Re: Oops in 2.6.10-rc1

[Linus Torvalds <torvalds@osdl.org>]

On Tue, 9 Nov 2004, Christian Kujau wrote:
> 
> >>>Looking at the list (appended), I don't see anything obvious, but hey, if 
> >>>it was obvious it wouldn't have been merged in the first place. 
> 
> yes, i'll look for changes regarding PCI. i've started to compile the -bk
> snapshots too. there i can do less wrong things. when i have the "bad" -bk
> snapshot i'll use "bk" itself again to find the detailed change leading to
> the oops.

Actually, looking a bit closer, I think the PCI merge we just looked at 
was the PCI merge that happened _after_ 2.6.10-rc1. And since 2.6.10-rc1 
already oopsed for you, it shouldn't be an issue.

I think the _real_ PCI merge we should have looked at is:

	ChangeSet@1.2000.1.7, 2004-10-19 16:59:19-07:00, torvalds@ppc970.osdl.org
	  Merge PCI updates

and in particular, that merged the PCI changes from

	ChangeSet@1.1988.2.81, 2004-10-19 14:48:04-07:00, greg@kroah.com
	  PCI: fix up pci_save/restore_state in via-agp due to api change.
	  
	  Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

with my pre-PCI-merge tree at:

	ChangeSet@1.2000.1.6, 2004-10-19 15:06:19-07:00, torvalds@ppc970.osdl.org
	  Merge bk://bart.bkbits.net/ide-2.6
	  into ppc970.osdl.org:/home/torvalds/v2.6/linux

(all of these revision numbers are relative to a pristine 2.6.10-rc1 
tree: remember that they change with merges, so they may not be the same 
in your tree. "bk changes -a" is your friend).

So what I'd like you to do is to take the pre-PCI-merge tree, and see if 
that works for you

	# assuming a 2.6.10-rc1 tree
	bk undo -a1.2000.1.6

and if that works, then try the post-PCI-merge tree:

	# assuming a 2.6.10-rc1 tree
	bk undo -a1.2000.1.7

(I just checked: the above numbers are actually valid even in the current
-bk tree, so you don't have to first go to 2.6.10-rc1, you can just start 
from a current tree)

Thanks for testing, and sorry for the confusion with the more recent PCI 
merge.

		Linus