Re: master 8c81818673a 6/7: Tune volatile in read_char

[Paul Eggert <eggert@cs.ucla.edu>]

On 2024-08-17 10:03, Pip Cet wrote:

> I think GCC, by its own logic, is right about this one. It doesn't know
> that local_getcjmp is local (and hasn't escaped), so it would be, in
> theory, possible for a signal handler to call 'longjmp' before
> c_volatile is initialized.

For what it's worth, gcc (GCC) 14.2.1 20240801 (Red Hat 14.2.1-1) does 
not issue the warning on x86-64, when Emacs is built with 
--enable-gcc-warnings. Either it's smarter about escaping, or it figures 
it's not worth issuing warnings about what random signal handlers might 
do, or (most likely) it's just the luck of that particular configuation.

Thanks for the patch; I installed it.


> I still think we should default to -Werror=clobbered to catch distributors
> performing LTO builds and such.

Not sure that'd be wise given the false positive that your patch fixes, 
More false positives are likely lurking in other GCC setups, and 
defaulting to -Werror=clobbered would likely break builds due to false 
alarms.