Merge branch 'work.csum_and_copy' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs

Pull copy_and_csum cleanups from Al Viro:
 "Saner calling conventions for csum_and_copy_..._user() and friends"

[ Removing 800+ lines of code and cleaning stuff up is good  - Linus ]

* 'work.csum_and_copy' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  ppc: propagate the calling conventions change down to csum_partial_copy_generic()
  amd64: switch csum_partial_copy_generic() to new calling conventions
  sparc64: propagate the calling convention changes down to __csum_partial_copy_...()
  xtensa: propagate the calling conventions change down into csum_partial_copy_generic()
  mips: propagate the calling convention change down into __csum_partial_copy_..._user()
  mips: __csum_partial_copy_kernel() has no users left
  mips: csum_and_copy_{to,from}_user() are never called under KERNEL_DS
  sparc32: propagate the calling conventions change down to __csum_partial_copy_sparc_generic()
  i386: propagate the calling conventions change down to csum_partial_copy_generic()
  sh: propage the calling conventions change down to csum_partial_copy_generic()
  m68k: get rid of zeroing destination on error in csum_and_copy_from_user()
  arm: propagate the calling convention changes down to csum_partial_copy_from_user()
  alpha: propagate the calling convention changes down to csum_partial_copy.c helpers
  saner calling conventions for csum_and_copy_..._user()
  csum_and_copy_..._user(): pass 0xffffffff instead of 0 as initial sum
  csum_partial_copy_nocheck(): drop the last argument
  unify generic instances of csum_partial_copy_nocheck()
  icmp_push_reply(): reorder adding the checksum up
  skb_copy_and_csum_bits(): don't bother with the last argument

3 files changed, 138 insertions, 205 deletions

diff --git a/arch/x86/lib/checksum_32.S b/arch/x86/lib/checksum_32.S
index d1d768912368..4304320e51f4 100644
--- a/arch/x86/lib/checksum_32.S
+++ b/arch/x86/lib/checksum_32.S
@@ -253,28 +253,17 @@ EXPORT_SYMBOL(csum_partial)
 
 /*
 unsigned int csum_partial_copy_generic (const char *src, char *dst,
-				  int len, int sum, int *src_err_ptr, int *dst_err_ptr)
+				  int len)
  */ 
 
 /*
  * Copy from ds while checksumming, otherwise like csum_partial
- *
- * The macros SRC and DST specify the type of access for the instruction.
- * thus we can call a custom exception handler for all access types.
- *
- * FIXME: could someone double-check whether I haven't mixed up some SRC and
- *	  DST definitions? It's damn hard to trigger all cases.  I hope I got
- *	  them all but there's no guarantee.
  */
 
-#define SRC(y...)			\
+#define EXC(y...)			\
 	9999: y;			\
 	_ASM_EXTABLE_UA(9999b, 6001f)
 
-#define DST(y...)			\
-	9999: y;			\
-	_ASM_EXTABLE_UA(9999b, 6002f)
-
 #ifndef CONFIG_X86_USE_PPRO_CHECKSUM
 
 #define ARGBASE 16		
@@ -285,20 +274,20 @@ SYM_FUNC_START(csum_partial_copy_generic)
 	pushl %edi
 	pushl %esi
 	pushl %ebx
-	movl ARGBASE+16(%esp),%eax	# sum
 	movl ARGBASE+12(%esp),%ecx	# len
 	movl ARGBASE+4(%esp),%esi	# src
 	movl ARGBASE+8(%esp),%edi	# dst
 
+	movl $-1, %eax			# sum
 	testl $2, %edi			# Check alignment. 
 	jz 2f				# Jump if alignment is ok.
 	subl $2, %ecx			# Alignment uses up two bytes.
 	jae 1f				# Jump if we had at least two bytes.
 	addl $2, %ecx			# ecx was < 2.  Deal with it.
 	jmp 4f
-SRC(1:	movw (%esi), %bx	)
+EXC(1:	movw (%esi), %bx	)
 	addl $2, %esi
-DST(	movw %bx, (%edi)	)
+EXC(	movw %bx, (%edi)	)
 	addl $2, %edi
 	addw %bx, %ax	
 	adcl $0, %eax
@@ -306,34 +295,34 @@ DST(	movw %bx, (%edi)	)
 	movl %ecx, FP(%esp)
 	shrl $5, %ecx
 	jz 2f
-	testl %esi, %esi
-SRC(1:	movl (%esi), %ebx	)
-SRC(	movl 4(%esi), %edx	)
+	testl %esi, %esi		# what's wrong with clc?
+EXC(1:	movl (%esi), %ebx	)
+EXC(	movl 4(%esi), %edx	)
 	adcl %ebx, %eax
-DST(	movl %ebx, (%edi)	)
+EXC(	movl %ebx, (%edi)	)
 	adcl %edx, %eax
-DST(	movl %edx, 4(%edi)	)
+EXC(	movl %edx, 4(%edi)	)
 
-SRC(	movl 8(%esi), %ebx	)
-SRC(	movl 12(%esi), %edx	)
+EXC(	movl 8(%esi), %ebx	)
+EXC(	movl 12(%esi), %edx	)
 	adcl %ebx, %eax
-DST(	movl %ebx, 8(%edi)	)
+EXC(	movl %ebx, 8(%edi)	)
 	adcl %edx, %eax
-DST(	movl %edx, 12(%edi)	)
+EXC(	movl %edx, 12(%edi)	)
 
-SRC(	movl 16(%esi), %ebx 	)
-SRC(	movl 20(%esi), %edx	)
+EXC(	movl 16(%esi), %ebx 	)
+EXC(	movl 20(%esi), %edx	)
 	adcl %ebx, %eax
-DST(	movl %ebx, 16(%edi)	)
+EXC(	movl %ebx, 16(%edi)	)
 	adcl %edx, %eax
-DST(	movl %edx, 20(%edi)	)
+EXC(	movl %edx, 20(%edi)	)
 
-SRC(	movl 24(%esi), %ebx	)
-SRC(	movl 28(%esi), %edx	)
+EXC(	movl 24(%esi), %ebx	)
+EXC(	movl 28(%esi), %edx	)
 	adcl %ebx, %eax
-DST(	movl %ebx, 24(%edi)	)
+EXC(	movl %ebx, 24(%edi)	)
 	adcl %edx, %eax
-DST(	movl %edx, 28(%edi)	)
+EXC(	movl %edx, 28(%edi)	)
 
 	lea 32(%esi), %esi
 	lea 32(%edi), %edi
@@ -345,9 +334,9 @@ DST(	movl %edx, 28(%edi)	)
 	andl $0x1c, %edx
 	je 4f
 	shrl $2, %edx			# This clears CF
-SRC(3:	movl (%esi), %ebx	)
+EXC(3:	movl (%esi), %ebx	)
 	adcl %ebx, %eax
-DST(	movl %ebx, (%edi)	)
+EXC(	movl %ebx, (%edi)	)
 	lea 4(%esi), %esi
 	lea 4(%edi), %edi
 	dec %edx
@@ -357,39 +346,24 @@ DST(	movl %ebx, (%edi)	)
 	jz 7f
 	cmpl $2, %ecx
 	jb 5f
-SRC(	movw (%esi), %cx	)
+EXC(	movw (%esi), %cx	)
 	leal 2(%esi), %esi
-DST(	movw %cx, (%edi)	)
+EXC(	movw %cx, (%edi)	)
 	leal 2(%edi), %edi
 	je 6f
 	shll $16,%ecx
-SRC(5:	movb (%esi), %cl	)
-DST(	movb %cl, (%edi)	)
+EXC(5:	movb (%esi), %cl	)
+EXC(	movb %cl, (%edi)	)
 6:	addl %ecx, %eax
 	adcl $0, %eax
 7:
-5000:
 
 # Exception handler:
 .section .fixup, "ax"							
 
 6001:
-	movl ARGBASE+20(%esp), %ebx	# src_err_ptr
-	movl $-EFAULT, (%ebx)
-
-	# zero the complete destination - computing the rest
-	# is too much work 
-	movl ARGBASE+8(%esp), %edi	# dst
-	movl ARGBASE+12(%esp), %ecx	# len
-	xorl %eax,%eax
-	rep ; stosb
-
-	jmp 5000b
-
-6002:
-	movl ARGBASE+24(%esp), %ebx	# dst_err_ptr
-	movl $-EFAULT,(%ebx)
-	jmp 5000b
+	xorl %eax, %eax
+	jmp 7b
 
 .previous
 
@@ -405,14 +379,14 @@ SYM_FUNC_END(csum_partial_copy_generic)
 /* Version for PentiumII/PPro */
 
 #define ROUND1(x) \
-	SRC(movl x(%esi), %ebx	)	;	\
+	EXC(movl x(%esi), %ebx	)	;	\
 	addl %ebx, %eax			;	\
-	DST(movl %ebx, x(%edi)	)	; 
+	EXC(movl %ebx, x(%edi)	)	;
 
 #define ROUND(x) \
-	SRC(movl x(%esi), %ebx	)	;	\
+	EXC(movl x(%esi), %ebx	)	;	\
 	adcl %ebx, %eax			;	\
-	DST(movl %ebx, x(%edi)	)	;
+	EXC(movl %ebx, x(%edi)	)	;
 
 #define ARGBASE 12
 		
@@ -423,7 +397,7 @@ SYM_FUNC_START(csum_partial_copy_generic)
 	movl ARGBASE+4(%esp),%esi	#src
 	movl ARGBASE+8(%esp),%edi	#dst	
 	movl ARGBASE+12(%esp),%ecx	#len
-	movl ARGBASE+16(%esp),%eax	#sum
+	movl $-1, %eax			#sum
 #	movl %ecx, %edx  
 	movl %ecx, %ebx  
 	movl %esi, %edx
@@ -439,7 +413,7 @@ SYM_FUNC_START(csum_partial_copy_generic)
 	JMP_NOSPEC ebx
 1:	addl $64,%esi
 	addl $64,%edi 
-	SRC(movb -32(%edx),%bl)	; SRC(movb (%edx),%bl)
+	EXC(movb -32(%edx),%bl)	; EXC(movb (%edx),%bl)
 	ROUND1(-64) ROUND(-60) ROUND(-56) ROUND(-52)	
 	ROUND (-48) ROUND(-44) ROUND(-40) ROUND(-36)	
 	ROUND (-32) ROUND(-28) ROUND(-24) ROUND(-20)	
@@ -453,29 +427,20 @@ SYM_FUNC_START(csum_partial_copy_generic)
 	jz 7f
 	cmpl $2, %edx
 	jb 5f
-SRC(	movw (%esi), %dx         )
+EXC(	movw (%esi), %dx         )
 	leal 2(%esi), %esi
-DST(	movw %dx, (%edi)         )
+EXC(	movw %dx, (%edi)         )
 	leal 2(%edi), %edi
 	je 6f
 	shll $16,%edx
 5:
-SRC(	movb (%esi), %dl         )
-DST(	movb %dl, (%edi)         )
+EXC(	movb (%esi), %dl         )
+EXC(	movb %dl, (%edi)         )
 6:	addl %edx, %eax
 	adcl $0, %eax
 7:
 .section .fixup, "ax"
-6001:	movl	ARGBASE+20(%esp), %ebx	# src_err_ptr	
-	movl $-EFAULT, (%ebx)
-	# zero the complete destination (computing the rest is too much work)
-	movl ARGBASE+8(%esp),%edi	# dst
-	movl ARGBASE+12(%esp),%ecx	# len
-	xorl %eax,%eax
-	rep; stosb
-	jmp 7b
-6002:	movl ARGBASE+24(%esp), %ebx	# dst_err_ptr
-	movl $-EFAULT, (%ebx)
+6001:	xorl %eax, %eax
 	jmp  7b			
 .previous				
 
