diff mbox series

[RFC,11/20] crypto: BLAKE2s - x86_64 implementation

Message ID 20190929173850.26055-12-ard.biesheuvel@linaro.org (mailing list archive)
State New, archived
Headers show
Series crypto: wireguard with crypto API library interface | expand

Commit Message

Ard Biesheuvel Sept. 29, 2019, 5:38 p.m. UTC 
From: "Jason A. Donenfeld" <Jason@zx2c4.com>

These implementations from Samuel Neves support AVX and AVX-512VL.
Originally this used AVX-512F, but Skylake thermal throttling made
AVX-512VL more attractive and possible to do with negligable difference.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Samuel Neves <sneves@dei.uc.pt>
Co-developed-by: Samuel Neves <sneves@dei.uc.pt>
[ardb: move to arch/x86/crypto, wire into lib/crypto framework]
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
---
 arch/x86/crypto/Makefile       |   2 +
 arch/x86/crypto/blake2s-core.S | 685 ++++++++++++++++++++
 arch/x86/crypto/blake2s-glue.c |  73 +++
 crypto/Kconfig                 |   6 +
 4 files changed, 766 insertions(+)

Comments

Jason A. Donenfeld Sept. 30, 2019, 2:42 a.m. UTC | #1 
Hi Sebastian, Thomas,

Take a look at the below snippet from this patch.

I had previously put quite some effort into the simd_get, simd_put,
simd_relax mechanism, so that the simd state could be persisted during
both several calls to the same function and within long loops like
below, with simd_relax existing to reenable preemption briefly if
things were getting out of hand. Ard got rid of this and has moved the
kernel_fpu_begin and kernel_fpu_end calls into the inner loop:

On Sun, Sep 29, 2019 at 7:39 PM Ard Biesheuvel
<ard.biesheuvel@linaro.org> wrote:
> +       for (;;) {
> +               const size_t blocks = min_t(size_t, nblocks,
> +                                           PAGE_SIZE / BLAKE2S_BLOCK_SIZE);
> +
> +               kernel_fpu_begin();
> +               if (IS_ENABLED(CONFIG_AS_AVX512) && blake2s_use_avx512)
> +                       blake2s_compress_avx512(state, block, blocks, inc);
> +               else
> +                       blake2s_compress_avx(state, block, blocks, inc);
> +               kernel_fpu_end();
> +
> +               nblocks -= blocks;
> +               if (!nblocks)
> +                       break;
> +               block += blocks * BLAKE2S_BLOCK_SIZE;
> +       }
> +       return true;
> +}

I'm wondering if on modern kernels this is actually fine and whether
my simd_get/put/relax thing no longer has a good use case.
Specifically, I recall last year there were a lot of patches and
discussions about doing FPU register restoration lazily -- on context
switch or the like. Did those land? Did the theory of action work out
in the end?

Regards,
Jason
Linus Torvalds Sept. 30, 2019, 2:51 a.m. UTC | #2 
On Sun, Sep 29, 2019 at 7:42 PM Jason A. Donenfeld <Jason@zx2c4.com> wrote:
>
> I had previously put quite some effort into the simd_get, simd_put,
> simd_relax mechanism, so that the simd state could be persisted during
> both several calls to the same function and within long loops like
> below, with simd_relax existing to reenable preemption briefly if
> things were getting out of hand. Ard got rid of this and has moved the
> kernel_fpu_begin and kernel_fpu_end calls into the inner loop:

Actually, that should be ok these days.

What has happened fairly recently (it got merged into 5.2 back in May)
is that we no longer do the FPU save/restore on each
kernel_fpu_begin/end.

Instead, we save it on kernel_fpu_begin(), and set a flag that it
needs to be restored when returning to user space.

So the kernel now on its own merges that FPU save/restore overhead
when you do it repeatedly.

The core change happened in

     5f409e20b794 ("x86/fpu: Defer FPU state load until return to userspace")

but there are a few commits around it for cleanups etc. The code was merged in

     8ff468c29e9a ("Merge branch 'x86-fpu-for-linus' of
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip")

if you want to see the whole series.

That said, it would be _lovely_ if you or somebody else actually
double-checked that it works as expected and that the numbers bear out
the improvements.

It should be superior to the old model of manually trying to merge FPU
use regions, both from a performance angle (because it will merge much
more), but also from a code simplicity angle and the whole preemption
latency worry also basically goes away.

             Linus
Sebastian Andrzej Siewior Sept. 30, 2019, 7:35 a.m. UTC | #3 
On 2019-09-30 04:42:06 [+0200], Jason A. Donenfeld wrote:
> Hi Sebastian, Thomas,
Hi Jason,

> On Sun, Sep 29, 2019 at 7:39 PM Ard Biesheuvel
> <ard.biesheuvel@linaro.org> wrote:
> > +       for (;;) {
> > +               const size_t blocks = min_t(size_t, nblocks,
> > +                                           PAGE_SIZE / BLAKE2S_BLOCK_SIZE);
> > +
> > +               kernel_fpu_begin();
> > +               if (IS_ENABLED(CONFIG_AS_AVX512) && blake2s_use_avx512)
> > +                       blake2s_compress_avx512(state, block, blocks, inc);
> > +               else
> > +                       blake2s_compress_avx(state, block, blocks, inc);
> > +               kernel_fpu_end();
> > +
> > +               nblocks -= blocks;
> > +               if (!nblocks)
> > +                       break;
> > +               block += blocks * BLAKE2S_BLOCK_SIZE;
> > +       }
> > +       return true;
> > +}
> 
> I'm wondering if on modern kernels this is actually fine and whether
> my simd_get/put/relax thing no longer has a good use case.
> Specifically, I recall last year there were a lot of patches and
> discussions about doing FPU register restoration lazily -- on context
> switch or the like. Did those land? Did the theory of action work out
> in the end?

That optimisation landed in v5.2. With that change (on x86)
kernel_fpu_end() does almost nothing (and so the following
kernel_fpu_begin()) and the FPU context will be restored once the task
returns to user land. (Note that this counts only for user-tasks because
we don't save the FPU state of a kernel thread.)
I haven't look at crypto code since that change was merged but looking
at the snippet from Ard is actually what I was aiming for.

> Regards,
> Jason

Sebastian
diff mbox series

Patch

diff --git a/arch/x86/crypto/Makefile b/arch/x86/crypto/Makefile
index 759b1a927826..db87c182ce0f 100644
--- a/arch/x86/crypto/Makefile
+++ b/arch/x86/crypto/Makefile
@@ -48,6 +48,7 @@  ifeq ($(avx_supported),yes)
 	obj-$(CONFIG_CRYPTO_CAST6_AVX_X86_64) += cast6-avx-x86_64.o
 	obj-$(CONFIG_CRYPTO_TWOFISH_AVX_X86_64) += twofish-avx-x86_64.o
 	obj-$(CONFIG_CRYPTO_SERPENT_AVX_X86_64) += serpent-avx-x86_64.o
+	obj-$(CONFIG_CRYPTO_LIB_BLAKE2S_X86) += blake2s-x86_64.o
 endif
 
 # These modules require assembler to support AVX2.
@@ -70,6 +71,7 @@  serpent-sse2-x86_64-y := serpent-sse2-x86_64-asm_64.o serpent_sse2_glue.o
 aegis128-aesni-y := aegis128-aesni-asm.o aegis128-aesni-glue.o
 
 nhpoly1305-sse2-y := nh-sse2-x86_64.o nhpoly1305-sse2-glue.o
+blake2s-x86_64-y := blake2s-core.o blake2s-glue.o
 
