon_callin
+	call	smp_callin
 	 nop
 
-	b,a	smp_do_cpu_idle
+	b,a	smp_panic
diff --git a/arch/sparc/kernel/traps_32.c b/arch/sparc/kernel/traps_32.c
index a5785ea2a85d..662982946a89 100644
--- a/arch/sparc/kernel/traps_32.c
+++ b/arch/sparc/kernel/traps_32.c
@@ -58,7 +58,7 @@ void die_if_kernel(char *str, struct pt_regs *regs)
 
 	printk("%s(%d): %s [#%d]\n", current->comm, task_pid_nr(current), str, ++die_counter);
 	show_regs(regs);
-	add_taint(TAINT_DIE);
+	add_taint(TAINT_DIE, LOCKDEP_NOW_UNRELIABLE);
 
 	__SAVE; __SAVE; __SAVE; __SAVE;
 	__SAVE; __SAVE; __SAVE; __SAVE;
diff --git a/arch/sparc/kernel/traps_64.c b/arch/sparc/kernel/traps_64.c
index b66a77968f35..8d38ca97aa23 100644
--- a/arch/sparc/kernel/traps_64.c
+++ b/arch/sparc/kernel/traps_64.c
@@ -2383,7 +2383,7 @@ void die_if_kernel(char *str, struct pt_regs *regs)
 	notify_die(DIE_OOPS, str, regs, 0, 255, SIGSEGV);
 	__asm__ __volatile__("flushw");
 	show_regs(regs);
-	add_taint(TAINT_DIE);
+	add_taint(TAINT_DIE, LOCKDEP_NOW_UNRELIABLE);
 	if (regs->tstate & TSTATE_PRIV) {
 		struct thread_info *tp = current_thread_info();
 		struct reg_window *rw = (struct reg_window *)
@@ -2688,8 +2688,8 @@ void __init trap_init(void)
 		     TI_PRE_COUNT != offsetof(struct thread_info,
 					      preempt_count) ||
 		     TI_NEW_CHILD != offsetof(struct thread_info, new_child) ||
-		     TI_SYS_NOERROR != offsetof(struct thread_info,
-						syscall_noerror) ||
+		     TI_CURRENT_DS != offsetof(struct thread_info,
+						current_ds) ||
 		     TI_RESTART_BLOCK != offsetof(struct thread_info,
 						  restart_block) ||
 		     TI_KUNA_REGS != offsetof(struct thread_info,
diff --git a/arch/sparc/kernel/tsb.S b/arch/sparc/kernel/tsb.S
index d4bdc7a62375..a313e4a9399b 100644
--- a/arch/sparc/kernel/tsb.S
+++ b/arch/sparc/kernel/tsb.S
@@ -136,12 +136,43 @@ tsb_miss_page_table_walk_sun4v_fastpath:
 	 nop
 
 	/* It is a huge page, use huge page TSB entry address we
-	 * calculated above.
+	 * calculated above.  If the huge page TSB has not been
+	 * allocated, setup a trap stack and call hugetlb_setup()
+	 * to do so, then return from the trap to replay the TLB
+	 * miss.
+	 *
+	 * This is necessary to handle the case of transparent huge
+	 * pages where we don't really have a non-atomic context
+	 * in which to allocate the hugepage TSB hash table.  When
+	 * the 'mm' faults in the hugepage for the first time, we
+	 * thus handle it here.  This also makes sure that we can
+	 * allocate the TSB hash table on the correct NUMA node.
 	 */
 	TRAP_LOAD_TRAP_BLOCK(%g7, %g2)
-	ldx		[%g7 + TRAP_PER_CPU_TSB_HUGE_TEMP], %g2
-	cmp		%g2, -1
-	movne		%xcc, %g2, %g1
+	ldx		[%g7 + TRAP_PER_CPU_TSB_HUGE_TEMP], %g1
+	cmp		%g1, -1
+	bne,pt		%xcc, 60f
+	 nop
+
+661:	rdpr		%pstate, %g5
+	wrpr		%g5, PSTATE_AG | PSTATE_MG, %pstate
+	.section	.sun4v_2insn_patch, "ax"
+	.word		661b
+	SET_GL(1)
+	nop
+	.previous
+
+	rdpr	%tl, %g3
+	cmp	%g3, 1
+	bne,pn	%xcc, winfix_trampoline
+	 nop
+	ba,pt	%xcc, etrap
+	 rd	%pc, %g7
+	call	hugetlb_setup
+	 add	%sp, PTREGS_OFF, %o0
+	ba,pt	%xcc, rtrap
+	 nop
+
 60:
 #endif
 
diff --git a/arch/sparc/kernel/unaligned_64.c b/arch/sparc/kernel/unaligned_64.c
index f81d038f7340..8201c25e7669 100644
--- a/arch/sparc/kernel/unaligned_64.c
+++ b/arch/sparc/kernel/unaligned_64.c
@@ -113,21 +113,24 @@ static inline long sign_extend_imm13(long imm)
 
