From: "Lucas A. M. Magalhaes" <lamm@linux.ibm.com>
To: libc-alpha@sourceware.org
Subject: [PATCH] nptl: Fix __futex_clocklock64 return error check [BZ #26964]
Date: Fri, 27 Nov 2020 11:33:59 -0300 [thread overview]
Message-ID: <20201127143359.1844560-1-lamm@linux.ibm.com> (raw)
The earlier implementation of this, __lll_clocklock, calls lll_clockwait
that doesn't return the futex syscall error codes. It always tries again
if that fails.
However in the current implementation, when the futex returns EAGAIN,
__futex_clocklock64 will also return EGAIN, even if the futex is taken.
This patch fixes the EAGAIN issue and also adds a check for EINTR. As
futex syscall can return EINTR if the thread is interrupted by a signal.
In this case I'm assuming the function should continue trying to lock as
there is no mention to about it on POSIX. Also add a test for both
scenarios.
---
nptl/Makefile | 2 +-
nptl/tst-pthread-timedlock-lockloop.c | 136 ++++++++++++++++++++++++++
sysdeps/nptl/futex-internal.h | 9 ++
3 files changed, 146 insertions(+), 1 deletion(-)
create mode 100644 nptl/tst-pthread-timedlock-lockloop.c
diff --git a/nptl/Makefile b/nptl/Makefile
index a48426a396..91324e09f2 100644
--- a/nptl/Makefile
+++ b/nptl/Makefile
@@ -298,7 +298,7 @@ tests = tst-attr2 tst-attr3 tst-default-attr \
tst-thread-affinity-sched \
tst-pthread-defaultattr-free \
tst-pthread-attr-sigmask \
-
+ tst-pthread-timedlock-lockloop \
tests-container = tst-pthread-getattr
diff --git a/nptl/tst-pthread-timedlock-lockloop.c b/nptl/tst-pthread-timedlock-lockloop.c
new file mode 100644
index 0000000000..abcc5724cf
--- /dev/null
+++ b/nptl/tst-pthread-timedlock-lockloop.c
@@ -0,0 +1,136 @@
+/* Make sure pthrea_mutex_timedlock doesn't return spurious error codes.
+
+ Copyright (C) 2020 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <https://www.gnu.org/licenses/>. */
+#include <errno.h>
+#include <pthread.h>
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+
+#define NANO_PER_SEC 1000000000LL
+#define TIMEOUT (NANO_PER_SEC / 1000LL)
+#define NUM_THREADS 50
+#define RETEST_TIMES 100
+
+struct worker_args
+{
+ pthread_mutex_t mtx;
+ int runs;
+};
+
+void
+signal_handler (int sig_num)
+{
+ if (sig_num != SIGUSR1)
+ printf ("Unexpected signal");
+}
+
+/* Call pthread_mutex_timedlock()/pthread_mutex_unlock() repetitively, hoping
+ that one of them returns EAGAIN or EINTR unexpectedly. */
+static void *
+worker (void *arg)
+{
+ pthread_mutex_t *mtx = &(((struct worker_args *) arg)->mtx);
+ struct timespec abs_time;
+ signal (SIGUSR1, signal_handler);
+
+ for (unsigned run = 0; run < ((struct worker_args *) arg)->runs; run++)
+ {
+ clock_gettime (CLOCK_REALTIME, &abs_time);
+ abs_time.tv_nsec += TIMEOUT;
+ if (abs_time.tv_nsec >= NANO_PER_SEC)
+ {
+ abs_time.tv_sec++;
+ abs_time.tv_nsec -= NANO_PER_SEC;
+ }
+
+ int ret = pthread_mutex_timedlock (mtx, &abs_time);
+
+ if (ret == 0)
+ pthread_mutex_unlock (mtx);
+
+ if (ret == EAGAIN || ret == EINTR)
+ {
+ printf ("Unexpected return %d\n", ret);
+ return (void *) 1;
+ }
+ }
+ return NULL;
+}
+
+static int
+do_test (void)
+{
+ pthread_t *workers =
+ (pthread_t *) malloc (NUM_THREADS * sizeof (pthread_t));
+
+ struct worker_args args;
+ pthread_mutex_init (&(args.mtx), NULL);
+
+ int *thread_ret = NULL, ret = 0;
+
+ /* Run the checks to catch an EAGAIN. */
+ /* As there is no way to ensure the error condition, just run the test many
+ times hoping to catch the error. */
+ args.runs = 100;
+ for (int run = 0; run < RETEST_TIMES; run++)
+ {
+ for (int i = 0; i < NUM_THREADS; i++)
+ {
+ pthread_create (&workers[i], NULL, worker, (void *) &args);
+ }
+ for (int i = 0; i < NUM_THREADS; i++)
+ {
+ pthread_join (workers[i], (void **) &thread_ret);
+ if (thread_ret != NULL)
+ ++ret;
+ }
+ if (ret != 0)
+ goto err;
+ }
+
+ /* Run the test to check if we catch an EINTR. */
+ /* As there is no way to ensure the error condition, just run the test many
+ times hoping to catch the error. */
+ pthread_t thread;
+ args.runs = 1;
+ for (int i = 0; i < RETEST_TIMES * 1000; i++)
+ {
+ if (pthread_mutex_lock (&(args.mtx)) != 0)
+ {
+ printf ("Mutex lock failed\n");
+ goto err;
+ }
+ pthread_create (&thread, NULL, worker, (void *) &args);
+ pthread_kill (thread, SIGUSR1);
+ pthread_mutex_unlock (&(args.mtx));
+ pthread_join (thread, (void **) &thread_ret);
+ if (thread_ret != NULL)
+ goto err;
+ }
+
+ free (workers);
+ return 0;
+
+err:
+ free (workers);
+ return 1;
+}
+
+#include <support/test-driver.c>
diff --git a/sysdeps/nptl/futex-internal.h b/sysdeps/nptl/futex-internal.h
index e67803952f..f16d26d994 100644
--- a/sysdeps/nptl/futex-internal.h
+++ b/sysdeps/nptl/futex-internal.h
@@ -424,10 +424,19 @@ __futex_clocklock64 (int *futex, clockid_t clockid,
{
while (atomic_exchange_acq (futex, 2) != 0)
{
+ /* At this point we tried to get the futex but failed and set its
+ value to 2. However the futex value can be changed by other
+ thread before this calls the futex syscall. If so the syscall
+ will return EAGAIN. */
err = __futex_abstimed_wait64 ((unsigned int *) futex, 2, clockid,
abstime, private);
if (err == EINVAL || err == ETIMEDOUT || err == EOVERFLOW)
break;
+ /* If EAGAIN or EINTR is returned here the error code should be reset
+ as we will try again to acquire the futex and it may success.
+ Otherwise the mutex will be locked and the return will not be 0. */
+ if (err == EAGAIN || err == EINTR)
+ err = 0;
}
}
return err;
--
2.26.2
next reply other threads:[~2020-11-27 14:34 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-11-27 14:33 Lucas A. M. Magalhaes [this message]
-- strict thread matches above, loose matches on Subject: below --
2020-11-27 14:33 Lucas A. M. Magalhaes
2020-11-27 17:27 ` Adhemerval Zanella
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=20201127143359.1844560-1-lamm@linux.ibm.com \
--to=lamm@linux.ibm.com \
--cc=libc-alpha@sourceware.org \
/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.
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 read-only IMAP folder(s) and NNTP newsgroup(s).