iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
2874c5fd28
linux/arch/arm64/crypto/aes-ce-ccm-core.S

225 lines
6.0 KiB
ArmAsm
Raw Normal View History

arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
/*
* aesce-ccm-core.S - AES-CCM transform for ARMv8 with Crypto Extensions
*
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
* Copyright (C) 2013 - 2017 Linaro Ltd <ard.biesheuvel@linaro.org>
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*/
#include <linux/linkage.h>
crypto: arm64/aes-ccm-ce: fix for big endian The AES-CCM implementation that uses ARMv8 Crypto Extensions instructions refers to the AES round keys as pairs of 64-bit quantities, which causes failures when building the code for big endian. In addition, it byte swaps the input counter unconditionally, while this is only required for little endian builds. So fix both issues. Fixes: 12ac3efe74f8 ("arm64/crypto: use crypto instructions to generate AES key schedule") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2016-10-11 21:15:17 +03:00
#include <asm/assembler.h>
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
.text
.arch armv8-a+crypto
/*
* void ce_aes_ccm_auth_data(u8 mac[], u8 const in[], u32 abytes,
* u32 *macp, u8 const rk[], u32 rounds);
*/
ENTRY(ce_aes_ccm_auth_data)
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
ldr w8, [x3] /* leftover from prev round? */
crypto: arm64/aes-ccm-ce: fix for big endian The AES-CCM implementation that uses ARMv8 Crypto Extensions instructions refers to the AES round keys as pairs of 64-bit quantities, which causes failures when building the code for big endian. In addition, it byte swaps the input counter unconditionally, while this is only required for little endian builds. So fix both issues. Fixes: 12ac3efe74f8 ("arm64/crypto: use crypto instructions to generate AES key schedule") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2016-10-11 21:15:17 +03:00
ld1 {v0.16b}, [x0] /* load mac */
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
cbz w8, 1f
sub w8, w8, #16
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
eor v1.16b, v1.16b, v1.16b
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
0: ldrb w7, [x1], #1 /* get 1 byte of input */
subs w2, w2, #1
add w8, w8, #1
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
ins v1.b[0], w7
ext v1.16b, v1.16b, v1.16b, #1 /* rotate in the input bytes */
beq 8f /* out of input? */
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
cbnz w8, 0b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
eor v0.16b, v0.16b, v1.16b
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
1: ld1 {v3.4s}, [x4] /* load first round key */
prfm pldl1strm, [x1]
cmp w5, #12 /* which key size? */
add x6, x4, #16
sub w7, w5, #2 /* modified # of rounds */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
bmi 2f
bne 5f
mov v5.16b, v3.16b
b 4f
2: mov v4.16b, v3.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
ld1 {v5.4s}, [x6], #16 /* load 2nd round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
3: aese v0.16b, v4.16b
aesmc v0.16b, v0.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
4: ld1 {v3.4s}, [x6], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aese v0.16b, v5.16b
aesmc v0.16b, v0.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
5: ld1 {v4.4s}, [x6], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
subs w7, w7, #3
aese v0.16b, v3.16b
aesmc v0.16b, v0.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
ld1 {v5.4s}, [x6], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
bpl 3b
aese v0.16b, v4.16b
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
subs w2, w2, #16 /* last data? */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
eor v0.16b, v0.16b, v5.16b /* final round */
bmi 6f
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
ld1 {v1.16b}, [x1], #16 /* load next input block */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
eor v0.16b, v0.16b, v1.16b /* xor with mac */
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
bne 1b
6: st1 {v0.16b}, [x0] /* store mac */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
beq 10f
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
adds w2, w2, #16
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
beq 10f
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
mov w8, w2
7: ldrb w7, [x1], #1
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
umov w6, v0.b[0]
eor w6, w6, w7
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
strb w6, [x0], #1
subs w2, w2, #1
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
beq 10f
ext v0.16b, v0.16b, v0.16b, #1 /* rotate out the mac bytes */
b 7b
crypto: arm64/aes-ccm - fix logical bug in AAD MAC handling The NEON MAC calculation routine fails to handle the case correctly where there is some data in the buffer, and the input fills it up exactly. In this case, we enter the loop at the end with w8 == 0, while a negative value is assumed, and so the loop carries on until the increment of the 32-bit counter wraps around, which is quite obviously wrong. So omit the loop altogether in this case, and exit right away. Reported-by: Eric Biggers <ebiggers@kernel.org> Fixes: a3fd82105b9d1 ("arm64/crypto: AES in CCM mode using ARMv8 Crypto ...") Cc: stable@vger.kernel.org Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2019-01-24 19:33:45 +03:00
8: cbz w8, 91f
mov w7, w8
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
add w8, w8, #16
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
9: ext v1.16b, v1.16b, v1.16b, #1
adds w7, w7, #1
bne 9b
crypto: arm64/aes-ccm - fix logical bug in AAD MAC handling The NEON MAC calculation routine fails to handle the case correctly where there is some data in the buffer, and the input fills it up exactly. In this case, we enter the loop at the end with w8 == 0, while a negative value is assumed, and so the loop carries on until the increment of the 32-bit counter wraps around, which is quite obviously wrong. So omit the loop altogether in this case, and exit right away. Reported-by: Eric Biggers <ebiggers@kernel.org> Fixes: a3fd82105b9d1 ("arm64/crypto: AES in CCM mode using ARMv8 Crypto ...") Cc: stable@vger.kernel.org Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2019-01-24 19:33:45 +03:00
91: eor v0.16b, v0.16b, v1.16b
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
st1 {v0.16b}, [x0]
10: str w8, [x3]
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
ret
ENDPROC(ce_aes_ccm_auth_data)
/*
* void ce_aes_ccm_final(u8 mac[], u8 const ctr[], u8 const rk[],
* u32 rounds);
*/
ENTRY(ce_aes_ccm_final)
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
ld1 {v3.4s}, [x2], #16 /* load first round key */
crypto: arm64/aes-ccm-ce: fix for big endian The AES-CCM implementation that uses ARMv8 Crypto Extensions instructions refers to the AES round keys as pairs of 64-bit quantities, which causes failures when building the code for big endian. In addition, it byte swaps the input counter unconditionally, while this is only required for little endian builds. So fix both issues. Fixes: 12ac3efe74f8 ("arm64/crypto: use crypto instructions to generate AES key schedule") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2016-10-11 21:15:17 +03:00
ld1 {v0.16b}, [x0] /* load mac */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
cmp w3, #12 /* which key size? */
sub w3, w3, #2 /* modified # of rounds */
crypto: arm64/aes-ccm-ce: fix for big endian The AES-CCM implementation that uses ARMv8 Crypto Extensions instructions refers to the AES round keys as pairs of 64-bit quantities, which causes failures when building the code for big endian. In addition, it byte swaps the input counter unconditionally, while this is only required for little endian builds. So fix both issues. Fixes: 12ac3efe74f8 ("arm64/crypto: use crypto instructions to generate AES key schedule") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2016-10-11 21:15:17 +03:00
ld1 {v1.16b}, [x1] /* load 1st ctriv */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
bmi 0f
bne 3f
mov v5.16b, v3.16b
b 2f
0: mov v4.16b, v3.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
1: ld1 {v5.4s}, [x2], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aese v0.16b, v4.16b
aesmc v0.16b, v0.16b
arm64/crypto: issue aese/aesmc instructions in pairs This changes the AES core transform implementations to issue aese/aesmc (and aesd/aesimc) in pairs. This enables a micro-architectural optimization in recent Cortex-A5x cores that improves performance by 50-90%. Measured performance in cycles per byte (Cortex-A57): CBC enc CBC dec CTR before 3.64 1.34 1.32 after 1.95 0.85 0.93 Note that this results in a ~5% performance decrease for older cores. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 21:05:13 +03:00
aese v1.16b, v4.16b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aesmc v1.16b, v1.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
2: ld1 {v3.4s}, [x2], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aese v0.16b, v5.16b
aesmc v0.16b, v0.16b
arm64/crypto: issue aese/aesmc instructions in pairs This changes the AES core transform implementations to issue aese/aesmc (and aesd/aesimc) in pairs. This enables a micro-architectural optimization in recent Cortex-A5x cores that improves performance by 50-90%. Measured performance in cycles per byte (Cortex-A57): CBC enc CBC dec CTR before 3.64 1.34 1.32 after 1.95 0.85 0.93 Note that this results in a ~5% performance decrease for older cores. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 21:05:13 +03:00
aese v1.16b, v5.16b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aesmc v1.16b, v1.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
3: ld1 {v4.4s}, [x2], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
subs w3, w3, #3
aese v0.16b, v3.16b
aesmc v0.16b, v0.16b
arm64/crypto: issue aese/aesmc instructions in pairs This changes the AES core transform implementations to issue aese/aesmc (and aesd/aesimc) in pairs. This enables a micro-architectural optimization in recent Cortex-A5x cores that improves performance by 50-90%. Measured performance in cycles per byte (Cortex-A57): CBC enc CBC dec CTR before 3.64 1.34 1.32 after 1.95 0.85 0.93 Note that this results in a ~5% performance decrease for older cores. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 21:05:13 +03:00
aese v1.16b, v3.16b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aesmc v1.16b, v1.16b
bpl 1b
aese v0.16b, v4.16b
aese v1.16b, v4.16b
/* final round key cancels out */
eor v0.16b, v0.16b, v1.16b /* en-/decrypt the mac */
crypto: arm64/aes-ccm-ce: fix for big endian The AES-CCM implementation that uses ARMv8 Crypto Extensions instructions refers to the AES round keys as pairs of 64-bit quantities, which causes failures when building the code for big endian. In addition, it byte swaps the input counter unconditionally, while this is only required for little endian builds. So fix both issues. Fixes: 12ac3efe74f8 ("arm64/crypto: use crypto instructions to generate AES key schedule") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2016-10-11 21:15:17 +03:00
st1 {v0.16b}, [x0] /* store result */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
ret
ENDPROC(ce_aes_ccm_final)
.macro aes_ccm_do_crypt,enc
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
ldr x8, [x6, #8] /* load lower ctr */
ld1 {v0.16b}, [x5] /* load mac */
CPU_LE( rev x8, x8 ) /* keep swabbed ctr in reg */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
0: /* outer loop */
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
ld1 {v1.8b}, [x6] /* load upper ctr */
prfm pldl1strm, [x1]
add x8, x8, #1
rev x9, x8
cmp w4, #12 /* which key size? */
sub w7, w4, #2 /* get modified # of rounds */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
ins v1.d[1], x9 /* no carry in lower ctr */
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
ld1 {v3.4s}, [x3] /* load first round key */
add x10, x3, #16
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
bmi 1f
bne 4f
mov v5.16b, v3.16b
b 3f
1: mov v4.16b, v3.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
ld1 {v5.4s}, [x10], #16 /* load 2nd round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
2: /* inner loop: 3 rounds, 2x interleaved */
aese v0.16b, v4.16b
aesmc v0.16b, v0.16b
arm64/crypto: issue aese/aesmc instructions in pairs This changes the AES core transform implementations to issue aese/aesmc (and aesd/aesimc) in pairs. This enables a micro-architectural optimization in recent Cortex-A5x cores that improves performance by 50-90%. Measured performance in cycles per byte (Cortex-A57): CBC enc CBC dec CTR before 3.64 1.34 1.32 after 1.95 0.85 0.93 Note that this results in a ~5% performance decrease for older cores. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 21:05:13 +03:00
aese v1.16b, v4.16b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aesmc v1.16b, v1.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
3: ld1 {v3.4s}, [x10], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aese v0.16b, v5.16b
aesmc v0.16b, v0.16b
arm64/crypto: issue aese/aesmc instructions in pairs This changes the AES core transform implementations to issue aese/aesmc (and aesd/aesimc) in pairs. This enables a micro-architectural optimization in recent Cortex-A5x cores that improves performance by 50-90%. Measured performance in cycles per byte (Cortex-A57): CBC enc CBC dec CTR before 3.64 1.34 1.32 after 1.95 0.85 0.93 Note that this results in a ~5% performance decrease for older cores. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 21:05:13 +03:00
aese v1.16b, v5.16b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aesmc v1.16b, v1.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
4: ld1 {v4.4s}, [x10], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
subs w7, w7, #3
aese v0.16b, v3.16b
aesmc v0.16b, v0.16b
arm64/crypto: issue aese/aesmc instructions in pairs This changes the AES core transform implementations to issue aese/aesmc (and aesd/aesimc) in pairs. This enables a micro-architectural optimization in recent Cortex-A5x cores that improves performance by 50-90%. Measured performance in cycles per byte (Cortex-A57): CBC enc CBC dec CTR before 3.64 1.34 1.32 after 1.95 0.85 0.93 Note that this results in a ~5% performance decrease for older cores. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 21:05:13 +03:00
aese v1.16b, v3.16b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
aesmc v1.16b, v1.16b
crypto: arm64/aes-ce-cipher - match round key endianness with generic code In order to be able to reuse the generic AES code as a fallback for situations where the NEON may not be used, update the key handling to match the byte order of the generic code: it stores round keys as sequences of 32-bit quantities rather than streams of bytes, and so our code needs to be updated to reflect that. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2017-07-24 13:28:10 +03:00
ld1 {v5.4s}, [x10], #16 /* load next round key */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
bpl 2b
aese v0.16b, v4.16b
aese v1.16b, v4.16b
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
subs w2, w2, #16
bmi 6f /* partial block? */
ld1 {v2.16b}, [x1], #16 /* load next input block */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
.if \enc == 1
eor v2.16b, v2.16b, v5.16b /* final round enc+mac */
eor v1.16b, v1.16b, v2.16b /* xor with crypted ctr */
.else
eor v2.16b, v2.16b, v1.16b /* xor with crypted ctr */
eor v1.16b, v2.16b, v5.16b /* final round enc */
.endif
eor v0.16b, v0.16b, v2.16b /* xor mac with pt ^ rk[last] */
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
st1 {v1.16b}, [x0], #16 /* write output block */
bne 0b
CPU_LE( rev x8, x8 )
st1 {v0.16b}, [x5] /* store mac */
str x8, [x6, #8] /* store lsb end of ctr (BE) */
5: ret
6: eor v0.16b, v0.16b, v5.16b /* final round mac */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
eor v1.16b, v1.16b, v5.16b /* final round enc */
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
st1 {v0.16b}, [x5] /* store mac */
add w2, w2, #16 /* process partial tail block */
7: ldrb w9, [x1], #1 /* get 1 byte of input */
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
umov w6, v1.b[0] /* get top crypted ctr byte */
umov w7, v0.b[0] /* get top mac byte */
.if \enc == 1
eor w7, w7, w9
eor w9, w9, w6
.else
eor w9, w9, w6
eor w7, w7, w9
.endif
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
strb w9, [x0], #1 /* store out byte */
strb w7, [x5], #1 /* store mac byte */
subs w2, w2, #1
beq 5b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
ext v0.16b, v0.16b, v0.16b, #1 /* shift out mac byte */
ext v1.16b, v1.16b, v1.16b, #1 /* shift out ctr byte */
crypto: arm64 - revert NEON yield for fast AEAD implementations As it turns out, checking the TIF_NEED_RESCHED flag after each iteration results in a significant performance regression (~10%) when running fast algorithms (i.e., ones that use special instructions and operate in the < 4 cycles per byte range) on in-order cores with comparatively slow memory accesses such as the Cortex-A53. Given the speed of these ciphers, and the fact that the page based nature of the AEAD scatterwalk API guarantees that the core NEON transform is never invoked with more than a single page's worth of input, we can estimate the worst case duration of any resulting scheduling blackout: on a 1 GHz Cortex-A53 running with 64k pages, processing a page's worth of input at 4 cycles per byte results in a delay of ~250 us, which is a reasonable upper bound. So let's remove the yield checks from the fused AES-CCM and AES-GCM routines entirely. This reverts commit 7b67ae4d5ce8e2f912377f5fbccb95811a92097f and partially reverts commit 7c50136a8aba8784f07fb66a950cc61a7f3d2ee3. Fixes: 7c50136a8aba ("crypto: arm64/aes-ghash - yield NEON after every ...") Fixes: 7b67ae4d5ce8 ("crypto: arm64/aes-ccm - yield NEON after every ...") Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2018-07-29 17:52:30 +03:00
b 7b
arm64/crypto: AES in CCM mode using ARMv8 Crypto Extensions This patch adds support for the AES-CCM encryption algorithm for CPUs that have support for the AES part of the ARM v8 Crypto Extensions. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
2014-02-10 14:26:29 +04:00
.endm
/*
* void ce_aes_ccm_encrypt(u8 out[], u8 const in[], u32 cbytes,
* u8 const rk[], u32 rounds, u8 mac[],
* u8 ctr[]);
* void ce_aes_ccm_decrypt(u8 out[], u8 const in[], u32 cbytes,
* u8 const rk[], u32 rounds, u8 mac[],
* u8 ctr[]);
*/
ENTRY(ce_aes_ccm_encrypt)
aes_ccm_do_crypt 1
ENDPROC(ce_aes_ccm_encrypt)
ENTRY(ce_aes_ccm_decrypt)
aes_ccm_do_crypt 0
ENDPROC(ce_aes_ccm_decrypt)
Reference in New Issue Copy Permalink