 static unsigned long fetch_reg(unsigned int reg, struct pt_regs *regs)
 {
-	unsigned long value;
+	unsigned long value, fp;
 	
 	if (reg < 16)
 		return (!reg ? 0 : regs->u_regs[reg]);
+
+	fp = regs->u_regs[UREG_FP];
+
 	if (regs->tstate & TSTATE_PRIV) {
 		struct reg_window *win;
-		win = (struct reg_window *)(regs->u_regs[UREG_FP] + STACK_BIAS);
+		win = (struct reg_window *)(fp + STACK_BIAS);
 		value = win->locals[reg - 16];
-	} else if (test_thread_flag(TIF_32BIT)) {
+	} else if (!test_thread_64bit_stack(fp)) {
 		struct reg_window32 __user *win32;
-		win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP]));
+		win32 = (struct reg_window32 __user *)((unsigned long)((u32)fp));
 		get_user(value, &win32->locals[reg - 16]);
 	} else {
 		struct reg_window __user *win;
-		win = (struct reg_window __user *)(regs->u_regs[UREG_FP] + STACK_BIAS);
+		win = (struct reg_window __user *)(fp + STACK_BIAS);
 		get_user(value, &win->locals[reg - 16]);
 	}
 	return value;
@@ -135,19 +138,24 @@ static unsigned long fetch_reg(unsigned int reg, struct pt_regs *regs)
 
 static unsigned long *fetch_reg_addr(unsigned int reg, struct pt_regs *regs)
 {
+	unsigned long fp;
+
 	if (reg < 16)
 		return &regs->u_regs[reg];
+
+	fp = regs->u_regs[UREG_FP];
+
 	if (regs->tstate & TSTATE_PRIV) {
 		struct reg_window *win;
-		win = (struct reg_window *)(regs->u_regs[UREG_FP] + STACK_BIAS);
+		win = (struct reg_window *)(fp + STACK_BIAS);
 		return &win->locals[reg - 16];
-	} else if (test_thread_flag(TIF_32BIT)) {
+	} else if (!test_thread_64bit_stack(fp)) {
 		struct reg_window32 *win32;
-		win32 = (struct reg_window32 *)((unsigned long)((u32)regs->u_regs[UREG_FP]));
+		win32 = (struct reg_window32 *)((unsigned long)((u32)fp));
 		return (unsigned long *)&win32->locals[reg - 16];
 	} else {
 		struct reg_window *win;
-		win = (struct reg_window *)(regs->u_regs[UREG_FP] + STACK_BIAS);
+		win = (struct reg_window *)(fp + STACK_BIAS);
 		return &win->locals[reg - 16];
 	}
 }
@@ -392,13 +400,15 @@ int handle_popc(u32 insn, struct pt_regs *regs)
 		if (rd)
 			regs->u_regs[rd] = ret;
 	} else {
-		if (test_thread_flag(TIF_32BIT)) {
+		unsigned long fp = regs->u_regs[UREG_FP];
+
+		if (!test_thread_64bit_stack(fp)) {
 			struct reg_window32 __user *win32;
-			win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP]));
+			win32 = (struct reg_window32 __user *)((unsigned long)((u32)fp));
 			put_user(ret, &win32->locals[rd - 16]);
 		} else {
 			struct reg_window __user *win;
-			win = (struct reg_window __user *)(regs->u_regs[UREG_FP] + STACK_BIAS);
+			win = (struct reg_window __user *)(fp + STACK_BIAS);
 			put_user(ret, &win->locals[rd - 16]);
 		}
 	}
@@ -554,7 +564,7 @@ void handle_ld_nf(u32 insn, struct pt_regs *regs)
 		reg[0] = 0;
 		if ((insn & 0x780000) == 0x180000)
 			reg[1] = 0;
-	} else if (test_thread_flag(TIF_32BIT)) {
+	} else if (!test_thread_64bit_stack(regs->u_regs[UREG_FP])) {
 		put_user(0, (int __user *) reg);
 		if ((insn & 0x780000) == 0x180000)
 			put_user(0, ((int __user *) reg) + 1);
diff --git a/arch/sparc/kernel/visemul.c b/arch/sparc/kernel/visemul.c
index 08e074b7eb6a..c096c624ac4d 100644
--- a/arch/sparc/kernel/visemul.c
+++ b/arch/sparc/kernel/visemul.c
@@ -149,21 +149,24 @@ static inline void maybe_flush_windows(unsigned int rs1, unsigned int rs2,
 
 static unsigned long fetch_reg(unsigned int reg, struct pt_regs *regs)
 {
-	unsigned long value;
+	unsigned long value, fp;
 	
 	if (reg < 16)
 		return (!reg ? 0 : regs->u_regs[reg]);
+
+	fp = regs->u_regs[UREG_FP];
+
 	if (regs->tstate & TSTATE_PRIV) {
 		struct reg_window *win;
-		win = (struct reg_window *)(regs->u_regs[UREG_FP] + STACK_BIAS);
+		win = (struct reg_window *)(fp + STACK_BIAS);
 		value = win->locals[reg - 16];
-	} else if (test_thread_flag(TIF_32BIT)) {
+	} else if (!test_thread_64bit_stack(fp)) {
 		struct reg_window32 __user *win32;
-		win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP]));
+		win32 = (struct reg_window32 __user *)((unsigned long)((u32)fp));
 		get_user(value, &win32->locals[reg - 16]);
 	} else {
 		struct reg_window __user *win;
-		win = (struct reg_window __user *)(regs->u_regs[UREG_FP] + STACK_BIAS);
+		win = (struct reg_window __user *)(fp + STACK_BIAS);
 		get_user(value, &win->locals[reg - 16]);
 	}
 	return value;
@@ -172,16 +175,18 @@ static unsigned long fetch_reg(unsigned int reg, struct pt_regs *regs)
 static inline unsigned long __user *__fetch_reg_addr_user(unsigned int reg,
 							  struct pt_regs *regs)
 {
+	unsigned long fp = regs->u_regs[UREG_FP];
+
 	BUG_ON(reg < 16);
 	BUG_ON(regs->tstate & TSTATE_PRIV);
 
-	if (test_thread_flag(TIF_32BIT)) {
+	if (!test_thread_64bit_stack(fp)) {
 		struct reg_window32 __user *win32;
-		win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP]));
+		win32 = (struct reg_window32 __user *)((unsigned long)((u32)fp));
 		return (unsigned long __user *)&win32->locals[reg - 16];
 	} else {
 		struct reg_window __user *win;
-		win = (struct reg_window __user *)(regs->u_regs[UREG_FP] + STACK_BIAS);
+		win = (struct reg_window __user *)(fp + STACK_BIAS);
 		return &win->locals[reg - 16];
 	}
 }
@@ -204,7 +209,7 @@ static void store_reg(struct pt_regs *regs, unsigned long val, unsigned long rd)
 	} else {
 		unsigned long __user *rd_user = __fetch_reg_addr_user(rd, regs);
 
-		if (test_thread_flag(TIF_32BIT))
+		if (!test_thread_64bit_stack(regs->u_regs[UREG_FP]))
 			__put_user((u32)val, (u32 __user *)rd_user);
 		else
 			__put_user(val, rd_user);
diff --git a/arch/sparc/kernel/vmlinux.lds.S b/arch/sparc/kernel/vmlinux.lds.S
index 89c2c29f154b..0bacceb19150 100644
--- a/arch/sparc/kernel/vmlinux.lds.S
+++ b/arch/sparc/kernel/vmlinux.lds.S
@@ -132,6 +132,11 @@ SECTIONS
 		*(.popc_6insn_patch)
 		__popc_6insn_patch_end = .;
 	}
+	.pause_3insn_patch : {
+		__pause_3insn_patch = .;
+		*(.pause_3insn_patch)
+		__pause_3insn_patch_end = .;
+	}
 	PERCPU_SECTION(SMP_CACHE_BYTES)
 
 	. = ALIGN(PAGE_SIZE);
diff --git a/arch/sparc/kernel/winfixup.S b/arch/sparc/kernel/winfixup.S
index a6b0863c27df..1e67ce958369 100644
--- a/arch/sparc/kernel/winfixup.S
+++ b/arch/sparc/kernel/winfixup.S
@@ -43,6 +43,8 @@ spill_fixup_mna:
 spill_fixup_dax:
 	TRAP_LOAD_THREAD_REG(%g6, %g1)
 	ldx	[%g6 + TI_FLAGS], %g1
+	andcc	%sp, 0x1, %g0
+	movne	%icc, 0, %g1
 	andcc	%g1, _TIF_32BIT, %g0
 	ldub	[%g6 + TI_WSAVED], %g1
 	sll	%g1, 3, %g3
diff --git a/arch/sparc/lib/atomic_64.S b/arch/sparc/lib/atomic_64.S
index 4d502da3de78..85c233d0a340 100644
--- a/arch/sparc/lib/atomic_64.S
+++ b/arch/sparc/lib/atomic_64.S
@@ -1,6 +1,6 @@
 /* atomic.S: These things are too big to do inline.
  *
- * Copyright (C) 1999, 2007 David S. Miller (davem@davemloft.net)
+ * Copyright (C) 1999, 2007 2012 David S. Miller (davem@davemloft.net)
  */
 
