Operating System - HP-UX
1745915 Members
4410 Online
108723 Solutions
New Discussion

HPUX MPT drivers version that fixed vmunix function arg tix

 
SOLVED
Go to solution
Mauro Livi
Valued Contributor

HPUX MPT drivers version that fixed vmunix function arg tix

Hi,

I seem to recall an issue in older versions of an MPT/SCSI drivers for a scsiU320 that if a system was left up something like over a year, then it would generate negative ticks value so you would see something like:

vmunix: function == 0x961918, arg == 0x41ED6400, ticks = 0xFFD9DFD0,  flags == 0x0

in syslog over and over.

Does this ring a bell with anyone? I see a server that has been up nearly over a year, but I do believe that it has a firmware version where the issue mentioned above has been fixed, so it should not be an issue. The firmware version on the server is B.11.31.1109.

Can anyone confirm that this version has the fix for the problem mentioned above? I believe it does

Thanks in advance and any response is appreciated. Kudos points for any replies...naturally. 

Mauro

4 REPLIES 4
Bill Hassell
Honored Contributor

Re: HPUX MPT drivers version that fixed vmunix function arg tix

Yes, that was a very annoying bug in the driver. Your version is quite old. Get the latest from:

https://h20392.www2.hpe.com/portal/swdepot/displayProductInfo.do?productNumber=scsiU320-00



Bill Hassell, sysadmin
Mauro Livi
Valued Contributor

Re: HPUX MPT drivers version that fixed vmunix function arg tix

Thanks Bill,

Yeah I know the firmware might want to be upgraded...but any idea if what I have resolves that issue? I believe HP had resolved that prior to my version (I guess I can check with them). Kudos assigned.

Thanks again

Mauro

Mauro Livi
Valued Contributor
Solution

Re: HPUX MPT drivers version that fixed vmunix function arg tix

For anyone interested, here was the answer from HP along with all the 11.11, 11.23, and 11.31 versions that fixed the problem.

 

HP-UX 11i Operating Environments - "vmunix: Timeout called with negative time." and "vmunix: function == 0x..., arg == 0x..., ticks == 0x..., flags == 0x0" Issue

HP Integrity and PA-RISC server running HP-UX 11.I v1 (11.11) or 11.I v2 (11.23) may report the following messages in /var/adm/syslog/syslog.log and dmesg :

Nov  9 16:12:11 rx4640 vmunix: function == 0xE0000000004F9060, arg == 0xE00000010F9FF400, ticks == 0xFF5F1197,  flags == 0x0 Nov  9 16:12:11 rx4640 vmunix: function == 0xE0000000004F9060, arg == 0xE00000010F9FF400, ticks == 0xFF5F1196,  flags == 0... Nov  9 16:12:13 rx4640 vmunix: function == 0xE0000000004F9060, arg == 0xE00000010F9FF400, ticks == 0xFF5F1179,  flags == 0x0 Nov  9 16:12:13 rx4640 vmunix: Timeout called with negative time. Nov  9 16:12:13 rx4640  above message repeats 57530 times ...

Solution

Use the following command to check which kernel function the messages are related too:

For HP-UX 11.I v1 (11.11) :

  # echo '<function number>/a' | adb /stand/vmunix /dev/kmem 

For HP-UX 11.I v2 (11.23) :

  # echo "<function number>/a"  | adb -o    /stand/vmunix /dev/kmem

For HP-UX 11.I v3 (11.31) :

The fix for this issue is to use the tick_since_boot_whole timer instead of the default 32-bit LBOLT value in the kernel space. This fix will be released with the B.11.31.0709 MPT driver with 0709 Fusion.

The fix from JAGag11960 on 11.23 has been ported into the 11.31 MPT driver. The fix will be released with 0709 Fusion.

Click here to access the technical article at http://psweb1.cup.hp.com/~projects/pq/cgi-bin/goto_cr.cgi?number=QXCR1000575700.

Example for HP-UX 11.I v2 (11.23):

  # echo "0xE0000000004F9060/a"  | adb -o    /stand/vmunix /dev/kmem   mpt_os_callback:        mpt_os_callback:  mpt_os_callback: 

This example shows that a mpt function is called.

In case a different kernel function reports the error, figure out to which kernel subsystem or driver this belongs, and check for any known issue and a fix in a newer version.

There is a known issue like this for mpt (scsiU320) driver older than B.11.23.0612 (or B.11.11.0701 for HP-UX 11i v1):

Enhanced I/O handling has been added to the Ultra320 SCSI mpt driver which will prevent SCSI I/O's from timing out prematurely on systems which have been online longer than 400-days without a system reboot.

See also HP-UX Ultra320 SCSI mpt driver release notes:

Version 11i v2 Revision B.11.23.0612:

Click here to access the technical article at: http://g4u0419c.houston.hp.com/en/J6373-90021/index.html.

Check with following command which mpt driver is installed:

# swlist -l bundle | grep scsiU320-00

Example:

  # swlist -l bundle | grep scsiU320-00   scsiU320-00        B.11.23.0603   PCI-X SCSI U320; Supptd HW=A7173A/AB290A 

In this case the installed version of mpt driver is older than the one with the fix (B.11.23.0612), and the server is up for more than 500 days. The HP-UX command uptime can be used to check uptime.

The mpt (scsiU320-00) driver needs to be updated.

Latest mpt driver can be downloaded by searching scsiU320-00 from the follwing website:

Click here to download the software utility/driver at: http://software.hp.com.

Direct link:

Click here to access the technical article at: https://h20392.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=scsiU320-00.

 

Dennis Handly
Acclaimed Contributor