2006-05-07 22:49:38

by Nathan Scott

[permalink] [raw]
Subject: Re: XFS regression on 2.6.17 (Does not happen on 2.6.16.XX)

On Sun, May 07, 2006 at 07:36:50PM -0300, Andr? Goddard Rosa wrote:
> Hi, XFS maintainers!

Hi there,

> I got an error when booting 2.6.17-rc3-git13 (I have this same error
> since 2.6.17) that prevents be from booting normally.
>
> This never happened before until 2.6.17 and is reproducible. I already
> run xfs_check and xfs_repair but they did nothing.

This is likely to be related to write barriers. I sent a barrier fix
which is probably going to resolve this on last week (I say "probably"
as this is not exactly the same symptoms as before), but looks like
its not yet merged, I'll resend today.

> What data do you need to help me on this issue?

(/me squints -- did you take that photo in a dark room? :)

Can you try the current -mm tree to make sure the problem is fixed
there already?

thanks.

--
Nathan


2006-05-07 22:59:09

by André Goddard Rosa

[permalink] [raw]
Subject: Re: XFS regression on 2.6.17 (Does not happen on 2.6.16.XX)

On 5/7/06, Nathan Scott <[email protected]> wrote:
> On Sun, May 07, 2006 at 07:36:50PM -0300, Andr? Goddard Rosa wrote:
> > Hi, XFS maintainers!
>
> Hi there,
>
> > I got an error when booting 2.6.17-rc3-git13 (I have this same error
> > since 2.6.17) that prevents be from booting normally.
> >
> > This never happened before until 2.6.17 and is reproducible. I already
> > run xfs_check and xfs_repair but they did nothing.
>
> This is likely to be related to write barriers. I sent a barrier fix
> which is probably going to resolve this on last week (I say "probably"
> as this is not exactly the same symptoms as before), but looks like
> its not yet merged, I'll resend today.
>
> > What data do you need to help me on this issue?
>
> (/me squints -- did you take that photo in a dark room? :)

DuP! Yes, now it is dark here, I can do better next time! :)

> Can you try the current -mm tree to make sure the problem is fixed
> there already?

Yes, I do not have this problem in -mm.

Thanks you so much for the quick response!
--
[]s,
Andr? Goddard