[llvm-bugs] [Bug 41553] New: a loop with a pointer increment is folded into a form that it fails to optimize further

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=41553

            Bug ID: 41553
           Summary: a loop with a pointer increment is folded into a form
                    that it fails to optimize further
           Product: new-bugs
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Keywords: code-quality
          Severity: normal
          Priority: P
         Component: new bugs
          Assignee: unassignedbugs at nondot.org
          Reporter: nok.raven at gmail.com
                CC: htmldeveloper at gmail.com, llvm-bugs at lists.llvm.org

The foo should be equivalent to bar when sizeof(n) < sizeof(p)

int* foo(int* p, unsigned n)
{
    for (unsigned i = 0; i < n; ++i)
        ++p;
    return p;
}

int* bar(int* p, unsigned n)
{
    p += n;
    return p;
}

Produced IR:

define i32* @foo(i32* readnone, i32) {
  %3 = icmp eq i32 %1, 0
  %4 = add i32 %1, -1
  %5 = zext i32 %4 to i64
  %6 = add nuw nsw i64 %5, 1
  %7 = getelementptr i32, i32* %0, i64 %6
  %8 = select i1 %3, i32* %0, i32* %7
  ret i32* %8
}

define i32* @bar(i32* readnone, i32) {
  %3 = zext i32 %1 to i64
  %4 = getelementptr inbounds i32, i32* %0, i64 %3
  ret i32* %4
}

The branch looks strange, but even if I rewrite it:

Pre: C0 > 0
  %a = add i32 C0, -1
  %b = zext i32 %a to i64
  %r = add nuw nsw i64 %b, 1
=>
  %r = zext i32 C0 to i64

https://rise4fun.com/Alive/wiV

define i32* @foo2(i32* readnone, i32) {
  %3 = icmp eq i32 %1, 0
  %4 = zext i32 %1 to i64
  %5 = getelementptr i32, i32* %0, i64 %4
  %6 = select i1 %3, i32* %0, i32* %5
  ret i32* %6
}

It is not optimized further.

-- 
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/20190422/f2e98b31/attachment.html>