 #include <linux/linkage.h>
@@ -117,3 +117,17 @@ ENTRY(atomic64_sub_ret) /* %o0 = decrement, %o1 = atomic_ptr */
 	 sub	%g1, %o0, %o0
 2:	BACKOFF_SPIN(%o2, %o3, 1b)
 ENDPROC(atomic64_sub_ret)
+
+ENTRY(atomic64_dec_if_positive) /* %o0 = atomic_ptr */
+	BACKOFF_SETUP(%o2)
+1:	ldx	[%o0], %g1
+	brlez,pn %g1, 3f
+	 sub	%g1, 1, %g7
+	casx	[%o0], %g1, %g7
+	cmp	%g1, %g7
+	bne,pn	%xcc, BACKOFF_LABEL(2f, 1b)
+	 nop
+3:	retl
+	 sub	%g1, 1, %o0
+2:	BACKOFF_SPIN(%o2, %o3, 1b)
+ENDPROC(atomic64_dec_if_positive)
diff --git a/arch/sparc/lib/ksyms.c b/arch/sparc/lib/ksyms.c
index ee31b884c61b..0c4e35e522fa 100644
--- a/arch/sparc/lib/ksyms.c
+++ b/arch/sparc/lib/ksyms.c
@@ -116,6 +116,7 @@ EXPORT_SYMBOL(atomic64_add);
 EXPORT_SYMBOL(atomic64_add_ret);
 EXPORT_SYMBOL(atomic64_sub);
 EXPORT_SYMBOL(atomic64_sub_ret);
+EXPORT_SYMBOL(atomic64_dec_if_positive);
 
 /* Atomic bit operations. */
 EXPORT_SYMBOL(test_and_set_bit);
diff --git a/arch/sparc/math-emu/math_64.c b/arch/sparc/math-emu/math_64.c
index 1704068da928..034aadbff036 100644
--- a/arch/sparc/math-emu/math_64.c
+++ b/arch/sparc/math-emu/math_64.c
@@ -320,7 +320,7 @@ int do_mathemu(struct pt_regs *regs, struct fpustate *f, bool illegal_insn_trap)
 					XR = 0;
 				else if (freg < 16)
 					XR = regs->u_regs[freg];
-				else if (test_thread_flag(TIF_32BIT)) {
+				else if (!test_thread_64bit_stack(regs->u_regs[UREG_FP])) {
 					struct reg_window32 __user *win32;
 					flushw_user ();
 					win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP]));
diff --git a/arch/sparc/mm/fault_64.c b/arch/sparc/mm/fault_64.c
index 097aee763af3..5062ff389e83 100644
--- a/arch/sparc/mm/fault_64.c
+++ b/arch/sparc/mm/fault_64.c
@@ -472,8 +472,13 @@ good_area:
 #if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
 	mm_rss = mm->context.huge_pte_count;
 	if (unlikely(mm_rss >
-		     mm->context.tsb_block[MM_TSB_HUGE].tsb_rss_limit))
-		tsb_grow(mm, MM_TSB_HUGE, mm_rss);
+		     mm->context.tsb_block[MM_TSB_HUGE].tsb_rss_limit)) {
+		if (mm->context.tsb_block[MM_TSB_HUGE].tsb)
+			tsb_grow(mm, MM_TSB_HUGE, mm_rss);
+		else
+			hugetlb_setup(regs);
+
+	}
 #endif
 	return;
 
diff --git a/arch/sparc/mm/gup.c b/arch/sparc/mm/gup.c
index 42c55df3aec3..01ee23dd724d 100644
--- a/arch/sparc/mm/gup.c
+++ b/arch/sparc/mm/gup.c
@@ -66,6 +66,56 @@ static noinline int gup_pte_range(pmd_t pmd, unsigned long addr,
 	return 1;
 }
 
+static int gup_huge_pmd(pmd_t *pmdp, pmd_t pmd, unsigned long addr,
+			unsigned long end, int write, struct page **pages,
+			int *nr)
+{
+	struct page *head, *page, *tail;
+	u32 mask;
+	int refs;
+
+	mask = PMD_HUGE_PRESENT;
+	if (write)
+		mask |= PMD_HUGE_WRITE;
+	if ((pmd_val(pmd) & mask) != mask)
+		return 0;
+
+	refs = 0;
+	head = pmd_page(pmd);
+	page = head + ((addr & ~PMD_MASK) >> PAGE_SHIFT);
+	tail = page;
+	do {
+		VM_BUG_ON(compound_head(page) != head);
+		pages[*nr] = page;
+		(*nr)++;
+		page++;
+		refs++;
+	} while (addr += PAGE_SIZE, addr != end);
+
+	if (!page_cache_add_speculative(head, refs)) {
+		*nr -= refs;
+		return 0;
+	}
+
+	if (unlikely(pmd_val(pmd) != pmd_val(*pmdp))) {
+		*nr -= refs;
+		while (refs--)
+			put_page(head);
+		return 0;
+	}
+
+	/* Any tail page need their mapcount reference taken before we
+	 * return.
+	 */
+	while (refs--) {
+		if (PageTail(tail))
+			get_huge_page_tail(tail);
+		tail++;
+	}
+
+	return 1;
+}
+
 static int gup_pmd_range(pud_t pud, unsigned long addr, unsigned long end,
 		int write, struct page **pages, int *nr)
 {
@@ -77,9 +127,14 @@ static int gup_pmd_range(pud_t pud, unsigned long addr, unsigned long end,
 		pmd_t pmd = *pmdp;
 