 ifeq ($(avx_supported),yes)
 	camellia-aesni-avx-x86_64-y := camellia-aesni-avx-asm_64.o \
diff --git a/arch/x86/crypto/blake2s-core.S b/arch/x86/crypto/blake2s-core.S
new file mode 100644
index 000000000000..675288fa4cca
--- /dev/null
+++ b/arch/x86/crypto/blake2s-core.S
@@ -0,0 +1,685 @@ 
+/* SPDX-License-Identifier: GPL-2.0 OR MIT */
+/*
+ * Copyright (C) 2015-2019 Jason A. Donenfeld <Jason@zx2c4.com>. All Rights Reserved.
+ * Copyright (C) 2017 Samuel Neves <sneves@dei.uc.pt>. All Rights Reserved.
+ */
+
+#include <linux/linkage.h>
+
+.section .rodata.cst32.BLAKE2S_IV, "aM", @progbits, 32
+.align 32
+IV:	.octa 0xA54FF53A3C6EF372BB67AE856A09E667
+	.octa 0x5BE0CD191F83D9AB9B05688C510E527F
+.section .rodata.cst16.ROT16, "aM", @progbits, 16
+.align 16
+ROT16:	.octa 0x0D0C0F0E09080B0A0504070601000302
+.section .rodata.cst16.ROR328, "aM", @progbits, 16
+.align 16
+ROR328:	.octa 0x0C0F0E0D080B0A090407060500030201
+#ifdef CONFIG_AS_AVX512
+.section .rodata.cst64.BLAKE2S_SIGMA, "aM", @progbits, 640
+.align 64
+SIGMA:
+.long 0, 2, 4, 6, 1, 3, 5, 7, 8, 10, 12, 14, 9, 11, 13, 15
+.long 11, 2, 12, 14, 9, 8, 15, 3, 4, 0, 13, 6, 10, 1, 7, 5
+.long 10, 12, 11, 6, 5, 9, 13, 3, 4, 15, 14, 2, 0, 7, 8, 1
+.long 10, 9, 7, 0, 11, 14, 1, 12, 6, 2, 15, 3, 13, 8, 5, 4
+.long 4, 9, 8, 13, 14, 0, 10, 11, 7, 3, 12, 1, 5, 6, 15, 2
+.long 2, 10, 4, 14, 13, 3, 9, 11, 6, 5, 7, 12, 15, 1, 8, 0
+.long 4, 11, 14, 8, 13, 10, 12, 5, 2, 1, 15, 3, 9, 7, 0, 6
+.long 6, 12, 0, 13, 15, 2, 1, 10, 4, 5, 11, 14, 8, 3, 9, 7
+.long 14, 5, 4, 12, 9, 7, 3, 10, 2, 0, 6, 15, 11, 1, 13, 8
+.long 11, 7, 13, 10, 12, 14, 0, 15, 4, 5, 6, 9, 2, 1, 8, 3
+#endif /* CONFIG_AS_AVX512 */
+
+.text
+#ifdef CONFIG_AS_AVX
+ENTRY(blake2s_compress_avx)
+	movl		%ecx, %ecx
+	testq		%rdx, %rdx
+	je		.Lendofloop
+	.align 32
+.Lbeginofloop:
+	addq		%rcx, 32(%rdi)
+	vmovdqu		IV+16(%rip), %xmm1
+	vmovdqu		(%rsi), %xmm4
+	vpxor		32(%rdi), %xmm1, %xmm1
+	vmovdqu		16(%rsi), %xmm3
+	vshufps		$136, %xmm3, %xmm4, %xmm6
+	vmovdqa		ROT16(%rip), %xmm7
+	vpaddd		(%rdi), %xmm6, %xmm6
+	vpaddd		16(%rdi), %xmm6, %xmm6
+	vpxor		%xmm6, %xmm1, %xmm1
+	vmovdqu		IV(%rip), %xmm8
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vmovdqu		48(%rsi), %xmm5
+	vpaddd		%xmm1, %xmm8, %xmm8
+	vpxor		16(%rdi), %xmm8, %xmm9
+	vmovdqu		32(%rsi), %xmm2
+	vpblendw	$12, %xmm3, %xmm5, %xmm13
+	vshufps		$221, %xmm5, %xmm2, %xmm12
+	vpunpckhqdq	%xmm2, %xmm4, %xmm14
+	vpslld		$20, %xmm9, %xmm0
+	vpsrld		$12, %xmm9, %xmm9
+	vpxor		%xmm0, %xmm9, %xmm0
+	vshufps		$221, %xmm3, %xmm4, %xmm9
+	vpaddd		%xmm9, %xmm6, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vmovdqa		ROR328(%rip), %xmm6
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm8, %xmm8
+	vpxor		%xmm8, %xmm0, %xmm0
+	vpshufd		$147, %xmm1, %xmm1
+	vpshufd		$78, %xmm8, %xmm8
+	vpslld		$25, %xmm0, %xmm10
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm10, %xmm0, %xmm0
+	vshufps		$136, %xmm5, %xmm2, %xmm10
+	vpshufd		$57, %xmm0, %xmm0
+	vpaddd		%xmm10, %xmm9, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpaddd		%xmm12, %xmm9, %xmm9
+	vpblendw	$12, %xmm2, %xmm3, %xmm12
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm8, %xmm8
+	vpxor		%xmm8, %xmm0, %xmm10
+	vpslld		$20, %xmm10, %xmm0
+	vpsrld		$12, %xmm10, %xmm10
+	vpxor		%xmm0, %xmm10, %xmm0
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm8, %xmm8
+	vpxor		%xmm8, %xmm0, %xmm0
+	vpshufd		$57, %xmm1, %xmm1
+	vpshufd		$78, %xmm8, %xmm8
+	vpslld		$25, %xmm0, %xmm10
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm10, %xmm0, %xmm0
+	vpslldq		$4, %xmm5, %xmm10
+	vpblendw	$240, %xmm10, %xmm12, %xmm12
+	vpshufd		$147, %xmm0, %xmm0
+	vpshufd		$147, %xmm12, %xmm12
+	vpaddd		%xmm9, %xmm12, %xmm12
+	vpaddd		%xmm0, %xmm12, %xmm12
+	vpxor		%xmm12, %xmm1, %xmm1
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm8, %xmm8
+	vpxor		%xmm8, %xmm0, %xmm11
+	vpslld		$20, %xmm11, %xmm9
+	vpsrld		$12, %xmm11, %xmm11
+	vpxor		%xmm9, %xmm11, %xmm0
+	vpshufd		$8, %xmm2, %xmm9
