mutex held in a thread which is cancelled stays busy
deischen at freebsd.org
Wed Aug 7 03:38:54 UTC 2019
> On Aug 6, 2019, at 9:55 PM, Erich Dollansky <freebsd.ed.lists at sumeritec.com> wrote:
> On Tue, 6 Aug 2019 20:58:30 -0400
> Daniel Eischen <deischen at freebsd.org> wrote:
>>> On Aug 6, 2019, at 4:54 AM, Erich Dollansky
>>> <freebsd.ed.lists at sumeritec.com> wrote:
>>> for testing purpose, I did the following.
>>> Start a thread, initialise a mutex in a global variable, lock the
>>> mutex and wait in that thread.
>>> Wait in the main program until above's thread waits and cancel it.
>>> Clean up behind the cancelled thread but leave intentional the mutex
>>> I would have expected now to get an error like 'EOWNERDEAD' doing
>>> operations with that mutex. But I get 'EBUSY' as the error.
>> Are you initializing the mutex as a robust mutex, via
>> pthread_mutexattr_setrobust()? Are you using _lock() or _trylock()?
>> For _trylock(), you only get EOWNERDEAD for robust mutexes. It seems
>> that you should get EOWNERDEAD for _lock() in this case, so if that's
>> what you're doing, it sounds like it might be a bug.
> I did both. One time with initialising the mutex with its defaults by
> handing over NULL as the attribute setting and one time with the
> attributes set.
> I use this line to set the attribute:
> pres = pthread_mutexattr_setrobust (& Attr, PTHREAD_MUTEX_ROBUST);
> The following line:
> pthread_mutexattr_getrobust (& Attr, &pres);
> Sets pres also to 1.
> I am doing this on 12.0-STABLE FreeBSD 12.0-STABLE r350391 GENERIC
> amd64 with the systems standard compiler.
> Is this the corrent way of doing it?
Yes, I believe so. I'm curious if the bug also exists in -current.
More information about the freebsd-questions