Re: Linux 2.6.35.2

[Ed Tomlinson <edt@aei.ca>]

On Saturday 14 August 2010 17:56:58 Linus Torvalds wrote:
> On Sat, Aug 14, 2010 at 12:47 PM, Ed Tomlinson <edt@aei.ca> wrote:
> > On Saturday 14 August 2010 14:47:45 Linus Torvalds wrote:
> >>
> >> Just to verify: you have CONFIG_HIGHPTE enabled on a x86-32 kernel,
> >> right? I'd have expected this to not ever show up anywhere else, and
> >> I'm just verifying that there isn't anything else going on. I'm
> >> appending the commit message, the patch hasn't changed.
> >
> > No.  I have a x86-64 based kernel (amd64).
> 
> Hmm. On x86-64, pte_unmap() is a no-op. So moving it around should not
> matter one whit. We don't even have a debug mode to turn it into some
> kind of "check that we use it properly in an atomic region" which we
> _could_ do.
> 
> Can you send me the messages you saw without that patch? There's
> something odd going on here. I don't like the fact that the patch
> apparently makes a difference for you. It shouldn't.

Linus,

Not to worry - looks like the problems here were due to a bad inode.
I just did not hit it when restarting with the  patch added.  When I
rebuilt the kernel without the patch it hit the fs/inode problem during
boot.  A manual fsck seems to have fixed things.

Ed Tomlinson