+	vpblendw	$192, %xmm5, %xmm3, %xmm11
+	vpblendw	$240, %xmm11, %xmm9, %xmm9
+	vpshufd		$177, %xmm9, %xmm9
+	vpaddd		%xmm12, %xmm9, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm11
+	vpxor		%xmm11, %xmm1, %xmm1
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm8, %xmm8
+	vpxor		%xmm8, %xmm0, %xmm9
+	vpshufd		$147, %xmm1, %xmm1
+	vpshufd		$78, %xmm8, %xmm8
+	vpslld		$25, %xmm9, %xmm0
+	vpsrld		$7, %xmm9, %xmm9
+	vpxor		%xmm0, %xmm9, %xmm0
+	vpslldq		$4, %xmm3, %xmm9
+	vpblendw	$48, %xmm9, %xmm2, %xmm9
+	vpblendw	$240, %xmm9, %xmm4, %xmm9
+	vpshufd		$57, %xmm0, %xmm0
+	vpshufd		$177, %xmm9, %xmm9
+	vpaddd		%xmm11, %xmm9, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm8, %xmm11
+	vpxor		%xmm11, %xmm0, %xmm0
+	vpslld		$20, %xmm0, %xmm8
+	vpsrld		$12, %xmm0, %xmm0
+	vpxor		%xmm8, %xmm0, %xmm0
+	vpunpckhdq	%xmm3, %xmm4, %xmm8
+	vpblendw	$12, %xmm10, %xmm8, %xmm12
+	vpshufd		$177, %xmm12, %xmm12
+	vpaddd		%xmm9, %xmm12, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm11, %xmm11
+	vpxor		%xmm11, %xmm0, %xmm0
+	vpshufd		$57, %xmm1, %xmm1
+	vpshufd		$78, %xmm11, %xmm11
+	vpslld		$25, %xmm0, %xmm12
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm12, %xmm0, %xmm0
+	vpunpckhdq	%xmm5, %xmm2, %xmm12
+	vpshufd		$147, %xmm0, %xmm0
+	vpblendw	$15, %xmm13, %xmm12, %xmm12
+	vpslldq		$8, %xmm5, %xmm13
+	vpshufd		$210, %xmm12, %xmm12
+	vpaddd		%xmm9, %xmm12, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm11, %xmm11
+	vpxor		%xmm11, %xmm0, %xmm0
+	vpslld		$20, %xmm0, %xmm12
+	vpsrld		$12, %xmm0, %xmm0
+	vpxor		%xmm12, %xmm0, %xmm0
+	vpunpckldq	%xmm4, %xmm2, %xmm12
+	vpblendw	$240, %xmm4, %xmm12, %xmm12
+	vpblendw	$192, %xmm13, %xmm12, %xmm12
+	vpsrldq		$12, %xmm3, %xmm13
+	vpaddd		%xmm12, %xmm9, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm11, %xmm11
+	vpxor		%xmm11, %xmm0, %xmm0
+	vpshufd		$147, %xmm1, %xmm1
+	vpshufd		$78, %xmm11, %xmm11
+	vpslld		$25, %xmm0, %xmm12
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm12, %xmm0, %xmm0
+	vpblendw	$60, %xmm2, %xmm4, %xmm12
+	vpblendw	$3, %xmm13, %xmm12, %xmm12
+	vpshufd		$57, %xmm0, %xmm0
+	vpshufd		$78, %xmm12, %xmm12
+	vpaddd		%xmm9, %xmm12, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm11, %xmm11
+	vpxor		%xmm11, %xmm0, %xmm12
+	vpslld		$20, %xmm12, %xmm13
+	vpsrld		$12, %xmm12, %xmm0
+	vpblendw	$51, %xmm3, %xmm4, %xmm12
+	vpxor		%xmm13, %xmm0, %xmm0
+	vpblendw	$192, %xmm10, %xmm12, %xmm10
+	vpslldq		$8, %xmm2, %xmm12
+	vpshufd		$27, %xmm10, %xmm10
+	vpaddd		%xmm9, %xmm10, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm11, %xmm11
+	vpxor		%xmm11, %xmm0, %xmm0
+	vpshufd		$57, %xmm1, %xmm1
+	vpshufd		$78, %xmm11, %xmm11
+	vpslld		$25, %xmm0, %xmm10
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm10, %xmm0, %xmm0
+	vpunpckhdq	%xmm2, %xmm8, %xmm10
+	vpshufd		$147, %xmm0, %xmm0
+	vpblendw	$12, %xmm5, %xmm10, %xmm10
+	vpshufd		$210, %xmm10, %xmm10
+	vpaddd		%xmm9, %xmm10, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm11, %xmm11
+	vpxor		%xmm11, %xmm0, %xmm10
+	vpslld		$20, %xmm10, %xmm0
+	vpsrld		$12, %xmm10, %xmm10
+	vpxor		%xmm0, %xmm10, %xmm0
+	vpblendw	$12, %xmm4, %xmm5, %xmm10
+	vpblendw	$192, %xmm12, %xmm10, %xmm10
+	vpunpckldq	%xmm2, %xmm4, %xmm12
+	vpshufd		$135, %xmm10, %xmm10
+	vpaddd		%xmm9, %xmm10, %xmm9
+	vpaddd		%xmm0, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm1, %xmm1
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm11, %xmm13
+	vpxor		%xmm13, %xmm0, %xmm0
+	vpshufd		$147, %xmm1, %xmm1
+	vpshufd		$78, %xmm13, %xmm13
+	vpslld		$25, %xmm0, %xmm10
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm10, %xmm0, %xmm0
+	vpblendw	$15, %xmm3, %xmm4, %xmm10
+	vpblendw	$192, %xmm5, %xmm10, %xmm10
+	vpshufd		$57, %xmm0, %xmm0
+	vpshufd		$198, %xmm10, %xmm10
+	vpaddd		%xmm9, %xmm10, %xmm10
+	vpaddd		%xmm0, %xmm10, %xmm10
+	vpxor		%xmm10, %xmm1, %xmm1
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm13, %xmm13
+	vpxor		%xmm13, %xmm0, %xmm9
+	vpslld		$20, %xmm9, %xmm0
