[openssl-commits] [openssl] master update

[Andy Polyakov]

The branch master has been updated
       via  2c12f22c3364a1e562a19415f9929359945a7c21 (commit)
       via  f198cc43a0eca4bf1a8e7f60c51af560f4346dc8 (commit)
      from  0685b15a68db1cd24678432f38b88f217495cbb9 (commit)


- Log -----------------------------------------------------------------
commit 2c12f22c3364a1e562a19415f9929359945a7c21
Author: Andy Polyakov <appro at openssl.org>
Date:   Thu Jun 30 16:06:48 2016 +0200

    SPARC assembly pack: enforce V8+ ABI constraints.
    
    Reviewed-by: Richard Levitte <levitte at openssl.org>

commit f198cc43a0eca4bf1a8e7f60c51af560f4346dc8
Author: Andy Polyakov <appro at openssl.org>
Date:   Thu Jun 30 15:57:57 2016 +0200

    SPARC assembly pack: enforce V8+ ABI constraints.
    
    Even though it's hard to imagine, it turned out that upper half of
    arguments passed to V8+ subroutine can be non-zero.
    
    ["n" pseudo-instructions, such as srln being srl in 32-bit case and
    srlx in 64-bit one, were implemented in binutils 2.10. It's assumed
    that Solaris assembler implemented it around same time, i.e. 2000.]
    
    Reviewed-by: Richard Levitte <levitte at openssl.org>

-----------------------------------------------------------------------

Summary of changes:
 crypto/des/asm/dest4-sparcv9.pl         |  8 ++++----
 crypto/modes/asm/ghash-sparcv9.pl       |  2 ++
 crypto/perlasm/sparcv9_modes.pl         |  4 ++++
 crypto/poly1305/asm/poly1305-sparcv9.pl | 10 +++++-----
 4 files changed, 15 insertions(+), 9 deletions(-)

diff --git a/crypto/des/asm/dest4-sparcv9.pl b/crypto/des/asm/dest4-sparcv9.pl
index 1d4f0d1..4a6e29f 100644
--- a/crypto/des/asm/dest4-sparcv9.pl
+++ b/crypto/des/asm/dest4-sparcv9.pl
@@ -106,7 +106,7 @@ $code.=<<___;
 des_t4_cbc_encrypt:
 	cmp		$len, 0
 	be,pn		$::size_t_cc, .Lcbc_abort
-	nop
+	srln		$len, 0, $len		! needed on v8+, "nop" on v9
 	ld		[$ivec + 0], %f0	! load ivec
 	ld		[$ivec + 4], %f1
 
@@ -207,7 +207,7 @@ des_t4_cbc_encrypt:
 des_t4_cbc_decrypt:
 	cmp		$len, 0
 	be,pn		$::size_t_cc, .Lcbc_abort
-	nop
+	srln		$len, 0, $len		! needed on v8+, "nop" on v9
 	ld		[$ivec + 0], %f2	! load ivec
 	ld		[$ivec + 4], %f3
 
@@ -315,7 +315,7 @@ $code.=<<___;
 des_t4_ede3_cbc_encrypt:
 	cmp		$len, 0
 	be,pn		$::size_t_cc, .Lcbc_abort
-	nop
+	srln		$len, 0, $len		! needed on v8+, "nop" on v9
 	ld		[$ivec + 0], %f0	! load ivec
 	ld		[$ivec + 4], %f1
 
@@ -467,7 +467,7 @@ des_t4_ede3_cbc_encrypt:
 des_t4_ede3_cbc_decrypt:
 	cmp		$len, 0
 	be,pn		$::size_t_cc, .Lcbc_abort
-	nop
+	srln		$len, 0, $len		! needed on v8+, "nop" on v9
 	ld		[$ivec + 0], %f2	! load ivec
 	ld		[$ivec + 4], %f3
 
diff --git a/crypto/modes/asm/ghash-sparcv9.pl b/crypto/modes/asm/ghash-sparcv9.pl
index 6ca3a9b..c4eb3b1 100644
--- a/crypto/modes/asm/ghash-sparcv9.pl
+++ b/crypto/modes/asm/ghash-sparcv9.pl
@@ -453,6 +453,8 @@ gcm_gmult_vis3:
 .align	32
 gcm_ghash_vis3:
 	save	%sp,-$frame,%sp
+	nop
+	srln	$len,0,$len		! needed on v8+, "nop" on v9
 
 	ldx	[$Xip+8],$C2		! load Xi
 	ldx	[$Xip+0],$C3
diff --git a/crypto/perlasm/sparcv9_modes.pl b/crypto/perlasm/sparcv9_modes.pl
index 47c2e53..bfdada8 100644
--- a/crypto/perlasm/sparcv9_modes.pl
+++ b/crypto/perlasm/sparcv9_modes.pl
@@ -48,6 +48,7 @@ ${alg}${bits}_t4_cbc_encrypt:
 	save		%sp, -$::frame, %sp
 	cmp		$len, 0
 	be,pn		$::size_t_cc, .L${bits}_cbc_enc_abort
+	srln		$len, 0, $len		! needed on v8+, "nop" on v9
 	sub		$inp, $out, $blk_init	! $inp!=$out
 ___
 $::code.=<<___ if (!$::evp);
@@ -265,6 +266,7 @@ ${alg}${bits}_t4_cbc_decrypt:
 	save		%sp, -$::frame, %sp
 	cmp		$len, 0
 	be,pn		$::size_t_cc, .L${bits}_cbc_dec_abort
+	srln		$len, 0, $len		! needed on v8+, "nop" on v9
 	sub		$inp, $out, $blk_init	! $inp!=$out
 ___
 $::code.=<<___ if (!$::evp);
@@ -624,6 +626,7 @@ $::code.=<<___;
 .align	32
 ${alg}${bits}_t4_ctr32_encrypt:
 	save		%sp, -$::frame, %sp
+	srln		$len, 0, $len		! needed on v8+, "nop" on v9
 
 	prefetch	[$inp], 20
 	prefetch	[$inp + 63], 20
@@ -927,6 +930,7 @@ $::code.=<<___;
 .align	32
 ${alg}${bits}_t4_xts_${dir}crypt:
 	save		%sp, -$::frame-16, %sp
+	srln		$len, 0, $len		! needed on v8+, "nop" on v9
 
 	mov		$ivec, %o0
 	add		%fp, $::bias-16, %o1
diff --git a/crypto/poly1305/asm/poly1305-sparcv9.pl b/crypto/poly1305/asm/poly1305-sparcv9.pl
index c22735a..0bdd048 100755
--- a/crypto/poly1305/asm/poly1305-sparcv9.pl
+++ b/crypto/poly1305/asm/poly1305-sparcv9.pl
@@ -146,7 +146,7 @@ poly1305_init:
 .align	32
 poly1305_blocks:
 	save	%sp,-STACK_FRAME,%sp
-	andn	$len,15,$len
+	srln	$len,4,$len
 
 	brz,pn	$len,.Lno_data
 	nop
@@ -202,7 +202,7 @@ poly1305_blocks:
 	umul	$r1,$h0,$d1
 	umul	$r2,$h0,$d2
 	umul	$r3,$h0,$d3
-	 sub	$len,16,$len
+	 sub	$len,1,$len
 	 add	$inp,16,$inp
 
 	umul	$s3,$h1,$t0
@@ -285,7 +285,7 @@ $code.=<<___;
 .align	32
 poly1305_blocks_vis3:
 	save	%sp,-STACK_FRAME,%sp
-	andn	$len,15,$len
+	srln	$len,4,$len
 
 	brz,pn	$len,.Lno_data
 	nop
@@ -323,7 +323,7 @@ poly1305_blocks_vis3:
 
 .Linp_aligned_vis3:
 	addcc	$D0,$H0,$H0		! accumulate input
-	 sub	$len,16,$len
+	 sub	$len,1,$len
 	addxccc	$D1,$H1,$H1
 	 add	$inp,16,$inp
 
@@ -608,7 +608,7 @@ poly1305_init_fma:
 .align	32
 poly1305_blocks_fma:
 	save	%sp,-STACK_FRAME-48,%sp
-	srlx	$len,4,$len
+	srln	$len,4,$len
 
 	brz,pn	$len,.Labort
 	sub	$len,1,$len