From c81c5c1010482d247b7263844e02f7371fbd2565 Mon Sep 17 00:00:00 2001
From: Frank Denis <github@pureftpd.org>
Date: Wed, 7 May 2014 21:16:19 -0700
Subject: [PATCH] Sync stream_salsa20_amd64_xmm6.S with SUPERCOP 2014-04-25

---
 .../amd64_xmm6/stream_salsa20_amd64_xmm6.S    | 299 +++++++++---------
 1 file changed, 142 insertions(+), 157 deletions(-)

diff --git a/src/libsodium/crypto_stream/salsa20/amd64_xmm6/stream_salsa20_amd64_xmm6.S b/src/libsodium/crypto_stream/salsa20/amd64_xmm6/stream_salsa20_amd64_xmm6.S
index 940d38ca..8e29a506 100644
--- a/src/libsodium/crypto_stream/salsa20/amd64_xmm6/stream_salsa20_amd64_xmm6.S
+++ b/src/libsodium/crypto_stream/salsa20/amd64_xmm6/stream_salsa20_amd64_xmm6.S
@@ -13,15 +13,15 @@ crypto_stream_salsa20:
 _crypto_stream_salsa20:
 mov %rsp,%r11
 and $31,%r11
-add $480,%r11
+add $512,%r11
 sub %r11,%rsp
-movq %r11,352(%rsp)
-movq %r12,360(%rsp)
-movq %r13,368(%rsp)
-movq %r14,376(%rsp)
-movq %r15,384(%rsp)
-movq %rbx,392(%rsp)
-movq %rbp,400(%rsp)
+movq %r11,416(%rsp)
+movq %r12,424(%rsp)
+movq %r13,432(%rsp)
+movq %r14,440(%rsp)
+movq %r15,448(%rsp)
+movq %rbx,456(%rsp)
+movq %rbp,464(%rsp)
 mov  %rsi,%r9
 mov  %rdi,%rdi
 mov  %rdi,%rsi
@@ -29,7 +29,6 @@ mov  %rdx,%rdx
 mov  %rcx,%r10
 cmp  $0,%r9
 jbe ._done
-
 mov  $0,%rax
 mov  %r9,%rcx
 rep stosb
@@ -47,17 +46,18 @@ jmp ._start
 #endif
 crypto_stream_salsa20_xor:
 _crypto_stream_salsa20_xor:
+
 mov %rsp,%r11
 and $31,%r11
-add $480,%r11
+add $512,%r11
 sub %r11,%rsp
-movq %r11,352(%rsp)
-movq %r12,360(%rsp)
-movq %r13,368(%rsp)
-movq %r14,376(%rsp)
-movq %r15,384(%rsp)
-movq %rbx,392(%rsp)
-movq %rbp,400(%rsp)
+movq %r11,416(%rsp)
+movq %r12,424(%rsp)
+movq %r13,432(%rsp)
+movq %r14,440(%rsp)
+movq %r15,448(%rsp)
+movq %rbx,456(%rsp)
+movq %rbp,464(%rsp)
 mov  %rdi,%rdi
 mov  %rsi,%rsi
 mov  %rdx,%r9
@@ -71,125 +71,119 @@ movl   20(%r10),%ecx
 movl   0(%r10),%r8d
 movl   0(%rdx),%eax
 movl   16(%r10),%r11d
-movl %ecx,0(%rsp)
-movl %r8d,4+0(%rsp)
-movl %eax,8+0(%rsp)
-movl %r11d,12+0(%rsp)
+movl %ecx,64(%rsp)
+movl %r8d,4+64(%rsp)
+movl %eax,8+64(%rsp)
