Re: flsym

[Mike Hibler <mike@fast.cs.utah.edu>]

Ah, thanks.  This is the key part:

0x0257669 in (0x0257650) __assert()
0x022b5f1 in (0x022b5c8) ani_demuxkey_removechannel()
0x02285e6 in (0x02281dc) ani_chan_init()
0x022094a in (0x022078c) an_outchan_create()
0x0101f9b in (0x0101de0) getUDPOutFromContext()
0x021e7bd in (0x021e754) jsiGetPlexOutLinkFromContext()
0x021f8d8 in (0x021f810) plexSetup()
0x021fd88 in (0x021fd30) jsiPlexWindowRecvPacket()
0x010184f in (0x01017a4) jsiCallBack()
0x02318ce in (0x0231408) ani_thread_loop()
0x023c5c8 in (0x023c584) pthread_start_thread()

which shows that an attempt to create a thread failed and it tried to
clean up.  This is exactly the case that should have been fixed by the
patch that Tim sent.  Something fishy is going on.  It sounds like there
might be a problem with our Makefiles and maybe the file didn't get
recompiled, or the recompiled module didn't get put into libmoab, or
the kernel didn't get relinked, etc.

You can verify this by sticking in a printf like:

	void
	ani_demuxkey_removechannel(ani_demuxkey_t *dkey, ani_chan_t *chan)
	{
	printf("In removechannel, dkey=%p, chan=%p\n", dkey, chan);
		assert(dkey->chan == chan);
		dkey->chan = 0;
	}


If it still panics and you don't see the printf, then the change never
made it into your kernel.