[sles-beta] Problems with mcelog on HP-Proliant Server with AMD Opteron processors

[urs.frey at post.ch]

Hi

Installation of SLES11-SP3 beta2 works smoothly.
I am testing on:
HP Proliant DL385-G5
HP Proliant BL465c-G7 Blade: AMD Opteron(tm) Processor 6172
HP Proliant BL465c-Gen8 Blade : AMD Opteron(TM) Processor 6220
HP Proliant BL460c-Gen8 Blade : Intel(R) Xeon(R) CPU E5-2660 0 @ 2.20GHz

The only thing which irritates me, is that mcelog failed message on my console:

On all AMD Opteron Processors models I get this failure:  "CPU is unsupported"

h04syy:/etc/init.d # /etc/init.d/mcelog start
Starting mcelog... CPU is unsupported
startproc:  exit status of parent of /usr/sbin/mcelog: 1                                       failed
h04syy:/etc/init.d #

I mean under SLES11-SP2 mcelog works with no failure.
On SLES11-SP2 there is an older release of mcelog

h04wwl:~ # uname -a
Linux h04wwl 3.0.58-0.6.6-default #1 SMP Tue Feb 19 11:07:00 UTC 2013 (1576ecd) x86_64 x86_64 x86_64 GNU/Linux
h04wwl:~ # cat /etc/SuSE-release
SUSE Linux Enterprise Server 11 (x86_64)
VERSION = 11
PATCHLEVEL = 2
h04wwl:~ #
h04wwl:~ # cat /proc/cpuinfo | grep "model name"
model name      : AMD Opteron(tm) Processor 6172
h04wwl:~ # ps -ef | grep mcelog
root      5603     1  0 Feb27 ?        00:00:00 /usr/sbin/mcelog --daemon --config-file /etc/mcelog/mcelog.conf
root     31534 31503  0 10:27 pts/0    00:00:00 grep mcelog
h04wwl:~ # cat /var/run/mcelog.pid
5603
h04wwl:~ #
freyu at h04wwl:~> rpm -q --changelog mcelog-1.0.2011.06.08-0.11.1 | less
* Tue Apr 17 2012 trenn at suse.de
- Add Ivy Bridge support (bnc#748484)
- Add forgotten CPU model 0x25 (bnc#742716)

On Intel mcelog works on SLES11-SP3 beta2
model name      : Intel(R) Xeon(R) CPU E5-2660 0 @ 2.20GHz
h05cni:~ #
h05cni:~ # uname -a
Linux h05cni 3.0.65-0.9-default #1 SMP Mon Feb 25 07:21:23 UTC 2013 (055263a) x86_64 x86_64 x86_64 GNU/Linux
h05cni:~ # cat /etc/SuSE-release
SUSE Linux Enterprise Server 11 (x86_64)
VERSION = 11
PATCHLEVEL = 3
h05cni:~ # ps -ef | grep mcelog | grep -v grep
root      4949     1  0 10:33 ?        00:00:00 /usr/sbin/mcelog --daemon --config-file /etc/mcelog/mcelog.conf
h05cni:~ # cat /var/run/mcelog.pid
4949
h05cni:~ #

But on AMD processors mcelog does not work anymore on SLES11-SP3 beta2
model name      : AMD Opteron(TM) Processor 6220
h05cnh:~ # ps -ef | grep mcelog | grep -v grep
h05cnh:~ # uname -a
Linux h05cnh 3.0.65-0.9-default #1 SMP Mon Feb 25 07:21:23 UTC 2013 (055263a) x86_64 x86_64 x86_64 GNU/Linux
h05cnh:~ # cat /etc/SuSE-release
SUSE Linux Enterprise Server 11 (x86_64)
VERSION = 11
PATCHLEVEL = 3
h05cnh:~ #


With SLES11-SP3 a new release of mcelog has been included: This is new in the RPM changelog:
h05cnh:~ # rpm -qa | grep mcelog
mcelog-1.0.2013.01.18-0.7.7
h05cnh:~ # rpm -q --changelog mcelog-1.0.2013.01.18-0.7.7 | less
* Mon Jan 21 2013 trenn at suse.de
- Updated to latest git state (2013-01-18) (fate#313746)
  and removed patches which went upstream meanwhile (for example
  enabling ivybridge, fate#313745). This one has latest IvyBridge EX/EP
  CPU decoding support
- Removed wrongly used getdomainname call and replaced it with
  a dnslookup call as done in factory for a while

QUESTION:
Is there now a special configuration needed for AMD Opteron processors?
When I diff /etc/mcelog/mcelog.conf from SLES11-SP2 and SLES11-SP3 there is no difference at all.

So why is there the message "CPU is unsupported" only on Amd Opteron processors?
Can it be, that with the removal of patches, some AMD specific code has been removed?

Thanks for your feedback
Best regards

Urs Frey
Die Schweizerische Post
Services
Informationstechnologie
Webergutstrasse 12
3030 Bern (Zollikofen)
Telefon : ++41 (0)58 338 58 70
FAX     : ++41 (0)58 667 30 07
E-Mail:   urs.frey at post.ch<mailto:urs.frey at post.ch>