https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125446
--- Comment #4 from bill <w.langdon at cs dot ucl.ac.uk> ---
Dear Drea Pinski,
Thank you for investigating.
I agree it must be frustrating to find the fault not reproducible.
Is there any further information or examples you need from this end
which might help explain why it fails here but not at your end.
Perhaps I should have explained that the problem started in
a bigger program and I have tried to find a minimal example
which still manifests the problem.
We are running Rocky 9.7
g++ was run on /tmp/ (df -k says 677050900 1K-blocks Available)
Top said g++ changed process id during the failing run.
I am guessing, but would g++ have created a new process
in order to write /tmp/cc47cICb.out
Many thanks for looking into this and please do not hesitate to
let me know if I can help
Bill
ps: sometime g++ produces an extra line in the error report:
during RTL pass: expand