+	vpsrld		$12, %xmm9, %xmm9
+	vpxor		%xmm0, %xmm9, %xmm0
+	vpunpckhdq	%xmm2, %xmm3, %xmm9
+	vpunpcklqdq	%xmm12, %xmm9, %xmm15
+	vpunpcklqdq	%xmm12, %xmm8, %xmm12
+	vpblendw	$15, %xmm5, %xmm8, %xmm8
+	vpaddd		%xmm15, %xmm10, %xmm15
+	vpaddd		%xmm0, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm1, %xmm1
+	vpshufd		$141, %xmm8, %xmm8
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm13, %xmm13
+	vpxor		%xmm13, %xmm0, %xmm0
+	vpshufd		$57, %xmm1, %xmm1
+	vpshufd		$78, %xmm13, %xmm13
+	vpslld		$25, %xmm0, %xmm10
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm10, %xmm0, %xmm0
+	vpunpcklqdq	%xmm2, %xmm3, %xmm10
+	vpshufd		$147, %xmm0, %xmm0
+	vpblendw	$51, %xmm14, %xmm10, %xmm14
+	vpshufd		$135, %xmm14, %xmm14
+	vpaddd		%xmm15, %xmm14, %xmm14
+	vpaddd		%xmm0, %xmm14, %xmm14
+	vpxor		%xmm14, %xmm1, %xmm1
+	vpunpcklqdq	%xmm3, %xmm4, %xmm15
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm13, %xmm13
+	vpxor		%xmm13, %xmm0, %xmm0
+	vpslld		$20, %xmm0, %xmm11
+	vpsrld		$12, %xmm0, %xmm0
+	vpxor		%xmm11, %xmm0, %xmm0
+	vpunpckhqdq	%xmm5, %xmm3, %xmm11
+	vpblendw	$51, %xmm15, %xmm11, %xmm11
+	vpunpckhqdq	%xmm3, %xmm5, %xmm15
+	vpaddd		%xmm11, %xmm14, %xmm11
+	vpaddd		%xmm0, %xmm11, %xmm11
+	vpxor		%xmm11, %xmm1, %xmm1
+	vpshufb		%xmm6, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm13, %xmm13
+	vpxor		%xmm13, %xmm0, %xmm0
+	vpshufd		$147, %xmm1, %xmm1
+	vpshufd		$78, %xmm13, %xmm13
+	vpslld		$25, %xmm0, %xmm14
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm14, %xmm0, %xmm14
+	vpunpckhqdq	%xmm4, %xmm2, %xmm0
+	vpshufd		$57, %xmm14, %xmm14
+	vpblendw	$51, %xmm15, %xmm0, %xmm15
+	vpaddd		%xmm15, %xmm11, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm1, %xmm1
+	vpshufb		%xmm7, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm13, %xmm13
+	vpxor		%xmm13, %xmm14, %xmm14
+	vpslld		$20, %xmm14, %xmm11
+	vpsrld		$12, %xmm14, %xmm14
+	vpxor		%xmm11, %xmm14, %xmm14
+	vpblendw	$3, %xmm2, %xmm4, %xmm11
+	vpslldq		$8, %xmm11, %xmm0
+	vpblendw	$15, %xmm5, %xmm0, %xmm0
+	vpshufd		$99, %xmm0, %xmm0
+	vpaddd		%xmm15, %xmm0, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm1, %xmm0
+	vpaddd		%xmm12, %xmm15, %xmm15
+	vpshufb		%xmm6, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm13, %xmm13
+	vpxor		%xmm13, %xmm14, %xmm14
+	vpshufd		$57, %xmm0, %xmm0
+	vpshufd		$78, %xmm13, %xmm13
+	vpslld		$25, %xmm14, %xmm1
+	vpsrld		$7, %xmm14, %xmm14
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpblendw	$3, %xmm5, %xmm4, %xmm1
+	vpshufd		$147, %xmm14, %xmm14
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpshufb		%xmm7, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm13, %xmm13
+	vpxor		%xmm13, %xmm14, %xmm14
+	vpslld		$20, %xmm14, %xmm12
+	vpsrld		$12, %xmm14, %xmm14
+	vpxor		%xmm12, %xmm14, %xmm14
+	vpsrldq		$4, %xmm2, %xmm12
+	vpblendw	$60, %xmm12, %xmm1, %xmm1
+	vpaddd		%xmm1, %xmm15, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpblendw	$12, %xmm4, %xmm3, %xmm1
+	vpshufb		%xmm6, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm13, %xmm13
+	vpxor		%xmm13, %xmm14, %xmm14
+	vpshufd		$147, %xmm0, %xmm0
+	vpshufd		$78, %xmm13, %xmm13
+	vpslld		$25, %xmm14, %xmm12
+	vpsrld		$7, %xmm14, %xmm14
+	vpxor		%xmm12, %xmm14, %xmm14
+	vpsrldq		$4, %xmm5, %xmm12
+	vpblendw	$48, %xmm12, %xmm1, %xmm1
+	vpshufd		$33, %xmm5, %xmm12
+	vpshufd		$57, %xmm14, %xmm14
+	vpshufd		$108, %xmm1, %xmm1
+	vpblendw	$51, %xmm12, %xmm10, %xmm12
+	vpaddd		%xmm15, %xmm1, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpaddd		%xmm12, %xmm15, %xmm15
+	vpshufb		%xmm7, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm13, %xmm1
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpslld		$20, %xmm14, %xmm13
+	vpsrld		$12, %xmm14, %xmm14
+	vpxor		%xmm13, %xmm14, %xmm14
+	vpslldq		$12, %xmm3, %xmm13
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpshufb		%xmm6, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpshufd		$57, %xmm0, %xmm0
+	vpshufd		$78, %xmm1, %xmm1
+	vpslld		$25, %xmm14, %xmm12
+	vpsrld		$7, %xmm14, %xmm14
+	vpxor		%xmm12, %xmm14, %xmm14
+	vpblendw	$51, %xmm5, %xmm4, %xmm12
+	vpshufd		$147, %xmm14, %xmm14
+	vpblendw	$192, %xmm13, %xmm12, %xmm12
+	vpaddd		%xmm12, %xmm15, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpsrldq		$4, %xmm3, %xmm12
+	vpshufb		%xmm7, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpslld		$20, %xmm14, %xmm13
+	vpsrld		$12, %xmm14, %xmm14
+	vpxor		%xmm13, %xmm14, %xmm14
+	vpblendw	$48, %xmm2, %xmm5, %xmm13
+	vpblendw	$3, %xmm12, %xmm13, %xmm13
+	vpshufd		$156, %xmm13, %xmm13
+	vpaddd		%xmm15, %xmm13, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpshufb		%xmm6, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpshufd		$147, %xmm0, %xmm0
+	vpshufd		$78, %xmm1, %xmm1
+	vpslld		$25, %xmm14, %xmm13
+	vpsrld		$7, %xmm14, %xmm14
+	vpxor		%xmm13, %xmm14, %xmm14
+	vpunpcklqdq	%xmm2, %xmm4, %xmm13
+	vpshufd		$57, %xmm14, %xmm14
+	vpblendw	$12, %xmm12, %xmm13, %xmm12
+	vpshufd		$180, %xmm12, %xmm12
+	vpaddd		%xmm15, %xmm12, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpshufb		%xmm7, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpslld		$20, %xmm14, %xmm12
+	vpsrld		$12, %xmm14, %xmm14
+	vpxor		%xmm12, %xmm14, %xmm14
+	vpunpckhqdq	%xmm9, %xmm4, %xmm12
+	vpshufd		$198, %xmm12, %xmm12
+	vpaddd		%xmm15, %xmm12, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpaddd		%xmm15, %xmm8, %xmm15
+	vpshufb		%xmm6, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpshufd		$57, %xmm0, %xmm0
+	vpshufd		$78, %xmm1, %xmm1
+	vpslld		$25, %xmm14, %xmm12
+	vpsrld		$7, %xmm14, %xmm14
+	vpxor		%xmm12, %xmm14, %xmm14
+	vpsrldq		$4, %xmm4, %xmm12
+	vpshufd		$147, %xmm14, %xmm14
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm15, %xmm0, %xmm0
+	vpshufb		%xmm7, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpslld		$20, %xmm14, %xmm8
+	vpsrld		$12, %xmm14, %xmm14
+	vpxor		%xmm14, %xmm8, %xmm14
+	vpblendw	$48, %xmm5, %xmm2, %xmm8
+	vpblendw	$3, %xmm12, %xmm8, %xmm8
+	vpunpckhqdq	%xmm5, %xmm4, %xmm12
+	vpshufd		$75, %xmm8, %xmm8
+	vpblendw	$60, %xmm10, %xmm12, %xmm10
+	vpaddd		%xmm15, %xmm8, %xmm15
+	vpaddd		%xmm14, %xmm15, %xmm15
+	vpxor		%xmm0, %xmm15, %xmm0
+	vpshufd		$45, %xmm10, %xmm10
+	vpshufb		%xmm6, %xmm0, %xmm0
+	vpaddd		%xmm15, %xmm10, %xmm15
+	vpaddd		%xmm0, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm14, %xmm14
+	vpshufd		$147, %xmm0, %xmm0
+	vpshufd		$78, %xmm1, %xmm1
+	vpslld		$25, %xmm14, %xmm8
+	vpsrld		$7, %xmm14, %xmm14
+	vpxor		%xmm14, %xmm8, %xmm8
+	vpshufd		$57, %xmm8, %xmm8
+	vpaddd		%xmm8, %xmm15, %xmm15
+	vpxor		%xmm0, %xmm15, %xmm0
+	vpshufb		%xmm7, %xmm0, %xmm0
+	vpaddd		%xmm0, %xmm1, %xmm1
+	vpxor		%xmm8, %xmm1, %xmm8
+	vpslld		$20, %xmm8, %xmm10
+	vpsrld		$12, %xmm8, %xmm8
+	vpxor		%xmm8, %xmm10, %xmm10
+	vpunpckldq	%xmm3, %xmm4, %xmm8
+	vpunpcklqdq	%xmm9, %xmm8, %xmm9
+	vpaddd		%xmm9, %xmm15, %xmm9
+	vpaddd		%xmm10, %xmm9, %xmm9
+	vpxor		%xmm0, %xmm9, %xmm8
+	vpshufb		%xmm6, %xmm8, %xmm8
+	vpaddd		%xmm8, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm10, %xmm10
+	vpshufd		$57, %xmm8, %xmm8
+	vpshufd		$78, %xmm1, %xmm1
+	vpslld		$25, %xmm10, %xmm12
+	vpsrld		$7, %xmm10, %xmm10
+	vpxor		%xmm10, %xmm12, %xmm10
+	vpblendw	$48, %xmm4, %xmm3, %xmm12
+	vpshufd		$147, %xmm10, %xmm0
+	vpunpckhdq	%xmm5, %xmm3, %xmm10
+	vpshufd		$78, %xmm12, %xmm12
+	vpunpcklqdq	%xmm4, %xmm10, %xmm10
+	vpblendw	$192, %xmm2, %xmm10, %xmm10
+	vpshufhw	$78, %xmm10, %xmm10
+	vpaddd		%xmm10, %xmm9, %xmm10
+	vpaddd		%xmm0, %xmm10, %xmm10
+	vpxor		%xmm8, %xmm10, %xmm8
+	vpshufb		%xmm7, %xmm8, %xmm8
+	vpaddd		%xmm8, %xmm1, %xmm1
+	vpxor		%xmm0, %xmm1, %xmm9
+	vpslld		$20, %xmm9, %xmm0
+	vpsrld		$12, %xmm9, %xmm9
+	vpxor		%xmm9, %xmm0, %xmm0
+	vpunpckhdq	%xmm5, %xmm4, %xmm9
+	vpblendw	$240, %xmm9, %xmm2, %xmm13
+	vpshufd		$39, %xmm13, %xmm13
+	vpaddd		%xmm10, %xmm13, %xmm10
+	vpaddd		%xmm0, %xmm10, %xmm10
+	vpxor		%xmm8, %xmm10, %xmm8
+	vpblendw	$12, %xmm4, %xmm2, %xmm13
+	vpshufb		%xmm6, %xmm8, %xmm8
+	vpslldq		$4, %xmm13, %xmm13
+	vpblendw	$15, %xmm5, %xmm13, %xmm13
+	vpaddd		%xmm8, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm0, %xmm0
+	vpaddd		%xmm13, %xmm10, %xmm13
+	vpshufd		$147, %xmm8, %xmm8
+	vpshufd		$78, %xmm1, %xmm1
+	vpslld		$25, %xmm0, %xmm14
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm0, %xmm14, %xmm14
+	vpshufd		$57, %xmm14, %xmm14
+	vpaddd		%xmm14, %xmm13, %xmm13
+	vpxor		%xmm8, %xmm13, %xmm8
+	vpaddd		%xmm13, %xmm12, %xmm12
+	vpshufb		%xmm7, %xmm8, %xmm8
+	vpaddd		%xmm8, %xmm1, %xmm1
+	vpxor		%xmm14, %xmm1, %xmm14
+	vpslld		$20, %xmm14, %xmm10
+	vpsrld		$12, %xmm14, %xmm14
+	vpxor		%xmm14, %xmm10, %xmm10
+	vpaddd		%xmm10, %xmm12, %xmm12
+	vpxor		%xmm8, %xmm12, %xmm8
+	vpshufb		%xmm6, %xmm8, %xmm8
+	vpaddd		%xmm8, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm10, %xmm0
+	vpshufd		$57, %xmm8, %xmm8
+	vpshufd		$78, %xmm1, %xmm1
+	vpslld		$25, %xmm0, %xmm10
+	vpsrld		$7, %xmm0, %xmm0
+	vpxor		%xmm0, %xmm10, %xmm10
+	vpblendw	$48, %xmm2, %xmm3, %xmm0
+	vpblendw	$15, %xmm11, %xmm0, %xmm0
+	vpshufd		$147, %xmm10, %xmm10
+	vpshufd		$114, %xmm0, %xmm0
+	vpaddd		%xmm12, %xmm0, %xmm0
+	vpaddd		%xmm10, %xmm0, %xmm0
+	vpxor		%xmm8, %xmm0, %xmm8
+	vpshufb		%xmm7, %xmm8, %xmm8
+	vpaddd		%xmm8, %xmm1, %xmm1
+	vpxor		%xmm10, %xmm1, %xmm10
+	vpslld		$20, %xmm10, %xmm11
+	vpsrld		$12, %xmm10, %xmm10
+	vpxor		%xmm10, %xmm11, %xmm10
+	vpslldq		$4, %xmm4, %xmm11
+	vpblendw	$192, %xmm11, %xmm3, %xmm3
+	vpunpckldq	%xmm5, %xmm4, %xmm4
+	vpshufd		$99, %xmm3, %xmm3
+	vpaddd		%xmm0, %xmm3, %xmm3
+	vpaddd		%xmm10, %xmm3, %xmm3
+	vpxor		%xmm8, %xmm3, %xmm11
+	vpunpckldq	%xmm5, %xmm2, %xmm0
+	vpblendw	$192, %xmm2, %xmm5, %xmm2
+	vpshufb		%xmm6, %xmm11, %xmm11
+	vpunpckhqdq	%xmm0, %xmm9, %xmm0
+	vpblendw	$15, %xmm4, %xmm2, %xmm4
+	vpaddd		%xmm11, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm10, %xmm10
+	vpshufd		$147, %xmm11, %xmm11
+	vpshufd		$201, %xmm0, %xmm0
+	vpslld		$25, %xmm10, %xmm8
+	vpsrld		$7, %xmm10, %xmm10
+	vpxor		%xmm10, %xmm8, %xmm10
+	vpshufd		$78, %xmm1, %xmm1
+	vpaddd		%xmm3, %xmm0, %xmm0
+	vpshufd		$27, %xmm4, %xmm4
+	vpshufd		$57, %xmm10, %xmm10
+	vpaddd		%xmm10, %xmm0, %xmm0
+	vpxor		%xmm11, %xmm0, %xmm11
+	vpaddd		%xmm0, %xmm4, %xmm0
+	vpshufb		%xmm7, %xmm11, %xmm7
+	vpaddd		%xmm7, %xmm1, %xmm1
+	vpxor		%xmm10, %xmm1, %xmm10
+	vpslld		$20, %xmm10, %xmm8
+	vpsrld		$12, %xmm10, %xmm10
+	vpxor		%xmm10, %xmm8, %xmm8
+	vpaddd		%xmm8, %xmm0, %xmm0
+	vpxor		%xmm7, %xmm0, %xmm7
+	vpshufb		%xmm6, %xmm7, %xmm6
+	vpaddd		%xmm6, %xmm1, %xmm1
+	vpxor		%xmm1, %xmm8, %xmm8
+	vpshufd		$78, %xmm1, %xmm1
+	vpshufd		$57, %xmm6, %xmm6
+	vpslld		$25, %xmm8, %xmm2
+	vpsrld		$7, %xmm8, %xmm8
+	vpxor		%xmm8, %xmm2, %xmm8
+	vpxor		(%rdi), %xmm1, %xmm1
+	vpshufd		$147, %xmm8, %xmm8
+	vpxor		%xmm0, %xmm1, %xmm0
+	vmovups		%xmm0, (%rdi)
+	vpxor		16(%rdi), %xmm8, %xmm0
+	vpxor		%xmm6, %xmm0, %xmm6
+	vmovups		%xmm6, 16(%rdi)
+	addq		$64, %rsi
+	decq		%rdx
+	jnz .Lbeginofloop
+.Lendofloop:
+	ret
+ENDPROC(blake2s_compress_avx)
+#endif /* CONFIG_AS_AVX */
+
+#ifdef CONFIG_AS_AVX512
+ENTRY(blake2s_compress_avx512)
+	vmovdqu		(%rdi),%xmm0
+	vmovdqu		0x10(%rdi),%xmm1
+	vmovdqu		0x20(%rdi),%xmm4
+	vmovq		%rcx,%xmm5
+	vmovdqa		IV(%rip),%xmm14
+	vmovdqa		IV+16(%rip),%xmm15
+	jmp		.Lblake2s_compress_avx512_mainloop
+.align 32
+.Lblake2s_compress_avx512_mainloop:
+	vmovdqa		%xmm0,%xmm10
+	vmovdqa		%xmm1,%xmm11
+	vpaddq		%xmm5,%xmm4,%xmm4
+	vmovdqa		%xmm14,%xmm2
+	vpxor		%xmm15,%xmm4,%xmm3
+	vmovdqu		(%rsi),%ymm6
+	vmovdqu		0x20(%rsi),%ymm7
+	addq		$0x40,%rsi
+	leaq		SIGMA(%rip),%rax
+	movb		$0xa,%cl
+.Lblake2s_compress_avx512_roundloop:
+	addq		$0x40,%rax
+	vmovdqa		-0x40(%rax),%ymm8
+	vmovdqa		-0x20(%rax),%ymm9
+	vpermi2d	%ymm7,%ymm6,%ymm8
+	vpermi2d	%ymm7,%ymm6,%ymm9
+	vmovdqa		%ymm8,%ymm6
+	vmovdqa		%ymm9,%ymm7
+	vpaddd		%xmm8,%xmm0,%xmm0
+	vpaddd		%xmm1,%xmm0,%xmm0
+	vpxor		%xmm0,%xmm3,%xmm3
+	vprord		$0x10,%xmm3,%xmm3
+	vpaddd		%xmm3,%xmm2,%xmm2
+	vpxor		%xmm2,%xmm1,%xmm1