+movl %r11d,12+64(%rsp)
 mov  $0,%rcx
 movl   24(%r10),%r8d
 movl   4(%r10),%eax
 movl   4(%rdx),%edx
-movl %ecx,16(%rsp)
-movl %r8d,4+16(%rsp)
-movl %eax,8+16(%rsp)
-movl %edx,12+16(%rsp)
+movq %rcx,472(%rsp)
+movl %ecx,80(%rsp)
+movl %r8d,4+80(%rsp)
+movl %eax,8+80(%rsp)
+movl %edx,12+80(%rsp)
 movl   12(%r10),%edx
 mov  $0,%rcx
 movl   28(%r10),%r8d
 movl   8(%r10),%eax
-movl %edx,32(%rsp)
-movl %ecx,4+32(%rsp)
-movl %r8d,8+32(%rsp)
-movl %eax,12+32(%rsp)
+movl %edx,96(%rsp)
+movl %ecx,4+96(%rsp)
+movl %r8d,8+96(%rsp)
+movl %eax,12+96(%rsp)
 mov  $1634760805,%rdx
 mov  $857760878,%rcx
 mov  $2036477234,%r8
 mov  $1797285236,%rax
-movl %edx,48(%rsp)
-movl %ecx,4+48(%rsp)
-movl %r8d,8+48(%rsp)
-movl %eax,12+48(%rsp)
+movl %edx,112(%rsp)
+movl %ecx,4+112(%rsp)
+movl %r8d,8+112(%rsp)
+movl %eax,12+112(%rsp)
 cmp  $256,%r9
 jb ._bytesbetween1and255
-
-movdqa 48(%rsp),%xmm0
+movdqa 112(%rsp),%xmm0
 pshufd $0x55,%xmm0,%xmm1
 pshufd $0xaa,%xmm0,%xmm2
 pshufd $0xff,%xmm0,%xmm3
 pshufd $0x00,%xmm0,%xmm0
-movdqa %xmm1,64(%rsp)
-movdqa %xmm2,80(%rsp)
-movdqa %xmm3,96(%rsp)
-movdqa %xmm0,112(%rsp)
-movdqa 0(%rsp),%xmm0
-pshufd $0xaa,%xmm0,%xmm1
-pshufd $0xff,%xmm0,%xmm2
-pshufd $0x00,%xmm0,%xmm3
-pshufd $0x55,%xmm0,%xmm0
 movdqa %xmm1,128(%rsp)
 movdqa %xmm2,144(%rsp)
 movdqa %xmm3,160(%rsp)
 movdqa %xmm0,176(%rsp)
-movdqa 16(%rsp),%xmm0
+movdqa 64(%rsp),%xmm0
+pshufd $0xaa,%xmm0,%xmm1
+pshufd $0xff,%xmm0,%xmm2
+pshufd $0x00,%xmm0,%xmm3
+pshufd $0x55,%xmm0,%xmm0
+movdqa %xmm1,192(%rsp)
+movdqa %xmm2,208(%rsp)
+movdqa %xmm3,224(%rsp)
+movdqa %xmm0,240(%rsp)
+movdqa 80(%rsp),%xmm0
 pshufd $0xff,%xmm0,%xmm1
 pshufd $0x55,%xmm0,%xmm2
 pshufd $0xaa,%xmm0,%xmm0
-movdqa %xmm1,192(%rsp)
-movdqa %xmm2,208(%rsp)
-movdqa %xmm0,224(%rsp)
-movdqa 32(%rsp),%xmm0
+movdqa %xmm1,256(%rsp)
+movdqa %xmm2,272(%rsp)
+movdqa %xmm0,288(%rsp)
+movdqa 96(%rsp),%xmm0
 pshufd $0x00,%xmm0,%xmm1
 pshufd $0xaa,%xmm0,%xmm2
 pshufd $0xff,%xmm0,%xmm0
-movdqa %xmm1,240(%rsp)
-movdqa %xmm2,256(%rsp)
-movdqa %xmm0,272(%rsp)
+movdqa %xmm1,304(%rsp)
+movdqa %xmm2,320(%rsp)
+movdqa %xmm0,336(%rsp)
 
