diff options
Diffstat (limited to 'mac-x86/crypto/aes/aesni-x86.S')
| -rw-r--r-- | mac-x86/crypto/aes/aesni-x86.S | 793 |
1 files changed, 524 insertions, 269 deletions
diff --git a/mac-x86/crypto/aes/aesni-x86.S b/mac-x86/crypto/aes/aesni-x86.S index 9000478..07719ba 100644 --- a/mac-x86/crypto/aes/aesni-x86.S +++ b/mac-x86/crypto/aes/aesni-x86.S @@ -22,7 +22,10 @@ L000enc1_loop_1: leal 16(%edx),%edx jnz L000enc1_loop_1 .byte 102,15,56,221,209 + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 movups %xmm2,(%eax) + pxor %xmm2,%xmm2 ret .globl _aesni_decrypt .private_extern _aesni_decrypt @@ -45,7 +48,10 @@ L001dec1_loop_2: leal 16(%edx),%edx jnz L001dec1_loop_2 .byte 102,15,56,223,209 + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 movups %xmm2,(%eax) + pxor %xmm2,%xmm2 ret .private_extern __aesni_encrypt2 .align 4 @@ -252,17 +258,15 @@ __aesni_encrypt6: negl %ecx .byte 102,15,56,220,225 pxor %xmm0,%xmm7 + movups (%edx,%ecx,1),%xmm0 addl $16,%ecx -.byte 102,15,56,220,233 -.byte 102,15,56,220,241 -.byte 102,15,56,220,249 - movups -16(%edx,%ecx,1),%xmm0 - jmp L_aesni_encrypt6_enter + jmp L008_aesni_encrypt6_inner .align 4,0x90 -L008enc6_loop: +L009enc6_loop: .byte 102,15,56,220,209 .byte 102,15,56,220,217 .byte 102,15,56,220,225 +L008_aesni_encrypt6_inner: .byte 102,15,56,220,233 .byte 102,15,56,220,241 .byte 102,15,56,220,249 @@ -276,7 +280,7 @@ L_aesni_encrypt6_enter: .byte 102,15,56,220,240 .byte 102,15,56,220,248 movups -16(%edx,%ecx,1),%xmm0 - jnz L008enc6_loop + jnz L009enc6_loop .byte 102,15,56,220,209 .byte 102,15,56,220,217 .byte 102,15,56,220,225 @@ -307,17 +311,15 @@ __aesni_decrypt6: negl %ecx .byte 102,15,56,222,225 pxor %xmm0,%xmm7 + movups (%edx,%ecx,1),%xmm0 addl $16,%ecx -.byte 102,15,56,222,233 -.byte 102,15,56,222,241 -.byte 102,15,56,222,249 - movups -16(%edx,%ecx,1),%xmm0 - jmp L_aesni_decrypt6_enter + jmp L010_aesni_decrypt6_inner .align 4,0x90 -L009dec6_loop: +L011dec6_loop: .byte 102,15,56,222,209 .byte 102,15,56,222,217 .byte 102,15,56,222,225 +L010_aesni_decrypt6_inner: .byte 102,15,56,222,233 .byte 102,15,56,222,241 .byte 102,15,56,222,249 @@ -331,7 +333,7 @@ L_aesni_decrypt6_enter: .byte 102,15,56,222,240 .byte 102,15,56,222,248 movups -16(%edx,%ecx,1),%xmm0 - jnz L009dec6_loop + jnz L011dec6_loop .byte 102,15,56,222,209 .byte 102,15,56,222,217 .byte 102,15,56,222,225 @@ -360,14 +362,14 @@ L_aesni_ecb_encrypt_begin: movl 32(%esp),%edx movl 36(%esp),%ebx andl $-16,%eax - jz L010ecb_ret + jz L012ecb_ret movl 240(%edx),%ecx testl %ebx,%ebx - jz L011ecb_decrypt + jz L013ecb_decrypt movl %edx,%ebp movl %ecx,%ebx cmpl $96,%eax - jb L012ecb_enc_tail + jb L014ecb_enc_tail movdqu (%esi),%xmm2 movdqu 16(%esi),%xmm3 movdqu 32(%esi),%xmm4 @@ -376,9 +378,9 @@ L_aesni_ecb_encrypt_begin: movdqu 80(%esi),%xmm7 leal 96(%esi),%esi subl $96,%eax - jmp L013ecb_enc_loop6_enter + jmp L015ecb_enc_loop6_enter .align 4,0x90 -L014ecb_enc_loop6: +L016ecb_enc_loop6: movups %xmm2,(%edi) movdqu (%esi),%xmm2 movups %xmm3,16(%edi) @@ -393,12 +395,12 @@ L014ecb_enc_loop6: leal 96(%edi),%edi movdqu 80(%esi),%xmm7 leal 96(%esi),%esi -L013ecb_enc_loop6_enter: +L015ecb_enc_loop6_enter: call __aesni_encrypt6 movl %ebp,%edx movl %ebx,%ecx subl $96,%eax - jnc L014ecb_enc_loop6 + jnc L016ecb_enc_loop6 movups %xmm2,(%edi) movups %xmm3,16(%edi) movups %xmm4,32(%edi) @@ -407,18 +409,18 @@ L013ecb_enc_loop6_enter: movups %xmm7,80(%edi) leal 96(%edi),%edi addl $96,%eax - jz L010ecb_ret -L012ecb_enc_tail: + jz L012ecb_ret +L014ecb_enc_tail: movups (%esi),%xmm2 cmpl $32,%eax - jb L015ecb_enc_one + jb L017ecb_enc_one movups 16(%esi),%xmm3 - je L016ecb_enc_two + je L018ecb_enc_two movups 32(%esi),%xmm4 cmpl $64,%eax - jb L017ecb_enc_three + jb L019ecb_enc_three movups 48(%esi),%xmm5 - je L018ecb_enc_four + je L020ecb_enc_four movups 64(%esi),%xmm6 xorps %xmm7,%xmm7 call __aesni_encrypt6 @@ -427,49 +429,49 @@ L012ecb_enc_tail: movups %xmm4,32(%edi) movups %xmm5,48(%edi) movups %xmm6,64(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L015ecb_enc_one: +L017ecb_enc_one: movups (%edx),%xmm0 movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L019enc1_loop_3: +L021enc1_loop_3: .byte 102,15,56,220,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L019enc1_loop_3 + jnz L021enc1_loop_3 .byte 102,15,56,221,209 movups %xmm2,(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L016ecb_enc_two: +L018ecb_enc_two: call __aesni_encrypt2 movups %xmm2,(%edi) movups %xmm3,16(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L017ecb_enc_three: +L019ecb_enc_three: call __aesni_encrypt3 movups %xmm2,(%edi) movups %xmm3,16(%edi) movups %xmm4,32(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L018ecb_enc_four: +L020ecb_enc_four: call __aesni_encrypt4 movups %xmm2,(%edi) movups %xmm3,16(%edi) movups %xmm4,32(%edi) movups %xmm5,48(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L011ecb_decrypt: +L013ecb_decrypt: movl %edx,%ebp movl %ecx,%ebx cmpl $96,%eax - jb L020ecb_dec_tail + jb L022ecb_dec_tail movdqu (%esi),%xmm2 movdqu 16(%esi),%xmm3 movdqu 32(%esi),%xmm4 @@ -478,9 +480,9 @@ L011ecb_decrypt: movdqu 80(%esi),%xmm7 leal 96(%esi),%esi subl $96,%eax - jmp L021ecb_dec_loop6_enter + jmp L023ecb_dec_loop6_enter .align 4,0x90 -L022ecb_dec_loop6: +L024ecb_dec_loop6: movups %xmm2,(%edi) movdqu (%esi),%xmm2 movups %xmm3,16(%edi) @@ -495,12 +497,12 @@ L022ecb_dec_loop6: leal 96(%edi),%edi movdqu 80(%esi),%xmm7 leal 96(%esi),%esi -L021ecb_dec_loop6_enter: +L023ecb_dec_loop6_enter: call __aesni_decrypt6 movl %ebp,%edx movl %ebx,%ecx subl $96,%eax - jnc L022ecb_dec_loop6 + jnc L024ecb_dec_loop6 movups %xmm2,(%edi) movups %xmm3,16(%edi) movups %xmm4,32(%edi) @@ -509,18 +511,18 @@ L021ecb_dec_loop6_enter: movups %xmm7,80(%edi) leal 96(%edi),%edi addl $96,%eax - jz L010ecb_ret -L020ecb_dec_tail: + jz L012ecb_ret +L022ecb_dec_tail: movups (%esi),%xmm2 cmpl $32,%eax - jb L023ecb_dec_one + jb L025ecb_dec_one movups 16(%esi),%xmm3 - je L024ecb_dec_two + je L026ecb_dec_two movups 32(%esi),%xmm4 cmpl $64,%eax - jb L025ecb_dec_three + jb L027ecb_dec_three movups 48(%esi),%xmm5 - je L026ecb_dec_four + je L028ecb_dec_four movups 64(%esi),%xmm6 xorps %xmm7,%xmm7 call __aesni_decrypt6 @@ -529,43 +531,51 @@ L020ecb_dec_tail: movups %xmm4,32(%edi) movups %xmm5,48(%edi) movups %xmm6,64(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L023ecb_dec_one: +L025ecb_dec_one: movups (%edx),%xmm0 movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L027dec1_loop_4: +L029dec1_loop_4: .byte 102,15,56,222,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L027dec1_loop_4 + jnz L029dec1_loop_4 .byte 102,15,56,223,209 movups %xmm2,(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L024ecb_dec_two: +L026ecb_dec_two: call __aesni_decrypt2 movups %xmm2,(%edi) movups %xmm3,16(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L025ecb_dec_three: +L027ecb_dec_three: call __aesni_decrypt3 movups %xmm2,(%edi) movups %xmm3,16(%edi) movups %xmm4,32(%edi) - jmp L010ecb_ret + jmp L012ecb_ret .align 4,0x90 -L026ecb_dec_four: +L028ecb_dec_four: call __aesni_decrypt4 movups %xmm2,(%edi) movups %xmm3,16(%edi) movups %xmm4,32(%edi) movups %xmm5,48(%edi) -L010ecb_ret: +L012ecb_ret: + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 + pxor %xmm2,%xmm2 + pxor %xmm3,%xmm3 + pxor %xmm4,%xmm4 + pxor %xmm5,%xmm5 + pxor %xmm6,%xmm6 + pxor %xmm7,%xmm7 popl %edi popl %esi popl %ebx @@ -611,7 +621,7 @@ L_aesni_ccm64_encrypt_blocks_begin: leal 32(%edx,%ecx,1),%edx subl %ecx,%ebx .byte 102,15,56,0,253 -L028ccm64_enc_outer: +L030ccm64_enc_outer: movups (%ebp),%xmm0 movl %ebx,%ecx movups (%esi),%xmm6 @@ -620,7 +630,7 @@ L028ccm64_enc_outer: xorps %xmm6,%xmm0 xorps %xmm0,%xmm3 movups 32(%ebp),%xmm0 -L029ccm64_enc2_loop: +L031ccm64_enc2_loop: .byte 102,15,56,220,209 .byte 102,15,56,220,217 movups (%edx,%ecx,1),%xmm1 @@ -628,7 +638,7 @@ L029ccm64_enc2_loop: .byte 102,15,56,220,208 .byte 102,15,56,220,216 movups -16(%edx,%ecx,1),%xmm0 - jnz L029ccm64_enc2_loop + jnz L031ccm64_enc2_loop .byte 102,15,56,220,209 .byte 102,15,56,220,217 paddq 16(%esp),%xmm7 @@ -641,10 +651,18 @@ L029ccm64_enc2_loop: movups %xmm6,(%edi) .byte 102,15,56,0,213 leal 16(%edi),%edi - jnz L028ccm64_enc_outer + jnz L030ccm64_enc_outer movl 48(%esp),%esp movl 40(%esp),%edi movups %xmm3,(%edi) + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 + pxor %xmm2,%xmm2 + pxor %xmm3,%xmm3 + pxor %xmm4,%xmm4 + pxor %xmm5,%xmm5 + pxor %xmm6,%xmm6 + pxor %xmm7,%xmm7 popl %edi popl %esi popl %ebx @@ -691,12 +709,12 @@ L_aesni_ccm64_decrypt_blocks_begin: movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L030enc1_loop_5: +L032enc1_loop_5: .byte 102,15,56,220,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L030enc1_loop_5 + jnz L032enc1_loop_5 .byte 102,15,56,221,209 shll $4,%ebx movl $16,%ecx @@ -706,16 +724,16 @@ L030enc1_loop_5: subl %ebx,%ecx leal 32(%ebp,%ebx,1),%edx movl %ecx,%ebx - jmp L031ccm64_dec_outer + jmp L033ccm64_dec_outer .align 4,0x90 -L031ccm64_dec_outer: +L033ccm64_dec_outer: xorps %xmm2,%xmm6 movdqa %xmm7,%xmm2 movups %xmm6,(%edi) leal 16(%edi),%edi .byte 102,15,56,0,213 subl $1,%eax - jz L032ccm64_dec_break + jz L034ccm64_dec_break movups (%ebp),%xmm0 movl %ebx,%ecx movups 16(%ebp),%xmm1 @@ -723,7 +741,7 @@ L031ccm64_dec_outer: xorps %xmm0,%xmm2 xorps %xmm6,%xmm3 movups 32(%ebp),%xmm0 -L033ccm64_dec2_loop: +L035ccm64_dec2_loop: .byte 102,15,56,220,209 .byte 102,15,56,220,217 movups (%edx,%ecx,1),%xmm1 @@ -731,7 +749,7 @@ L033ccm64_dec2_loop: .byte 102,15,56,220,208 .byte 102,15,56,220,216 movups -16(%edx,%ecx,1),%xmm0 - jnz L033ccm64_dec2_loop + jnz L035ccm64_dec2_loop movups (%esi),%xmm6 paddq 16(%esp),%xmm7 .byte 102,15,56,220,209 @@ -739,9 +757,9 @@ L033ccm64_dec2_loop: .byte 102,15,56,221,208 .byte 102,15,56,221,216 leal 16(%esi),%esi - jmp L031ccm64_dec_outer + jmp L033ccm64_dec_outer .align 4,0x90 -L032ccm64_dec_break: +L034ccm64_dec_break: movl 240(%ebp),%ecx movl %ebp,%edx movups (%edx),%xmm0 @@ -749,16 +767,24 @@ L032ccm64_dec_break: xorps %xmm0,%xmm6 leal 32(%edx),%edx xorps %xmm6,%xmm3 -L034enc1_loop_6: +L036enc1_loop_6: .byte 102,15,56,220,217 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L034enc1_loop_6 + jnz L036enc1_loop_6 .byte 102,15,56,221,217 movl 48(%esp),%esp movl 40(%esp),%edi movups %xmm3,(%edi) + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 + pxor %xmm2,%xmm2 + pxor %xmm3,%xmm3 + pxor %xmm4,%xmm4 + pxor %xmm5,%xmm5 + pxor %xmm6,%xmm6 + pxor %xmm7,%xmm7 popl %edi popl %esi popl %ebx @@ -783,7 +809,7 @@ L_aesni_ctr32_encrypt_blocks_begin: andl $-16,%esp movl %ebp,80(%esp) cmpl $1,%eax - je L035ctr32_one_shortcut + je L037ctr32_one_shortcut movdqu (%ebx),%xmm7 movl $202182159,(%esp) movl $134810123,4(%esp) @@ -821,7 +847,7 @@ L_aesni_ctr32_encrypt_blocks_begin: pshufd $192,%xmm0,%xmm2 pshufd $128,%xmm0,%xmm3 cmpl $6,%eax - jb L036ctr32_tail + jb L038ctr32_tail pxor %xmm6,%xmm7 shll $4,%ecx movl $16,%ebx @@ -830,9 +856,9 @@ L_aesni_ctr32_encrypt_blocks_begin: subl %ecx,%ebx leal 32(%edx,%ecx,1),%edx subl $6,%eax - jmp L037ctr32_loop6 + jmp L039ctr32_loop6 .align 4,0x90 -L037ctr32_loop6: +L039ctr32_loop6: pshufd $64,%xmm0,%xmm4 movdqa 32(%esp),%xmm0 pshufd $192,%xmm1,%xmm5 @@ -886,27 +912,27 @@ L037ctr32_loop6: leal 96(%edi),%edi pshufd $128,%xmm0,%xmm3 subl $6,%eax - jnc L037ctr32_loop6 + jnc L039ctr32_loop6 addl $6,%eax - jz L038ctr32_ret + jz L040ctr32_ret movdqu (%ebp),%xmm7 movl %ebp,%edx pxor 32(%esp),%xmm7 movl 240(%ebp),%ecx -L036ctr32_tail: +L038ctr32_tail: por %xmm7,%xmm2 cmpl $2,%eax - jb L039ctr32_one + jb L041ctr32_one pshufd $64,%xmm0,%xmm4 por %xmm7,%xmm3 - je L040ctr32_two + je L042ctr32_two pshufd $192,%xmm1,%xmm5 por %xmm7,%xmm4 cmpl $4,%eax - jb L041ctr32_three + jb L043ctr32_three pshufd $128,%xmm1,%xmm6 por %xmm7,%xmm5 - je L042ctr32_four + je L044ctr32_four por %xmm7,%xmm6 call __aesni_encrypt6 movups (%esi),%xmm1 @@ -924,29 +950,29 @@ L036ctr32_tail: movups %xmm4,32(%edi) movups %xmm5,48(%edi) movups %xmm6,64(%edi) - jmp L038ctr32_ret + jmp L040ctr32_ret .align 4,0x90 -L035ctr32_one_shortcut: +L037ctr32_one_shortcut: movups (%ebx),%xmm2 movl 240(%edx),%ecx -L039ctr32_one: +L041ctr32_one: movups (%edx),%xmm0 movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L043enc1_loop_7: +L045enc1_loop_7: .byte 102,15,56,220,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L043enc1_loop_7 + jnz L045enc1_loop_7 .byte 102,15,56,221,209 movups (%esi),%xmm6 xorps %xmm2,%xmm6 movups %xmm6,(%edi) - jmp L038ctr32_ret + jmp L040ctr32_ret .align 4,0x90 -L040ctr32_two: +L042ctr32_two: call __aesni_encrypt2 movups (%esi),%xmm5 movups 16(%esi),%xmm6 @@ -954,9 +980,9 @@ L040ctr32_two: xorps %xmm6,%xmm3 movups %xmm2,(%edi) movups %xmm3,16(%edi) - jmp L038ctr32_ret + jmp L040ctr32_ret .align 4,0x90 -L041ctr32_three: +L043ctr32_three: call __aesni_encrypt3 movups (%esi),%xmm5 movups 16(%esi),%xmm6 @@ -967,9 +993,9 @@ L041ctr32_three: xorps %xmm7,%xmm4 movups %xmm3,16(%edi) movups %xmm4,32(%edi) - jmp L038ctr32_ret + jmp L040ctr32_ret .align 4,0x90 -L042ctr32_four: +L044ctr32_four: call __aesni_encrypt4 movups (%esi),%xmm6 movups 16(%esi),%xmm7 @@ -983,7 +1009,18 @@ L042ctr32_four: xorps %xmm0,%xmm5 movups %xmm4,32(%edi) movups %xmm5,48(%edi) -L038ctr32_ret: +L040ctr32_ret: + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 + pxor %xmm2,%xmm2 + pxor %xmm3,%xmm3 + pxor %xmm4,%xmm4 + movdqa %xmm0,32(%esp) + pxor %xmm5,%xmm5 + movdqa %xmm0,48(%esp) + pxor %xmm6,%xmm6 + movdqa %xmm0,64(%esp) + pxor %xmm7,%xmm7 movl 80(%esp),%esp popl %edi popl %esi @@ -1007,12 +1044,12 @@ L_aesni_xts_encrypt_begin: movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L044enc1_loop_8: +L046enc1_loop_8: .byte 102,15,56,220,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L044enc1_loop_8 + jnz L046enc1_loop_8 .byte 102,15,56,221,209 movl 20(%esp),%esi movl 24(%esp),%edi @@ -1036,14 +1073,14 @@ L044enc1_loop_8: movl %edx,%ebp movl %ecx,%ebx subl $96,%eax - jc L045xts_enc_short + jc L047xts_enc_short shll $4,%ecx movl $16,%ebx subl %ecx,%ebx leal 32(%edx,%ecx,1),%edx - jmp L046xts_enc_loop6 + jmp L048xts_enc_loop6 .align 4,0x90 -L046xts_enc_loop6: +L048xts_enc_loop6: pshufd $19,%xmm0,%xmm2 pxor %xmm0,%xmm0 movdqa %xmm1,(%esp) @@ -1132,23 +1169,23 @@ L046xts_enc_loop6: pcmpgtd %xmm1,%xmm0 pxor %xmm2,%xmm1 subl $96,%eax - jnc L046xts_enc_loop6 + jnc L048xts_enc_loop6 movl 240(%ebp),%ecx movl %ebp,%edx movl %ecx,%ebx -L045xts_enc_short: +L047xts_enc_short: addl $96,%eax - jz L047xts_enc_done6x + jz L049xts_enc_done6x movdqa %xmm1,%xmm5 cmpl $32,%eax - jb L048xts_enc_one + jb L050xts_enc_one pshufd $19,%xmm0,%xmm2 pxor %xmm0,%xmm0 paddq %xmm1,%xmm1 pand %xmm3,%xmm2 pcmpgtd %xmm1,%xmm0 pxor %xmm2,%xmm1 - je L049xts_enc_two + je L051xts_enc_two pshufd $19,%xmm0,%xmm2 pxor %xmm0,%xmm0 movdqa %xmm1,%xmm6 @@ -1157,7 +1194,7 @@ L045xts_enc_short: pcmpgtd %xmm1,%xmm0 pxor %xmm2,%xmm1 cmpl $64,%eax - jb L050xts_enc_three + jb L052xts_enc_three pshufd $19,%xmm0,%xmm2 pxor %xmm0,%xmm0 movdqa %xmm1,%xmm7 @@ -1167,7 +1204,7 @@ L045xts_enc_short: pxor %xmm2,%xmm1 movdqa %xmm5,(%esp) movdqa %xmm6,16(%esp) - je L051xts_enc_four + je L053xts_enc_four movdqa %xmm7,32(%esp) pshufd $19,%xmm0,%xmm7 movdqa %xmm1,48(%esp) @@ -1199,9 +1236,9 @@ L045xts_enc_short: movups %xmm5,48(%edi) movups %xmm6,64(%edi) leal 80(%edi),%edi - jmp L052xts_enc_done + jmp L054xts_enc_done .align 4,0x90 -L048xts_enc_one: +L050xts_enc_one: movups (%esi),%xmm2 leal 16(%esi),%esi xorps %xmm5,%xmm2 @@ -1209,20 +1246,20 @@ L048xts_enc_one: movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L053enc1_loop_9: +L055enc1_loop_9: .byte 102,15,56,220,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L053enc1_loop_9 + jnz L055enc1_loop_9 .byte 102,15,56,221,209 xorps %xmm5,%xmm2 movups %xmm2,(%edi) leal 16(%edi),%edi movdqa %xmm5,%xmm1 - jmp L052xts_enc_done + jmp L054xts_enc_done .align 4,0x90 -L049xts_enc_two: +L051xts_enc_two: movaps %xmm1,%xmm6 movups (%esi),%xmm2 movups 16(%esi),%xmm3 @@ -1236,9 +1273,9 @@ L049xts_enc_two: movups %xmm3,16(%edi) leal 32(%edi),%edi movdqa %xmm6,%xmm1 - jmp L052xts_enc_done + jmp L054xts_enc_done .align 4,0x90 -L050xts_enc_three: +L052xts_enc_three: movaps %xmm1,%xmm7 movups (%esi),%xmm2 movups 16(%esi),%xmm3 @@ -1256,9 +1293,9 @@ L050xts_enc_three: movups %xmm4,32(%edi) leal 48(%edi),%edi movdqa %xmm7,%xmm1 - jmp L052xts_enc_done + jmp L054xts_enc_done .align 4,0x90 -L051xts_enc_four: +L053xts_enc_four: movaps %xmm1,%xmm6 movups (%esi),%xmm2 movups 16(%esi),%xmm3 @@ -1280,28 +1317,28 @@ L051xts_enc_four: movups %xmm5,48(%edi) leal 64(%edi),%edi movdqa %xmm6,%xmm1 - jmp L052xts_enc_done + jmp L054xts_enc_done .align 4,0x90 -L047xts_enc_done6x: +L049xts_enc_done6x: movl 112(%esp),%eax andl $15,%eax - jz L054xts_enc_ret + jz L056xts_enc_ret movdqa %xmm1,%xmm5 movl %eax,112(%esp) - jmp L055xts_enc_steal + jmp L057xts_enc_steal .align 4,0x90 -L052xts_enc_done: +L054xts_enc_done: movl 112(%esp),%eax pxor %xmm0,%xmm0 andl $15,%eax - jz L054xts_enc_ret + jz L056xts_enc_ret pcmpgtd %xmm1,%xmm0 movl %eax,112(%esp) pshufd $19,%xmm0,%xmm5 paddq %xmm1,%xmm1 pand 96(%esp),%xmm5 pxor %xmm1,%xmm5 -L055xts_enc_steal: +L057xts_enc_steal: movzbl (%esi),%ecx movzbl -16(%edi),%edx leal 1(%esi),%esi @@ -1309,7 +1346,7 @@ L055xts_enc_steal: movb %dl,(%edi) leal 1(%edi),%edi subl $1,%eax - jnz L055xts_enc_steal + jnz L057xts_enc_steal subl 112(%esp),%edi movl %ebp,%edx movl %ebx,%ecx @@ -1319,16 +1356,30 @@ L055xts_enc_steal: movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L056enc1_loop_10: +L058enc1_loop_10: .byte 102,15,56,220,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L056enc1_loop_10 + jnz L058enc1_loop_10 .byte 102,15,56,221,209 xorps %xmm5,%xmm2 movups %xmm2,-16(%edi) -L054xts_enc_ret: +L056xts_enc_ret: + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 + pxor %xmm2,%xmm2 + movdqa %xmm0,(%esp) + pxor %xmm3,%xmm3 + movdqa %xmm0,16(%esp) + pxor %xmm4,%xmm4 + movdqa %xmm0,32(%esp) + pxor %xmm5,%xmm5 + movdqa %xmm0,48(%esp) + pxor %xmm6,%xmm6 + movdqa %xmm0,64(%esp) + pxor %xmm7,%xmm7 + movdqa %xmm0,80(%esp) movl 116(%esp),%esp popl %edi popl %esi @@ -1352,12 +1403,12 @@ L_aesni_xts_decrypt_begin: movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L057enc1_loop_11: +L059enc1_loop_11: .byte 102,15,56,220,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L057enc1_loop_11 + jnz L059enc1_loop_11 .byte 102,15,56,221,209 movl 20(%esp),%esi movl 24(%esp),%edi @@ -1386,14 +1437,14 @@ L057enc1_loop_11: pcmpgtd %xmm1,%xmm0 andl $-16,%eax subl $96,%eax - jc L058xts_dec_short + jc L060xts_dec_short shll $4,%ecx movl $16,%ebx subl %ecx,%ebx leal 32(%edx,%ecx,1),%edx - jmp L059xts_dec_loop6 + jmp L061xts_dec_loop6 .align 4,0x90 -L059xts_dec_loop6: +L061xts_dec_loop6: pshufd $19,%xmm0,%xmm2 pxor %xmm0,%xmm0 movdqa %xmm1,(%esp) @@ -1482,23 +1533,23 @@ L059xts_dec_loop6: pcmpgtd %xmm1,%xmm0 pxor %xmm2,%xmm1 subl $96,%eax - jnc L059xts_dec_loop6 + jnc L061xts_dec_loop6 movl 240(%ebp),%ecx movl %ebp,%edx movl %ecx,%ebx -L058xts_dec_short: +L060xts_dec_short: addl $96,%eax - jz L060xts_dec_done6x + jz L062xts_dec_done6x movdqa %xmm1,%xmm5 cmpl $32,%eax - jb L061xts_dec_one + jb L063xts_dec_one pshufd $19,%xmm0,%xmm2 pxor %xmm0,%xmm0 paddq %xmm1,%xmm1 pand %xmm3,%xmm2 pcmpgtd %xmm1,%xmm0 pxor %xmm2,%xmm1 - je L062xts_dec_two + je L064xts_dec_two pshufd $19,%xmm0,%xmm2 pxor %xmm0,%xmm0 movdqa %xmm1,%xmm6 @@ -1507,7 +1558,7 @@ L058xts_dec_short: pcmpgtd %xmm1,%xmm0 pxor %xmm2,%xmm1 cmpl $64,%eax - jb L063xts_dec_three + jb L065xts_dec_three pshufd $19,%xmm0,%xmm2 pxor %xmm0,%xmm0 movdqa %xmm1,%xmm7 @@ -1517,7 +1568,7 @@ L058xts_dec_short: pxor %xmm2,%xmm1 movdqa %xmm5,(%esp) movdqa %xmm6,16(%esp) - je L064xts_dec_four + je L066xts_dec_four movdqa %xmm7,32(%esp) pshufd $19,%xmm0,%xmm7 movdqa %xmm1,48(%esp) @@ -1549,9 +1600,9 @@ L058xts_dec_short: movups %xmm5,48(%edi) movups %xmm6,64(%edi) leal 80(%edi),%edi - jmp L065xts_dec_done + jmp L067xts_dec_done .align 4,0x90 -L061xts_dec_one: +L063xts_dec_one: movups (%esi),%xmm2 leal 16(%esi),%esi xorps %xmm5,%xmm2 @@ -1559,20 +1610,20 @@ L061xts_dec_one: movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L066dec1_loop_12: +L068dec1_loop_12: .byte 102,15,56,222,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L066dec1_loop_12 + jnz L068dec1_loop_12 .byte 102,15,56,223,209 xorps %xmm5,%xmm2 movups %xmm2,(%edi) leal 16(%edi),%edi movdqa %xmm5,%xmm1 - jmp L065xts_dec_done + jmp L067xts_dec_done .align 4,0x90 -L062xts_dec_two: +L064xts_dec_two: movaps %xmm1,%xmm6 movups (%esi),%xmm2 movups 16(%esi),%xmm3 @@ -1586,9 +1637,9 @@ L062xts_dec_two: movups %xmm3,16(%edi) leal 32(%edi),%edi movdqa %xmm6,%xmm1 - jmp L065xts_dec_done + jmp L067xts_dec_done .align 4,0x90 -L063xts_dec_three: +L065xts_dec_three: movaps %xmm1,%xmm7 movups (%esi),%xmm2 movups 16(%esi),%xmm3 @@ -1606,9 +1657,9 @@ L063xts_dec_three: movups %xmm4,32(%edi) leal 48(%edi),%edi movdqa %xmm7,%xmm1 - jmp L065xts_dec_done + jmp L067xts_dec_done .align 4,0x90 -L064xts_dec_four: +L066xts_dec_four: movaps %xmm1,%xmm6 movups (%esi),%xmm2 movups 16(%esi),%xmm3 @@ -1630,20 +1681,20 @@ L064xts_dec_four: movups %xmm5,48(%edi) leal 64(%edi),%edi movdqa %xmm6,%xmm1 - jmp L065xts_dec_done + jmp L067xts_dec_done .align 4,0x90 -L060xts_dec_done6x: +L062xts_dec_done6x: movl 112(%esp),%eax andl $15,%eax - jz L067xts_dec_ret + jz L069xts_dec_ret movl %eax,112(%esp) - jmp L068xts_dec_only_one_more + jmp L070xts_dec_only_one_more .align 4,0x90 -L065xts_dec_done: +L067xts_dec_done: movl 112(%esp),%eax pxor %xmm0,%xmm0 andl $15,%eax - jz L067xts_dec_ret + jz L069xts_dec_ret pcmpgtd %xmm1,%xmm0 movl %eax,112(%esp) pshufd $19,%xmm0,%xmm2 @@ -1653,7 +1704,7 @@ L065xts_dec_done: pand %xmm3,%xmm2 pcmpgtd %xmm1,%xmm0 pxor %xmm2,%xmm1 -L068xts_dec_only_one_more: +L070xts_dec_only_one_more: pshufd $19,%xmm0,%xmm5 movdqa %xmm1,%xmm6 paddq %xmm1,%xmm1 @@ -1667,16 +1718,16 @@ L068xts_dec_only_one_more: movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L069dec1_loop_13: +L071dec1_loop_13: .byte 102,15,56,222,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L069dec1_loop_13 + jnz L071dec1_loop_13 .byte 102,15,56,223,209 xorps %xmm5,%xmm2 movups %xmm2,(%edi) -L070xts_dec_steal: +L072xts_dec_steal: movzbl 16(%esi),%ecx movzbl (%edi),%edx leal 1(%esi),%esi @@ -1684,7 +1735,7 @@ L070xts_dec_steal: movb %dl,16(%edi) leal 1(%edi),%edi subl $1,%eax - jnz L070xts_dec_steal + jnz L072xts_dec_steal subl 112(%esp),%edi movl %ebp,%edx movl %ebx,%ecx @@ -1694,16 +1745,30 @@ L070xts_dec_steal: movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L071dec1_loop_14: +L073dec1_loop_14: .byte 102,15,56,222,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L071dec1_loop_14 + jnz L073dec1_loop_14 .byte 102,15,56,223,209 xorps %xmm6,%xmm2 movups %xmm2,(%edi) -L067xts_dec_ret: +L069xts_dec_ret: + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 + pxor %xmm2,%xmm2 + movdqa %xmm0,(%esp) + pxor %xmm3,%xmm3 + movdqa %xmm0,16(%esp) + pxor %xmm4,%xmm4 + movdqa %xmm0,32(%esp) + pxor %xmm5,%xmm5 + movdqa %xmm0,48(%esp) + pxor %xmm6,%xmm6 + movdqa %xmm0,64(%esp) + pxor %xmm7,%xmm7 + movdqa %xmm0,80(%esp) movl 116(%esp),%esp popl %edi popl %esi @@ -1728,7 +1793,7 @@ L_aesni_cbc_encrypt_begin: movl 32(%esp),%edx movl 36(%esp),%ebp testl %eax,%eax - jz L072cbc_abort + jz L074cbc_abort cmpl $0,40(%esp) xchgl %esp,%ebx movups (%ebp),%xmm7 @@ -1736,14 +1801,14 @@ L_aesni_cbc_encrypt_begin: movl %edx,%ebp movl %ebx,16(%esp) movl %ecx,%ebx - je L073cbc_decrypt + je L075cbc_decrypt movaps %xmm7,%xmm2 cmpl $16,%eax - jb L074cbc_enc_tail + jb L076cbc_enc_tail subl $16,%eax - jmp L075cbc_enc_loop + jmp L077cbc_enc_loop .align 4,0x90 -L075cbc_enc_loop: +L077cbc_enc_loop: movups (%esi),%xmm7 leal 16(%esi),%esi movups (%edx),%xmm0 @@ -1751,24 +1816,25 @@ L075cbc_enc_loop: xorps %xmm0,%xmm7 leal 32(%edx),%edx xorps %xmm7,%xmm2 -L076enc1_loop_15: +L078enc1_loop_15: .byte 102,15,56,220,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L076enc1_loop_15 + jnz L078enc1_loop_15 .byte 102,15,56,221,209 movl %ebx,%ecx movl %ebp,%edx movups %xmm2,(%edi) leal 16(%edi),%edi subl $16,%eax - jnc L075cbc_enc_loop + jnc L077cbc_enc_loop addl $16,%eax - jnz L074cbc_enc_tail + jnz L076cbc_enc_tail movaps %xmm2,%xmm7 - jmp L077cbc_ret -L074cbc_enc_tail: + pxor %xmm2,%xmm2 + jmp L079cbc_ret +L076cbc_enc_tail: movl %eax,%ecx .long 2767451785 movl $16,%ecx @@ -1779,20 +1845,20 @@ L074cbc_enc_tail: movl %ebx,%ecx movl %edi,%esi movl %ebp,%edx - jmp L075cbc_enc_loop + jmp L077cbc_enc_loop .align 4,0x90 -L073cbc_decrypt: +L075cbc_decrypt: cmpl $80,%eax - jbe L078cbc_dec_tail + jbe L080cbc_dec_tail movaps %xmm7,(%esp) subl $80,%eax - jmp L079cbc_dec_loop6_enter + jmp L081cbc_dec_loop6_enter .align 4,0x90 -L080cbc_dec_loop6: +L082cbc_dec_loop6: movaps %xmm0,(%esp) movups %xmm7,(%edi) leal 16(%edi),%edi -L079cbc_dec_loop6_enter: +L081cbc_dec_loop6_enter: movdqu (%esi),%xmm2 movdqu 16(%esi),%xmm3 movdqu 32(%esi),%xmm4 @@ -1822,28 +1888,28 @@ L079cbc_dec_loop6_enter: movups %xmm6,64(%edi) leal 80(%edi),%edi subl $96,%eax - ja L080cbc_dec_loop6 + ja L082cbc_dec_loop6 movaps %xmm7,%xmm2 movaps %xmm0,%xmm7 addl $80,%eax - jle L081cbc_dec_tail_collected + jle L083cbc_dec_clear_tail_collected movups %xmm2,(%edi) leal 16(%edi),%edi -L078cbc_dec_tail: +L080cbc_dec_tail: movups (%esi),%xmm2 movaps %xmm2,%xmm6 cmpl $16,%eax - jbe L082cbc_dec_one + jbe L084cbc_dec_one movups 16(%esi),%xmm3 movaps %xmm3,%xmm5 cmpl $32,%eax - jbe L083cbc_dec_two + jbe L085cbc_dec_two movups 32(%esi),%xmm4 cmpl $48,%eax - jbe L084cbc_dec_three + jbe L086cbc_dec_three movups 48(%esi),%xmm5 cmpl $64,%eax - jbe L085cbc_dec_four + jbe L087cbc_dec_four movups 64(%esi),%xmm6 movaps %xmm7,(%esp) movups (%esi),%xmm2 @@ -1861,55 +1927,62 @@ L078cbc_dec_tail: xorps %xmm0,%xmm6 movups %xmm2,(%edi) movups %xmm3,16(%edi) + pxor %xmm3,%xmm3 movups %xmm4,32(%edi) + pxor %xmm4,%xmm4 movups %xmm5,48(%edi) + pxor %xmm5,%xmm5 leal 64(%edi),%edi movaps %xmm6,%xmm2 + pxor %xmm6,%xmm6 subl $80,%eax - jmp L081cbc_dec_tail_collected + jmp L088cbc_dec_tail_collected .align 4,0x90 -L082cbc_dec_one: +L084cbc_dec_one: movups (%edx),%xmm0 movups 16(%edx),%xmm1 leal 32(%edx),%edx xorps %xmm0,%xmm2 -L086dec1_loop_16: +L089dec1_loop_16: .byte 102,15,56,222,209 decl %ecx movups (%edx),%xmm1 leal 16(%edx),%edx - jnz L086dec1_loop_16 + jnz L089dec1_loop_16 .byte 102,15,56,223,209 xorps %xmm7,%xmm2 movaps %xmm6,%xmm7 subl $16,%eax - jmp L081cbc_dec_tail_collected + jmp L088cbc_dec_tail_collected .align 4,0x90 -L083cbc_dec_two: +L085cbc_dec_two: call __aesni_decrypt2 xorps %xmm7,%xmm2 xorps %xmm6,%xmm3 movups %xmm2,(%edi) movaps %xmm3,%xmm2 + pxor %xmm3,%xmm3 leal 16(%edi),%edi movaps %xmm5,%xmm7 subl $32,%eax - jmp L081cbc_dec_tail_collected + jmp L088cbc_dec_tail_collected .align 4,0x90 -L084cbc_dec_three: +L086cbc_dec_three: call __aesni_decrypt3 xorps %xmm7,%xmm2 xorps %xmm6,%xmm3 xorps %xmm5,%xmm4 movups %xmm2,(%edi) movaps %xmm4,%xmm2 + pxor %xmm4,%xmm4 movups %xmm3,16(%edi) + pxor %xmm3,%xmm3 leal 32(%edi),%edi movups 32(%esi),%xmm7 subl $48,%eax - jmp L081cbc_dec_tail_collected + jmp L088cbc_dec_tail_collected .align 4,0x90 -L085cbc_dec_four: +L087cbc_dec_four: call __aesni_decrypt4 movups 16(%esi),%xmm1 movups 32(%esi),%xmm0 @@ -1919,28 +1992,44 @@ L085cbc_dec_four: movups %xmm2,(%edi) xorps %xmm1,%xmm4 movups %xmm3,16(%edi) + pxor %xmm3,%xmm3 xorps %xmm0,%xmm5 movups %xmm4,32(%edi) + pxor %xmm4,%xmm4 leal 48(%edi),%edi movaps %xmm5,%xmm2 + pxor %xmm5,%xmm5 subl $64,%eax -L081cbc_dec_tail_collected: + jmp L088cbc_dec_tail_collected +.align 4,0x90 +L083cbc_dec_clear_tail_collected: + pxor %xmm3,%xmm3 + pxor %xmm4,%xmm4 + pxor %xmm5,%xmm5 + pxor %xmm6,%xmm6 +L088cbc_dec_tail_collected: andl $15,%eax - jnz L087cbc_dec_tail_partial + jnz