Summary:	ASTERISK-01811: System panics - I think its to do with ZAP.
Reporter:	deon (deon)	Labels:
Date Opened:	2004-06-12 23:07:54	Date Closed:	2004-09-25 02:42:34
Priority:	Major	Regression?	No
Status:	Closed/Complete	Components:	Core/General
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:
Description:	Hi Guys, I'm not having a lot of luck with my new Opteron :-( I've submitted a bug, when trying to run asterisk compiled in x86_64 -(its bug 1820). So, I reinstalled fedora core 2 and am running in 32bit mode. I compiled the zaptel drives (CVS date 20040610) and the drivers detect my X101P and TDM400P - everything there seems good. I then ran for a little while, a compiled version of asterkisk (1.0-stable) that was compiled on my old pentium III 450Mhz SMP box (RH 7.3) - asterisk runs very well (only tried for a day at most). I made a small change to the source (changed #define DEFAULT_CIDRINGS 2 in chan_zap.c as I'm in Australia - saw this in google somewhere) and recompiled on Fedora Core 2 (32bit). When I run my (FC2) compiled asterisk (1.0-stable) - it'll only run for a few minutes (sometimes 5mins sometimes 20mins), then my server panics - and it seems to panic more frequently when I'm using my FXS phone. Unfortunatley, the panic doesnt make it to syslog, as the system locks up, and I need to reboot. Looking at the console, it seems there are many lines of panic output - so I cant even capture how the panic started. When I revert back to my RH 7.3 compiled version of asterisk (1.0-stable), it runs very well. I'm happy to help you guys any way I can to help you debug this - if you dont have a solution already. (Wishing I could be running 64bit linux... :-) ****** ADDITIONAL INFORMATION ****** System is: AMD Opteron on a Tyan S2882 board - 512MB X101P & TDM400P (with 1 FXS). Fedora Core 2 - 32bit - no updates so kernel 2.6.5-1.358 gcc-3.3.3-7 Anything else I can give you?
Comments:	By: Mark Spencer (markster) 2004-06-13 01:56:47 We've already seen a bug about this and the solution was to turn off optimizations beyond "-O" in the zaptel Makefile.  Seems to be some compiler bug with respect to x86-64. By: zoa (zoa) 2004-06-13 08:14:32 i'm running a dual opteron on Tyan MOBO with kernel 2.6.5 and TE410p and did not see any of problems once i got it to compile. (i'm on suse64) Its not that fast though. By: Mark Spencer (markster) 2004-06-15 15:31:53 Any more input here? By: deon (deon) 2004-06-17 01:40:35 markster - this might have been fixed by bug 1820 - I've recompiled (after the work you guys did for 1820) and it works a treat... (I didnt make any changes to the makefiles for zaptel or asterisk...) By: Malcolm Davenport (mdavenport) 2004-06-17 09:41:44 Resolving pending further notice.