Subj : Re: threaded ticker design problem
To   : comp.programming.threads
From : Loic Domaigne
Date : Tue Jan 04 2005 08:18 pm

Hi Paul!

> I'm attempting to solve a problem I am having that deals with the
> interaction of both threads and signals.  (Yeah!)

[snip]

> However, it is the update function, which is called from a signal
> handler, that presents a problem.  Looking at the list of functions
> which are safe to call from a signal handler (POSIX), there aren't any
> pthread_*() calls available.

Correct.

> I do note, however, the presence of sem_post().  I'm thinking it might
> be safe to actually have *two* threads which handle the timer.  One
> thread waits on the semaphore while the signal handler does a
> sem_post() to notify it to increment the counter.

Avoid sema if you can.


> Anyway, thoughts on how this might be architected differently/cleanly
> are requested.

Yes, perhaps. Block with pthread_sigmask() the targeted signal in all 
threads. Dedicated a special thread responsible for the handling of the 
signal. In that thread, waits synchronously on the signal using sigwait().


HTH,
Loic.

.