L090cbc_dec_tail_partial movups %xmm2,(%edi) - jmp L077cbc_ret + pxor %xmm0,%xmm0 + jmp L079cbc_ret .align 4,0x90 -L087cbc_dec_tail_partial: +L090cbc_dec_tail_partial: movaps %xmm2,(%esp) + pxor %xmm0,%xmm0 movl $16,%ecx movl %esp,%esi subl %eax,%ecx .long 2767451785 -L077cbc_ret: + movdqa %xmm2,(%esp) +L079cbc_ret: movl 16(%esp),%esp movl 36(%esp),%ebp + pxor %xmm2,%xmm2 + pxor %xmm1,%xmm1 movups %xmm7,(%ebp) -L072cbc_abort: + pxor %xmm7,%xmm7 +L074cbc_abort: popl %edi popl %esi popl %ebx @@ -1949,52 +2038,62 @@ L072cbc_abort: .private_extern __aesni_set_encrypt_key .align 4 __aesni_set_encrypt_key: + pushl %ebp + pushl %ebx testl %eax,%eax - jz L088bad_pointer + jz L091bad_pointer testl %edx,%edx - jz L088bad_pointer + jz L091bad_pointer + call L092pic +L092pic: + popl %ebx + leal Lkey_const-L092pic(%ebx),%ebx + movl L_OPENSSL_ia32cap_P$non_lazy_ptr-Lkey_const(%ebx),%ebp movups (%eax),%xmm0 xorps %xmm4,%xmm4 + movl 4(%ebp),%ebp leal 16(%edx),%edx + andl $268437504,%ebp cmpl $256,%ecx - je L08914rounds + je L09314rounds cmpl $192,%ecx - je L09012rounds + je L09412rounds cmpl $128,%ecx - jne L091bad_keybits + jne L095bad_keybits .align 4,0x90 -L09210rounds: +L09610rounds: + cmpl $268435456,%ebp + je L09710rounds_alt movl $9,%ecx movups %xmm0,-16(%edx) .byte 102,15,58,223,200,1 - call L093key_128_cold + call L098key_128_cold .byte 102,15,58,223,200,2 - call L094key_128 + call L099key_128 .byte 102,15,58,223,200,4 - call L094key_128 + call L099key_128 .byte 102,15,58,223,200,8 - call L094key_128 + call L099key_128 .byte 102,15,58,223,200,16 - call L094key_128 + call L099key_128 .byte 102,15,58,223,200,32 - call L094key_128 + call L099key_128 .byte 102,15,58,223,200,64 - call L094key_128 + call L099key_128 .byte 102,15,58,223,200,128 - call L094key_128 + call L099key_128 .byte 102,15,58,223,200,27 - call L094key_128 + call L099key_128 .byte 102,15,58,223,200,54 - call L094key_128 + call L099key_128 movups %xmm0,(%edx) movl %ecx,80(%edx) - xorl %eax,%eax - ret + jmp L100good_key .align 4,0x90 -L094key_128: +L099key_128: movups %xmm0,(%edx) leal 16(%edx),%edx -L093key_128_cold: +L098key_128_cold: shufps $16,%xmm0,%xmm4 xorps %xmm4,%xmm0 shufps $140,%xmm0,%xmm4 @@ -2003,38 +2102,91 @@ L093key_128_cold: xorps %xmm1,%xmm0 ret .align 4,0x90 -L09012rounds: +L09710rounds_alt: + movdqa (%ebx),%xmm5 + movl $8,%ecx + movdqa 32(%ebx),%xmm4 + movdqa %xmm0,%xmm2 + movdqu %xmm0,-16(%edx) +L101loop_key128: +.byte 102,15,56,0,197 +.byte 102,15,56,221,196 + pslld $1,%xmm4 + leal 16(%edx),%edx + movdqa %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm3,%xmm2 + pxor %xmm2,%xmm0 + movdqu %xmm0,-16(%edx) + movdqa %xmm0,%xmm2 + decl %ecx + jnz L101loop_key128 + movdqa 48(%ebx),%xmm4 +.byte 102,15,56,0,197 +.byte 102,15,56,221,196 + pslld $1,%xmm4 + movdqa %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm3,%xmm2 + pxor %xmm2,%xmm0 + movdqu %xmm0,(%edx) + movdqa %xmm0,%xmm2 +.byte 102,15,56,0,197 +.byte 102,15,56,221,196 + movdqa %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm2,%xmm3 + pslldq $4,%xmm2 + pxor %xmm3,%xmm2 + pxor %xmm2,%xmm0 + movdqu %xmm0,16(%edx) + movl $9,%ecx + movl %ecx,96(%edx) + jmp L100good_key +.align 4,0x90 +L09412rounds: movq 16(%eax),%xmm2 + cmpl $268435456,%ebp + je L10212rounds_alt movl $11,%ecx movups %xmm0,-16(%edx) .byte 102,15,58,223,202,1 - call L095key_192a_cold + call L103key_192a_cold .byte 102,15,58,223,202,2 - call L096key_192b + call L104key_192b .byte 102,15,58,223,202,4 - call L097key_192a + call L105key_192a .byte 102,15,58,223,202,8 - call L096key_192b + call L104key_192b .byte 102,15,58,223,202,16 - call L097key_192a + call L105key_192a .byte 102,15,58,223,202,32 - call L096key_192b + call L104key_192b .byte 102,15,58,223,202,64 - call L097key_192a + call L105key_192a .byte 102,15,58,223,202,128 - call L096key_192b + call L104key_192b movups %xmm0,(%edx) movl %ecx,48(%edx) - xorl %eax,%eax - ret + jmp L100good_key .align 4,0x90 -L097key_192a: +L105key_192a: movups %xmm0,(%edx) leal 16(%edx),%edx .align 4,0x90 -L095key_192a_cold: +L103key_192a_cold: movaps %xmm2,%xmm5 -L098key_192b_warm: +L106key_192b_warm: shufps $16,%xmm0,%xmm4 movdqa %xmm2,%xmm3 xorps %xmm4,%xmm0 @@ -2048,56 +2200,90 @@ L098key_192b_warm: pxor %xmm3,%xmm2 ret .align 4,0x90 -L096key_192b: +L104key_192b: movaps %xmm0,%xmm3 shufps $68,%xmm0,%xmm5 movups %xmm5,(%edx) shufps $78,%xmm2,%xmm3 movups %xmm3,16(%edx) leal 32(%edx),%edx - jmp L098key_192b_warm + jmp L106key_192b_warm .align 4,0x90 -L08914rounds: +L10212rounds_alt: + movdqa 16(%ebx),%xmm5 + movdqa 32(%ebx),%xmm4 + movl $8,%ecx + movdqu %xmm0,-16(%edx) +L107loop_key192: + movq %xmm2,(%edx) + movdqa %xmm2,%xmm1 +.byte 102,15,56,0,213 +.byte 102,15,56,221,212 + pslld $1,%xmm4 + leal 24(%edx),%edx + movdqa %xmm0,%xmm3 + pslldq $4,%xmm0 + pxor %xmm0,%xmm3 + pslldq $4,%xmm0 + pxor %xmm0,%xmm3 + pslldq $4,%xmm0 + pxor %xmm3,%xmm0 + pshufd $255,%xmm0,%xmm3 + pxor %xmm1,%xmm3 + pslldq $4,%xmm1 + pxor %xmm1,%xmm3 + pxor %xmm2,%xmm0 + pxor %xmm3,%xmm2 + movdqu %xmm0,-16(%edx) + decl %ecx + jnz L107loop_key192 + movl $11,%ecx + movl %ecx,32(%edx) + jmp L100good_key +.align 4,0x90 +L09314rounds: movups 16(%eax),%xmm2 - movl $13,%ecx leal 16(%edx),%edx + cmpl $268435456,%ebp + je L10814rounds_alt + movl $13,%ecx movups %xmm0,-32(%edx) movups %xmm2,-16(%edx) .byte 102,15,58,223,202,1 - call L099key_256a_cold + call L109key_256a_cold .byte 102,15,58,223,200,1 - call L100key_256b + call L110key_256b .byte 102,15,58,223,202,2 - call L101key_256a + call L111key_256a .byte 102,15,58,223,200,2 - call L100key_256b + call L110key_256b .byte 102,15,58,223,202,4 - call L101key_256a + call L111key_256a .byte 102,15,58,223,200,4 - call L100key_256b + call L110key_256b .byte 102,15,58,223,202,8 - call L101key_256a + call L111key_256a .byte 102,15,58,223,200,8 - call L100key_256b + call L110key_256b .byte 102,15,58,223,202,16 - call L101key_256a + call L111key_256a .byte 102,15,58,223,200,16 - call L100key_256b + call L110key_256b .byte 102,15,58,223,202,32 - call L101key_256a + call L111key_256a .byte 102,15,58,223,200,32 - call L100key_256b + call L110key_256b .byte 102,15,58,223,202,64 - call L101key_256a + call L111key_256a movups %xmm0,(%edx) movl %ecx,16(%edx) xorl %eax,%eax - ret + jmp L100good_key .align 4,0x90 -L101key_256a: +L111key_256a: movups %xmm2,(%edx) leal 16(%edx),%edx -L099key_256a_cold: +L109key_256a_cold: shufps $16,%xmm0,%xmm4 xorps %xmm4,%xmm0 shufps $140,%xmm0,%xmm4 @@ -2106,7 +2292,7 @@ L099key_256a_cold: xorps %xmm1,%xmm0 ret .align 4,0x90 -L100key_256b: +L110key_256b: movups %xmm0,(%edx) leal 16(%edx),%edx shufps $16,%xmm2,%xmm4 @@ -2116,13 +2302,70 @@ L100key_256b: shufps $170,%xmm1,%xmm1 xorps %xmm1,%xmm2 ret +.align 4,0x90 +L10814rounds_alt: + movdqa (%ebx),%xmm5 + movdqa 32(%ebx),%xmm4 + movl $7,%ecx + movdqu %xmm0,-32(%edx) + movdqa %xmm2,%xmm1 + movdqu %xmm2,-16(%edx) +L112loop_key256: +.byte 102,15,56,0,213 +.byte 102,15,56,221,212 + movdqa %xmm0,%xmm3 + pslldq $4,%xmm0 + pxor %xmm0,%xmm3 + pslldq $4,%xmm0 + pxor %xmm0,%xmm3 + pslldq $4,%xmm0 + pxor %xmm3,%xmm0 + pslld $1,%xmm4 + pxor %xmm2,%xmm0 + movdqu %xmm0,(%edx) + decl %ecx + jz L113done_key256 + pshufd $255,%xmm0,%xmm2 + pxor %xmm3,%xmm3 +.byte 102,15,56,221,211 + movdqa %xmm1,%xmm3 + pslldq $4,%xmm1 + pxor %xmm1,%xmm3 + pslldq $4,%xmm1 + pxor %xmm1,%xmm3 + pslldq $4,%xmm1 + pxor %xmm3,%xmm1 + pxor %xmm1,%xmm2 + movdqu %xmm2,16(%edx) + leal 32(%edx),%edx + movdqa %xmm2,%xmm1 + jmp L112loop_key256 +L113done_key256: + movl $13,%ecx + movl %ecx,16(%edx) +L100good_key: + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 + pxor %xmm2,%xmm2 + pxor %xmm3,%xmm3 + pxor %xmm4,%xmm4 + pxor %xmm5,%xmm5 + xorl %eax,%eax + popl %ebx + popl %ebp + ret .align 2,0x90 -L088bad_pointer: +L091bad_pointer: movl $-1,%eax + popl %ebx + popl %ebp ret .align 2,0x90 -L091bad_keybits: +L095bad_keybits: + pxor %xmm0,%xmm0 movl $-2,%eax + popl %ebx + popl %ebp ret .globl _aesni_set_encrypt_key .private_extern _aesni_set_encrypt_key @@ -2146,7 +2389,7 @@ L_aesni_set_decrypt_key_begin: movl 12(%esp),%edx shll $4,%ecx testl %eax,%eax - jnz L102dec_key_ret + jnz L114dec_key_ret leal 16(%edx,%ecx,1),%eax movups (%edx),%xmm0 movups (%eax),%xmm1 @@ -2154,7 +2397,7 @@ L_aesni_set_decrypt_key_begin: movups %xmm1,(%edx) leal 16(%edx),%edx leal -16(%eax),%eax -L103dec_key_inverse: +L115dec_key_inverse: movups (%edx),%xmm0 movups (%eax),%xmm1 .byte 102,15,56,219,192 @@ -2164,15 +2407,27 @@ L103dec_key_inverse: movups %xmm0,16(%eax) movups %xmm1,-16(%edx) cmpl %edx,%eax - ja L103dec_key_inverse + ja L115dec_key_inverse movups (%edx),%xmm0 .byte 102,15,56,219,192 movups %xmm0,(%edx) + pxor %xmm0,%xmm0 + pxor %xmm1,%xmm1 xorl %eax,%eax -L102dec_key_ret: +L114dec_key_ret: ret +.align 6,0x90 +Lkey_const: +.long 202313229,202313229,202313229,202313229 +.long 67569157,67569157,67569157,67569157 +.long 1,1,1,1 +.long 27,27,27,27 .byte 65,69,83,32,102,111,114,32,73,110,116,101,108,32,65,69 .byte 83,45,78,73,44,32,67,82,89,80,84,79,71,65,77,83 .byte 32,98,121,32,60,97,112,112,114,111,64,111,112,101,110,115 .byte 115,108,46,111,114,103,62,0 +.section __IMPORT,__pointers,non_lazy_symbol_pointers +L_OPENSSL_ia32cap_P$non_lazy_ptr: +.indirect_symbol _OPENSSL_ia32cap_P +.long 0 #endif |