SPARC Solaris and Linux assemblers treat .align directive differently.

PR: 1547

crypto/bn/asm/sparcv9-mont.pl

@@ -260,24 +260,25 @@ $fname:
 
 	srl	$npj,30,%o0		! boundary condition...
 	brz,pn	%o0,.Lcopy		! ... is met
-	subcc	%g0,%g0,%g0		! clear %icc.c
+	nop
 
-.align	16,0x1000000
+	ba	.Lsub
+	subcc	%g0,%g0,%g0		! clear %icc.c
+.align	16
 .Lsub:
 	ld	[$tp+%o7],%o0
 	ld	[$np+%o7],%o1
 	subccc	%o0,%o1,%o1		! tp[j]-np[j]
-	st	%o1,[$rp+%o7]
+	add	$rp,%o7,$i
 	add	%o7,4,%o7
 	brnz	%o7,.Lsub
-	nop
+	st	%o1,[$i]
 	subc	$car2,0,$car2		! handle upmost overflow bit
 	and	$tp,$car2,$ap
 	andn	$rp,$car2,$np
 	or	$ap,$np,$ap
 	sub	%g0,$num,%o7
 
-.align	16,0x1000000
 .Lcopy:
 	ld	[$ap+%o7],%o0		! copy or in-place refresh
 	st	%g0,[$tp+%o7]		! zap tp
@@ -603,6 +604,7 @@ $code.=<<___;
 .type	$fname,#function
 .size	$fname,(.-$fname)
 .asciz	"Montgomery Multipltication for SPARCv9, CRYPTOGAMS by <appro\@openssl.org>"
+.align	32
 ___
 $code =~ s/\`([^\`]*)\`/eval($1)/gem;
 print $code;

crypto/bn/asm/sparcv9a-mont.pl

@@ -353,7 +353,7 @@ $fname:
 	bz,pn	%icc,.L1stskip
 	std	$nlod,[%sp+$bias+$frame+24]
 
-.align	32,0x1000000
+.align	32			! incidentally already aligned !
 .L1st:
 	add	$ap,$j,%o4
 	add	$np,$j,%o5
@@ -805,8 +805,10 @@ $fname:
 	orn	%g0,%g0,%g4
 	brz,pn	%o1,.Lcopy		! ... is met
 	sub	%g0,$num,%o7		! n=-num
-	
-.align	32,0x1000000
+	ba	.Lsub
+	nop
+
+.align	32
 .Lsub:
 	ldx	[$tp+%o7],%o0
 	add	$np,%o7,%g1
@@ -822,8 +824,10 @@ $fname:
 	st	%o3,[%g1+4]
 	subc	$carry,0,%g4
 	sub	%g0,$num,%o7		! n=-num
+	ba	.Lcopy
+	nop
 
-.align	32,0x1000000
+.align	32
 .Lcopy:
 	ldx	[$tp+%o7],%o0
 	add	$rp,%o7,%g1
@@ -842,7 +846,6 @@ $fname:
 	st	%o1,[%g1+4]
 	sub	%g0,$num,%o7		! n=-num
 
-.align	32,0x1000000
 .Lzap:
 	stx	%g0,[$ap_l+%o7]
 	stx	%g0,[$ap_h+%o7]
@@ -862,6 +865,7 @@ $fname:
 .type   $fname,#function
 .size	$fname,(.-$fname)
 .asciz	"Montgomery Multipltication for UltraSPARC, CRYPTOGAMS by <appro\@openssl.org>"
+.align	32
 ___
 
 $code =~ s/\`([^\`]*)\`/eval($1)/gem;