+	vprord		$0xc,%xmm1,%xmm1
+	vextracti128	$0x1,%ymm8,%xmm8
+	vpaddd		%xmm8,%xmm0,%xmm0
+	vpaddd		%xmm1,%xmm0,%xmm0
+	vpxor		%xmm0,%xmm3,%xmm3
+	vprord		$0x8,%xmm3,%xmm3
+	vpaddd		%xmm3,%xmm2,%xmm2
+	vpxor		%xmm2,%xmm1,%xmm1
+	vprord		$0x7,%xmm1,%xmm1
+	vpshufd		$0x39,%xmm1,%xmm1
+	vpshufd		$0x4e,%xmm2,%xmm2
+	vpshufd		$0x93,%xmm3,%xmm3
+	vpaddd		%xmm9,%xmm0,%xmm0
+	vpaddd		%xmm1,%xmm0,%xmm0
+	vpxor		%xmm0,%xmm3,%xmm3
+	vprord		$0x10,%xmm3,%xmm3
+	vpaddd		%xmm3,%xmm2,%xmm2
+	vpxor		%xmm2,%xmm1,%xmm1
+	vprord		$0xc,%xmm1,%xmm1
+	vextracti128	$0x1,%ymm9,%xmm9
+	vpaddd		%xmm9,%xmm0,%xmm0
+	vpaddd		%xmm1,%xmm0,%xmm0
+	vpxor		%xmm0,%xmm3,%xmm3
+	vprord		$0x8,%xmm3,%xmm3
+	vpaddd		%xmm3,%xmm2,%xmm2
+	vpxor		%xmm2,%xmm1,%xmm1
+	vprord		$0x7,%xmm1,%xmm1
+	vpshufd		$0x93,%xmm1,%xmm1
+	vpshufd		$0x4e,%xmm2,%xmm2
+	vpshufd		$0x39,%xmm3,%xmm3
+	decb		%cl
+	jne		.Lblake2s_compress_avx512_roundloop
+	vpxor		%xmm10,%xmm0,%xmm0
+	vpxor		%xmm11,%xmm1,%xmm1
+	vpxor		%xmm2,%xmm0,%xmm0
+	vpxor		%xmm3,%xmm1,%xmm1
+	decq		%rdx
+	jne		.Lblake2s_compress_avx512_mainloop
+	vmovdqu		%xmm0,(%rdi)
+	vmovdqu		%xmm1,0x10(%rdi)
+	vmovdqu		%xmm4,0x20(%rdi)
+	vzeroupper
+	retq
+ENDPROC(blake2s_compress_avx512)
+#endif /* CONFIG_AS_AVX512 */
diff --git a/arch/x86/crypto/blake2s-glue.c b/arch/x86/crypto/blake2s-glue.c
new file mode 100644
index 000000000000..f3b80ffa513a
--- /dev/null
+++ b/arch/x86/crypto/blake2s-glue.c
@@ -0,0 +1,73 @@ 
+// SPDX-License-Identifier: GPL-2.0 OR MIT
+/*
+ * Copyright (C) 2015-2019 Jason A. Donenfeld <Jason@zx2c4.com>. All Rights Reserved.
+ */
+
+#include <crypto/blake2s.h>
+#include <crypto/internal/simd.h>
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+
+#include <asm/cpufeature.h>
+#include <asm/fpu/api.h>
+#include <asm/processor.h>
+#include <asm/simd.h>
+
+asmlinkage void blake2s_compress_avx(struct blake2s_state *state,
+				     const u8 *block, const size_t nblocks,
+				     const u32 inc);
+asmlinkage void blake2s_compress_avx512(struct blake2s_state *state,
+					const u8 *block, const size_t nblocks,
+					const u32 inc);
+
+static bool blake2s_use_avx __ro_after_init;
+static bool blake2s_use_avx512 __ro_after_init;
+
+bool blake2s_compress_arch(struct blake2s_state *state,
+			   const u8 *block, size_t nblocks,
+			   const u32 inc)
+{
+	/* SIMD disables preemption, so relax after processing each page. */
+	BUILD_BUG_ON(PAGE_SIZE / BLAKE2S_BLOCK_SIZE < 8);
+
+	if (!blake2s_use_avx || !crypto_simd_usable())
+		return false;
+
+	for (;;) {
+		const size_t blocks = min_t(size_t, nblocks,
+					    PAGE_SIZE / BLAKE2S_BLOCK_SIZE);
+
+		kernel_fpu_begin();
+		if (IS_ENABLED(CONFIG_AS_AVX512) && blake2s_use_avx512)
+			blake2s_compress_avx512(state, block, blocks, inc);
+		else
+			blake2s_compress_avx(state, block, blocks, inc);
+		kernel_fpu_end();
+
+		nblocks -= blocks;
+		if (!nblocks)
+			break;
+		block += blocks * BLAKE2S_BLOCK_SIZE;
+	}
+	return true;
+}
+
+static int __init blake2s_mod_init(void)
+{
+	blake2s_use_avx =
+		boot_cpu_has(X86_FEATURE_AVX) &&
+		cpu_has_xfeatures(XFEATURE_MASK_SSE | XFEATURE_MASK_YMM, NULL);
+	blake2s_use_avx512 =
+		boot_cpu_has(X86_FEATURE_AVX) &&
+		boot_cpu_has(X86_FEATURE_AVX2) &&
+		boot_cpu_has(X86_FEATURE_AVX512F) &&
+		boot_cpu_has(X86_FEATURE_AVX512VL) &&
+		cpu_has_xfeatures(XFEATURE_MASK_SSE | XFEATURE_MASK_YMM |
+				  XFEATURE_MASK_AVX512, NULL);
+
+	return 0;
+}
+
+module_init(blake2s_mod_init);
diff --git a/crypto/Kconfig b/crypto/Kconfig
index 7e1c794b570e..5055e05e8f80 100644
--- a/crypto/Kconfig
+++ b/crypto/Kconfig
@@ -1010,6 +1010,12 @@  config CRYPTO_LIB_BLAKE2S
 	tristate "BLAKE2s hash function library"
 	depends on CRYPTO_ARCH_HAVE_LIB_BLAKE2S || !CRYPTO_ARCH_HAVE_LIB_BLAKE2S
 
+config CRYPTO_LIB_BLAKE2S_X86
+	tristate "BLAKE2s hash function library (x86 accelerated version)"
+	depends on X86 && 64BIT
+	select CRYPTO_LIB_BLAKE2S
+	select CRYPTO_ARCH_HAVE_LIB_BLAKE2S
+
 comment "Ciphers"
 
 config CRYPTO_LIB_AES