[llvm-bugs] [Bug 30773] New: cmp should be known true (via CVP?)

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=30773

            Bug ID: 30773
           Summary: cmp should be known true (via CVP?)
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Scalar Optimizations
          Assignee: unassignedbugs at nondot.org
          Reporter: spatel+llvm at rotateright.com
                CC: llvm-bugs at lists.llvm.org
    Classification: Unclassified

$ cat cvp.c
void g(int * __attribute((aligned(256))) a, int * __attribute((aligned(256)))
b, int * __attribute((aligned(256))) c, int n) {
  for (int i = 0; i < n * 16; ++i)
    c[i] = a[i] + b[i];
}

$ ./clang -O2 cvp.c -S -emit-llvm -o -
define void @g(i32* nocapture readonly %a, i32* nocapture readonly %b, i32*
nocapture %c, i32 %n) local_unnamed_addr #0 {
entry:
  %cmp10 = icmp sgt i32 %n, 0
  br i1 %cmp10, label %for.body.preheader, label %for.cond.cleanup

for.body.preheader:                        
  %mul = shl nsw i32 %n, 4  <--- %n is positive; %mul must be positive (nsw)
  %0 = sext i32 %mul to i64 
  %1 = icmp sgt i64 %0, 1   <--- so this must be true
  %smax = select i1 %1, i64 %0, i64 1 <--- and this shouldn't be needed
...

Is this a deficiency of correlated value propagation?

-- 
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/20161024/043561f5/attachment.html>