diff --git a/arch/x86/lib/csum-copy_64.S b/arch/x86/lib/csum-copy_64.S
index 3394a8ff7fd0..1fbd8ee9642d 100644
--- a/arch/x86/lib/csum-copy_64.S
+++ b/arch/x86/lib/csum-copy_64.S
@@ -18,9 +18,6 @@
  * rdi  source
  * rsi  destination
  * edx  len (32bit)
- * ecx  sum (32bit)
- * r8   src_err_ptr (int)
- * r9   dst_err_ptr (int)
  *
  * Output
  * eax  64bit sum. undefined in case of exception.
@@ -31,44 +28,32 @@
 
 	.macro source
 10:
-	_ASM_EXTABLE_UA(10b, .Lbad_source)
+	_ASM_EXTABLE_UA(10b, .Lfault)
 	.endm
 
 	.macro dest
 20:
-	_ASM_EXTABLE_UA(20b, .Lbad_dest)
+	_ASM_EXTABLE_UA(20b, .Lfault)
 	.endm
 
-	/*
-	 * No _ASM_EXTABLE_UA; this is used for intentional prefetch on a
-	 * potentially unmapped kernel address.
-	 */
-	.macro ignore L=.Lignore
-30:
-	_ASM_EXTABLE(30b, \L)
-	.endm
-
-
 SYM_FUNC_START(csum_partial_copy_generic)
-	cmpl	$3*64, %edx
-	jle	.Lignore
-
-.Lignore:
-	subq  $7*8, %rsp
-	movq  %rbx, 2*8(%rsp)
-	movq  %r12, 3*8(%rsp)
-	movq  %r14, 4*8(%rsp)
-	movq  %r13, 5*8(%rsp)
-	movq  %r15, 6*8(%rsp)
+	subq  $5*8, %rsp
+	movq  %rbx, 0*8(%rsp)
+	movq  %r12, 1*8(%rsp)
+	movq  %r14, 2*8(%rsp)
+	movq  %r13, 3*8(%rsp)
+	movq  %r15, 4*8(%rsp)
 
-	movq  %r8, (%rsp)
-	movq  %r9, 1*8(%rsp)
-
-	movl  %ecx, %eax
+	movl  $-1, %eax
+	xorl  %r9d, %r9d
 	movl  %edx, %ecx
+	cmpl  $8, %ecx
+	jb    .Lshort
 
-	xorl  %r9d, %r9d
-	movq  %rcx, %r12
+	testb  $7, %sil
+	jne   .Lunaligned
+.Laligned:
+	movl  %ecx, %r12d
 
 	shrq  $6, %r12
 	jz	.Lhandle_tail       /* < 64 */
@@ -99,7 +84,12 @@ SYM_FUNC_START(csum_partial_copy_generic)
 	source
 	movq  56(%rdi), %r13
 
-	ignore 2f
+30:
+	/*
+	 * No _ASM_EXTABLE_UA; this is used for intentional prefetch on a
+	 * potentially unmapped kernel address.
+	 */
+	_ASM_EXTABLE(30b, 2f)
 	prefetcht0 5*64(%rdi)
 2:
 	adcq  %rbx, %rax
@@ -131,8 +121,6 @@ SYM_FUNC_START(csum_partial_copy_generic)
 	dest
 	movq %r13, 56(%rsi)
 
-3:
-
 	leaq 64(%rdi), %rdi
 	leaq 64(%rsi), %rsi
 
@@ -142,8 +130,8 @@ SYM_FUNC_START(csum_partial_copy_generic)
 