 ._bytesatleast256:
-movl 16(%rsp),%edx
-movl 4+32(%rsp),%ecx
-movl %edx,288(%rsp)
-movl %ecx,304(%rsp)
-add  $1,%rdx
-shl  $32,%rcx
-add  %rcx,%rdx
+movq 472(%rsp),%rdx
 mov  %rdx,%rcx
 shr  $32,%rcx
-movl %edx,4+288(%rsp)
-movl %ecx,4+304(%rsp)
+movl %edx,352(%rsp)
+movl %ecx,368(%rsp)
 add  $1,%rdx
-shl  $32,%rcx
-add  %rcx,%rdx
 mov  %rdx,%rcx
 shr  $32,%rcx
-movl %edx,8+288(%rsp)
-movl %ecx,8+304(%rsp)
+movl %edx,4+352(%rsp)
+movl %ecx,4+368(%rsp)
 add  $1,%rdx
-shl  $32,%rcx
-add  %rcx,%rdx
 mov  %rdx,%rcx
 shr  $32,%rcx
-movl %edx,12+288(%rsp)
-movl %ecx,12+304(%rsp)
+movl %edx,8+352(%rsp)
+movl %ecx,8+368(%rsp)
 add  $1,%rdx
-shl  $32,%rcx
-add  %rcx,%rdx
 mov  %rdx,%rcx
 shr  $32,%rcx
-movl %edx,16(%rsp)
-movl %ecx,4+32(%rsp)
-movq %r9,408(%rsp)
+movl %edx,12+352(%rsp)
+movl %ecx,12+368(%rsp)
+add  $1,%rdx
+mov  %rdx,%rcx
+shr  $32,%rcx
+movl %edx,80(%rsp)
+movl %ecx,4+96(%rsp)
+movq %rdx,472(%rsp)
+movq %r9,480(%rsp)
 mov  $20,%rdx
-movdqa 64(%rsp),%xmm0
-movdqa 80(%rsp),%xmm1
-movdqa 96(%rsp),%xmm2
-movdqa 256(%rsp),%xmm3
-movdqa 272(%rsp),%xmm4
-movdqa 128(%rsp),%xmm5
-movdqa 144(%rsp),%xmm6
-movdqa 176(%rsp),%xmm7
-movdqa 192(%rsp),%xmm8
-movdqa 208(%rsp),%xmm9
-movdqa 224(%rsp),%xmm10
-movdqa 304(%rsp),%xmm11
-movdqa 112(%rsp),%xmm12
-movdqa 160(%rsp),%xmm13
-movdqa 240(%rsp),%xmm14
-movdqa 288(%rsp),%xmm15
+movdqa 128(%rsp),%xmm0
+movdqa 144(%rsp),%xmm1
+movdqa 160(%rsp),%xmm2
+movdqa 320(%rsp),%xmm3
+movdqa 336(%rsp),%xmm4
+movdqa 192(%rsp),%xmm5
+movdqa 208(%rsp),%xmm6
+movdqa 240(%rsp),%xmm7
+movdqa 256(%rsp),%xmm8
+movdqa 272(%rsp),%xmm9
+movdqa 288(%rsp),%xmm10
+movdqa 368(%rsp),%xmm11
+movdqa 176(%rsp),%xmm12
+movdqa 224(%rsp),%xmm13
+movdqa 304(%rsp),%xmm14
+movdqa 352(%rsp),%xmm15
 
