Bug 200409

Summary: [olpc] panic in ext3 code on geode using i686 code
Product: [Fedora] Fedora Reporter: David Zeuthen <davidz>
Component: kernelAssignee: Kernel Maintainer List <kernel-maint>
Status: CLOSED RAWHIDE QA Contact: Brian Brock <bbrock>
Severity: high Docs Contact:
Priority: high    
Version: rawhideCC: aoliva, jakub, mclasen, wtogami
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2006-08-01 23:30:31 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:
Bug Depends On:    
Bug Blocks: 150223, 150224, 191929, 191931    
Attachments:
Description Flags
photograph of panic none

Description David Zeuthen 2006-07-27 15:11:00 UTC 
Description of problem:

Kernel panic. See attached photograph. This might very well be related to i686
weirdness as described in bug 200330. My guess right now is binutils weirdness
breaking i686 for the AMD Geode.

Version-Release number of selected component (if applicable):

2.6.17-1.2449.fc6

How reproducible:

Run some file system intensive code like 'yum -y install wget'. Reproducible on
build45 on OLPC available here
 http://olpc.download.redhat.com/olpc/streams/development/build45-20060726_1718/images-rpm/olpc-stream-development-45-20060726_1718-rpm-ext3.img.bz2

bunzip2 the image, dd it to a USB stick (e.g. /dev/sda) with at least 512MB of
space and boot it on a OLPC developer boards. We have one such board in the
Westford office.
Comment 1 David Zeuthen 2006-07-27 15:11:02 UTC 
Created attachment 133158 [details]
photograph of panic
Comment 2 Alexandre Oliva 2006-07-27 18:35:03 UTC 
Doesn't appear to be related.  The code at the crash point is:

c04748d4:       e8 be f7 ff ff          call   c0474097 <unmap_underlying_metadata>
c04748d9:       8b 5b 04                mov    0x4(%ebx),%ebx   <=== crashes here

I'd guess that %ebx ended up with an invalid address, for whatever reason. 
Without the top of the Oops, it's hard to tell more.  It's definitely not an
illegal instruction.
Comment 3 Alexandre Oliva 2006-07-27 18:38:14 UTC 
It might actually have to do with i686 instructions, though, if some replacement
chunk of code for instructions missing on the Geode turn out to corrupt %ebx or
whatever memory slot it happened to be preserved in by the callee.
Comment 4 Christopher Blizzard 2006-08-01 23:22:56 UTC 
Do we still see this panic after the rebuild with the new binutils?
Comment 5 David Zeuthen 2006-08-01 23:30:31 UTC 
No, this seems to be fixed. Closing. Thanks.