Land BoringSSL roll on master

third_party/boringssl/win-x86_64/crypto/bn/x86_64-mont.asm

 default rel
 %define XMMWORD
 %define YMMWORD
 %define ZMMWORD
 section .text code align=64
 
 
 EXTERN  OPENSSL_ia32cap_P
 
 global  bn_mul_mont
@@ skipping 659 matching lines @@
         mov     r10d,r9d
         shl     r9d,3
         shl     r10,3+2
         neg     r9
 
 
 
 
 
 
-        lea     r11,[((-64))+r9*4+rsp]
+        lea     r11,[((-64))+r9*2+rsp]
         mov     r8,QWORD[r8]
         sub     r11,rsi
         and     r11,4095
         cmp     r10,r11
         jb      NEAR $L$sqr8x_sp_alt
         sub     rsp,r11
-        lea     rsp,[((-64))+r9*4+rsp]
+        lea     rsp,[((-64))+r9*2+rsp]
         jmp     NEAR $L$sqr8x_sp_done
 
 ALIGN   32
 $L$sqr8x_sp_alt:
-        lea     r10,[((4096-64))+r9*4]
-        lea     rsp,[((-64))+r9*4+rsp]
+        lea     r10,[((4096-64))+r9*2]
+        lea     rsp,[((-64))+r9*2+rsp]
         sub     r11,r10
         mov     r10,0
         cmovc   r11,r10
         sub     rsp,r11
 $L$sqr8x_sp_done:
         and     rsp,-64
         mov     r10,r9
         neg     r9
 
-        lea     r11,[64+r9*2+rsp]
         mov     QWORD[32+rsp],r8
         mov     QWORD[40+rsp],rax
 $L$sqr8x_body:
 
-        mov     rbp,r9
-DB      102,73,15,110,211
-        shr     rbp,3+2
-        mov     eax,DWORD[((OPENSSL_ia32cap_P+8))]
-        jmp     NEAR $L$sqr8x_copy_n
-
-ALIGN   32
-$L$sqr8x_copy_n:
-        movq    xmm0,QWORD[rcx]
-        movq    xmm1,QWORD[8+rcx]
-        movq    xmm3,QWORD[16+rcx]
-        movq    xmm4,QWORD[24+rcx]
-        lea     rcx,[32+rcx]
-        movdqa  XMMWORD[r11],xmm0
-        movdqa  XMMWORD[16+r11],xmm1
-        movdqa  XMMWORD[32+r11],xmm3
-        movdqa  XMMWORD[48+r11],xmm4
-        lea     r11,[64+r11]
-        dec     rbp
-        jnz     NEAR $L$sqr8x_copy_n
-
+DB      102,72,15,110,209
         pxor    xmm0,xmm0
 DB      102,72,15,110,207
 DB      102,73,15,110,218
         call    bn_sqr8x_internal
 
-        pxor    xmm0,xmm0
-        lea     rax,[48+rsp]
-        lea     rdx,[64+r9*2+rsp]
-        shr     r9,3+2
-        mov     rsi,QWORD[40+rsp]
-        jmp     NEAR $L$sqr8x_zero
+
+
+
+        lea     rbx,[r9*1+rdi]
+        mov     rcx,r9
+        mov     rdx,r9
+DB      102,72,15,126,207
+        sar     rcx,3+2
+        jmp     NEAR $L$sqr8x_sub
 
 ALIGN   32
-$L$sqr8x_zero:
-        movdqa  XMMWORD[rax],xmm0
-        movdqa  XMMWORD[16+rax],xmm0
-        movdqa  XMMWORD[32+rax],xmm0
-        movdqa  XMMWORD[48+rax],xmm0
-        lea     rax,[64+rax]
-        movdqa  XMMWORD[rdx],xmm0
-        movdqa  XMMWORD[16+rdx],xmm0
-        movdqa  XMMWORD[32+rdx],xmm0
-        movdqa  XMMWORD[48+rdx],xmm0
-        lea     rdx,[64+rdx]
-        dec     r9
-        jnz     NEAR $L$sqr8x_zero
+$L$sqr8x_sub:
+        mov     r12,QWORD[rbx]
+        mov     r13,QWORD[8+rbx]
+        mov     r14,QWORD[16+rbx]
+        mov     r15,QWORD[24+rbx]
+        lea     rbx,[32+rbx]
+        sbb     r12,QWORD[rbp]
+        sbb     r13,QWORD[8+rbp]
+        sbb     r14,QWORD[16+rbp]
+        sbb     r15,QWORD[24+rbp]
+        lea     rbp,[32+rbp]
+        mov     QWORD[rdi],r12
+        mov     QWORD[8+rdi],r13
+        mov     QWORD[16+rdi],r14
+        mov     QWORD[24+rdi],r15
+        lea     rdi,[32+rdi]
+        inc     rcx
+        jnz     NEAR $L$sqr8x_sub
+
+        sbb     rax,0
+        lea     rbx,[r9*1+rbx]
+        lea     rdi,[r9*1+rdi]
+
+DB      102,72,15,110,200
+        pxor    xmm0,xmm0
+        pshufd  xmm1,xmm1,0
+        mov     rsi,QWORD[40+rsp]
+        jmp     NEAR $L$sqr8x_cond_copy
+
+ALIGN   32
+$L$sqr8x_cond_copy:
+        movdqa  xmm2,XMMWORD[rbx]
+        movdqa  xmm3,XMMWORD[16+rbx]
+        lea     rbx,[32+rbx]
+        movdqu  xmm4,XMMWORD[rdi]
+        movdqu  xmm5,XMMWORD[16+rdi]
+        lea     rdi,[32+rdi]
+        movdqa  XMMWORD[(-32)+rbx],xmm0
+        movdqa  XMMWORD[(-16)+rbx],xmm0
+        movdqa  XMMWORD[(-32)+rdx*1+rbx],xmm0
+        movdqa  XMMWORD[(-16)+rdx*1+rbx],xmm0
+        pcmpeqd xmm0,xmm1
+        pand    xmm2,xmm1
+        pand    xmm3,xmm1
+        pand    xmm4,xmm0
+        pand    xmm5,xmm0
+        pxor    xmm0,xmm0
+        por     xmm4,xmm2
+        por     xmm5,xmm3
+        movdqu  XMMWORD[(-32)+rdi],xmm4
+        movdqu  XMMWORD[(-16)+rdi],xmm5
+        add     r9,32
+        jnz     NEAR $L$sqr8x_cond_copy
 
         mov     rax,1
         mov     r15,QWORD[((-48))+rsi]
         mov     r14,QWORD[((-40))+rsi]
         mov     r13,QWORD[((-32))+rsi]
         mov     r12,QWORD[((-24))+rsi]
         mov     rbp,QWORD[((-16))+rsi]
         mov     rbx,QWORD[((-8))+rsi]
         lea     rsp,[rsi]
 $L$sqr8x_epilogue:
@@ skipping 167 matching lines @@
         DD      mul_handler wrt ..imagebase
         DD      $L$mul_body wrt ..imagebase,$L$mul_epilogue wrt ..imagebase
 $L$SEH_info_bn_mul4x_mont:
 DB      9,0,0,0
         DD      mul_handler wrt ..imagebase
         DD      $L$mul4x_body wrt ..imagebase,$L$mul4x_epilogue wrt ..imagebase
 $L$SEH_info_bn_sqr8x_mont:
 DB      9,0,0,0
         DD      sqr_handler wrt ..imagebase
         DD      $L$sqr8x_body wrt ..imagebase,$L$sqr8x_epilogue wrt ..imagebase