 ._mainloop1:
-movdqa %xmm1,320(%rsp)
-movdqa %xmm2,336(%rsp)
+movdqa %xmm1,384(%rsp)
+movdqa %xmm2,400(%rsp)
 movdqa %xmm13,%xmm1
 paddd %xmm12,%xmm1
 movdqa %xmm1,%xmm2
@@ -239,8 +233,8 @@ pslld $18,%xmm1
 pxor  %xmm1,%xmm12
 psrld $14,%xmm2
 pxor  %xmm2,%xmm12
-movdqa 320(%rsp),%xmm1
-movdqa %xmm12,320(%rsp)
+movdqa 384(%rsp),%xmm1
+movdqa %xmm12,384(%rsp)
 movdqa %xmm9,%xmm2
 paddd %xmm7,%xmm2
 movdqa %xmm2,%xmm12
@@ -255,8 +249,8 @@ pslld $7,%xmm2
 pxor  %xmm2,%xmm3
 psrld $25,%xmm12
 pxor  %xmm12,%xmm3
-movdqa 336(%rsp),%xmm2
-movdqa %xmm0,336(%rsp)
+movdqa 400(%rsp),%xmm2
+movdqa %xmm0,400(%rsp)
 movdqa %xmm6,%xmm0
 paddd %xmm2,%xmm0
 movdqa %xmm0,%xmm12
@@ -299,8 +293,8 @@ pslld $18,%xmm0
 pxor  %xmm0,%xmm1
 psrld $14,%xmm12
 pxor  %xmm12,%xmm1
-movdqa 320(%rsp),%xmm0
-movdqa %xmm1,320(%rsp)
+movdqa 384(%rsp),%xmm0
+movdqa %xmm1,384(%rsp)
 movdqa %xmm4,%xmm1
 paddd %xmm0,%xmm1
 movdqa %xmm1,%xmm12
@@ -315,8 +309,8 @@ pslld $18,%xmm1
 pxor  %xmm1,%xmm2
 psrld $14,%xmm12
 pxor  %xmm12,%xmm2
-movdqa 336(%rsp),%xmm12
-movdqa %xmm2,336(%rsp)
+movdqa 400(%rsp),%xmm12
+movdqa %xmm2,400(%rsp)
 movdqa %xmm14,%xmm1
 paddd %xmm12,%xmm1
 movdqa %xmm1,%xmm2
@@ -359,8 +353,8 @@ pslld $18,%xmm1
 pxor  %xmm1,%xmm0
 psrld $14,%xmm2
 pxor  %xmm2,%xmm0
-movdqa 320(%rsp),%xmm1
-movdqa %xmm0,320(%rsp)
+movdqa 384(%rsp),%xmm1
+movdqa %xmm0,384(%rsp)
 movdqa %xmm8,%xmm0
 paddd %xmm14,%xmm0
 movdqa %xmm0,%xmm2
@@ -375,8 +369,8 @@ pslld $7,%xmm0
 pxor  %xmm0,%xmm6
 psrld $25,%xmm2
 pxor  %xmm2,%xmm6
-movdqa 336(%rsp),%xmm2
-movdqa %xmm12,336(%rsp)
+movdqa 400(%rsp),%xmm2
+movdqa %xmm12,400(%rsp)
 movdqa %xmm3,%xmm0
 paddd %xmm2,%xmm0
 movdqa %xmm0,%xmm12
@@ -426,15 +420,14 @@ pslld $18,%xmm0
 pxor  %xmm0,%xmm2
 psrld $14,%xmm12
 pxor  %xmm12,%xmm2
-movdqa 320(%rsp),%xmm12
-movdqa 336(%rsp),%xmm0
+movdqa 384(%rsp),%xmm12
+movdqa 400(%rsp),%xmm0
 sub  $2,%rdx
 ja ._mainloop1
-
-paddd 112(%rsp),%xmm12
-paddd 176(%rsp),%xmm7
-paddd 224(%rsp),%xmm10
-paddd 272(%rsp),%xmm4
+paddd 176(%rsp),%xmm12
+paddd 240(%rsp),%xmm7
+paddd 288(%rsp),%xmm10
+paddd 336(%rsp),%xmm4
 movd   %xmm12,%rdx
 movd   %xmm7,%rcx
 movd   %xmm10,%r8
@@ -495,10 +488,10 @@ movl   %edx,192(%rdi)
 movl   %ecx,196(%rdi)
 movl   %r8d,200(%rdi)
 movl   %r9d,204(%rdi)
-paddd 240(%rsp),%xmm14
-paddd 64(%rsp),%xmm0
-paddd 128(%rsp),%xmm5
-paddd 192(%rsp),%xmm8
+paddd 304(%rsp),%xmm14
+paddd 128(%rsp),%xmm0
+paddd 192(%rsp),%xmm5
+paddd 256(%rsp),%xmm8
 movd   %xmm14,%rdx
 movd   %xmm0,%rcx
 movd   %xmm5,%r8
@@ -559,10 +552,10 @@ movl   %edx,208(%rdi)
 movl   %ecx,212(%rdi)
 movl   %r8d,216(%rdi)
 movl   %r9d,220(%rdi)
-paddd 288(%rsp),%xmm15
-paddd 304(%rsp),%xmm11
-paddd 80(%rsp),%xmm1
-paddd 144(%rsp),%xmm6
+paddd 352(%rsp),%xmm15
+paddd 368(%rsp),%xmm11
+paddd 144(%rsp),%xmm1
+paddd 208(%rsp),%xmm6
 movd   %xmm15,%rdx
 movd   %xmm11,%rcx
 movd   %xmm1,%r8
@@ -623,10 +616,10 @@ movl   %edx,224(%rdi)
 movl   %ecx,228(%rdi)
 movl   %r8d,232(%rdi)
 movl   %r9d,236(%rdi)
-paddd 160(%rsp),%xmm13
-paddd 208(%rsp),%xmm9
-paddd 256(%rsp),%xmm3
-paddd 96(%rsp),%xmm2
+paddd 224(%rsp),%xmm13
+paddd 272(%rsp),%xmm9
+paddd 320(%rsp),%xmm3
+paddd 160(%rsp),%xmm2
 movd   %xmm13,%rdx
 movd   %xmm9,%rcx
 movd   %xmm3,%r8
@@ -687,33 +680,31 @@ movl   %edx,240(%rdi)
 movl   %ecx,244(%rdi)
 movl   %r8d,248(%rdi)
 movl   %r9d,252(%rdi)
-movq 408(%rsp),%r9
+movq 480(%rsp),%r9
 sub  $256,%r9
 add  $256,%rsi
 add  $256,%rdi
 cmp  $256,%r9
 jae ._bytesatleast256
-
 cmp  $0,%r9
 jbe ._done
 
