Hey Mark, does the order of .cfi_startproc and endbranch matter at all,
I noticed a difference between your openssl/3.1 and 3.2 patches and was
just wondering, it's reversed in gcm_init_avx.
-Bryan.
diff -u -p -r1.1.1.1 patch-crypto_modes_asm_ghash-x86_64_pl
--- security/openssl/3.1/patches/patch-crypto_modes_asm_ghash-x86_64_pl
14 Mar 2023 18:54:42 -0000 1.1.1.1
+++ security/openssl/3.1/patches/patch-crypto_modes_asm_ghash-x86_64_pl
18 Feb 2024 22:25:50 -0000
@@ -1,7 +1,23 @@
Index: crypto/modes/asm/ghash-x86_64.pl
--- crypto/modes/asm/ghash-x86_64.pl.orig
+++ crypto/modes/asm/ghash-x86_64.pl
-@@ -1609,6 +1609,7 @@ ___
+@@ -534,6 +534,7 @@ $code.=<<___;
+ .align 16
+ gcm_init_clmul:
+ .cfi_startproc
++ endbranch
+ .L_init_clmul:
+ ___
+ $code.=<<___ if ($win64);
+@@ -1026,6 +1027,7 @@ $code.=<<___;
+ .type gcm_init_avx,\@abi-omnipotent
+ .align 32
+ gcm_init_avx:
++ endbranch
+ .cfi_startproc
+ ___
+ if ($avx) {
+@@ -1609,6 +1611,7 @@ ___
}
$code.=<<___;
@@ -9,7 +25,7 @@ Index: crypto/modes/asm/ghash-x86_64.pl
.align 64
.Lbswap_mask:
.byte 15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0
-@@ -1662,6 +1663,7 @@ $code.=<<___;
+@@ -1662,6 +1665,7 @@ $code.=<<___;
.asciz "GHASH for x86_64, CRYPTOGAMS by <appro\@openssl.org>"
.align 64
