https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125804
Bug ID: 125804
Summary: [17 Regression] 403.gcc miscompares
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Keywords: needs-bisection, wrong-code
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: pheeck at gcc dot gnu.org
Blocks: 26163
Target Milestone: ---
Host: aarch64-gnu-linux, x86_64-pc-linux-gnu
Target: aarch64-gnu-linux, x86_64-pc-linux-gnu
Compiling the 403.gcc SPEC CPU 2006 benchmark with trunk GCC results in a
miscompare
*** Miscompare of 166.s; for details see
/home/gcc/buildworker/source/cpu2006/benchspec/CPU2006/403.gcc/run/run_peak_ref_amd64-m64-mine.0000/166.s.mis
1020: je .L14338
je .L14341
^
1483: .L14338:
.L14341:
^
1914: je .L14340
je .L14343
^
2084: je .L14342
je .L14345
^
2196: .L14342:
.L14345:
^
2289: .L14340:
.L14343:
^
2501: je .L14339
je .L14342
^
2655: je .L14341
je .L14344
^
2767: .L14341:
.L14344:
^
2841: .L14339:
.L14342:
^
I've seen this on x86_64 -O2 -march=native on both Intel and AMD machines
(Zen{2,3,4,5}, Ice Lake (3rd generation Xeon)). I've also seen this on Aarch64
machines with -O2 -flto generic march (Ampere Altra Neoverse N1). I've also
seen this with many other combinations of cmdline flags.
Started happening some time around 28th March. I'll bisect.
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26163
[Bug 26163] [meta-bug] missed optimization in SPEC (2026, 2k17, 2k and 2k6 and
95)
