On Mon, Nov 17, 2025 at 02:45:28PM +0100, Claudio Jeker wrote: > This tripped over gcc15 in c23 mode. Our pthreads mutexes > (pthread_mutex_t) are typedef-ed to volatile struct pthread_mutex * > so this should do the same.
It's probably fine without a bump, but I'd probably still do it. Also the other hunk can go away once we switch to gcc 15 :) ok tb > > -- > :wq Claudio > > Index: patches/patch-threads_h > =================================================================== > RCS file: /cvs/ports/devel/libstdthreads/patches/patch-threads_h,v > diff -u -p -r1.1 patch-threads_h > --- patches/patch-threads_h 2 Sep 2024 16:39:12 -0000 1.1 > +++ patches/patch-threads_h 30 Oct 2025 14:18:50 -0000 > @@ -1,4 +1,5 @@ > __has_extension support added in gcc >= 14 > +Fixup mtx_t definition to mirror pthread_mutex_t > > Index: threads.h > --- threads.h.orig > @@ -14,3 +15,12 @@ Index: threads.h > #if __has_extension(c_thread_safety_attributes) > #define __lock_annotate(x) __attribute__((x)) > #else > +@@ -97,7 +101,7 @@ > + */ > + > + typedef struct pthread_cond *cnd_t; > +-typedef struct pthread_mutex *mtx_t; > ++typedef volatile struct pthread_mutex *mtx_t; > + typedef struct pthread *thrd_t; > + typedef int tss_t; > + >
