linux-next: revert of "debug: add notifier chain debugging"

linux-next: revert of "debug: add notifier chain debugging"

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 591 bytes --]

Hi all,

I have reverted commit 16f9b13de93c8bfdac16b4d15577af2c132358ef ("debug:
add notifier chain debugging") for today hate typing thas it prevents Powerpc kernels
from booting.  There is (I think) a bug there (__devinit routines are not
put in the right section as pointed otu by David Woodhouse), but not a
fatal bug and the above commit makes it one.

Ingo, Arjan sent you a new version of this patch on Aug 27.

I will have a go at fixing the powerpc bug tomorrow.
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: linux-next: revert of "debug: add notifier chain debugging"

[Arjan van de Ven]

Stephen Rothwell wrote:
> Hi all,
> 
> I have reverted commit 16f9b13de93c8bfdac16b4d15577af2c132358ef ("debug:
> add notifier chain debugging") for today hate typing thas it prevents Powerpc kernels
> from booting.  There is (I think) a bug there (__devinit routines are not
> put in the right section as pointed otu by David Woodhouse), but not a
> fatal bug and the above commit makes it one.

well it's a pretty bad bug since there are many places that depend on the kernel text stuff to work.
(backtracing, lockdep, some of the security checks etc). PPC needs to be fixed rather urgently.
but yeah the system kept plodding on.
David also sent a patch to fix it fwiw.

I'm a bit sad to see a debug patch get reverted just because it caught bugs ;-)
at least I assume this is very temporary.

Re: linux-next: revert of "debug: add notifier chain debugging"

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 766 bytes --]

Hi Arjan,

On Mon, 01 Sep 2008 07:23:58 -0700 Arjan van de Ven <arjan@linux.intel.com> wrote:
>
> well it's a pretty bad bug since there are many places that depend on the kernel text stuff to work.
> (backtracing, lockdep, some of the security checks etc). PPC needs to be fixed rather urgently.
> but yeah the system kept plodding on.
> David also sent a patch to fix it fwiw.
> 
> I'm a bit sad to see a debug patch get reverted just because it caught bugs ;-)
> at least I assume this is very temporary.

Today I still reverted the patch, but I applied the new version as well
as applying the simple powerpc fix patch, so we shall see ...

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: linux-next: revert of "debug: add notifier chain debugging"

[David Woodhouse]

On Mon, 2008-09-01 at 07:23 -0700, Arjan van de Ven wrote:
> 
> well it's a pretty bad bug since there are many places that depend on
> the kernel text stuff to work.
> (backtracing, lockdep, some of the security checks etc). PPC needs to
> be fixed rather urgently.
> but yeah the system kept plodding on.
> David also sent a patch to fix it fwiw.

My patch broke allyesconfig though -- we need .text and .fixup to be
close together, otherwise the branches back from fixup sections are too
long to fit in a relative jump.

Because TEXT_TEXT doesn't include .fixup, 
 there was a _reason_ for not using TEXT_TEXT, and it's to ensure that
the branches back to .text sections within the .fixup code are short
enough. If they're too far away, then the final link fails.

It might be sufficient to put the .fixup section immediately
_before_ .text rather than a long way after it as my patch did.

> I'm a bit sad to see a debug patch get reverted just because it caught
> bugs ;-)
> at least I assume this is very temporary.

Yeah, I'm sure it will be.
http://moss.amr.ith.intel.com/sites/Purchasing/CPC/EMEA_ApplyUpdateCard/EMEA_Apply_Update_Card.aspx
-- 
David Woodhouse                            Open Source Technology Centre
David.Woodhouse@intel.com                              Intel Corporation

Re: linux-next: revert of "debug: add notifier chain debugging"

[David Woodhouse]

On Tue, 2008-09-02 at 18:57 +0100, David Woodhouse wrote:

... some half-composed and out of date nonsense, which had since been
used as a cut/paste scratch buffer.

Sorry, please ignore that -- I hate it when I accidentally click the
mouse button twice, and end up hitting 'send' on something completely
random that happens to be lying around my desktop...

-- 
dwmw2