[Bug tree-optimization/88398] vectorization failure for a small loop to do byte comparison
wilco at gcc dot gnu.org
gcc-bugzilla@gcc.gnu.org
Thu May 28 12:44:30 GMT 2020
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88398
--- Comment #34 from Wilco <wilco at gcc dot gnu.org> ---
(In reply to Jiu Fu Guo from comment #33)
> It would be relatively easy if the target supports unaligned access. like
> read64ne in
> https://git.tukaani.org/?p=xz.git;a=blob;f=src/liblzma/common/memcmplen.h
> Then the alignment issue is relaxed. It may be safer if we can
> prove/assume the underlying buffer is enough, like array accessing or
> pointer+index accessing in a loop.
Yes, without unaligned support you can't use a wider access.
If we can't prove the buffer bounds then we'd have to use a page cross check
before every unaligned access.
More information about the Gcc-bugs
mailing list