From: Eric Biggers <ebiggers@kernel.org>
To: linux-crypto@vger.kernel.org
Cc: Ard Biesheuvel <ardb@kernel.org>,
"Jason A . Donenfeld" <Jason@zx2c4.com>,
Herbert Xu <herbert@gondor.apana.org.au>,
Martin Willi <martin@strongswan.org>,
Antonio Quartulli <antonio@mandelbit.com>,
Eric Biggers <ebiggers@kernel.org>
Subject: [PATCH] lib/crypto: x86/chacha: Drop redundant moves from AVX2 code
Date: Wed, 11 Nov 2026 14:34:55 -0800 [thread overview]
Message-ID: <20261111223455.89467-1-ebiggers@kernel.org> (raw)
The AVX2 optimized ChaCha code has many instances of moving one register
to another, then left-shifting the destination register with 'vpslld'.
This pattern likely was copied from the SSSE3 code, which uses 'pslld'.
However, like all the other AVX instructions, 'vpslld' has separate
source and destination registers. Just utilize that instead.
Signed-off-by: Eric Biggers <ebiggers@kernel.org>
---
lib/crypto/x86/chacha-avx2-x86_64.S | 36 ++++++++++-------------------
1 file changed, 12 insertions(+), 24 deletions(-)
diff --git a/lib/crypto/x86/chacha-avx2-x86_64.S b/lib/crypto/x86/chacha-avx2-x86_64.S
index 18ae03e05074f..8aa3227f7302b 100644
--- a/lib/crypto/x86/chacha-avx2-x86_64.S
+++ b/lib/crypto/x86/chacha-avx2-x86_64.S
@@ -74,8 +74,7 @@ SYM_FUNC_START(chacha_1to2blocks_avx2)
# x2 += x3, x1 = rotl32(x1 ^ x2, 12)
vpaddd %ymm3,%ymm2,%ymm2
vpxor %ymm2,%ymm1,%ymm1
- vmovdqa %ymm1,%ymm6
- vpslld $12,%ymm6,%ymm6
+ vpslld $12,%ymm1,%ymm6
vpsrld $20,%ymm1,%ymm1
vpor %ymm6,%ymm1,%ymm1
@@ -87,8 +86,7 @@ SYM_FUNC_START(chacha_1to2blocks_avx2)
# x2 += x3, x1 = rotl32(x1 ^ x2, 7)
vpaddd %ymm3,%ymm2,%ymm2
vpxor %ymm2,%ymm1,%ymm1
- vmovdqa %ymm1,%ymm7
- vpslld $7,%ymm7,%ymm7
+ vpslld $7,%ymm1,%ymm7
vpsrld $25,%ymm1,%ymm1
vpor %ymm7,%ymm1,%ymm1
@@ -107,8 +105,7 @@ SYM_FUNC_START(chacha_1to2blocks_avx2)
# x2 += x3, x1 = rotl32(x1 ^ x2, 12)
vpaddd %ymm3,%ymm2,%ymm2
vpxor %ymm2,%ymm1,%ymm1
- vmovdqa %ymm1,%ymm6
- vpslld $12,%ymm6,%ymm6
+ vpslld $12,%ymm1,%ymm6
vpsrld $20,%ymm1,%ymm1
vpor %ymm6,%ymm1,%ymm1
@@ -120,8 +117,7 @@ SYM_FUNC_START(chacha_1to2blocks_avx2)
# x2 += x3, x1 = rotl32(x1 ^ x2, 7)
vpaddd %ymm3,%ymm2,%ymm2
vpxor %ymm2,%ymm1,%ymm1
- vmovdqa %ymm1,%ymm7
- vpslld $7,%ymm7,%ymm7
+ vpslld $7,%ymm1,%ymm7
vpsrld $25,%ymm1,%ymm1
vpor %ymm7,%ymm1,%ymm1
@@ -251,15 +247,13 @@ SYM_FUNC_START(chacha_2to4blocks_avx2)
# x2 += x3, x1 = rotl32(x1 ^ x2, 12)
vpaddd %ymm3,%ymm2,%ymm2
vpxor %ymm2,%ymm1,%ymm1
- vmovdqa %ymm1,%ymm10
- vpslld $12,%ymm10,%ymm10
+ vpslld $12,%ymm1,%ymm10
vpsrld $20,%ymm1,%ymm1
vpor %ymm10,%ymm1,%ymm1
vpaddd %ymm7,%ymm6,%ymm6
vpxor %ymm6,%ymm5,%ymm5
- vmovdqa %ymm5,%ymm10
- vpslld $12,%ymm10,%ymm10
+ vpslld $12,%ymm5,%ymm10
vpsrld $20,%ymm5,%ymm5
vpor %ymm10,%ymm5,%ymm5
@@ -275,15 +269,13 @@ SYM_FUNC_START(chacha_2to4blocks_avx2)
# x2 += x3, x1 = rotl32(x1 ^ x2, 7)
vpaddd %ymm3,%ymm2,%ymm2
vpxor %ymm2,%ymm1,%ymm1
- vmovdqa %ymm1,%ymm10
- vpslld $7,%ymm10,%ymm10
+ vpslld $7,%ymm1,%ymm10
vpsrld $25,%ymm1,%ymm1
vpor %ymm10,%ymm1,%ymm1
vpaddd %ymm7,%ymm6,%ymm6
vpxor %ymm6,%ymm5,%ymm5
- vmovdqa %ymm5,%ymm10
- vpslld $7,%ymm10,%ymm10
+ vpslld $7,%ymm5,%ymm10
vpsrld $25,%ymm5,%ymm5
vpor %ymm10,%ymm5,%ymm5
@@ -309,15 +301,13 @@ SYM_FUNC_START(chacha_2to4blocks_avx2)
# x2 += x3, x1 = rotl32(x1 ^ x2, 12)
vpaddd %ymm3,%ymm2,%ymm2
vpxor %ymm2,%ymm1,%ymm1
- vmovdqa %ymm1,%ymm10
- vpslld $12,%ymm10,%ymm10
+ vpslld $12,%ymm1,%ymm10
vpsrld $20,%ymm1,%ymm1
vpor %ymm10,%ymm1,%ymm1
vpaddd %ymm7,%ymm6,%ymm6
vpxor %ymm6,%ymm5,%ymm5
- vmovdqa %ymm5,%ymm10
- vpslld $12,%ymm10,%ymm10
+ vpslld $12,%ymm5,%ymm10
vpsrld $20,%ymm5,%ymm5
vpor %ymm10,%ymm5,%ymm5
@@ -333,15 +323,13 @@ SYM_FUNC_START(chacha_2to4blocks_avx2)
# x2 += x3, x1 = rotl32(x1 ^ x2, 7)
vpaddd %ymm3,%ymm2,%ymm2
vpxor %ymm2,%ymm1,%ymm1
- vmovdqa %ymm1,%ymm10
- vpslld $7,%ymm10,%ymm10
+ vpslld $7,%ymm1,%ymm10
vpsrld $25,%ymm1,%ymm1
vpor %ymm10,%ymm1,%ymm1
vpaddd %ymm7,%ymm6,%ymm6
vpxor %ymm6,%ymm5,%ymm5
- vmovdqa %ymm5,%ymm10
- vpslld $7,%ymm10,%ymm10
+ vpslld $7,%ymm5,%ymm10
vpsrld $25,%ymm5,%ymm5
vpor %ymm10,%ymm5,%ymm5
base-commit: 33b4f31b1aea3f62b61c892daefc0e6a77859032
--
2.56.1
next reply other threads:[~2026-11-11 22:35 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-11 22:34 Eric Biggers [this message]
2026-11-11 23:37 ` [PATCH] lib/crypto: x86/chacha: Drop redundant moves from AVX2 code Antonio Quartulli
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20261111223455.89467-1-ebiggers@kernel.org \
--to=ebiggers@kernel.org \
--cc=Jason@zx2c4.com \
--cc=antonio@mandelbit.com \
--cc=ardb@kernel.org \
--cc=herbert@gondor.apana.org.au \
--cc=linux-crypto@vger.kernel.org \
--cc=martin@strongswan.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox