[Backport] Chromium-side changes for BoringSSL cherry-pick.

BoringSSL cherry-picks a little complex due to generated files spanning
two repositories. (We'll need to come up with a better story here...)

This is the result of running generate_build_files.py at this revision:
https://boringssl.googlesource.com/boringssl/+/a20bb7ff8bb5057065a2e7941249773f9676cf45

Conveniently, all affected files in this commit are Chromium-side, but
I'll do a DEPS update too, just to keep it all in sync.

Bug: 793030
Reviewed-on: https://chromium-review.googlesource.com/820333
Reviewed-by: Steven Valdez <svaldez@chromium.org>
Cr-Commit-Position: refs/branch-heads/3282@{#145}
Cr-Branched-From: 5fdc0fab22ce7efd32532ee989b223fa12f8171e-refs/heads/master@{#520840}

Change-Id: I6fea7106d5d1a667bb48cd90da54289339296cab
Reviewed-by: Michael Brüning <michael.bruning@qt.io>

3 files changed, 18 insertions, 15 deletions

diff --git a/chromium/third_party/boringssl/linux-x86_64/crypto/fipsmodule/rsaz-avx2.S b/chromium/third_party/boringssl/linux-x86_64/crypto/fipsmodule/rsaz-avx2.S
index cd3048260e1..e6db7f6ef59 100644
--- a/chromium/third_party/boringssl/linux-x86_64/crypto/fipsmodule/rsaz-avx2.S
+++ b/chromium/third_party/boringssl/linux-x86_64/crypto/fipsmodule/rsaz-avx2.S
@@ -77,7 +77,7 @@ rsaz_1024_sqr_avx2:
 	vmovdqu	256-128(%rsi),%ymm8
 
 	leaq	192(%rsp),%rbx
-	vpbroadcastq	.Land_mask(%rip),%ymm15
+	vmovdqu	.Land_mask(%rip),%ymm15
 	jmp	.LOOP_GRANDE_SQR_1024
 
 .align	32
@@ -829,10 +829,10 @@ rsaz_1024_mul_avx2:
 	vpmuludq	192-128(%rcx),%ymm11,%ymm12
 	vpaddq	%ymm12,%ymm6,%ymm6
 	vpmuludq	224-128(%rcx),%ymm11,%ymm13
-	vpblendd	$3,%ymm14,%ymm9,%ymm9
+	vpblendd	$3,%ymm14,%ymm9,%ymm12
 	vpaddq	%ymm13,%ymm7,%ymm7
 	vpmuludq	256-128(%rcx),%ymm11,%ymm0
-	vpaddq	%ymm9,%ymm3,%ymm3
+	vpaddq	%ymm12,%ymm3,%ymm3
 	vpaddq	%ymm0,%ymm8,%ymm8
 
 	movq	%rbx,%rax
@@ -845,7 +845,9 @@ rsaz_1024_mul_avx2:
 	vmovdqu	-8+64-128(%rsi),%ymm13
 
 	movq	%r10,%rax
+	vpblendd	$0xfc,%ymm14,%ymm9,%ymm9
 	imull	%r8d,%eax
+	vpaddq	%ymm9,%ymm4,%ymm4
 	andl	$0x1fffffff,%eax
 
 	imulq	16-128(%rsi),%rbx
@@ -1074,7 +1076,6 @@ rsaz_1024_mul_avx2:
 
 	decl	%r14d
 	jnz	.Loop_mul_1024
-	vpermq	$0,%ymm15,%ymm15
 	vpaddq	(%rsp),%ymm12,%ymm0
 
 	vpsrlq	$29,%ymm0,%ymm12
@@ -1735,7 +1736,7 @@ rsaz_avx2_eligible:
 
 .align	64
 .Land_mask:
-.quad	0x1fffffff,0x1fffffff,0x1fffffff,-1
+.quad	0x1fffffff,0x1fffffff,0x1fffffff,0x1fffffff
 .Lscatter_permd:
 .long	0,2,4,6,7,7,7,7
 .Lgather_permd:
diff --git a/chromium/third_party/boringssl/mac-x86_64/crypto/fipsmodule/rsaz-avx2.S b/chromium/third_party/boringssl/mac-x86_64/crypto/fipsmodule/rsaz-avx2.S
index d5202250eb4..c82c9d6f948 100644
--- a/chromium/third_party/boringssl/mac-x86_64/crypto/fipsmodule/rsaz-avx2.S
+++ b/chromium/third_party/boringssl/mac-x86_64/crypto/fipsmodule/rsaz-avx2.S
@@ -77,7 +77,7 @@ L$sqr_1024_no_n_copy:
 	vmovdqu	256-128(%rsi),%ymm8
 
 	leaq	192(%rsp),%rbx
-	vpbroadcastq	L$and_mask(%rip),%ymm15
+	vmovdqu	L$and_mask(%rip),%ymm15
 	jmp	L$OOP_GRANDE_SQR_1024
 
