Messages in this thread | | | From | Tim Schmielau <> | Subject | Re: BUG: soft lockup detected on Phenom with Debian 2.6.24-4 | Date | Wed, 2 Apr 2008 20:15:00 +0100 |
| |
[apologies if you receive this email twice; apparently the exchange server this message was sent through previously triggered one of lkml's taboo expressions]
On Sat, 08 Mar 2008 23:00:15 +0100, Laurent GUERBY wrote:
> I have a system with an "AMD64 Phenom 9500" quad core cpu, 4GB RAM, > "ASUS M3A32 MVP Deluxe wifi" motherboard with latest vendor BIOS (0801). > > I tried stock debian etch kernel (Debian 2.6.18.dfsg.1-18etch1), machine > froze with no message, debian etch backport kernel same, and then > Debian 2.6.24-4 from unstable and I got some messages: machine > is not frozen but some userland processes are (ps says "Dl" state > with child in "Zs" state) and "events/3" is taking 100% cpu > according to top: > > 18 root 15 -5 0 0 0 R 100 0.0 74:59.46 events/3 > > Got to the same state with ubuntu hardy 2.6.24-8-server kernel. All > kernels are untainted, no X running anyway. > > It takes a few hours of doing some stuff, in my case bootstraping or > testing GCC at -j 4, and then the problem happens. > > I did 32 hours of memtest without issue on this system, temperatures > are very low and the case has plenty of airflow, making memory > issue less likely.
I have a very similar, if not the same issue:
I just bought a HP Pavillion 6332 with a Phenom 9500 quad core cpu and 3GB RAM on some ASUS-like looking mainboard with NVidia MCP61 chipset (actually my first PC not assembled from components myself). I installed OpenSUSE 10.3 64bit and tried various kernels (2.6.24.4, 2.6.23.17 + the erratum298-workaround from AMD, OpenSUSE's default 2.6.22.5-31 and 2.6.22.17-0.1), but the machine will hang within less than an hour of intense OpenMP load over all four cores (using a homemade scientific application).
The symptoms of the hang are similar to what Laurent saw: In xosview, the load display one cpu would get stuck (not necessarily at 100%), and usually (but not always) ps would hang in the middle of its output. I could still login (no X running here either) and use the remaining cores, but the OpenMP application would hang in an unkillable state (don't know which as ps gets stuck). Unlike Laurent, I however don't see anything relevant in the syslog.
Occasionally I also had wrong output from the program (i.e., running it twice gives different results, one of them bogus), although that was on 32bit OpenSUSE 10.3, as far as I remember. I know due to the nature of OpenMP this could as well be a bug in my program, but it has not yet happened on any other machine.
The system is well vented and very cool, but as of yet I only had time to run memtest for 6 hours (no errors found).
If anybody has an idea how to debug this, please ask for more information, otherwise I'll just return this machine for refund.
Thanks, Tim
| |