Latest update

crypto/modes/asm/ghash-armv4.pl

@@ -142,7 +142,6 @@ ___
 $code=<<___;
 #include "arm_arch.h"
 
-.text
 #if defined(__thumb2__) || defined(__clang__)
 .syntax	unified
 #define ldrplb  ldrbpl
@@ -154,6 +153,8 @@ $code=<<___;
 .code	32
 #endif
 
+.text
+
 .type	rem_4bit,%object
 .align	5
 rem_4bit:

crypto/modes/asm/ghash-x86_64.pl

@@ -529,6 +529,7 @@ $code.=<<___;
 .type	gcm_init_clmul,\@abi-omnipotent
 .align	16
 gcm_init_clmul:
+.cfi_startproc
 .L_init_clmul:
 ___
 $code.=<<___ if ($win64);
@@ -598,6 +599,7 @@ $code.=<<___ if ($win64);
 ___
 $code.=<<___;
 	ret
+.cfi_endproc
 .size	gcm_init_clmul,.-gcm_init_clmul
 ___
 }
@@ -609,6 +611,7 @@ $code.=<<___;
 .type	gcm_gmult_clmul,\@abi-omnipotent
 .align	16
 gcm_gmult_clmul:
+.cfi_startproc
 .L_gmult_clmul:
 	movdqu		($Xip),$Xi
 	movdqa		.Lbswap_mask(%rip),$T3
@@ -645,6 +648,7 @@ $code.=<<___;
 	pshufb		$T3,$Xi
 	movdqu		$Xi,($Xip)
 	ret
+.cfi_endproc
 .size	gcm_gmult_clmul,.-gcm_gmult_clmul
 ___
 }
@@ -658,6 +662,7 @@ $code.=<<___;
 .type	gcm_ghash_clmul,\@abi-omnipotent
 .align	32
 gcm_ghash_clmul:
+.cfi_startproc
 .L_ghash_clmul:
 ___
 $code.=<<___ if ($win64);
@@ -1005,6 +1010,7 @@ $code.=<<___ if ($win64);
 ___
 $code.=<<___;
 	ret
+.cfi_endproc
 .size	gcm_ghash_clmul,.-gcm_ghash_clmul
 ___
 }
@@ -1014,6 +1020,7 @@ $code.=<<___;
 .type	gcm_init_avx,\@abi-omnipotent
 .align	32
 gcm_init_avx:
+.cfi_startproc
 ___
 if ($avx) {
 my ($Htbl,$Xip)=@_4args;
@@ -1142,11 +1149,13 @@ $code.=<<___ if ($win64);
 ___
 $code.=<<___;
 	ret
+.cfi_endproc
 .size	gcm_init_avx,.-gcm_init_avx
 ___
 } else {
 $code.=<<___;
 	jmp	.L_init_clmul
+.cfi_endproc
 .size	gcm_init_avx,.-gcm_init_avx
 ___
 }
@@ -1156,7 +1165,9 @@ $code.=<<___;
 .type	gcm_gmult_avx,\@abi-omnipotent
 .align	32
 gcm_gmult_avx:
+.cfi_startproc
 	jmp	.L_gmult_clmul
+.cfi_endproc
 .size	gcm_gmult_avx,.-gcm_gmult_avx
 ___
 
@@ -1165,6 +1176,7 @@ $code.=<<___;
 .type	gcm_ghash_avx,\@abi-omnipotent
 .align	32
 gcm_ghash_avx:
+.cfi_startproc
 ___
 if ($avx) {
 my ($Xip,$Htbl,$inp,$len)=@_4args;
@@ -1577,11 +1589,13 @@ $code.=<<___ if ($win64);
 ___
 $code.=<<___;
 	ret
+.cfi_endproc
 .size	gcm_ghash_avx,.-gcm_ghash_avx
 ___
 } else {
 $code.=<<___;
 	jmp	.L_ghash_clmul
+.cfi_endproc
 .size	gcm_ghash_avx,.-gcm_ghash_avx
 ___
 }

crypto/modes/asm/ghashv8-armx.pl

@@ -66,18 +66,26 @@ $inc="x12";
 {
 my ($Xl,$Xm,$Xh,$IN)=map("q$_",(0..3));
 my ($t0,$t1,$t2,$xC2,$H,$Hhl,$H2)=map("q$_",(8..14));
+my $_byte = ($flavour =~ /win/ ? "DCB" : ".byte");
 
 $code=<<___;
 #include "arm_arch.h"
 
 #if __ARM_MAX_ARCH__>=7
-.text
 ___
-$code.=".arch	armv8-a+crypto\n"	if ($flavour =~ /64/);
-$code.=<<___				if ($flavour !~ /64/);
+$code.=".arch	armv8-a+crypto\n.text\n"	if ($flavour =~ /64/);
+$code.=<<___					if ($flavour !~ /64/);
 .fpu	neon
-.code	32
-#undef	__thumb2__
+#ifdef __thumb2__
+.syntax        unified
+.thumb
+# define INST(a,b,c,d) $_byte  c,0xef,a,b
+#else
+.code  32
+# define INST(a,b,c,d) $_byte  a,b,c,0xf2
+#endif
+
+.text
 ___
 
 ################################################################################
@@ -752,7 +760,7 @@ if ($flavour =~ /64/) {			######## 64-bit code
 	    # since ARMv7 instructions are always encoded little-endian.
 	    # correct solution is to use .inst directive, but older
 	    # assemblers don't implement it:-(
-	    sprintf ".byte\t0x%02x,0x%02x,0x%02x,0x%02x\t@ %s %s",
+	    sprintf "INST(0x%02x,0x%02x,0x%02x,0x%02x)\t@ %s %s",
 			$word&0xff,($word>>8)&0xff,
 			($word>>16)&0xff,($word>>24)&0xff,
 			$mnemonic,$arg;
@@ -767,13 +775,17 @@ if ($flavour =~ /64/) {			######## 64-bit code
 	# fix up remaining new-style suffixes
 	s/\],#[0-9]+/]!/o;
 
-	s/cclr\s+([^,]+),\s*([a-z]+)/mov$2	$1,#0/o			or
+	s/cclr\s+([^,]+),\s*([a-z]+)/mov.$2	$1,#0/o			or
 	s/vdup\.32\s+(.*)/unvdup32($1)/geo				or
 	s/v?(pmull2?)\.p64\s+(.*)/unvpmullp64($1,$2)/geo		or
 	s/\bq([0-9]+)#(lo|hi)/sprintf "d%d",2*$1+($2 eq "hi")/geo	or
 	s/^(\s+)b\./$1b/o						or
 	s/^(\s+)ret/$1bx\tlr/o;
 
+	if (s/^(\s+)mov\.([a-z]+)/$1mov$2/) {
+	    print "     it      $2\n";
+	}
+
 	print $_,"\n";
     }
 }