 	/* do last up to 56 bytes */
 .Lhandle_tail:
-	/* ecx:	count */
-	movl %ecx, %r10d
+	/* ecx:	count, rcx.63: the end result needs to be rol8 */
+	movq %rcx, %r10
 	andl $63, %ecx
 	shrl $3, %ecx
 	jz	.Lfold
@@ -172,6 +160,7 @@ SYM_FUNC_START(csum_partial_copy_generic)
 .Lhandle_7:
 	movl %r10d, %ecx
 	andl $7, %ecx
+.L1:				/* .Lshort rejoins the common path here */
 	shrl $1, %ecx
 	jz   .Lhandle_1
 	movl $2, %edx
@@ -203,26 +192,65 @@ SYM_FUNC_START(csum_partial_copy_generic)
 	adcl %r9d, %eax		/* carry */
 
 .Lende:
-	movq 2*8(%rsp), %rbx
-	movq 3*8(%rsp), %r12
-	movq 4*8(%rsp), %r14
-	movq 5*8(%rsp), %r13
-	movq 6*8(%rsp), %r15
-	addq $7*8, %rsp
+	testq %r10, %r10
+	js  .Lwas_odd
+.Lout:
+	movq 0*8(%rsp), %rbx
+	movq 1*8(%rsp), %r12
+	movq 2*8(%rsp), %r14
+	movq 3*8(%rsp), %r13
+	movq 4*8(%rsp), %r15
+	addq $5*8, %rsp
 	ret
+.Lshort:
+	movl %ecx, %r10d
+	jmp  .L1
+.Lunaligned:
+	xorl %ebx, %ebx
+	testb $1, %sil
+	jne  .Lodd
+1:	testb $2, %sil
+	je   2f
+	source
+	movw (%rdi), %bx
+	dest
+	movw %bx, (%rsi)
+	leaq 2(%rdi), %rdi
+	subq $2, %rcx
+	leaq 2(%rsi), %rsi
+	addq %rbx, %rax
+2:	testb $4, %sil
+	je .Laligned
+	source
+	movl (%rdi), %ebx
+	dest
+	movl %ebx, (%rsi)
+	leaq 4(%rdi), %rdi
+	subq $4, %rcx
+	leaq 4(%rsi), %rsi
+	addq %rbx, %rax
+	jmp .Laligned
+
+.Lodd:
+	source
+	movb (%rdi), %bl
+	dest
+	movb %bl, (%rsi)
+	leaq 1(%rdi), %rdi
+	leaq 1(%rsi), %rsi
+	/* decrement, set MSB */
+	leaq -1(%rcx, %rcx), %rcx
+	rorq $1, %rcx
+	shll $8, %ebx
+	addq %rbx, %rax
+	jmp 1b
+
+.Lwas_odd:
+	roll $8, %eax
+	jmp .Lout
 
-	/* Exception handlers. Very simple, zeroing is done in the wrappers */
-.Lbad_source:
-	movq (%rsp), %rax
-	testq %rax, %rax
-	jz   .Lende
-	movl $-EFAULT, (%rax)
-	jmp  .Lende
-
-.Lbad_dest:
-	movq 8(%rsp), %rax
-	testq %rax, %rax
-	jz   .Lende
-	movl $-EFAULT, (%rax)
-	jmp .Lende
+	/* Exception: just return 0 */
+.Lfault:
+	xorl %eax, %eax
+	jmp  .Lout
 SYM_FUNC_END(csum_partial_copy_generic)
diff --git a/arch/x86/lib/csum-wrappers_64.c b/arch/x86/lib/csum-wrappers_64.c
index ee63d7576fd2..189344924a2b 100644
--- a/arch/x86/lib/csum-wrappers_64.c
+++ b/arch/x86/lib/csum-wrappers_64.c
@@ -21,52 +21,16 @@
  * src and dst are best aligned to 64bits.
  */
 __wsum
-csum_and_copy_from_user(const void __user *src, void *dst,
-			    int len, __wsum isum, int *errp)
+csum_and_copy_from_user(const void __user *src, void *dst, int len)
 {
-	might_sleep();
-	*errp = 0;
+	__wsum sum;
 
