| Summary: | kernel gentoo-dev-sources-2.6.4-rc1 Oops'es somewhere in the VFS/FS code - causes FS corruption!! | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Adrian Head <adrian.head> |
| Component: | [OLD] Core system | Assignee: | x86-kernel (DEPRECATED) <x86-kernel> |
| Status: | RESOLVED TEST-REQUEST | ||
| Severity: | critical | CC: | m.debruijne |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | x86 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: |
gentoo-dev-sources-2.6.4-rc1 .config file.
/var/log/messages that contains the Oops'es |
||
|
Description
Adrian Head
2004-03-20 17:20:53 UTC
Created attachment 27710 [details]
gentoo-dev-sources-2.6.4-rc1 .config file.
This is the gentoo-dev-sources-2.6.4-rc1 .config file that was used to compile
the kernel.
Please attach the oops, otherwise it is pretty much impossible to help. Created attachment 27711 [details]
/var/log/messages that contains the Oops'es
This file is an edited version of /var/log/messages that contains the Oops'es.
Also included in this file is the messages log during bootup, and the
corruption of the messages file. (There is KDE source code in the middle of
the messages file for no real reason).
I think this now brings me back to where I was before Konq crashed. Do I need to supply any extra information? PS: It seems that any file that was being written to during when the machine was Oops'ing - is corrupt. Thankfully at the moment it doesn't seem like many files. /var/log/messages is the worst. If you disable CONFIG_PREEMPT, does this bug go away? This is really an upstream bug, you should check 2.6.5-rc2 to see if it is still present or not. If it is, please let the developers on the linux-kernel mailing list know about it. Also, 2.6.4-rc1 is a bit old :) OK - I've recompiled without CONFIG_PREEMPT and you seem to be correct in that it doesn't Oops any more. I was lead to believe that the gentoo-dev-sources-2.6.4-rc1 were different from the vanilla 2.6.4 kernels. Is this correct? If so; what would using the 2.6.5-rc2 vanilla actually prove? Does any of Gentoo's patches touch the preempt code path? 2.6.4-rc1 is older than a 2.6.4 kernel. The current kernel is 2.6.5-rc2 right now. And no, there are no patches in the gentoo kernel for preempt, but there are still some known issues with preempt enabled in the main kernel. Oops - seems to be my mistake (typo) with regard to the ebuild. It is in fact:
heady portage # ACCEPT_KEYWORDS="~x86" emerge -s gentoo-dev-sources
Searching...
[ Results for search key : gentoo-dev-sources ]
[ Applications found : 1 ]
* sys-kernel/gentoo-dev-sources
Latest version available: 2.6.4-r1
Latest version installed: 2.6.4-r1
Size of downloaded files: 34,008 kB
Homepage: http://www.kernel.org/ http://www.gentoo.org/
Description: Full sources including the gentoo patchset for the 2.6 kernel tree
License: GPL-2
So the question is now: where to from here - what can I do to help the "gentoo-ized" 2.6.x kernel reach stability?
This bug report was more to bring to the attention of the gentoo kernel folks that the current unstable dentoo-dev-sources is infact that - unstable with CONFIG_PREEMPT.
The other alternative is to try the development-sources; however, I'm not sure how different the "Gentoo-ized" gentoo-dev-sources is compared to development-sources. For the un-initiated would anyone like to elaborate?
|
