[llvm-bugs] [Bug 48224] New: [AssumeBundles] ValueTracking cannot use alignment assumptions

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=48224

            Bug ID: 48224
           Summary: [AssumeBundles] ValueTracking cannot use alignment
                    assumptions
           Product: libraries
           Version: trunk
          Hardware: All
                OS: All
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Global Analyses
          Assignee: unassignedbugs at nondot.org
          Reporter: nikolai.bozhenov at intel.com
                CC: llvm-bugs at lists.llvm.org

This is a regression caused by switch to using operand bundles to encode
alignment assumptions
(https://reviews.llvm.org/rG78de7297abe2e8fa782682168989c70e3cb34a5c).

It looks like the ValueTracking (computeKnownBits) cannot understand the new
format of alignment assumptions. As an example, consider compilation of the
following reproducer with clang-11 (old @llvm.assume format) and clang-trunk
(@llvm.assume with bundles):

  #include <stdint.h>
  int foo(int *p) {
    __builtin_assume_aligned(p, 32);
    return ((intptr_t) p) & 4;
  }

clang-11 (https://godbolt.org/z/xGh1e4) computes the returned value statically:

  ret i32 0

clang-trunk (https://godbolt.org/z/1jWe5j) fails to optimize the code:

  %3 = ptrtoint i32* %0 to i64
  %4 = trunc i64 %3 to i32
  %5 = and i32 %4, 4
  ret i32 %5

As one can see, after switching to operand bundles, clang-trunk lost its
ability to reason about low bits in aligned pointers.

-- 
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/20201119/600ead70/attachment.html>