 		next = pmd_addr_end(addr, end);
-		if (pmd_none(pmd))
+		if (pmd_none(pmd) || pmd_trans_splitting(pmd))
 			return 0;
-		if (!gup_pte_range(pmd, addr, next, write, pages, nr))
+		if (unlikely(pmd_large(pmd))) {
+			if (!gup_huge_pmd(pmdp, pmd, addr, next,
+					  write, pages, nr))
+				return 0;
+		} else if (!gup_pte_range(pmd, addr, next, write,
+					  pages, nr))
 			return 0;
 	} while (pmdp++, addr = next, addr != end);
 
diff --git a/arch/sparc/mm/hugetlbpage.c b/arch/sparc/mm/hugetlbpage.c
index f76f83d5ac63..d2b59441ebdd 100644
--- a/arch/sparc/mm/hugetlbpage.c
+++ b/arch/sparc/mm/hugetlbpage.c
@@ -30,55 +30,28 @@ static unsigned long hugetlb_get_unmapped_area_bottomup(struct file *filp,
 							unsigned long pgoff,
 							unsigned long flags)
 {
-	struct mm_struct *mm = current->mm;
-	struct vm_area_struct * vma;
 	unsigned long task_size = TASK_SIZE;
-	unsigned long start_addr;
+	struct vm_unmapped_area_info info;
 
 	if (test_thread_flag(TIF_32BIT))
 		task_size = STACK_TOP32;
-	if (unlikely(len >= VA_EXCLUDE_START))
-		return -ENOMEM;
 
-	if (len > mm->cached_hole_size) {
-	        start_addr = addr = mm->free_area_cache;
-	} else {
-	        start_addr = addr = TASK_UNMAPPED_BASE;
-	        mm->cached_hole_size = 0;
+	info.flags = 0;
+	info.length = len;
+	info.low_limit = TASK_UNMAPPED_BASE;
+	info.high_limit = min(task_size, VA_EXCLUDE_START);
+	info.align_mask = PAGE_MASK & ~HPAGE_MASK;
+	info.align_offset = 0;
+	addr = vm_unmapped_area(&info);
+
+	if ((addr & ~PAGE_MASK) && task_size > VA_EXCLUDE_END) {
+		VM_BUG_ON(addr != -ENOMEM);
+		info.low_limit = VA_EXCLUDE_END;
+		info.high_limit = task_size;
+		addr = vm_unmapped_area(&info);
 	}
 
-	task_size -= len;
-
-full_search:
-	addr = ALIGN(addr, HPAGE_SIZE);
-
-	for (vma = find_vma(mm, addr); ; vma = vma->vm_next) {
-		/* At this point:  (!vma || addr < vma->vm_end). */
-		if (addr < VA_EXCLUDE_START &&
-		    (addr + len) >= VA_EXCLUDE_START) {
-			addr = VA_EXCLUDE_END;
-			vma = find_vma(mm, VA_EXCLUDE_END);
-		}
-		if (unlikely(task_size < addr)) {
-			if (start_addr != TASK_UNMAPPED_BASE) {
-				start_addr = addr = TASK_UNMAPPED_BASE;
-				mm->cached_hole_size = 0;
-				goto full_search;
-			}
-			return -ENOMEM;
-		}
-		if (likely(!vma || addr + len <= vma->vm_start)) {
-			/*
-			 * Remember the place where we stopped the search:
-			 */
-			mm->free_area_cache = addr + len;
-			return addr;
-		}
-		if (addr + mm->cached_hole_size < vma->vm_start)
-		        mm->cached_hole_size = vma->vm_start - addr;
-
-		addr = ALIGN(vma->vm_end, HPAGE_SIZE);
-	}
+	return addr;
 }
 
 static unsigned long
@@ -87,71 +60,34 @@ hugetlb_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
 				  const unsigned long pgoff,
 				  const unsigned long flags)
 {
-	struct vm_area_struct *vma;
 	struct mm_struct *mm = current->mm;
 	unsigned long addr = addr0;
+	struct vm_unmapped_area_info info;
 
 	/* This should only ever run for 32-bit processes.  */
 	BUG_ON(!test_thread_flag(TIF_32BIT));
 
