Corey,

On Sat, Jul 25, 2009 at 3:07 AM, Corey J Ashford <[email protected]>wrote:

> stephane eranian <[email protected]> wrote on 07/24/2009 05:31:16 PM:
>
> [snip]
> > But you run into the problem if the monitored process is using signals.
> > For instance, if the program is using SIGALRM, then SIGIO may be
> > delivered to the wrong thread. If your program does not use any signal
> > then, you may be okay (assuming libpthread does not use signals
> internally).
> >
> > I do not have a good fix for this now but should have one by next week.
> >
>
> It sounds like the fix might require changing signal-handling code in the
> kernel, is that right?  If so, this might be a nice feature to request
> upstream, to be able to say, "this signal must be queued to this thread."
>

They have that API already, e.g., for tkill() or some synchronous signals.
But here we want an asynchronous signal delivered to a specific thread.

I think we just need a new internal interface or a parameter to
kill_fasync() to indicate which queue to use.

BTW, PCL suffers from that same problem. I will post something
about this on LKML on Monday.
------------------------------------------------------------------------------
_______________________________________________
perfmon2-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel

Reply via email to