[llvm-bugs] [Bug 49939] New: Optimizer does wrong load elimination after a branch

via llvm-bugs llvm-bugs at lists.llvm.org
Mon Apr 12 14:02:48 PDT 2021


https://bugs.llvm.org/show_bug.cgi?id=49939

            Bug ID: 49939
           Summary: Optimizer does wrong load elimination after a branch
           Product: clang
           Version: 11.0
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: LLVM Codegen
          Assignee: unassignedclangbugs at nondot.org
          Reporter: mdzar5 at gmail.com
                CC: llvm-bugs at lists.llvm.org, neeilans at live.com,
                    richard-llvm at metafoo.co.uk

Created attachment 24746
  --> https://bugs.llvm.org/attachment.cgi?id=24746&action=edit
Wrong displacement in "mov" instruction highlighted

I don't know what exactly goes wrong here. Looks like LLVM 11.0.0 optimizer
generates wrong IR, if Clang is asked to compile this program:

clang -v -std=c99 -masm=intel -O3 -finline-hint-functions -S bug.c

// bug.c
#include <stdint.h>

typedef unsigned char byte;
typedef uint32_t      udword;

typedef struct {
    udword x, y, z;
} A;


udword foo(A *s, udword n) {
    return s->x + n;
}

void bar(byte *ptr, A *s) {
    if (s->y == s->z) 
        *(ptr + 1) = foo(s, s->z * 3); // -O3 calculates s->y * 3 instead
    *ptr = foo(s, s->y);
}

-- 
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/20210412/e8fd3aa1/attachment.html>


More information about the llvm-bugs mailing list