repeatable system lockups on 3 TC DEC Alpha's

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

repeatable system lockups on 3 TC DEC Alpha's

Alastair Boyanich
Hi, I'll keep this brief, I'm sure you're busy. I have a number of alpha's.
I'm getting system lockups under 3.8 on:

3000/600  - Sandpiper  175mhz with 96mb ram + 4.3gb
3000/700  - Sandpiper45  225mhz with 128mb ram + 4.3gb
3000/700  - Sandpiper45  225mhz with 128mb ram + 36.1gb

It's highly repeatable (every time) and using the default kernel. I can and
have compiled another minimalistic kernel on a stable machine (200 4/166
mustang) and moved it over. Same deal. Under very heavy system loads with a
small amount of swap space used (~21mb) after a while it will lock solid and
drop me to the debugger. I have swapped memory around completely, removed
all turbo channel expansion cards. Nothing exotic. Running off serial
console only. VMS runs stabilly as does Tru64 on these actual machines.
Grappled with this problem for a few weeks now.

They all give the same error when they lock solid.
Stopped at      Debugger+0x4:   ret     zero,(ra)

Attached is a log of the trace and ps + friends from the debugger. Anything
I can do to help or provide further details please don't hesitate to ask. To
crash it, I log in, fire up ssh, thrash the disk by setting it checking out
/usr/src to /tmp, check out pkgsrc to /tmp/pkg from netbsd, set it compiling
/usr/src also with make build, then randomly pick pkgsrc packages to build.
Give it 10mins and it's locked solid. Thank you for your time.

Regs Al Boyanich.
Perth, Western Australia

[demime 1.01d removed an attachment of type application/octet-stream which had a name of crash.log]