-	/* check if free_area_cache is useful for us */
-	if (len <= mm->cached_hole_size) {
- 	        mm->cached_hole_size = 0;
- 		mm->free_area_cache = mm->mmap_base;
- 	}
-
-	/* either no address requested or can't fit in requested address hole */
-	addr = mm->free_area_cache & HPAGE_MASK;
-
-	/* make sure it can fit in the remaining address space */
-	if (likely(addr > len)) {
-		vma = find_vma(mm, addr-len);
-		if (!vma || addr <= vma->vm_start) {
-			/* remember the address as a hint for next time */
-			return (mm->free_area_cache = addr-len);
-		}
-	}
-
-	if (unlikely(mm->mmap_base < len))
-		goto bottomup;
-
-	addr = (mm->mmap_base-len) & HPAGE_MASK;
-
-	do {
-		/*
-		 * Lookup failure means no vma is above this address,
-		 * else if new region fits below vma->vm_start,
-		 * return with success:
-		 */
-		vma = find_vma(mm, addr);
-		if (likely(!vma || addr+len <= vma->vm_start)) {
-			/* remember the address as a hint for next time */
-			return (mm->free_area_cache = addr);
-		}
-
- 		/* remember the largest hole we saw so far */
- 		if (addr + mm->cached_hole_size < vma->vm_start)
- 		        mm->cached_hole_size = vma->vm_start - addr;
-
-		/* try just below the current vma->vm_start */
-		addr = (vma->vm_start-len) & HPAGE_MASK;
-	} while (likely(len < vma->vm_start));
+	info.flags = VM_UNMAPPED_AREA_TOPDOWN;
+	info.length = len;
+	info.low_limit = PAGE_SIZE;
+	info.high_limit = mm->mmap_base;
+	info.align_mask = PAGE_MASK & ~HPAGE_MASK;
+	info.align_offset = 0;
+	addr = vm_unmapped_area(&info);
 
-bottomup:
 	/*
 	 * A failed mmap() very likely causes application failure,
 	 * so fall back to the bottom-up function here. This scenario
 	 * can happen with large stack limits and large mmap()
 	 * allocations.
 	 */
-	mm->cached_hole_size = ~0UL;
-  	mm->free_area_cache = TASK_UNMAPPED_BASE;
-	addr = arch_get_unmapped_area(filp, addr0, len, pgoff, flags);
-	/*
-	 * Restore the topdown base:
-	 */
-	mm->free_area_cache = mm->mmap_base;
-	mm->cached_hole_size = ~0UL;
+	if (addr & ~PAGE_MASK) {
+		VM_BUG_ON(addr != -ENOMEM);
+		info.flags = 0;
+		info.low_limit = TASK_UNMAPPED_BASE;
+		info.high_limit = STACK_TOP32;
+		addr = vm_unmapped_area(&info);
+	}
 
 	return addr;
 }
diff --git a/arch/sparc/mm/init_32.c b/arch/sparc/mm/init_32.c
index dde85ef1c56d..48e0c030e8f5 100644
--- a/arch/sparc/mm/init_32.c
+++ b/arch/sparc/mm/init_32.c
@@ -57,7 +57,7 @@ void show_mem(unsigned int filter)
 	printk("Mem-info:\n");
 	show_free_areas(filter);
 	printk("Free swap:       %6ldkB\n",
-	       nr_swap_pages << (PAGE_SHIFT-10));
+	       get_nr_swap_pages() << (PAGE_SHIFT-10));
 	printk("%ld pages of RAM\n", totalram_pages);
 	printk("%ld free pages\n", nr_free_pages());
 }
diff --git a/arch/sparc/mm/init_64.c b/arch/sparc/mm/init_64.c
index 9e28a118e6a4..1588d33d5492 100644
--- a/arch/sparc/mm/init_64.c
+++ b/arch/sparc/mm/init_64.c
@@ -87,8 +87,8 @@ static unsigned long cpu_pgsz_mask;
 
 #define MAX_BANKS	32
 
-static struct linux_prom64_registers pavail[MAX_BANKS] __devinitdata;
-static int pavail_ents __devinitdata;
+static struct linux_prom64_registers pavail[MAX_BANKS];
+static int pavail_ents;
 
 static int cmp_p64(const void *a, const void *b)
 {
@@ -314,16 +314,31 @@ static void __update_mmu_tsb_insert(struct mm_struct *mm, unsigned long tsb_inde
 	struct tsb *tsb = mm->context.tsb_block[tsb_index].tsb;
 	unsigned long tag;
 
+	if (unlikely(!tsb))
+		return;
+
 	tsb += ((address >> tsb_hash_shift) &
 		(mm->context.tsb_block[tsb_index].tsb_nentries - 1UL));
 	tag = (address >> 22UL);
 	tsb_insert(tsb, tag, tte);
 }
 
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
+static inline bool is_hugetlb_pte(pte_t pte)
+{
+	if ((tlb_type == hypervisor &&
+	     (pte_val(pte) & _PAGE_SZALL_4V) == _PAGE_SZHUGE_4V) ||
+	    (tlb_type != hypervisor &&
+	     (pte_val(pte) & _PAGE_SZALL_4U) == _PAGE_SZHUGE_4U))
+		return true;
+	return false;
+}
+#endif
+
 void update_mmu_cache(struct vm_area_struct *vma, unsigned long address, pte_t *ptep)
 {
-	unsigned long tsb_index, tsb_hash_shift, flags;
 	struct mm_struct *mm;
+	unsigned long flags;
 	pte_t pte = *ptep;
 
 	if (tlb_type != hypervisor) {
@@ -335,25 +350,16 @@ void update_mmu_cache(struct vm_area_struct *vma, unsigned long address, pte_t *
 
 	mm = vma->vm_mm;
 
