https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128081
Bug ID: 128081
Summary: [17 Regression] Incorrect code (?) after folding pow
(-Inf, 0.5)
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: law at gcc dot gnu.org
Target Milestone: ---
This was a bit of a surprise when I bisected. So be at least somewhat
skeptical.
After this change:
commit a327f4cf83765f53e54c93523bbdb3cebec9d1fc
Author: Roger Sayle <[email protected]>
Date: Wed Oct 21 13:26:48 2026 +0100
fold-const: Constant fold pow(-Inf,0.5) at compile-time
I'm seeing these execution failures on rv32gcv:
Tests that now fail, but worked before (40 tests):
unix//-march=rv32gcv: gcc:
gcc.target/riscv/rvv/autovec/unop/math-llceilf-run-0.c -ftree-vectorize -O2
-mrvv-max-lmul=dynamic execution test
[ more failures related to math-llceilf-run-0, just with different flags ]
I do see that there is a -Inf test in there. So maybe I shouldn't have been so
surprised. Anyway, I haven't debugged, just bisected and then peeked at the
testcase.
