Hi! This UB is mine, sorry for the hassle;
I'm currently trying to look into libcpp more deeply
before submitting a patch to fix it and break another
behaviour again.
For transparency, here are my current approaches:
1. Revert to the old behaviour, but change glue_header_name
to not consume EOF (i.e. _cpp_get_token_no_padding). This
would fix the crash and the warnings, but I'm trying to
see if it's idiomatic and wouldn't break anything else.
or
2. Understand why for OpenMP pragmas (pfile->state.in_directive)
is 0, and validate if it makes sense to early return only if
not inside that as well (i.e. treating __has_include as a stray
identifier when not in any preprocessor directive)
I hope to send out a patch for that still in the current week.
Best regards,
Léo
On 2026-09-29 15:49, Aldy Hernandez wrote:
On Tue, Sep 29, 2026 at 11:40:47AM -0700, Andrea Pinski wrote:
> Unless I'm missing something...
You are not. See
https://gcc.gnu.org/pipermail/gcc-patches/2026-September/732195.html .
Since it is only c-c++-common/gomp/has-include-1.c that has been a
spurious test failure. I have not worked on it yet.
Ughh, I'm only a week late ;-). Serves me right for not keeping up
with the lists.
Thanks.
Aldy
