HashKey computation only needs to happen once per scatter/gather operation, save it between calls in gcm_context struct instead of on the stack. Since the asm no longer stores anything on the stack, we can use %rsp directly, and clean up the frame save/restore macros a bit. Hashkeys actually only need to be calculated once per key and could be moved to when set_key is called, however, the current glue code falls back to generic aes code if fpu is disabled. Signed-off-by: Dave Watson <davejwatson@xxxxxx> --- arch/x86/crypto/aesni-intel_asm.S | 205 ++++++++++++++++++++------------------ 1 file changed, 106 insertions(+), 99 deletions(-) diff --git a/arch/x86/crypto/aesni-intel_asm.S b/arch/x86/crypto/aesni-intel_asm.S index 37b1cee..3ada06b 100644 --- a/arch/x86/crypto/aesni-intel_asm.S +++ b/arch/x86/crypto/aesni-intel_asm.S @@ -93,23 +93,6 @@ ALL_F: .octa 0xffffffffffffffffffffffffffffffff #define STACK_OFFSET 8*3 -#define HashKey 16*0 // store HashKey <<1 mod poly here -#define HashKey_2 16*1 // store HashKey^2 <<1 mod poly here -#define HashKey_3 16*2 // store HashKey^3 <<1 mod poly here -#define HashKey_4 16*3 // store HashKey^4 <<1 mod poly here -#define HashKey_k 16*4 // store XOR of High 64 bits and Low 64 - // bits of HashKey <<1 mod poly here - //(for Karatsuba purposes) -#define HashKey_2_k 16*5 // store XOR of High 64 bits and Low 64 - // bits of HashKey^2 <<1 mod poly here - // (for Karatsuba purposes) -#define HashKey_3_k 16*6 // store XOR of High 64 bits and Low 64 - // bits of HashKey^3 <<1 mod poly here - // (for Karatsuba purposes) -#define HashKey_4_k 16*7 // store XOR of High 64 bits and Low 64 - // bits of HashKey^4 <<1 mod poly here - // (for Karatsuba purposes) -#define VARIABLE_OFFSET 16*8 #define AadHash 16*0 #define AadLen 16*1 @@ -118,6 +101,22 @@ ALL_F: .octa 0xffffffffffffffffffffffffffffffff #define OrigIV 16*3 #define CurCount 16*4 #define PBlockLen 16*5 +#define HashKey 16*6 // store HashKey <<1 mod poly here +#define HashKey_2 16*7 // store HashKey^2 <<1 mod poly here +#define HashKey_3 16*8 // store HashKey^3 <<1 mod poly here +#define HashKey_4 16*9 // store HashKey^4 <<1 mod poly here +#define HashKey_k 16*10 // store XOR of High 64 bits and Low 64 + // bits of HashKey <<1 mod poly here + //(for Karatsuba purposes) +#define HashKey_2_k 16*11 // store XOR of High 64 bits and Low 64 + // bits of HashKey^2 <<1 mod poly here + // (for Karatsuba purposes) +#define HashKey_3_k 16*12 // store XOR of High 64 bits and Low 