arlad dies on NetBSD/sparc64 3.0

[Jonathan A. Kollasch]

On Tue, Jan 24, 2006 at 01:36:29PM -0600, Jonathan A. Kollasch wrote:
> Hi,
> 	I've been testing arla on NetBSD/sparc64 3.0 in hopes of using
> AFS for the $HOME space on my primary headless-workstation.  Anyway
> arla compiles from pkgsrc fine.  It even runs fine until you put
> a good load (cksum(1)ing various distfiles in AFS in this case) on it,
> then arla gives syslog a message and dies.  Obviously not suitable
> for production use.  The "error" is:
> Jan 24 13:15:56  arla[371]: worker_setdebuginfo: Could not find worker for tag 1021868741
> This tag number seems to look about the same every time.
> 
> I'm using pkgsrc-2005Q4 and netbsd-3-0-RELEASE sources, and the stock
> GENERIC kernel on an Ultra 10.
> 
> Any insight or suggestions of things to try would be most appreciated.

Here's the synopsis of the BT:

Core was generated by 'arlad'.
Program terminated with signal 10, Bus error.
...
#0 0x127e00 in LWP_WaitProcess
#1 0x128738 in LWP_WaitProcess
#2 0x126f3c in LWP_CreateProcess
#3 0x128264 in LWP_WaitProcess
#4 0x1278ec in LWP_WaitProcess
#5 0x128730 in LWP_WaitProcess
#6 0x129898 in IOMGR_Initialize
Cannot access memory at address 0x72

ISTR that pthreaded apps (such as Firefox and Apache) on NetBSD/sparc64
suffer stability problems.  This may also be troubling arlad if the LWP
functions are similar to (or wrappers around?) the pthread stuff.

Is signal 10 consistent with crashes of firefox and apache?

	Jonathan Kollasch
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 186 bytes
Desc: not available
Url : http://lists.stacken.kth.se/pipermail/arla-drinkers/attachments/20060125/36b1ef2e/attachment.bin