+	might_sleep();
 	if (!user_access_begin(src, len))
-		goto out_err;
-
-	/*
-	 * Why 6, not 7? To handle odd addresses aligned we
-	 * would need to do considerable complications to fix the
-	 * checksum which is defined as an 16bit accumulator. The
-	 * fix alignment code is primarily for performance
-	 * compatibility with 32bit and that will handle odd
-	 * addresses slowly too.
-	 */
-	if (unlikely((unsigned long)src & 6)) {
-		while (((unsigned long)src & 6) && len >= 2) {
-			__u16 val16;
-
-			unsafe_get_user(val16, (const __u16 __user *)src, out);
-
-			*(__u16 *)dst = val16;
-			isum = (__force __wsum)add32_with_carry(
-					(__force unsigned)isum, val16);
-			src += 2;
-			dst += 2;
-			len -= 2;
-		}
-	}
-	isum = csum_partial_copy_generic((__force const void *)src,
-				dst, len, isum, errp, NULL);
-	user_access_end();
-	if (unlikely(*errp))
-		goto out_err;
-
-	return isum;
-
-out:
+		return 0;
+	sum = csum_partial_copy_generic((__force const void *)src, dst, len);
 	user_access_end();
-out_err:
-	*errp = -EFAULT;
-	memset(dst, 0, len);
-
-	return isum;
+	return sum;
 }
 EXPORT_SYMBOL(csum_and_copy_from_user);
 
@@ -82,40 +46,16 @@ EXPORT_SYMBOL(csum_and_copy_from_user);
  * src and dst are best aligned to 64bits.
  */
 __wsum
-csum_and_copy_to_user(const void *src, void __user *dst,
-			  int len, __wsum isum, int *errp)
+csum_and_copy_to_user(const void *src, void __user *dst, int len)
 {
-	__wsum ret;
+	__wsum sum;
 
 	might_sleep();
-
-	if (!user_access_begin(dst, len)) {
-		*errp = -EFAULT;
+	if (!user_access_begin(dst, len))
 		return 0;
-	}
-
-	if (unlikely((unsigned long)dst & 6)) {
-		while (((unsigned long)dst & 6) && len >= 2) {
-			__u16 val16 = *(__u16 *)src;
-
-			isum = (__force __wsum)add32_with_carry(
-					(__force unsigned)isum, val16);
-			unsafe_put_user(val16, (__u16 __user *)dst, out);
-			src += 2;
-			dst += 2;
-			len -= 2;
-		}
-	}
-
-	*errp = 0;
-	ret = csum_partial_copy_generic(src, (void __force *)dst,
-					len, isum, NULL, errp);
-	user_access_end();
-	return ret;
-out:
+	sum = csum_partial_copy_generic(src, (void __force *)dst, len);
 	user_access_end();
-	*errp = -EFAULT;
-	return isum;
+	return sum;
 }
 EXPORT_SYMBOL(csum_and_copy_to_user);
 
@@ -129,9 +69,9 @@ EXPORT_SYMBOL(csum_and_copy_to_user);
  * Returns an 32bit unfolded checksum of the buffer.
  */
 __wsum
-csum_partial_copy_nocheck(const void *src, void *dst, int len, __wsum sum)
+csum_partial_copy_nocheck(const void *src, void *dst, int len)
 {
-	return csum_partial_copy_generic(src, dst, len, sum, NULL, NULL);
+	return csum_partial_copy_generic(src, dst, len);
 }
 EXPORT_SYMBOL(csum_partial_copy_nocheck);