topic Re: HP supplied QLogic FC driver taints kernel on RHEL5? in Operating System - Linux

HP supplied QLogic FC driver taints kernel on RHEL5?

Howard Palmer — Mon, 22 Sep 2008 09:52:22 GMT

Using the QLogic driver pack for RedHat supplied by HP, under RHEL5 the kernel is tainted. This does not happen under RHEL4. It seems to be something to do with "intermodule" which I guess is some sort of kernel module wrapper or something for RHEL5.

Have tried the following versions: -

hp_qla2x00src-8.01.07.25-1
hp_qla2x00src-8.02.11-1

# cat /proc/sys/kernel/tainted
66

# dmesg | grep taint
qla2xxx_conf: no version for "inter_module_unregister" found: kernel tainted.

Of course the driver works fine but could cause us issues if we have to log RedHat support call against these servers.

Does anyone else have this/know of a fix?

Re: HP supplied QLogic FC driver taints kernel on RHEL5?

Howard Palmer — Mon, 22 Sep 2008 10:01:31 GMT

Forgot to say that there's no insurmountable reason why the kernel should be tainted. Generally that's for binary modules, but this is delivered as source. I assume it's just a coding oversight somewhere, but kernel module development isn't my thing!

Re: HP supplied QLogic FC driver taints kernel on RHEL5?

Nick Orant — Tue, 13 Jan 2009 10:45:55 GMT

We also have a customer with the same problem:-

# rpm -qa | grep qla
hp_qla2x00src-8.01.07.25.02-1
# cat /proc/sys/kernel/tainted
66
# dmesg | grep taint
qla2xxx_conf: no version for "inter_module_unregister" found: kernel tainted.

Does anybody have any ideas for a fix

Re: HP supplied QLogic FC driver taints kernel on RHEL5?

Louis Bouchard — Fri, 23 Jan 2009 13:02:27 GMT

This is normal behaviour. This is how the '66' value is decoded :

# echo "obase=2;66" | bc
1000010

Bits #1 and #6 are asserted. This mean that the flags TAINT_FORCED_MODULE and TAINT_UNSIGNED_MODULE are set.

This is explained by the fact that the both modules qla2xxx and qla2xxx_conf are not GPG signed (TAINT_UNSIGNED_MODULE) and that the module qla2xxx_conf has no version field set.

This is not a defect and should not pose a problem.

Re: HP supplied QLogic FC driver taints kernel on RHEL5?

Nick Orant — Fri, 23 Jan 2009 13:13:24 GMT

Whilst I accept that this may not pose a problem with the running of the server, we have an issue with Oracle where we can't get support while the kernel is tainted.

Re: HP supplied QLogic FC driver taints kernel on RHEL5?

Louis Bouchard — Fri, 23 Jan 2009 13:20:14 GMT

Hello,

Looks like I'm the L2 engineer who is assigned to your issue internally.

My ITRC answer is part of the full answer I provided to the L1 engineer. The rest of the answer should get to you shortly.

Kind Regards,

...Louis