On Wed, Jul 10, 2024 at 9:31 PM Samuel Thibault <samuel.thiba...@gnu.org> wrote:
> Sergey Bugaev, le mer. 10 juil. 2024 21:07:21 +0300, a ecrit: > > Do you expect the process itself to query its threads' names locally, > > or do you expect another process (GDB) to issue thread_get_name > > remotely? > > We'll want e.g. gdb to use thread_get_name to get it indeed. > > > On Wed, Jul 10, 2024 at 7:04 PM Flavio Cruz <flavioc...@gmail.com> > wrote: > > > diff --git a/htl/pt-internal.h b/htl/pt-internal.h > > > index 85a7d905..b6ccbe12 100644 > > > --- a/htl/pt-internal.h > > > +++ b/htl/pt-internal.h > > > + char *thread_name; > > > > Does it make sense to store the name locally? Could pthread_getname_np > > call thread_get_name also, if we don't expect it to be a frequent > > operation? > > > > Not saying you shouldn't store the name, but let's think of the space > > vs performance trade-off and document it. > > I don't think pthread_getname_np will be a frequent operation, so it's > probably not worth storing it inside glibc. That'll make the > implementation much simpler and avoid the question of the mutex :) > Thanks Sergey and Samuel for the reviews. I have published a second version with a new thread_get_name RPC that is used to implement pthread_getname_np. > Samuel >
