diff options
Diffstat (limited to 'crypto/aes/asm/aes-586.pl')
-rwxr-xr-x | crypto/aes/asm/aes-586.pl | 12 |
1 files changed, 6 insertions, 6 deletions
diff --git a/crypto/aes/asm/aes-586.pl b/crypto/aes/asm/aes-586.pl index 1ba356508a..61bdce865c 100755 --- a/crypto/aes/asm/aes-586.pl +++ b/crypto/aes/asm/aes-586.pl @@ -123,7 +123,7 @@ # words every cache-line is *guaranteed* to be accessed within ~50 # cycles window. Why just SSE? Because it's needed on hyper-threading # CPU! Which is also why it's prefetched with 64 byte stride. Best -# part is that it has no negative effect on performance:-) +# part is that it has no negative effect on performance:-) # # Version 4.3 implements switch between compact and non-compact block # functions in AES_cbc_encrypt depending on how much data was asked @@ -585,7 +585,7 @@ sub enctransform() # +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+ # | mm4 | mm0 | # +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+ -# | s3 | s2 | s1 | s0 | +# | s3 | s2 | s1 | s0 | # +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+ # |15|14|13|12|11|10| 9| 8| 7| 6| 5| 4| 3| 2| 1| 0| # +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+ @@ -805,7 +805,7 @@ sub encstep() if ($i==3) { $tmp=$s[3]; &mov ($s[2],$__s1); }##%ecx elsif($i==2){ &movz ($tmp,&HB($s[3])); }#%ebx[2] - else { &mov ($tmp,$s[3]); + else { &mov ($tmp,$s[3]); &shr ($tmp,24) } &xor ($out,&DWP(1,$te,$tmp,8)); if ($i<2) { &mov (&DWP(4+4*$i,"esp"),$out); } @@ -1558,7 +1558,7 @@ sub sse_deccompact() &pxor ("mm1","mm3"); &pxor ("mm5","mm7"); # tp4 &pshufw ("mm3","mm1",0xb1); &pshufw ("mm7","mm5",0xb1); &pxor ("mm0","mm1"); &pxor ("mm4","mm5"); # ^= tp4 - &pxor ("mm0","mm3"); &pxor ("mm4","mm7"); # ^= ROTATE(tp4,16) + &pxor ("mm0","mm3"); &pxor ("mm4","mm7"); # ^= ROTATE(tp4,16) &pxor ("mm3","mm3"); &pxor ("mm7","mm7"); &pcmpgtb("mm3","mm1"); &pcmpgtb("mm7","mm5"); @@ -2028,7 +2028,7 @@ sub declast() { # stack frame layout # -4(%esp) # return address 0(%esp) -# 0(%esp) # s0 backing store 4(%esp) +# 0(%esp) # s0 backing store 4(%esp) # 4(%esp) # s1 backing store 8(%esp) # 8(%esp) # s2 backing store 12(%esp) # 12(%esp) # s3 backing store 16(%esp) @@ -2738,7 +2738,7 @@ sub enckey() &mov (&DWP(80,"edi"),10); # setup number of rounds &xor ("eax","eax"); &jmp (&label("exit")); - + &set_label("12rounds"); &mov ("eax",&DWP(0,"esi")); # copy first 6 dwords &mov ("ebx",&DWP(4,"esi")); |