Changes in crc64.h

d142ab35ee0b7f9e84115fe3e4c3de4a9ac35f5e - Merge tag 'crc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux

Tue, 14 Apr 2026 00:36:04 +0000

Merge tag 'crc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linuxPull CRC updates from Eric Biggers: - Several improvements related to crc_kunit, to align with the standard KUnit conventions and make it easier for developers and CI systems to run this test suite - Add an arm64-optimized implementation of CRC64-NVME - Remove unused code for big endian arm64* tag 'crc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux: lib/crc: arm64: Simplify intrinsics implementation lib/crc: arm64: Use existing macros for kernel-mode FPU cflags lib/crc: arm64: Drop unnecessary chunking logic from crc64 lib/crc: arm64: Assume a little-endian kernel lib/crc: arm64: add NEON accelerated CRC64-NVMe implementation lib/crc: arm64: Drop check for CONFIG_KERNEL_MODE_NEON crypto: crc32c - Remove another outdated comment crypto: crc32c - Remove more outdated usage information kunit: configs: Enable all CRC tests in all_tests.config lib/crc: tests: Add a .kunitconfig file lib/crc: tests: Add CRC_ENABLE_ALL_FOR_KUNIT lib/crc: tests: Make crc_kunit test only the enabled CRC variants List of files: /linux/lib/crc/arm64/crc64.h

e0718ed60d60299840cfc2a408eb26042a20d186 - lib/crc: arm64: Drop unnecessary chunking logic from crc64

Mon, 30 Mar 2026 14:46:32 +0000

lib/crc: arm64: Drop unnecessary chunking logic from crc64On arm64, kernel mode NEON executes with preemption enabled, so there isno need to chunk the input by hand.Signed-off-by: Ard Biesheuvel Link: https://lore.kernel.org/r/20260330144630.33026-8-ardb@kernel.orgSigned-off-by: Eric Biggers List of files: /linux/lib/crc/arm64/crc64.h

63432fd625372a0e79fb00a4009af204f4edc013 - lib/crc: arm64: add NEON accelerated CRC64-NVMe implementation

Sun, 29 Mar 2026 07:43:38 +0000

lib/crc: arm64: add NEON accelerated CRC64-NVMe implementationImplement an optimized CRC64 (NVMe) algorithm for ARM64 using NEONPolynomial Multiply Long (PMULL) instructions. The generic shift-and-XORsoftware implementation is slow, which creates a bottleneck in NVMe andother storage subsystems.The acceleration is implemented using C intrinsics () ratherthan raw assembly for better readability and maintainability.Key highlights of this implementation:- Uses 4KB chunking inside scoped_ksimd() to avoid preemption latency spikes on large buffers.- Pre-calculates and loads fold constants via vld1q_u64() to minimize register spilling.- Benchmarks show the break-even point against the generic implementation is around 128 bytes. The PMULL path is enabled only for len >= 128.Performance results (kunit crc_benchmark on Cortex-A72):- Generic (len=4096): ~268 MB/s- PMULL (len=4096): ~1556 MB/s (nearly 6x improvement)Signed-off-by: Demian Shulhan Link: https://lore.kernel.org/r/20260329074338.1053550-1-demyansh@gmail.comSigned-off-by: Eric Biggers List of files: /linux/lib/crc/arm64/crc64.h