From: Rich Felker <dalias@libc.org>
To: musl@lists.openwall.com
Subject: Re: [PATCH v2 2/2] pthread: implement try/timed join variants
Date: Sun, 11 Oct 2015 22:30:20 -0400 [thread overview]
Message-ID: <20151012023020.GM8645@brightrain.aerifal.cx> (raw)
In-Reply-To: <1443988530-23978-2-git-send-email-koorogi@koorogi.info>
On Sun, Oct 04, 2015 at 02:55:30PM -0500, Bobby Bingham wrote:
> ---
> include/pthread.h | 2 ++
> src/thread/pthread_join.c | 29 ++++++++++++++++++++++-------
> 2 files changed, 24 insertions(+), 7 deletions(-)
>
> diff --git a/include/pthread.h b/include/pthread.h
> index af70b73..3d2e0c4 100644
> --- a/include/pthread.h
> +++ b/include/pthread.h
> @@ -214,6 +214,8 @@ struct cpu_set_t;
> int pthread_getaffinity_np(pthread_t, size_t, struct cpu_set_t *);
> int pthread_setaffinity_np(pthread_t, size_t, const struct cpu_set_t *);
> int pthread_getattr_np(pthread_t, pthread_attr_t *);
> +int pthread_tryjoin_np(pthread_t, void **);
> +int pthread_timedjoin_np(pthread_t, void **, const struct timespec *);
> #endif
>
> #ifdef __cplusplus
> diff --git a/src/thread/pthread_join.c b/src/thread/pthread_join.c
> index 694d377..b041ff8 100644
> --- a/src/thread/pthread_join.c
> +++ b/src/thread/pthread_join.c
> @@ -5,18 +5,33 @@ int __munmap(void *, size_t);
> void __pthread_testcancel(void);
> int __pthread_setcancelstate(int, int *);
>
> -int __pthread_join(pthread_t t, void **res)
> +int __pthread_tryjoin_np(pthread_t t, void **res)
> {
> - int tmp, cs;
> - __pthread_testcancel();
> - __pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &cs);
> - if (cs == PTHREAD_CANCEL_ENABLE) __pthread_setcancelstate(cs, 0);
> - while ((tmp = t->tid)) __timedwait_cp(&t->tid, tmp, 0, 0, 0);
> - __pthread_setcancelstate(cs, 0);
> + if (t->tid) return EBUSY;
> a_barrier();
> if (res) *res = t->result;
> if (t->map_base) __munmap(t->map_base, t->map_size);
> return 0;
> }
This can spuriously fail when lack of memory synchronization causes
the caller not to see that the target thread has exited. However this
condition does not seem observable; calling pthread_tryjoin_np from
multiple threads where the calls could race is certainly UB, and
otherwise there should be no way to distinguish between the state of
just-before-exiting and already-exited. So I _think_ this is okay.
> +int __pthread_timedjoin_np(pthread_t t, void **res, const struct timespec *at)
> +{
> + int tmp, cs, r = 0;
> + __pthread_testcancel();
> + __pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &cs);
> + if (cs == PTHREAD_CANCEL_ENABLE) __pthread_setcancelstate(cs, 0);
> + while ((tmp = t->tid) && r != ETIMEDOUT && r != EINVAL)
> + r = __timedwait_cp(&t->tid, tmp, CLOCK_REALTIME, at, 0);
> + __pthread_setcancelstate(cs, 0);
> + if (r == ETIMEDOUT || r == EINVAL) return r;
> + return __pthread_tryjoin_np(t, res);
> +}
> +
> +int __pthread_join(pthread_t t, void **res)
> +{
> + return __pthread_timedjoin_np(t, res, 0);
> +}
> +
> +weak_alias(__pthread_tryjoin_np, pthread_tryjoin_np);
> +weak_alias(__pthread_timedjoin_np, pthread_timedjoin_np);
> weak_alias(__pthread_join, pthread_join);
This all looks okay to me. I'll probably commit the patch as-is, as
long as nobody thinks the reasoning on the tryjoin issue above is
wrong.
Rich
next prev parent reply other threads:[~2015-10-12 2:30 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-10-04 19:55 [PATCH v2 1/2] pthread: add missing memory barrier Bobby Bingham
2015-10-04 19:55 ` [PATCH v2 2/2] pthread: implement try/timed join variants Bobby Bingham
2015-10-12 2:30 ` Rich Felker [this message]
2015-10-16 3:15 ` Rich Felker
2016-05-07 18:48 ` [PATCH v3] " Bobby Bingham
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20151012023020.GM8645@brightrain.aerifal.cx \
--to=dalias@libc.org \
--cc=musl@lists.openwall.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://git.vuxu.org/mirror/musl/
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).