[llvm-bugs] [Bug 37610] New: xor %a, C -> xor %a, -1 canonicalization

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=37610

            Bug ID: 37610
           Summary: xor %a, C -> xor %a, -1 canonicalization
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Scalar Optimizations
          Assignee: unassignedbugs at nondot.org
          Reporter: lebedev.ri at gmail.com
                CC: llvm-bugs at lists.llvm.org

Clang sometimes generates from

#include <cstdint>
uint8_t and_0(uint32_t cache, uint8_t count) {
  return uint8_t(cache) & uint8_t((uint8_t(1) << uint8_t(count)) - uint8_t(1));
}

this

define dso_local zeroext i8 @and_0(unsigned int, unsigned char)(i32, i8
zeroext) local_unnamed_addr #0 {
  %3 = zext i8 %1 to i32
  %4 = shl i32 1, %3
  %5 = add i32 %4, 255
  %6 = and i32 %5, %0
  %7 = trunc i32 %6 to i8
  ret i8 %7
}

Because the constant is not -1, the final code is suboptimal

https://godbolt.org/g/DeY8sr

Should we canonicalize it somehow?
https://rise4fun.com/Alive/sEB

-- 
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/20180528/323b2eeb/attachment.html>