64 + // bits of HashKey^3 <<1 mod poly here + // (for Karatsuba purposes) +#define HashKey_4_k 16*13 // store XOR of High 64 bits and Low 64 + // bits of HashKey^4 <<1 mod poly here + // (for Karatsuba purposes) #define arg1 rdi #define arg2 rsi @@ -125,11 +124,11 @@ ALL_F: .octa 0xffffffffffffffffffffffffffffffff #define arg4 rcx #define arg5 r8 #define arg6 r9 -#define arg7 STACK_OFFSET+8(%r14) -#define arg8 STACK_OFFSET+16(%r14) -#define arg9 STACK_OFFSET+24(%r14) -#define arg10 STACK_OFFSET+32(%r14) -#define arg11 STACK_OFFSET+40(%r14) +#define arg7 STACK_OFFSET+8(%rsp) +#define arg8 STACK_OFFSET+16(%rsp) +#define arg9 STACK_OFFSET+24(%rsp) +#define arg10 STACK_OFFSET+32(%rsp) +#define arg11 STACK_OFFSET+40(%rsp) #define keysize 2*15*16(%arg1) #endif @@ -183,28 +182,79 @@ ALL_F: .octa 0xffffffffffffffffffffffffffffffff push %r12 push %r13 push %r14 - mov %rsp, %r14 # # states of %xmm registers %xmm6:%xmm15 not saved # all %xmm registers are clobbered # - sub $VARIABLE_OFFSET, %rsp - and $~63, %rsp .endm .macro FUNC_RESTORE - mov %r14, %rsp pop %r14 pop %r13 pop %r12 .endm +# Precompute hashkeys. +# Input: Hash subkey. +# Output: HashKeys stored in gcm_context_data. Only needs to be called +# once per key. +# clobbers r12, and tmp xmm registers. +.macro PRECOMPUTE TMP1 TMP2 TMP3 TMP4 TMP5 TMP6 TMP7 + mov arg7, %r12 + movdqu (%r12), \TMP3 + movdqa SHUF_MASK(%rip), \TMP2 + PSHUFB_XMM \TMP2, \TMP3 + + # precompute HashKey<<1 mod poly from the HashKey (required for GHASH) + + movdqa \TMP3, \TMP2 + psllq $1, \TMP3 + psrlq $63, \TMP2 + movdqa \TMP2, \TMP1 + pslldq $8, \TMP2 + psrldq $8, \TMP1 + por \TMP2, \TMP3 + + # reduce HashKey<<1 + + pshufd $0x24, \TMP1, \TMP2 + pcmpeqd TWOONE(%rip), \TMP2 + pand POLY(%rip), \TMP2 + pxor \TMP2, \TMP3 + movdqa \TMP3, HashKey(%arg2) + + movdqa \TMP3, \TMP5 + pshufd $78, \TMP3, \TMP1 + pxor \TMP3, \TMP1 + movdqa \TMP1, HashKey_k(%arg2) + + GHASH_MUL \TMP5, \TMP3, \TMP1, \TMP2, \TMP4, \TMP6, \TMP7 +# TMP5 = HashKey^2<<1 (mod poly) + movdqa \TMP5, HashKey_2(%arg2) +# HashKey_2 = HashKey^2<<1 (mod poly) + pshufd $78, \TMP5, \TMP1 + pxor \TMP5, \TMP1 + movdqa \TMP1, HashKey_2_k(%arg2) + + GHASH_MUL \TMP5, \TMP3, \TMP1, \TMP2, \TMP4, \TMP6, \TMP7 +# TMP5 = HashKey^3<<1 (mod poly) + movdqa \TMP5, HashKey_3(%arg2) + pshufd $78, \TMP5, \TMP1 + pxor \TMP5, \TMP1 + movdqa \TMP1, HashKey_3_k(%arg2) + + GHASH_MUL \TMP5, \TMP3, \TMP1, \TMP2, \TMP4, \TMP6, \TMP7 +# TMP5 = HashKey^3<<1 (mod poly) + movdqa \TMP5, HashKey_4(%arg2) + pshufd $78, \TMP5, \TMP1 + pxor \TMP5, \TMP1 + movdqa \TMP1, HashKey_4_k(%arg2) +.endm # GCM_INIT initializes a gcm_context struct to prepare for encoding/decoding. # Clobbers rax, r10-r13 and xmm0-xmm6, %xmm13 .macro GCM_INIT - mov arg9, %r11 mov %r11, AadLen(%arg2) # ctx_data.aad_length = aad_length xor %r11, %r11 @@ -219,28 +269,8 @@ ALL_F: .octa 0xffffffffffffffffffffffffffffffff PSHUFB_XMM %xmm2, %xmm0 movdqu %xmm0, CurCount(%arg2) # ctx_data.current_counter = iv - mov arg7, %r12 - movdqu (%r12), %xmm13 - movdqa SHUF_MASK(%rip), %xmm2 - PSHUFB_XMM %xmm2, %xmm13 - - # precompute HashKey<<1 mod poly from the HashKey (required for GHASH) - - movdqa %xmm13, %xmm2 - psllq $1, %xmm13 - psrlq $63, %xmm2 - movdqa %xmm2, %xmm1 - pslldq $8, %xmm2 - psrldq $8, %xmm1 - por %xmm2, %xmm13 - - # reduce HashKey<<1 - - pshufd $0x24, %xmm1, %xmm2 - pcmpeqd TWOONE(%rip), %xmm2 - pand POLY(%rip), %xmm2 - pxor %xmm2, %xmm13 - movdqa %xmm13, HashKey(%rsp) + PRECOMPUTE %xmm1 %xmm2 %xmm3 %xmm4 %xmm5 %xmm6 %xmm7 + movdqa HashKey(%arg2), %xmm13 CALC_AAD_HASH %xmm13 %xmm0 %xmm1 %xmm2 %xmm3 %xmm4 \ %xmm5 %xmm6 @@ -252,7 +282,7 @@ ALL_F: .octa 0xffffffffffffffffffffffffffffffff # Clobbers rax, r10-r13, and xmm0-xmm15 .macro GCM_ENC_DEC operation movdqu AadHash(%arg2), %xmm8 - movdqu HashKey(%rsp), %xmm13 + movdqu HashKey(%arg2), %xmm13 add %arg5, InLen(%arg2) mov %arg5, %r13 # save the number of bytes and $-16, %r13 # %r13 = %r13 - (%r13 mod 16) @@ -376,7 +406,7 @@ _multiple_of_16_bytes_\@: # Clobbers rax, r10-r12, and xmm0, xmm1, xmm5-xmm15 .macro GCM_COMPLETE movdqu AadHash(%arg2), %xmm8 - movdqu HashKey(%rsp), %xmm13 + movdqu HashKey(%arg2), %xmm13 mov PBlockLen(%arg2), %r12 @@ -583,7 +613,7 @@ _get_AAD_done\@: * the ciphertext * %r10, %r11, %r12, %rax, %xmm5, %xmm6, %xmm7, %xmm8, %xmm9 registers * are clobbered -* arg1, %arg2, %arg3, %r14 are used as a pointer only, not modified +* arg1, %arg2, %arg3 are used as a pointer only, not modified */ @@ -694,17 +724,6 @@ aes_loop_initial_\@: pxor \TMP1, \XMM2 pxor \TMP1, \XMM3 pxor \TMP1, \XMM4 - movdqa \TMP3, \TMP5 - pshufd $78, \TMP3, \TMP1 - pxor \TMP3, \TMP1 - movdqa \TMP1, HashKey_k(%rsp) - GHASH_MUL \TMP5, \TMP3, \TMP1, \TMP2, \TMP4, \TMP6, \TMP7 -# TMP5 = HashKey^2<<1 (mod poly) - movdqa \TMP5, HashKey_2(%rsp) -# HashKey_2 = HashKey^2<<1 (mod poly) - pshufd $78, \TMP5, \TMP1 - pxor \TMP5, \TMP1 - movdqa \TMP1, HashKey_2_k(%rsp) .irpc index, 1234 # do 4 rounds movaps 0x10*\index(%arg1), \TMP1 AESENC \TMP1, \XMM1 @@ -712,12 +731,6 @@ aes_loop_initial_\@: AESENC \TMP1, \XMM3 AESENC \TMP1, \XMM4 .endr - GHASH_MUL \TMP5, \TMP3, \TMP1, \TMP2, \TMP4, \TMP6, \TMP7 -# TMP5 = HashKey^3<<1 (mod poly) - movdqa \TMP5, HashKey_3(%rsp) - pshufd $78, \TMP5, \TMP1 - pxor \TMP5, \TMP1 - movdqa \TMP1, HashKey_3_k(%rsp) .irpc index, 56789 # do next 5 rounds movaps 0x10*\index(%arg1), \TMP1 AESENC \TMP1, \XMM1 @@ -725,12 +738,6 @@ aes_loop_initial_\@: AESENC \TMP1, \XMM3 AESENC \TMP1, \XMM4 .endr - GHASH_MUL \TMP5, \TMP3, \TMP1, \TMP2, \TMP4, \TMP6, \TMP7 -# TMP5 = HashKey^3<<1 (mod poly) - movdqa \TMP5, HashKey_4(%rsp) - pshufd $78, \TMP5, \TMP1 - pxor \TMP5, \TMP1 - movdqa \TMP1, HashKey_4_k(%rsp) lea 0xa0(%arg1),%r10 mov keysize,%eax shr $2,%eax # 128->4, 192->6, 256->8 @@ -815,7 +822,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation pshufd $78, \XMM5, \TMP6 pxor \XMM5, \TMP6 paddd ONE(%rip), \XMM0 # INCR CNT - movdqa HashKey_4(%rsp), \TMP5 + movdqa HashKey_4(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP4 # TMP4 = a1*b1 movdqa \XMM0, \XMM1 paddd ONE(%rip), \XMM0 # INCR CNT @@ -834,7 +841,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation pxor (%arg1), \XMM2 pxor (%arg1), \XMM3 pxor (%arg1), \XMM4 - movdqa HashKey_4_k(%rsp), \TMP5 + movdqa HashKey_4_k(%arg2), \TMP5 PCLMULQDQ 0x00, \TMP5, \TMP6 # TMP6 = (a1+a0)*(b1+b0) movaps 0x10(%arg1), \TMP1 AESENC \TMP1, \XMM1 # Round 1 @@ -849,7 +856,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation movdqa \XMM6, \TMP1 pshufd $78, \XMM6, \TMP2 pxor \XMM6, \TMP2 - movdqa HashKey_3(%rsp), \TMP5 + movdqa HashKey_3(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP1 # TMP1 = a1 * b1 movaps 0x30(%arg1), \TMP3 AESENC \TMP3, \XMM1 # Round 3 @@ -862,7 +869,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation AESENC \TMP3, \XMM2 AESENC \TMP3, \XMM3 AESENC \TMP3, \XMM4 - movdqa HashKey_3_k(%rsp), \TMP5 + movdqa HashKey_3_k(%arg2), \TMP5 PCLMULQDQ 0x00, \TMP5, \TMP2 # TMP2 = (a1+a0)*(b1+b0) movaps 0x50(%arg1), \TMP3 AESENC \TMP3, \XMM1 # Round 5 @@ -876,7 +883,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation movdqa \XMM7, \TMP1 pshufd $78, \XMM7, \TMP2 pxor \XMM7, \TMP2 - movdqa HashKey_2(%rsp ), \TMP5 + movdqa HashKey_2(%arg2), \TMP5 # Multiply TMP5 * HashKey using karatsuba @@ -892,7 +899,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation AESENC \TMP3, \XMM2 AESENC \TMP3, \XMM3 AESENC \TMP3, \XMM4 - movdqa HashKey_2_k(%rsp), \TMP5 + movdqa HashKey_2_k(%arg2), \TMP5 PCLMULQDQ 0x00, \TMP5, \TMP2 # TMP2 = (a1+a0)*(b1+b0) movaps 0x80(%arg1), \TMP3 AESENC \TMP3, \XMM1 # Round 8 @@ -910,7 +917,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation movdqa \XMM8, \TMP1 pshufd $78, \XMM8, \TMP2 pxor \XMM8, \TMP2 - movdqa HashKey(%rsp), \TMP5 + movdqa HashKey(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP1 # TMP1 = a1*b1 movaps 0x90(%arg1), \TMP3 AESENC \TMP3, \XMM1 # Round 9 @@ -939,7 +946,7 @@ aes_loop_par_enc_done: AESENCLAST \TMP3, \XMM2 AESENCLAST \TMP3, \XMM3 AESENCLAST \TMP3, \XMM4 - movdqa HashKey_k(%rsp), \TMP5 + movdqa HashKey_k(%arg2), \TMP5 PCLMULQDQ 0x00, \TMP5, \TMP2 # TMP2 = (a1+a0)*(b1+b0) movdqu (%arg4,%r11,1), \TMP3 pxor \TMP3, \XMM1 # Ciphertext/Plaintext XOR EK @@ -1023,7 +1030,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation pshufd $78, \XMM5, \TMP6 pxor \XMM5, \TMP6 paddd ONE(%rip), \XMM0 # INCR CNT - movdqa HashKey_4(%rsp), \TMP5 + movdqa HashKey_4(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP4 # TMP4 = a1*b1 movdqa \XMM0, \XMM1 paddd ONE(%rip), \XMM0 # INCR CNT @@ -1042,7 +1049,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation pxor (%arg1), \XMM2 pxor (%arg1), \XMM3 pxor (%arg1), \XMM4 - movdqa HashKey_4_k(%rsp), \TMP5 + movdqa HashKey_4_k(%arg2), \TMP5 PCLMULQDQ 0x00, \TMP5, \TMP6 # TMP6 = (a1+a0)*(b1+b0) movaps 0x10(%arg1), \TMP1 AESENC \TMP1, \XMM1 # Round 1 @@ -1057,7 +1064,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation movdqa \XMM6, \TMP1 pshufd $78, \XMM6, \TMP2 pxor \XMM6, \TMP2 - movdqa HashKey_3(%rsp), \TMP5 + movdqa HashKey_3(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP1 # TMP1 = a1 * b1 movaps 0x30(%arg1), \TMP3 AESENC \TMP3, \XMM1 # Round 3 @@ -1070,7 +1077,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation AESENC \TMP3, \XMM2 AESENC \TMP3, \XMM3 AESENC \TMP3, \XMM4 - movdqa HashKey_3_k(%rsp), \TMP5 + movdqa HashKey_3_k(%arg2), \TMP5 PCLMULQDQ 0x00, \TMP5, \TMP2 # TMP2 = (a1+a0)*(b1+b0) movaps 0x50(%arg1), \TMP3 AESENC \TMP3, \XMM1 # Round 5 @@ -1084,7 +1091,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation movdqa \XMM7, \TMP1 pshufd $78, \XMM7, \TMP2 pxor \XMM7, \TMP2 - movdqa HashKey_2(%rsp ), \TMP5 + movdqa HashKey_2(%arg2), \TMP5 # Multiply TMP5 * HashKey using karatsuba @@ -1100,7 +1107,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation AESENC \TMP3, \XMM2 AESENC \TMP3, \XMM3 AESENC \TMP3, \XMM4 - movdqa HashKey_2_k(%rsp), \TMP5 + movdqa HashKey_2_k(%arg2), \TMP5 PCLMULQDQ 0x00, \TMP5, \TMP2 # TMP2 = (a1+a0)*(b1+b0) movaps 