-	tsb_index = MM_TSB_BASE;
-	tsb_hash_shift = PAGE_SHIFT;
-
 	spin_lock_irqsave(&mm->context.lock, flags);
 
 #if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
-	if (mm->context.tsb_block[MM_TSB_HUGE].tsb != NULL) {
-		if ((tlb_type == hypervisor &&
-		     (pte_val(pte) & _PAGE_SZALL_4V) == _PAGE_SZHUGE_4V) ||
-		    (tlb_type != hypervisor &&
-		     (pte_val(pte) & _PAGE_SZALL_4U) == _PAGE_SZHUGE_4U)) {
-			tsb_index = MM_TSB_HUGE;
-			tsb_hash_shift = HPAGE_SHIFT;
-		}
-	}
+	if (mm->context.huge_pte_count && is_hugetlb_pte(pte))
+		__update_mmu_tsb_insert(mm, MM_TSB_HUGE, HPAGE_SHIFT,
+					address, pte_val(pte));
+	else
 #endif
-
-	__update_mmu_tsb_insert(mm, tsb_index, tsb_hash_shift,
-				address, pte_val(pte));
+		__update_mmu_tsb_insert(mm, MM_TSB_BASE, PAGE_SHIFT,
+					address, pte_val(pte));
 
 	spin_unlock_irqrestore(&mm->context.lock, flags);
 }
@@ -624,7 +630,7 @@ static void __init inherit_prom_mappings(void)
 void prom_world(int enter)
 {
 	if (!enter)
-		set_fs((mm_segment_t) { get_thread_current_ds() });
+		set_fs(get_fs());
 
 	__asm__ __volatile__("flushw");
 }
@@ -1931,7 +1937,7 @@ void __init paging_init(void)
 	printk("Booting Linux...\n");
 }
 
-int __devinit page_in_phys_avail(unsigned long paddr)
+int page_in_phys_avail(unsigned long paddr)
 {
 	int i;
 
@@ -2021,6 +2027,16 @@ static void __init patch_tlb_miss_handler_bitmap(void)
 	flushi(&valid_addr_bitmap_insn[0]);
 }
 
+static void __init register_page_bootmem_info(void)
+{
+#ifdef CONFIG_NEED_MULTIPLE_NODES
+	int i;
+
+	for_each_online_node(i)
+		if (NODE_DATA(i)->node_spanned_pages)
+			register_page_bootmem_info_node(NODE_DATA(i));
+#endif
+}
 void __init mem_init(void)
 {
 	unsigned long codepages, datapages, initpages;
@@ -2038,20 +2054,8 @@ void __init mem_init(void)
 
 	high_memory = __va(last_valid_pfn << PAGE_SHIFT);
 
-#ifdef CONFIG_NEED_MULTIPLE_NODES
-	{
-		int i;
-		for_each_online_node(i) {
-			if (NODE_DATA(i)->node_spanned_pages != 0) {
-				totalram_pages +=
-					free_all_bootmem_node(NODE_DATA(i));
-			}
-		}
-		totalram_pages += free_low_memory_core_early(MAX_NUMNODES);
-	}
-#else
+	register_page_bootmem_info();
 	totalram_pages = free_all_bootmem();
-#endif
 
 	/* We subtract one to account for the mem_map_zero page
 	 * allocated below.
@@ -2231,6 +2235,11 @@ void __meminit vmemmap_populate_print_last(void)
 		node_start = 0;
 	}
 }
+
+void vmemmap_free(struct page *memmap, unsigned long nr_pages)
+{
+}
+
 #endif /* CONFIG_SPARSEMEM_VMEMMAP */
 
 static void prot_init_common(unsigned long page_none,
@@ -2712,14 +2721,28 @@ static void context_reload(void *__data)
 		load_secondary_context(mm);
 }
 
-void hugetlb_setup(struct mm_struct *mm)
+void hugetlb_setup(struct pt_regs *regs)
 {
-	struct tsb_config *tp = &mm->context.tsb_block[MM_TSB_HUGE];
+	struct mm_struct *mm = current->mm;
+	struct tsb_config *tp;
 
-	if (likely(tp->tsb != NULL))
-		return;
+	if (in_atomic() || !mm) {
+		const struct exception_table_entry *entry;
+
+		entry = search_exception_tables(regs->tpc);
+		if (entry) {
+			regs->tpc = entry->fixup;
+			regs->tnpc = regs->tpc + 4;
+			return;
+		}
+		pr_alert("Unexpected HugeTLB setup in atomic context.\n");
+		die_if_kernel("HugeTSB in atomic", regs);
+	}
+
+	tp = &mm->context.tsb_block[MM_TSB_HUGE];
+	if (likely(tp->tsb == NULL))
+		tsb_grow(mm, MM_TSB_HUGE, 0);
 
