OOPS (kswapd) on dual-celeron box
recently I set up new woody box for some service test.
Installation was finished well, but the box was very unstable.
Sometimes the box generate "Unable to handle kernel NULL pointer
dereference..." at process kswapd.
I got following oops data, and lost control to that box.
c013771c : __remove_from_queues
c0139e47 : try_to_free_buffers
c013853c : try_to_release_page
c012f4f4 : shrink_cache
c012f7A6 : shrink_cache
c012f80f : try_to_free_pages
c012f8a3 : kswapd_balance_pgdat
c012f8fe : kswapd_balance
c012fa0d : kswapd
c01055a4 : release_segments
Sorry for non-complete data, but after this oops, I cannot control the box
in any way.
so I wrote down process stack contents and manually checked against
System.map.
(Anybody knows about using ksymoops in this case, plz let me know...)
System spec:
QDI LegenX IV (440LX chipset) with dual Celeron 466MHz
physical memory 256MB
swap partition size 1GB
on-board AIC7880 wide scsi host adapter/i82557 Etherexpress 100B ethernet
adapter
IBM 18GB scsi hdd
there is no hdd on PCI-IDE interface and one CD-ROM drive attached.
Because this box was set up recently, there is few connection/user currently
and usually no swap used.
If boot dmesg is required, I'll send message again to this thread.
Any comment will be appreciated, QDI M/B issue, Celeron SMP issue, or
SCSI/NIC issue, etc.
I struggled with this box for almost a month, it sometimes runs for several
days well, or sometimes oops twice in fifteen minutes...:-(
-
Choe, Hyun-ho
Reply to: