[Bug tree-optimization/125214] [17 Regression] FAIL: gcc.target/powerpc/recip-[67].c - Starting from r17-370-g5dee3c6521c278

2026-05-07 Thread pinskia at gcc dot gnu.org via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125214

Drea Pinski  changed:

   What|Removed |Added

 CC||dhruvc at gcc dot gnu.org

--- Comment #7 from Drea Pinski  ---
*** Bug 125227 has been marked as a duplicate of this bug. ***

[Bug tree-optimization/125214] [17 Regression] FAIL: gcc.target/powerpc/recip-[67].c - Starting from r17-370-g5dee3c6521c278

2026-05-07 Thread pzheng at gcc dot gnu.org via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125214

Pengxuan Zheng  changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |FIXED

--- Comment #6 from Pengxuan Zheng  ---
Fixed.

[Bug tree-optimization/125214] [17 Regression] FAIL: gcc.target/powerpc/recip-[67].c - Starting from r17-370-g5dee3c6521c278

2026-05-07 Thread cvs-commit at gcc dot gnu.org via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125214

--- Comment #5 from GCC Commits  ---
The master branch has been updated by Pengxuan Zheng :

https://gcc.gnu.org/g:9e24eda38e9a6a6c64f80dd56737224bfd4619de

commit r17-384-g9e24eda38e9a6a6c64f80dd56737224bfd4619de
Author: Pengxuan Zheng 
Date:   Thu May 7 09:18:02 2026 -0700

match: Restrict "min|max(a+|-c,b+|-c)" folding to integral types [PR125214]

The original patch adding the "min|max(a+|-c,b+|-c) -> min|max(a,b)+|-c"
pattern
missed integral type checks and caused some powerpc tests to fail.

PR tree-optimization/125214

gcc/ChangeLog:

* match.pd (min|max(a+|-c,b+|-c)): Add integral type check.

gcc/testsuite/ChangeLog:

* gcc.dg/tree-ssa/pr125214.c: New test.

Signed-off-by: Pengxuan Zheng

[Bug tree-optimization/125214] [17 Regression] FAIL: gcc.target/powerpc/recip-[67].c - Starting from r17-370-g5dee3c6521c278

2026-05-07 Thread pinskia at gcc dot gnu.org via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125214

--- Comment #4 from Drea Pinski  ---
Most likely testcase:
```
double f(double a, double b, double c)
{
  return __builtin_fmax(a+c,b+c);
}
```

Compile with -Ofast