-	tsb_grow(mm, MM_TSB_HUGE, 0);
 	tsb_context_switch(mm);
 	smp_tsb_sync(mm);
 
diff --git a/arch/sparc/mm/tlb.c b/arch/sparc/mm/tlb.c
index 3e8fec391fe0..ba6ae7ffdc2c 100644
--- a/arch/sparc/mm/tlb.c
+++ b/arch/sparc/mm/tlb.c
@@ -135,8 +135,15 @@ void set_pmd_at(struct mm_struct *mm, unsigned long addr,
 			mm->context.huge_pte_count++;
 		else
 			mm->context.huge_pte_count--;
-		if (mm->context.huge_pte_count == 1)
-			hugetlb_setup(mm);
+
+		/* Do not try to allocate the TSB hash table if we
+		 * don't have one already.  We have various locks held
+		 * and thus we'll end up doing a GFP_KERNEL allocation
+		 * in an atomic context.
+		 *
+		 * Instead, we let the first TLB miss on a hugepage
+		 * take care of this.
+		 */
 	}
 
 	if (!pmd_none(orig)) {
diff --git a/arch/sparc/mm/tsb.c b/arch/sparc/mm/tsb.c
index 7f6474347491..428982b9becf 100644
--- a/arch/sparc/mm/tsb.c
+++ b/arch/sparc/mm/tsb.c
@@ -314,7 +314,7 @@ void tsb_grow(struct mm_struct *mm, unsigned long tsb_index, unsigned long rss)
 retry_tsb_alloc:
 	gfp_flags = GFP_KERNEL;
 	if (new_size > (PAGE_SIZE * 2))
-		gfp_flags = __GFP_NOWARN | __GFP_NORETRY;
+		gfp_flags |= __GFP_NOWARN | __GFP_NORETRY;
 
 	new_tsb = kmem_cache_alloc_node(tsb_caches[new_cache_index],
 					gfp_flags, numa_node_id());
diff --git a/arch/sparc/net/bpf_jit_comp.c b/arch/sparc/net/bpf_jit_comp.c
index 28368701ef79..3109ca684a99 100644
--- a/arch/sparc/net/bpf_jit_comp.c
+++ b/arch/sparc/net/bpf_jit_comp.c
@@ -3,6 +3,7 @@
 #include <linux/netdevice.h>
 #include <linux/filter.h>
 #include <linux/cache.h>
+#include <linux/if_vlan.h>
 
 #include <asm/cacheflush.h>
 #include <asm/ptrace.h>
@@ -312,6 +313,12 @@ do {	*prog++ = BR_OPC | WDISP22(OFF);		\
 #define emit_addi(R1, IMM, R3) \
 	*prog++ = (ADD | IMMED | RS1(R1) | S13(IMM) | RD(R3))
 
+#define emit_and(R1, R2, R3) \
+	*prog++ = (AND | RS1(R1) | RS2(R2) | RD(R3))
+
+#define emit_andi(R1, IMM, R3) \
+	*prog++ = (AND | IMMED | RS1(R1) | S13(IMM) | RD(R3))
+
 #define emit_alloc_stack(SZ) \
 	*prog++ = (SUB | IMMED | RS1(SP) | S13(SZ) | RD(SP))
 
@@ -415,6 +422,8 @@ void bpf_jit_compile(struct sk_filter *fp)
 		case BPF_S_ANC_IFINDEX:
 		case BPF_S_ANC_MARK:
 		case BPF_S_ANC_RXHASH:
+		case BPF_S_ANC_VLAN_TAG:
+		case BPF_S_ANC_VLAN_TAG_PRESENT:
 		case BPF_S_ANC_CPU:
 		case BPF_S_ANC_QUEUE:
 		case BPF_S_LD_W_ABS:
@@ -600,6 +609,16 @@ void bpf_jit_compile(struct sk_filter *fp)
 			case BPF_S_ANC_RXHASH:
 				emit_skb_load32(rxhash, r_A);
 				break;
+			case BPF_S_ANC_VLAN_TAG:
+			case BPF_S_ANC_VLAN_TAG_PRESENT:
+				emit_skb_load16(vlan_tci, r_A);
+				if (filter[i].code == BPF_S_ANC_VLAN_TAG) {
+					emit_andi(r_A, VLAN_VID_MASK, r_A);
+				} else {
+					emit_loadimm(VLAN_TAG_PRESENT, r_TMP);
+					emit_and(r_A, r_TMP, r_A);
+				}
+				break;
 
 			case BPF_S_LD_IMM:
 				emit_loadimm(K, r_A);