 .p2align	5
@@ -829,10 +829,10 @@ L$oop_mul_1024:
 	vpmuludq	192-128(%rcx),%ymm11,%ymm12
 	vpaddq	%ymm12,%ymm6,%ymm6
 	vpmuludq	224-128(%rcx),%ymm11,%ymm13
-	vpblendd	$3,%ymm14,%ymm9,%ymm9
+	vpblendd	$3,%ymm14,%ymm9,%ymm12
 	vpaddq	%ymm13,%ymm7,%ymm7
 	vpmuludq	256-128(%rcx),%ymm11,%ymm0
-	vpaddq	%ymm9,%ymm3,%ymm3
+	vpaddq	%ymm12,%ymm3,%ymm3
 	vpaddq	%ymm0,%ymm8,%ymm8
 
 	movq	%rbx,%rax
@@ -845,7 +845,9 @@ L$oop_mul_1024:
 	vmovdqu	-8+64-128(%rsi),%ymm13
 
 	movq	%r10,%rax
+	vpblendd	$0xfc,%ymm14,%ymm9,%ymm9
 	imull	%r8d,%eax
+	vpaddq	%ymm9,%ymm4,%ymm4
 	andl	$0x1fffffff,%eax
 
 	imulq	16-128(%rsi),%rbx
@@ -1074,7 +1076,6 @@ L$oop_mul_1024:
 
 	decl	%r14d
 	jnz	L$oop_mul_1024
-	vpermq	$0,%ymm15,%ymm15
 	vpaddq	(%rsp),%ymm12,%ymm0
 
 	vpsrlq	$29,%ymm0,%ymm12
@@ -1734,7 +1735,7 @@ _rsaz_avx2_eligible:
 
 .p2align	6
 L$and_mask:
-.quad	0x1fffffff,0x1fffffff,0x1fffffff,-1
+.quad	0x1fffffff,0x1fffffff,0x1fffffff,0x1fffffff
 L$scatter_permd:
 .long	0,2,4,6,7,7,7,7
 L$gather_permd:
diff --git a/chromium/third_party/boringssl/win-x86_64/crypto/fipsmodule/rsaz-avx2.asm b/chromium/third_party/boringssl/win-x86_64/crypto/fipsmodule/rsaz-avx2.asm
index 05b562e2de4..0c6f6e1f76f 100644
--- a/chromium/third_party/boringssl/win-x86_64/crypto/fipsmodule/rsaz-avx2.asm
+++ b/chromium/third_party/boringssl/win-x86_64/crypto/fipsmodule/rsaz-avx2.asm
@@ -103,7 +103,7 @@ $L$sqr_1024_no_n_copy:
 	vmovdqu	ymm8,YMMWORD[((256-128))+rsi]
 
 	lea	rbx,[192+rsp]
-	vpbroadcastq	ymm15,QWORD[$L$and_mask]
+	vmovdqu	ymm15,YMMWORD[$L$and_mask]
 	jmp	NEAR $L$OOP_GRANDE_SQR_1024
 
 ALIGN	32
@@ -891,10 +891,10 @@ $L$oop_mul_1024:
 	vpmuludq	ymm12,ymm11,YMMWORD[((192-128))+rcx]
 	vpaddq	ymm6,ymm6,ymm12
 	vpmuludq	ymm13,ymm11,YMMWORD[((224-128))+rcx]
-	vpblendd	ymm9,ymm9,ymm14,3
+	vpblendd	ymm12,ymm9,ymm14,3
 	vpaddq	ymm7,ymm7,ymm13
 	vpmuludq	ymm0,ymm11,YMMWORD[((256-128))+rcx]
-	vpaddq	ymm3,ymm3,ymm9
+	vpaddq	ymm3,ymm3,ymm12
 	vpaddq	ymm8,ymm8,ymm0
 
 	mov	rax,rbx
@@ -907,7 +907,9 @@ $L$oop_mul_1024:
 	vmovdqu	ymm13,YMMWORD[((-8+64-128))+rsi]
 
 	mov	rax,r10
+	vpblendd	ymm9,ymm9,ymm14,0xfc
 	imul	eax,r8d
+	vpaddq	ymm4,ymm4,ymm9
 	and	eax,0x1fffffff
 
 	imul	rbx,QWORD[((16-128))+rsi]
@@ -1136,7 +1138,6 @@ $L$oop_mul_1024:
 
 	dec	r14d
 	jnz	NEAR $L$oop_mul_1024
-	vpermq	ymm15,ymm15,0
 	vpaddq	ymm0,ymm12,YMMWORD[rsp]
 
 	vpsrlq	ymm12,ymm0,29
@@ -1828,7 +1829,7 @@ rsaz_avx2_eligible:
 
 ALIGN	64
 $L$and_mask:
-	DQ	0x1fffffff,0x1fffffff,0x1fffffff,-1
+	DQ	0x1fffffff,0x1fffffff,0x1fffffff,0x1fffffff
 $L$scatter_permd:
 	DD	0,2,4,6,7,7,7,7
 $L$gather_permd: