Skip to content

Wrong code at -O2 on x86_64-linux_gnu since 3ddd1ff #68260

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
shao-hua-li opened this issue Oct 4, 2023 · 2 comments
Closed

Wrong code at -O2 on x86_64-linux_gnu since 3ddd1ff #68260

shao-hua-li opened this issue Oct 4, 2023 · 2 comments
Assignees
Labels

Comments

@shao-hua-li
Copy link

clang at -O2 produced the wrong code.

Bisected to 3ddd1ff, which was committed by @nikic

Compiler explorer: https://godbolt.org/z/rEv69ovMj

% cat a.c
int printf(const char *, ...);
char *a;
long b, c, d, m;
int e, f, j, l;
short g;
int *k = &j;
int main() {
  int *n[] = {&f, &f, &f, &f, &f, &e};
  d = 0;
  for (; d <= 1; d++) {
    g = 0;
    for (; g <= 4; g++) {
      char h[] = {0, 0, 4};
      char *i = h;
      a = i;
      do {
        a++;
        b /= 10;
      } while (b);
      c = a - i;
      while (i < a){
        *i = *(i+1);
        i++;
      }
      *k ^= c;
      k = n[d + g];
      l ^= m;
    }
  }
  printf("%d\n", e);
}
%
% clang -O0 a.c && ./a.out
0
% clang -O2 a.c && ./a.out
1
%
@EugeneZelenko EugeneZelenko added the llvm:SCEV Scalar Evolution label Oct 5, 2023
@nikic
Copy link
Contributor

nikic commented Oct 6, 2023

I believe the problem here is that we have a load from an addrec based on a constant global. The SCEV for the load will indeed always be Unknown, but the SCEVAtScope may not be: We can determine the exit value of the addrec and based on that determine the exit value of the load.

@nikic
Copy link
Contributor

nikic commented Oct 6, 2023

I think this is a nice reduction:

; RUN: opt -S -passes='print<scalar-evolution>,loop(loop-unroll-full,indvars)' < %s
@g = constant [5 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4]

declare void @use(i32)

define void @test() {
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
  %iv.next = add i64 %iv, 1
  br label %loop2

loop2:
  %iv2 = phi i64 [ 0, %loop ], [ %iv2.next, %loop2 ]
  %iv2.next = add i64 %iv2, 1
  %idx = add i64 %iv, %iv2
  %gep = getelementptr i32, ptr @g, i64 %idx
  %load = load i32, ptr %gep
  %cmp2 = icmp ne i64 %iv2, 3
  br i1 %cmp2, label %loop2, label %loop.latch, !llvm.loop !0

loop.latch:
  %load.lcssa = phi i32 [ %load, %loop2 ]
  call void @use(i32 %load)
  %cmp = icmp ne i64 %iv, 1
  br i1 %cmp, label %loop, label %exit

exit:
  ret void
}

!0 = distinct !{!0, !1}
!1 = !{!"llvm.loop.unroll.disable", i1 true}

Results in:

@g = constant [5 x i32] [i32 0, i32 1, i32 2, i32 3, i32 4]

declare void @use(i32)

define void @test() {
entry:
  br label %loop

loop:                                             ; preds = %entry
  br label %loop2

loop2:                                            ; preds = %loop2, %loop
  br i1 false, label %loop2, label %loop.latch, !llvm.loop !0

loop.latch:                                       ; preds = %loop2
  call void @use(i32 4)
  br label %loop2.1

loop2.1:                                          ; preds = %loop2.1, %loop.latch
  br i1 false, label %loop2.1, label %loop.latch.1, !llvm.loop !0

loop.latch.1:                                     ; preds = %loop2.1
  call void @use(i32 4)
  ret void
}

!0 = distinct !{!0, !1}
!1 = !{!"llvm.loop.unroll.disable", i1 true}

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

4 participants