Hi,
bug was with a handling of temporarily failed concat. messages.
It should be fixed in cvs now.
Thanks,
Alex
Wilfried Goesgens wrote:
On Tue, Oct 25, 2005 at 11:30:57AM +0200, Wilfried Goesgens wrote:
On Tue, Oct 25, 2005 at 10:07:22AM +0100, Paul Keogh wrote:
I've atached to it with gdb to find out, that it starts consuming memory
and cpu while it is Sleeping!
That does'nt sound right - are you sure you're looking at the right thread
?
Sometimes I think gdb gets confused around threads.
the gwthread_sleep function, in the line where it actually
sleeps:
ret = poll(&pollfd, 1, milliseconds);
(line 750 in my case)
Because of i've not yet done much with the threadding
stuff, can somebody help me on that point?
What kind of file does it poll?
It is, as far as I remember, polling the read end of a pipe(). The logic
being that you can wake up the thread using gwthread_wakeup() by writing
to
the write end of the pipe().
What could go wrong here?
Not much, really.
Tia, Wilfried Goesgens
hm, the thread i've atached is main according to the threadinfo struct:
(gdb) print *threadinfo
$3 = {self = 2974619232, name = 0xb1a3b49c "main", func = 0,
number = 0, wakefd_recv = 4, wakefd_send = 5, joiners = 0x0,
pid = -1}
(gdb)
so it might be, gdb just ataches the main thread?
So again, how do i change threads?
Or find out, which one of those is the bad guy and what he's doing there?
Wilfried Goesgens
ok, it seems as if the opponent didn't like an emi message.
so it disconnected, and tried to discard the messages it had in it's spool.
that made the actual panic.
Maybe it's the fault of store dir?
