On Thu, 22 Jan 2026, Kalvis Duckmanton wrote:

> >   Have you run your fix through full regression-testing or is it only the
> > reproducer from the PR and presumably the build of GCC that the fix has
> > been verified against?  Regardless, I'll push it through verification in
> > my own environment and let you know of the outcome.
> Only the reproduction case from the PR and builds of GCC - both on
> NetBSD/amd64 and Linux/i686.

 OK.  The more verification the better as it increases the diversity of 
test environments, e.g. I use PPC64/Linux as the test host.  Plus sadly I 
might be the only one nowadays to ever run regression-testing with the 
VAX/NetBSD target, so the bus factor is pretty low.

> >   I'm curious as to what has caused this regression, or alternatively what
> > is different that causes the problem with your configuration but not mine,
> > as I was certainly able to build a GCC 15 snapshot just fine shortly after
> > GCC 14 was branched:
> > 
> > $ vax-netbsdelf-gcc --version
> > vax-netbsdelf-gcc (GCC) 15.0.0 20240525 (experimental)
> > [...]
> I suspect that it's some function of the language being compiled (C++ vs C),
> the complexity of the input, the optimisation level requested and the phase of
> the moon.  I do know that GCC 12.2.0 also terminated abruptly under the same
> circumstances and that the reproduction case was quite a bit smaller.  I'll
> send you some more details of my build environment off-list.

 Weird, but the issue does trigger here with your reproducer included with 
the PR, so it's good enough to work with.  Compilation succeeds with GCC 
11 though, so it's indeed a regression and might be related to the switch 
from CC0 to MODE_CC internal condition code representation.  I might try 
to bisect it.

 In any case I've built a fresh version of the compiler with no issue and 
regression-testing is under way.  As I recall it took ~72h to complete on 
the last occasion, so it'll take a while yet and then another run will be 
required with your change applied.

 I'll keep you posted.

  Maciej

Reply via email to