0x80(%arg1), \TMP3 AESENC \TMP3, \XMM1 # Round 8 @@ -1118,7 +1125,7 @@ TMP6 XMM0 XMM1 XMM2 XMM3 XMM4 XMM5 XMM6 XMM7 XMM8 operation movdqa \XMM8, \TMP1 pshufd $78, \XMM8, \TMP2 pxor \XMM8, \TMP2 - movdqa HashKey(%rsp), \TMP5 + movdqa HashKey(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP1 # TMP1 = a1*b1 movaps 0x90(%arg1), \TMP3 AESENC \TMP3, \XMM1 # Round 9 @@ -1147,7 +1154,7 @@ aes_loop_par_dec_done: AESENCLAST \TMP3, \XMM2 AESENCLAST \TMP3, \XMM3 AESENCLAST \TMP3, \XMM4 - movdqa HashKey_k(%rsp), \TMP5 + movdqa HashKey_k(%arg2), \TMP5 PCLMULQDQ 0x00, \TMP5, \TMP2 # TMP2 = (a1+a0)*(b1+b0) movdqu (%arg4,%r11,1), \TMP3 pxor \TMP3, \XMM1 # Ciphertext/Plaintext XOR EK @@ -1223,10 +1230,10 @@ TMP7 XMM1 XMM2 XMM3 XMM4 XMMDst movdqa \XMM1, \TMP6 pshufd $78, \XMM1, \TMP2 pxor \XMM1, \TMP2 - movdqa HashKey_4(%rsp), \TMP5 + movdqa HashKey_4(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP6 # TMP6 = a1*b1 PCLMULQDQ 0x00, \TMP5, \XMM1 # XMM1 = a0*b0 - movdqa HashKey_4_k(%rsp), \TMP4 + movdqa HashKey_4_k(%arg2), \TMP4 PCLMULQDQ 0x00, \TMP4, \TMP2 # TMP2 = (a1+a0)*(b1+b0) movdqa \XMM1, \XMMDst movdqa \TMP2, \XMM1 # result in TMP6, XMMDst, XMM1 @@ -1236,10 +1243,10 @@ TMP7 XMM1 XMM2 XMM3 XMM4 XMMDst movdqa \XMM2, \TMP1 pshufd $78, \XMM2, \TMP2 pxor \XMM2, \TMP2 - movdqa HashKey_3(%rsp), \TMP5 + movdqa HashKey_3(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP1 # TMP1 = a1*b1 PCLMULQDQ 0x00, \TMP5, \XMM2 # XMM2 = a0*b0 - movdqa HashKey_3_k(%rsp), \TMP4 + movdqa HashKey_3_k(%arg2), \TMP4 PCLMULQDQ 0x00, \TMP4, \TMP2 # TMP2 = (a1+a0)*(b1+b0) pxor \TMP1, \TMP6 pxor \XMM2, \XMMDst @@ -1251,10 +1258,10 @@ TMP7 XMM1 XMM2 XMM3 XMM4 XMMDst movdqa \XMM3, \TMP1 pshufd $78, \XMM3, \TMP2 pxor \XMM3, \TMP2 - movdqa HashKey_2(%rsp), \TMP5 + movdqa HashKey_2(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP1 # TMP1 = a1*b1 PCLMULQDQ 0x00, \TMP5, \XMM3 # XMM3 = a0*b0 - movdqa HashKey_2_k(%rsp), \TMP4 + movdqa HashKey_2_k(%arg2), \TMP4 PCLMULQDQ 0x00, \TMP4, \TMP2 # TMP2 = (a1+a0)*(b1+b0) pxor \TMP1, \TMP6 pxor \XMM3, \XMMDst @@ -1264,10 +1271,10 @@ TMP7 XMM1 XMM2 XMM3 XMM4 XMMDst movdqa \XMM4, \TMP1 pshufd $78, \XMM4, \TMP2 pxor \XMM4, \TMP2 - movdqa HashKey(%rsp), \TMP5 + movdqa HashKey(%arg2), \TMP5 PCLMULQDQ 0x11, \TMP5, \TMP1 # TMP1 = a1*b1 PCLMULQDQ 0x00, \TMP5, \XMM4 # XMM4 = a0*b0 - movdqa HashKey_k(%rsp), \TMP4 + movdqa HashKey_k(%arg2), \TMP4 PCLMULQDQ 0x00, \TMP4, \TMP2 # TMP2 = (a1+a0)*(b1+b0) pxor \TMP1, \TMP6 pxor \XMM4, \XMMDst -- 2.9.5