Does anyone know when FUTEX_WAIT can fail with EAGAIN?

Does anyone know when FUTEX_WAIT can fail with EAGAIN?

[Steven Stewart-Gallus]

Hello,

I'm trying to debug a hangup where my program loops with FUTEX_WAIT (actually
FUTEX_WAIT_PRIVATE but same thing)  endlessly erring out with EAGAIN. I would
like to know if anyone on the mailing list knows when FUTEX_WAIT can fail with
EAGAIN.

Some info for people who are interested in what I'm working on.

I'm working on sand boxing my program at gitorious.org/linted with namespaces
and chroot.

The error only appeared after changing unshare then fork code into a direct
clone call.

> | uname -a
> Linux alonzo 3.13.0-24-generic #1trisquel1 SMP Tue May 13 15:17:50 UTC 2014
x86_64 x86_64 x86_64 GNU/Linux

Thank you,
Steven Stewart-Gallus

Re: Does anyone know when FUTEX_WAIT can fail with EAGAIN?

[Eric Wong]

Steven Stewart-Gallus <sstewartgallus00@mylangara.bc.ca> wrote:
> I'm trying to debug a hangup where my program loops with FUTEX_WAIT (actually
> FUTEX_WAIT_PRIVATE but same thing)  endlessly erring out with EAGAIN. I would
> like to know if anyone on the mailing list knows when FUTEX_WAIT can fail with
> EAGAIN.

FUTEX_WAIT fails with EAGAIN if the value of *uaddr no longer matches
the expected val.
---
#include <linux/futex.h>
#include <sys/syscall.h>
#include <unistd.h>
#include <errno.h>
int main(void) {
	int op = FUTEX_WAIT;
	int exp = 1;
	int *addr = &exp;
	int val = 1; /* XXX change this to anything else to get EAGAIN */
	int rc = syscall(SYS_futex, addr, op, val, 0, 0, 0);

	if (rc < 0 && errno == EAGAIN)
		write(1, "EAGAIN\n", 7);

	return 0;
}
---
I just encountered this myself yesterday while implementing futex-based
locks/condvars for Ruby:
https://bugs.ruby-lang.org/issues/10009#change-48372

Re: Does anyone know when FUTEX_WAIT can fail with EAGAIN?

[Davidlohr Bueso]

On Tue, 2014-07-29 at 21:55 +0000, Steven Stewart-Gallus wrote:
> Hello,
> 
> I'm trying to debug a hangup where my program loops with FUTEX_WAIT (actually
> FUTEX_WAIT_PRIVATE but same thing)  endlessly erring out with EAGAIN. I would
> like to know if anyone on the mailing list knows when FUTEX_WAIT can fail with
> EAGAIN.

You're probably running into the EWOULDBLOCK return -- which on Linux is
just another name for EAGAIN. So if you didn't know this, and were
reading the code related to FUTEX_WAIT, you'd get mislead by this.

The check is there to avoid userspace misusing futexes and screwing up
applications. When the futex_wait() call is issued, a lot can happen in
the kernel before the task is blocked (key hashing, reference counting,
acquire internal spinlock, etc.). At any point while preparing to block,
another userspace thread could have changed the futex value, and thus
the condition for which futex_wait() call was issued in the first place
might no longer be true.

Thanks,
Davidlohr