 ._bytesbetween1and255:
 cmp  $64,%r9
 jae ._nocopy
-
 mov  %rdi,%rdx
-leaq 416(%rsp),%rdi
+leaq 0(%rsp),%rdi
 mov  %r9,%rcx
 rep movsb
-leaq 416(%rsp),%rdi
-leaq 416(%rsp),%rsi
+leaq 0(%rsp),%rdi
+leaq 0(%rsp),%rsi
 
 ._nocopy:
-movq %r9,408(%rsp)
-movdqa 48(%rsp),%xmm0
-movdqa 0(%rsp),%xmm1
-movdqa 16(%rsp),%xmm2
-movdqa 32(%rsp),%xmm3
+movq %r9,480(%rsp)
+movdqa 112(%rsp),%xmm0
+movdqa 64(%rsp),%xmm1
+movdqa 80(%rsp),%xmm2
+movdqa 96(%rsp),%xmm3
 movdqa %xmm1,%xmm4
 mov  $20,%rcx
 
@@ -845,11 +836,10 @@ pxor  %xmm5,%xmm0
 pshufd $0x39,%xmm3,%xmm3
 pxor  %xmm6,%xmm0
 ja ._mainloop2
-
-paddd 48(%rsp),%xmm0
-paddd 0(%rsp),%xmm1
-paddd 16(%rsp),%xmm2
-paddd 32(%rsp),%xmm3
+paddd 112(%rsp),%xmm0
+paddd 64(%rsp),%xmm1
+paddd 80(%rsp),%xmm2
+paddd 96(%rsp),%xmm3
 movd   %xmm0,%rcx
 movd   %xmm1,%r8
 movd   %xmm2,%r9
@@ -910,22 +900,17 @@ movl   %ecx,60(%rdi)
 movl   %r8d,44(%rdi)
 movl   %r9d,28(%rdi)
 movl   %eax,12(%rdi)
-movq 408(%rsp),%r9
-movl 16(%rsp),%ecx
-movl 4+32(%rsp),%r8d
+movq 480(%rsp),%r9
+movq 472(%rsp),%rcx
 add  $1,%rcx
-shl  $32,%r8
-add  %r8,%rcx
 mov  %rcx,%r8
 shr  $32,%r8
-movl %ecx,16(%rsp)
-movl %r8d,4+32(%rsp)
+movl %ecx,80(%rsp)
+movl %r8d,4+96(%rsp)
+movq %rcx,472(%rsp)
 cmp  $64,%r9
-
 ja ._bytesatleast65
-
 jae ._bytesatleast64
-
 mov  %rdi,%rsi
 mov  %rdx,%rdi
 mov  %r9,%rcx
@@ -933,16 +918,16 @@ rep movsb
 
 ._bytesatleast64:
 ._done:
-movq 352(%rsp),%r11
-movq 360(%rsp),%r12
-movq 368(%rsp),%r13
-movq 376(%rsp),%r14
-movq 384(%rsp),%r15
-movq 392(%rsp),%rbx
-movq 400(%rsp),%rbp
+movq 416(%rsp),%r11
+movq 424(%rsp),%r12
+movq 432(%rsp),%r13
+movq 440(%rsp),%r14
+movq 448(%rsp),%r15
+movq 456(%rsp),%rbx
+movq 464(%rsp),%rbp
 add %r11,%rsp
 xor %rax,%rax
-xor %rdx,%rdx
+mov %rsi,%rdx
 ret
 
 ._bytesatleast65: