public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* [PATCH v2 00/10] AEGIS x86 assembly tuning
@ 2024-10-17  0:00 Eric Biggers
  2024-10-17  0:00 ` [PATCH v2 01/10] crypto: x86/aegis128 - access 32-bit arguments as 32-bit Eric Biggers
                   ` (10 more replies)
  0 siblings, 11 replies; 12+ messages in thread
From: Eric Biggers @ 2024-10-17  0:00 UTC (permalink / raw)
  To: linux-crypto; +Cc: x86, Ondrej Mosnacek, linux-kernel

This series cleans up the AES-NI optimized implementation of AEGIS-128.

Performance is improved by 1-5% depending on the input lengths.  Binary
code size is reduced by about 20% (measuring glue + assembly combined),
and source code length is reduced by about 150 lines.

The first patch also fixes a bug which could theoretically cause
incorrect behavior but was seemingly not being encountered in practice.

Note: future optimizations for AEGIS-128 could involve adding AVX512 /
AVX10 optimized assembly code.  However, unfortunately due to the way
that AEGIS-128 is specified, its level of parallelism is limited, and it
can't really take advantage of vector lengths greater than 128 bits.
So, probably this would provide only another modest improvement, mostly
coming from being able to use the ternary logic instructions.

Changed in v2:
- Put assoclen and cryptlen in the correct order in the prototype of
  aegis128_aesni_final().
- Expanded commit message of "eliminate some indirect calls"
- Added Ondrej's Reviewed-by.

Eric Biggers (10):
  crypto: x86/aegis128 - access 32-bit arguments as 32-bit
  crypto: x86/aegis128 - remove no-op init and exit functions
  crypto: x86/aegis128 - eliminate some indirect calls
  crypto: x86/aegis128 - don't bother with special code for aligned data
  crypto: x86/aegis128 - optimize length block preparation using SSE4.1
  crypto: x86/aegis128 - improve assembly function prototypes
  crypto: x86/aegis128 - optimize partial block handling using SSE4.1
  crypto: x86/aegis128 - take advantage of block-aligned len
  crypto: x86/aegis128 - remove unneeded FRAME_BEGIN and FRAME_END
  crypto: x86/aegis128 - remove unneeded RETs

 arch/x86/crypto/Kconfig               |   4 +-
 arch/x86/crypto/aegis128-aesni-asm.S  | 532 ++++++++++----------------
 arch/x86/crypto/aegis128-aesni-glue.c | 145 ++++---
 3 files changed, 261 insertions(+), 420 deletions(-)

base-commit: 5c20772738e1d1d7bec41664eb9d61497e53c10e
-- 
2.47.0


^ permalink raw reply	[flat|nested] 12+ messages in thread

end of thread, other threads:[~2024-10-26  6:59 UTC | newest]

Thread overview: 12+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-10-17  0:00 [PATCH v2 00/10] AEGIS x86 assembly tuning Eric Biggers
2024-10-17  0:00 ` [PATCH v2 01/10] crypto: x86/aegis128 - access 32-bit arguments as 32-bit Eric Biggers
2024-10-17  0:00 ` [PATCH v2 02/10] crypto: x86/aegis128 - remove no-op init and exit functions Eric Biggers
2024-10-17  0:00 ` [PATCH v2 03/10] crypto: x86/aegis128 - eliminate some indirect calls Eric Biggers
2024-10-17  0:00 ` [PATCH v2 04/10] crypto: x86/aegis128 - don't bother with special code for aligned data Eric Biggers
2024-10-17  0:00 ` [PATCH v2 05/10] crypto: x86/aegis128 - optimize length block preparation using SSE4.1 Eric Biggers
2024-10-17  0:00 ` [PATCH v2 06/10] crypto: x86/aegis128 - improve assembly function prototypes Eric Biggers
2024-10-17  0:00 ` [PATCH v2 07/10] crypto: x86/aegis128 - optimize partial block handling using SSE4.1 Eric Biggers
2024-10-17  0:00 ` [PATCH v2 08/10] crypto: x86/aegis128 - take advantage of block-aligned len Eric Biggers
2024-10-17  0:00 ` [PATCH v2 09/10] crypto: x86/aegis128 - remove unneeded FRAME_BEGIN and FRAME_END Eric Biggers
2024-10-17  0:00 ` [PATCH v2 10/10] crypto: x86/aegis128 - remove unneeded RETs Eric Biggers
2024-10-26  6:59 ` [PATCH v2 00/10] AEGIS x86 assembly tuning Herbert Xu

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox