[libc-commits] [libc] [libc] Process words in inline_strcmp 8 bytes at a time (PR #214363)

[via libc-commits]

afnrow wrote:

For unaligned access, I already use __builtin_memcpy() such that we won't have any unaligned read or such, I agree with ASAN point thanks for bringing it to my attention, should we add an if condition to fall back to the loop version? Or how would you recommend we handle it, sure we can test it on architectures with non aligned load penalty like riscv

https://github.com/llvm/llvm-project/pull/214363