sha/asm/sha256-586.pl

112 $K256="ebp";
165 	 &mov	("esi",&DWP(0,$K256));
178 	&add	($K256,4);
192 	&blindpop($K256);
193 	&lea	($K256,&DWP(&label("K256")."-".&label("pic_point"),$K256));
205 	&picmeup("edx","OPENSSL_ia32cap_P",$K256,&label("K256"));
323 	&sub	($K256,4*64);			# rewind K
339 &set_label("K256",64);	# Yes! I keep it in the code segment!
340 @K256=(	0x428a2f98,0x71374491,0xb5c0fbcf,0xe9b5dba5,
356 &data_word(@K256);
364 my @AH=($A,$K256);
458 	 &lea	($E,&DWP(@K256[$i],$T,$E));	# T += Sigma1(1)+K[i]
536 	&lea		($K256,&DWP(0x80,$K256));
538 	&movdqa		($TMP,&QWP(0x100-0x80,$K256));	# byte swap mask
555 	&movdqa		($Wi,&QWP(0*16-0x80,$K256));
564 	&movdqa		($Wi,&QWP(1*16-0x80,$K256));
573 	&movdqa		($Wi,&QWP(2*16-0x80,$K256));
585 	&movdqa		($Wi,&QWP(3*16-0x80,$K256));
598 	&movdqa		($Wi,&QWP($i*16-0x80,$K256));
612 	&movdqa		($Wi,&QWP(13*16-0x80,$K256));
622 	&movdqa		($Wi,&QWP(14*16-0x80,$K256));
627 	&movdqa		($TMP,&QWP(0x100-0x80,$K256));	# byte swap mask
630 	&movdqa		($Wi,&QWP(15*16-0x80,$K256));
680 	&movdqa	($t3,&QWP(256,$K256));
684 	# load input, reverse byte order, add K256[0..15], save to stack
693 	&movdqa	($t0,&QWP(0,$K256));
695 	&movdqa	($t1,&QWP(16,$K256));
698 	&movdqa	($t2,&QWP(32,$K256));
700 	&movdqa	($t3,&QWP(48,$K256));
710 	&add		($K256,64);
850 	&movdqa		($t2,&QWP(16*$j,$K256));
918 	&cmp	(&DWP(16*$j,$K256),0x00010203);
961 	&movdqa	($t3,&QWP(64,$K256));
962 	&sub	($K256,3*64);			# rewind K
996 	&vmovdqa	($t3,&QWP(256,$K256));
1000 	# load input, reverse byte order, add K256[0..15], save to stack
1010 	&vpaddd		($t0,@X[0],&QWP(0,$K256));
1012 	&vpaddd		($t1,@X[1],&QWP(16,$K256));
1013 	&vpaddd		($t2,@X[2],&QWP(32,$K256));
1014 	&vpaddd		($t3,@X[3],&QWP(48,$K256));
1022 	&add		($K256,64);
1075 	&vpaddd		($t2,@X[0],&QWP(16*$j,$K256));
1084 	&cmp	(&DWP(16*$j,$K256),0x00010203);
1127 	&vmovdqa	($t3,&QWP(64,$K256));
1128 	&sub	($K256,3*64);			# rewind K
1193 	&vmovdqa	($t3,&QWP(256,$K256));
1197 	# load input, reverse byte order, add K256[0..15], save to stack
1207 	&vpaddd		($t0,@X[0],&QWP(0,$K256));
1209 	&vpaddd		($t1,@X[1],&QWP(16,$K256));
1210 	&vpaddd		($t2,@X[2],&QWP(32,$K256));
1211 	&vpaddd		($t3,@X[3],&QWP(48,$K256));
1219 	&add		($K256,64);
1225 	&cmp	(&DWP(16*$j,$K256),0x00010203);
1268 	&vmovdqa	($t3,&QWP(64,$K256));
1269 	&sub	($K256,3*64);			# rewind K