syslog growing with vmunix messages

[MartyB]

Hello, on one of my 11.11 systems, the syslog is growing VERY quickly. the message being repeated is:

vmunix: function == 0x8A38F0, arg == 0x40A1A400, ticks == 0xFE446010, flags == 0x0

which is apparently generated by what I'm guessing is the MPT driver:

$ echo 0x8A38F0/i | adb /stand/vmunix /dev/kmem
mpt_os_callback:
mpt_os_callback: STD r2,-16(r30)

Anyone know what is actually going wrong?

Thanks!

[Rita C Workman · Accepted solution]

Did anyone make any changes on that box recently?

Take a look here for Known Problems...maybe there is something that can help you...

http://docs.hp.com/en/J6373-90015/J6373-90015.pdf

Rgrds,
Rita

[Tim Nelson]

When is the last time that the server was rebooted ?

If over a year then you need a FW update for your SCSI card(s).

Might as well do a FW update for GSP/MP as well.

[MartyB]

lol, well, it's been awhile. 500 days to be exact...perhaps it's about time for some firmware updates. thanks!

[Don Morris_1]

Is there a line just ahead of that with "Timeout called with negative time." ? It looks to me as if there should be. (Note ticks is a 32-bit value starting with 0xF -- that's a negative for a signed 32-bit value).

If so -- well, that's what's happening. A function in the kernel is being called to schedule mpt_os_callback which takes a ticks argument as a time (in ticks) for timer expiration. Passing a negative value (i.e. asking to set an alarm for 5 minutes ago in concept) is unexpected and is sufficiently odd [because it is likely a coding error that may indicate other problems] that the owning function wants to call it out in this fashion.

Checking around -- it looks like this has been noted and fixed in 2007. As such, the latest version of the software (11.11.0803) should resolve the issue. Get it from
http://h20392.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=A7173A

[Don Morris_1]

Ah... thanks to Rita I narrowed things down. This was fixed as part of 8606472711 (it doesn't describe it in quite this way.. but it is part of that fix). Definitely in the 11.11.0701 bundle [I didn't catch the link to the Archived release notes the first time]. See:

http://docs.hp.com/en/J6373-90024/ch01s04.html

[Tim Nelson]

Don has stated the issue to perfection as usual.

some register in the scsi fw is being overflowed due to the uptime being so long.

I ran into this issue on some rx2620 and rx4640s after about 365 days of uptime.

The reboot will fix it for another year or so, the FW update is safe and completely fixes the issue.

The biz support center has the MP and SCSI FW updates, I believe they are even available via OS patch ( MP depends on model ).

[Ivan Krastev]

This issue happend after 400 days uptime. Changing the driver scsi-320 will fix it.

regards,
ivan

[MartyB]

Thanks for the help everyone!