On Tue, 25 Sep 2007, Jonathan Corbet wrote:

> One quick question:
>
> > Like the previous timerfd API implementation, read(2) and poll(2) are supported
> > (with the same interface).
>
> Looking at that interface, it appears that a process doing a read() on a
> timerfd with no timer set will block for a very long time. It's an
> obvious "don't do that" situation, but perhaps we could help an
> occasional developer get a clue by returning something like -EINVAL when
> the timer has not been set?

That is the same as you try to read once more after an expired timer. You
won't wake up until the next timer event will show up. That is, after at
most TP time for periodic timers, or after the time the next
timerfd_settime() will setup.
I'd like to keep the "timerfd not set yet" and the "timerfd already
expired and not re-armed" acting the same way. That is, wait till next
event happen (unless O_NONBLOCK of course).



- Davide


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Davide Libenzi wrote:
> On Tue, 25 Sep 2007, Jonathan Corbet wrote:
>
>> One quick question:
>>
>>> Like the previous timerfd API implementation, read(2) and poll(2) are supported
>>> (with the same interface).
>> Looking at that interface, it appears that a process doing a read() on a
>> timerfd with no timer set will block for a very long time. It's an
>> obvious "don't do that" situation, but perhaps we could help an
>> occasional developer get a clue by returning something like -EINVAL when
>> the timer has not been set?
>
> That is the same as you try to read once more after an expired timer. You
> won't wake up until the next timer event will show up. That is, after at
> most TP time for periodic timers, or after the time the next
> timerfd_settime() will setup.
> I'd like to keep the "timerfd not set yet" and the "timerfd already
> expired and not re-armed" acting the same way. That is, wait till next
> event happen (unless O_NONBLOCK of course).

Yes. The timer_settime() and read() might for example be done in separate
threads, and it would make sense for the read() to block until the timer
has been armed.

Cheers,

Michael


--
Michael Kerrisk
maintainer of Linux man pages Sections 2, 3, 4, 5, and 7

Want to help with man page maintenance? Grab the latest tarball at
http://www.kernel.org/pub/linux/docs/manpages/
read the HOWTOHELP file and grep the source files for 'FIXME'.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/