Strange behaviour

Hi all,

In our lab we have two B2-stepping BigSur (that is the codename for the 2-cpu ones isn't it ?) machines running slightly modified RH7.1 2.4.3 kernel with SMP support (Yes I know, it's unsupported hardware, but they've worked fine so far). When I do a rather long DMA transfer with a Dolphin SCI card the machine prints this to the serial console and stops :

gSelFlashStartAddr  :  ffe4ffb0
gSelLogLength       :      ffaf
gSelNextEntry       :      5386
gSelEntryCount      :         9
gSelNewEntriesCount :         1
gSelLastLogRead     :         0
gSelCurMaxRecordSize:      13b0
gSelFlags           :        28
gSelFullFlag        :         0
gSelValidFlag       :         1
gSelClearFlag       :         0
gSelControl         :         0
gFlashExists        :         1
gSelFlashStartAddr  :  ffe4ffb0
gSelLogLength       :      ffaf
gSelNextEntry       :      673c
gSelEntryCount      :         a
gSelNewEntriesCount :         2
gSelLastLogRead     :         0
gSelCurMaxRecordSize:      13b0
gSelFlags           :        28
gSelFullFlag        :         0
gSelValidFlag       :         1
gSelClearFlag       :         0
gSelControl         :         0
gFlashExists        :         1

No kernel oops, no kdb, no nothing. Only solution is to power the server off and then on again.

I'm curious to what this is, it seems to come from the machine firmware or something ? Is this related to the beta HW I have, Linux kernel version or is it something else ?

Thanks in advance,

  Steffen Persvold   | Scalable Linux Systems |   Try out the world's best
 mailto:[email protected] |  http://www.scali.com  | performing MPI implementation:
Tel: (+47) 2262 8950 |   Olaf Helsets vei 6   |      - ScaMPI 1.13.8 -
Fax: (+47) 2262 8951 |   N0621 Oslo, NORWAY   | >320MBytes/s and <4uS latency

Steffen> Hi all, In our lab we have two B2-stepping BigSur (that is Steffen> the codename for the 2-cpu ones isn't it ?) machines Steffen> running slightly modified RH7.1 2.4.3 kernel with SMP Steffen> support (Yes I know, it's unsupported hardware, but they've Steffen> worked fine so far). When I do a rather long DMA transfer Steffen> with a Dolphin SCI card the machine prints this to the Steffen> serial console and stops :

Steffen> gSelFlashStartAddr : ffe4ffb0

I believe this is a sign that a machine check (or "init") occurred. Given that B2 is pretty much ancient, I'm not sure there is much that can be done about it.

Steffen> I'm curious to what this is, it seems to come from the Steffen> machine firmware or something ? Is this related to the beta Steffen> HW I have, Linux kernel version or is it something else ?

It's coming from the firmware. Perhaps an older kernel with B2 support would work better, but then again, a lot of bugs got fixed in the kernel since then.

	--david