[llvm-bugs] [Bug 40976] New: arm tree vectorizer fails to optimize kernel raid/xor code

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=40976

            Bug ID: 40976
           Summary: arm tree vectorizer fails to optimize kernel raid/xor
                    code
           Product: clang
           Version: 8.0
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: -New Bugs
          Assignee: unassignedclangbugs at nondot.org
          Reporter: arnd at linaro.org
                CC: htmldeveloper at gmail.com, llvm-bugs at lists.llvm.org,
                    neeilans at live.com, richard-llvm at metafoo.co.uk

Created attachment 21569
  --> https://bugs.llvm.org/attachment.cgi?id=21569&action=edit
preprocessed arch/arm/lib/xor-neon.c

The RAID code in the Linux kernel is written in a way that should let the
compiler autovectorize it with good results. When compiling the kernel with
clang-8, this does not happen, only scalar code is emitted.

I have reproduced this with the attached file, also for comparison the same
file at https://godbolt.org/z/7bMthX, which shows the difference between the
gcc and the clang output.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20190306/e0cbcb50/attachment.html>