[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20250620111219.52182-1-cp0613@linux.alibaba.com>
Date: Fri, 20 Jun 2025 19:12:17 +0800
From: cp0613@...ux.alibaba.com
To: yury.norov@...il.com,
linux@...musvillemoes.dk,
arnd@...db.de,
paul.walmsley@...ive.com,
palmer@...belt.com,
aou@...s.berkeley.edu,
alex@...ti.fr
Cc: linux-riscv@...ts.infradead.org,
linux-arch@...r.kernel.org,
linux-kernel@...r.kernel.org,
Chen Pei <cp0613@...ux.alibaba.com>
Subject: [PATCH 0/2] Implementing bitops rotate using riscv Zbb extension
From: Chen Pei <cp0613@...ux.alibaba.com>
This patch series moves the ror*/rol* functions from include/linux/bitops.h
to include/asm-generic/bitops/rotate.h as a generic implementation.
At the same time, an optimized implementation is made based on the bitwise
rotation instructions provided by the RISC-V Zbb extension[1].
Based on the RISC-V processor XUANTIE C908, I tested the performance of
sha3_generic. Compared with the generic implementation, the RISC-V Zbb
instruction implementation performance increased by an average of 6.87%.
Test method:
1. CONFIG_CRYPTO_TEST=m
2. modprobe tcrypt mode=322 sec=3
Different parameters will be selected to test the performance of sha3-224.
[1] https://github.com/riscv/riscv-bitmanip/
Chen Pei (2):
bitops: generic rotate
bitops: rotate: Add riscv implementation using Zbb extension
arch/riscv/include/asm/bitops.h | 127 ++++++++++++++++++++++++++++
include/asm-generic/bitops.h | 2 +-
include/asm-generic/bitops/rotate.h | 97 +++++++++++++++++++++
include/linux/bitops.h | 80 ------------------
tools/include/asm-generic/bitops.h | 2 +-
5 files changed, 226 insertions(+), 82 deletions(-)
create mode 100644 include/asm-generic/bitops/rotate.h
--
2.49.0
Powered by blists - more mailing lists