On Wed, Aug 19, 2009 at 09:58:04PM -0700, Matt Helsley wrote:
> The robust futex test can hang if the kernel fails to properly set the
> robust list pointer. This currently happens during restart. The test should not
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Argh. Forgot to update the description and remove this sentence. I do
not wish to report a bug :).
Nonetheless, the patch is a nice way to ensure the test
doesn't hang under as many circumstances as possible.
> hang and instead should report failure.
>
> Use a timeout to ensure that hangs are caught and reported as failure.
> The timeout should return ETIMEDOUT. This limits the total amount of
> time checkpoint/restart can take so a suitable timeout is essential here.
>
> Signed-off-by: Matt Helsley <matthltc@xxxxxxxxxx>
> Cc: containers@xxxxxxxxxxxxxxxxxxxxxxxxxx
> ---
> futex/robust.c | 11 ++++++++---
> 1 files changed, 8 insertions(+), 3 deletions(-)
>
> diff --git a/futex/robust.c b/futex/robust.c
> index 1025515..304c35f 100644
> --- a/futex/robust.c
> +++ b/futex/robust.c
> @@ -103,6 +103,10 @@ void add_rfutex(struct futex *rf)
>
> void acquire_rfutex(struct futex *rf, pid_t tid)
> {
> + struct timespec timeout = {
> + .tv_sec = 5,
> + .tv_nsec = 0
> + };
> int oldval, newval, val = 0;
>
> rlist.list_op_pending = &rf->rlist; /* ARCH TODO make sure this assignment is atomic */
> @@ -126,7 +130,7 @@ void acquire_rfutex(struct futex *rf, pid_t tid)
> val = __sync_or_and_fetch(&rf->tid.counter, FUTEX_WAITERS);
> log("INFO", "futex(FUTEX_WAIT, %x)\n", val);
> if (futex(&rf->tid.counter, FUTEX_WAIT, val,
> - NULL, NULL, 0) == 0)
> + &timeout, NULL, 0) == 0)
> break;
> log("INFO", "futex returned with errno %d (%s).\n", errno, strerror(errno));
> switch(errno) {
> @@ -147,8 +151,9 @@ void acquire_rfutex(struct futex *rf, pid_t tid)
> log("WARN", "EINTR while sleeping on futex\n");
> continue;
> case ETIMEDOUT:
> - log("WARN", "ETIMEDOUT while sleeping on futex\n");
> - continue;
> + log("FAIL", "ETIMEDOUT while sleeping on futex.\n");
> + fail++;
> + return;
> case EACCES:
> log("FAIL", "FUTEX_WAIT EACCES - no read access to futex memory\n");
> fail++;
> --
> 1.5.6.3
>
>
_______________________________________________
Containers mailing list
Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/containers
