Bug 854931

Summary: mrg 2.1 kernel will not boot on x3690 X5 (and perhaps other X5 systems) with optimal firmware settings
Product: Red Hat Enterprise MRG Reporter: IBM Bug Proxy <bugproxy>
Component: realtime-kernelAssignee: John Kacur <jkacur>
Status: CLOSED CURRENTRELEASE QA Contact: David Sommerseth <davids>
Severity: high Docs Contact:
Priority: high    
Version: 2.1CC: bhu, jkachuck, jkacur, lgoncalv, ovasik, wgomerin, williams
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: All   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2012-10-05 14:06:46 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
x5 low latency UEFI options
none
Console output during failed boot none

Description IBM Bug Proxy 2012-09-06 10:30:31 UTC
== Comment: #0 - Michael Kravetz <mkravetz.com> - 2012-09-05 16:59:54 ==
Installed RHEL 6.2 and latest MRG 2.1 images on a x3690 X5.  All was well with default settings.  I then modified uEFI settings for lowest latency.  After this, the realtime kernel would no longer successfully boot.  It pauses for a loooooooong time after "megasas: FW now in Ready state" is printed to the console.

Two attachments will follow:
- uEFI settings
- console output from failed boot attempt

Note that the RHEL 6.2 kernel boots without issue with these uEFI settings.  In fact, these are the settings we recommend to Low Latency (not realtime) customers.  Also, if the uEFI settings are returned to their default values, the mrg/realtime kernel will boot.

Thanks

Comment 1 IBM Bug Proxy 2012-09-06 10:30:39 UTC
Created attachment 610252 [details]
x5 low latency UEFI options

Comment 2 IBM Bug Proxy 2012-09-06 10:30:45 UTC
Created attachment 610253 [details]
Console output during failed boot

Comment 3 IBM Bug Proxy 2012-09-06 20:40:42 UTC
------- Comment From mkravetz.com 2012-09-06 20:39 EDT-------
vanilla kernel fails in the same way as RT, so this may not be RT related.

off to build/attempt to boot mainline.

Comment 4 IBM Bug Proxy 2012-09-06 21:30:45 UTC
------- Comment From mkravetz.com 2012-09-06 21:29 EDT-------
3.0.42 (unmodified) also fails to boot.  Therefore, we can safely say this is NOT a RT issue.

Sorry for the noise Red Hat MRG team

Comment 5 IBM Bug Proxy 2012-09-06 22:00:35 UTC
------- Comment From mkravetz.com 2012-09-06 21:52 EDT-------
Just to verify, I restored uEFI settings to their defaults (F1 during system initialization).  With setting back at the defaults, 3.0.42 kernel boots.

Comment 6 IBM Bug Proxy 2012-09-07 23:40:31 UTC
------- Comment From mkravetz.com 2012-09-07 23:36 EDT-------
This e-mail thread appears to describe the issue:

https://lkml.org/lkml/2011/12/14/102

Comment 7 IBM Bug Proxy 2012-09-08 01:00:31 UTC
------- Comment From mkravetz.com 2012-09-08 00:58 EDT-------
Changed the UEFI setting "ProcessorXapic" to disabled and the MRG 2.1 kernel boots.

So, the question is should CONFIG_X86_X2APIC be enabled in the MRG 2.1 kernel(s) ?

Comment 8 IBM Bug Proxy 2012-09-11 22:50:38 UTC
------- Comment From mkravetz.com 2012-09-11 22:46 EDT-------
Just checked and 2.2 candidate kernel (i have access to) does have CONFIG_X86_X2APIC enabled.

Comment 9 Clark Williams 2012-09-12 16:55:26 UTC
I'd prefer not to change the MRG 2.1 configs as we're about to release MRG 2.2 (and that has X86_X2APIC enabled).