Bug#611946: Freezes apparently caused by ath9k

2011-02-24 Thread Jonathan Nieder
forwarded 611946 
http://thread.gmane.org/gmane.linux.kernel.wireless.general/65397
reassign 611946 linux-2.6
tags 611946 + upstream
found 611946 linux-2.6/2.6.37-1~experimental.1
found 611946 linux-2.6/2.6.37-1
# bisected to v2.6.35-rc1~473^2~167^2~346
found 611946 linux-2.6/2.6.35~rc4-1~experimental.1
quit

Tony Houghton wrote:

> I think I
> successfully found the "bad commit" with git bisect so I've cross-posted
> the result to this bug and linux-wireless.

Thanks!

> I was a bit puzzled by git
> bisect though because it seemed to be getting versions/tags out of
> order. For example, I found a commit after tag 2.6.34 to be "good", but
> eventually I homed in on 2.6.33. Is that normal, perhaps because they
> originated from different branches which were merged apparently out of
> order?

Exactly, it's from nonlinear history.  See


https://git.wiki.kernel.org/index.php/GitFaq#Why_does_.22git_bisect.22_make_me_test_versions_outside_the_.22good-bad.22_range.3F

and

http://thread.gmane.org/gmane.comp.version-control.git/99967/focus=99977



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/20110224195143.GA17412@elie



Bug#611946: Freezes apparently caused by ath9k

2011-02-24 Thread Tony Houghton
In the end I used a chroot to do the cross-compiling. I think I
successfully found the "bad commit" with git bisect so I've cross-posted
the result to this bug and linux-wireless. I was a bit puzzled by git
bisect though because it seemed to be getting versions/tags out of
order. For example, I found a commit after tag 2.6.34 to be "good", but
eventually I homed in on 2.6.33. Is that normal, perhaps because they
originated from different branches which were merged apparently out of
order?



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/20110224180414.7a7cb7d6@toddler



Bug#611946: Freezes apparently caused by ath9k

2011-02-24 Thread Tony Houghton
In the end I used a chroot to do the cross-compiling. I think I
successfully found the "bad commit" with git bisect so I've cross-posted
the result to this bug and linux-wireless. I was a bit puzzled by git
bisect though because it seemed to be getting versions/tags out of
order. For example, I found a commit after tag 2.6.34 to be "good", but
eventually I homed in on 2.6.33. Is that normal, perhaps because they
originated from different branches which were merged apparently out of
order?



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/20110224180607.34011d2c@toddler



Bug#611946: Freezes apparently caused by ath9k

2011-02-22 Thread Jonathan Nieder
maximilian attems wrote:

> only with newer 2.6.38-rc4 :
>
>  make ARCH=i386 deb-pkg

Wasn't that part of 2.6.38-rc1?  (v2.6.38-rc1~438^2~4, kbuild,
deb-pkg: select userland architecture based on UTS_MACHINE,
2010-11-05)

To do the same on older kernels:

git checkout origin/master -- scripts/package/builddeb
make ARCH=i386 deb-pkg

(localmodconfig won't work here but you can always copy over an
appropriate .config from the laptop.)

Hope that helps.



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/2011034535.GC26238@elie



Bug#611946: Freezes apparently caused by ath9k

2011-02-22 Thread Julien Cristau
On Tue, Feb 22, 2011 at 23:25:51 +, Tony Houghton wrote:

> That sounds like a good idea. Is there an easy way to cross-compile for
> 686 on an x86_64 system eg by adding a variable to the make deb-pkg
> line? It will save loads of time if I can do it on my quad core desktop
> PC instead of my netbook's weedy Atom.
> 
make deb-pkg CC='gcc -m32' might work.  Assuming you have gcc-multilib
installed.

Cheers,
Julien



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/2011034118.gw12...@radis.liafa.jussieu.fr



Bug#611946: Freezes apparently caused by ath9k

2011-02-22 Thread maximilian attems
On Tue, Feb 22, 2011 at 11:25:51PM +, Tony Houghton wrote:
> On Tue, 22 Feb 2011 15:50:25 -0600
> Jonathan Nieder  wrote:
> 
> > What is very helpful if you have time is
> > 
> >  1) describe the symptoms on the linux-wireless mailing list
> >  2) grab the kernel source:
> > 
> > git clone
> > git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
> > 
> >  3) bisect:
> > 
> > git bisect start
> > git bisect good v2.6.32
> > git bisect bad v2.6.35
> > make localmodconfig
> > make deb-pkg
> > ... install the resulting .deb, reboot, try it, and then ...
> > git bisect good;# if it works
> > git bisect bad; # if it doesn't work
> > 
> >  4) let the linux-wireless people know which turned out to be the
> > "first bad commit".
> 
> That sounds like a good idea. Is there an easy way to cross-compile for
> 686 on an x86_64 system eg by adding a variable to the make deb-pkg
> line? It will save loads of time if I can do it on my quad core desktop
> PC instead of my netbook's weedy Atom.

only with newer 2.6.38-rc4 :

 make ARCH=i386 deb-pkg

so currently only helps for linux-next or newer.



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/2011033634.gf3...@vostochny.stro.at



Bug#611946: Freezes apparently caused by ath9k

2011-02-22 Thread Tony Houghton
On Tue, 22 Feb 2011 15:50:25 -0600
Jonathan Nieder  wrote:

> What is very helpful if you have time is
> 
>  1) describe the symptoms on the linux-wireless mailing list
>  2) grab the kernel source:
> 
>   git clone
> git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
> 
>  3) bisect:
> 
>   git bisect start
>   git bisect good v2.6.32
>   git bisect bad v2.6.35
>   make localmodconfig
>   make deb-pkg
>   ... install the resulting .deb, reboot, try it, and then ...
>   git bisect good;# if it works
>   git bisect bad; # if it doesn't work
> 
>  4) let the linux-wireless people know which turned out to be the
> "first bad commit".

That sounds like a good idea. Is there an easy way to cross-compile for
686 on an x86_64 system eg by adding a variable to the make deb-pkg
line? It will save loads of time if I can do it on my quad core desktop
PC instead of my netbook's weedy Atom.



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/2011032551.17c144b6@toddler



Bug#611946: Freezes apparently caused by ath9k

2011-02-22 Thread Jonathan Nieder
(resending to the right bug number.  Sorry for the noise.)
Hi Tony,

Tony Houghton wrote:

> I've reassigned this bug to the latest unstable kernel because that's
> affected too. I think the problem is caused by the ath9k driver because
> any attempt to disable wireless or even disconnect usually causes
> similar crashes.
[...]
> 2.6.32 from squeeze does not have this problem, but I've tried Ubuntu
> Maverick which has kernel 2.6.35 and that exhibits similar symptoms. I
> think I should report this upstream but I don't know the best way
> to do that.

See .

What is very helpful if you have time is

 1) describe the symptoms on the linux-wireless mailing list
 2) grab the kernel source:

git clone 
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git

 3) bisect:

git bisect start
git bisect good v2.6.32
git bisect bad v2.6.35
make localmodconfig
make deb-pkg
... install the resulting .deb, reboot, try it, and then ...
git bisect good;# if it works
git bisect bad; # if it doesn't work

 4) let the linux-wireless people know which turned out to be the
"first bad commit".

Hope that helps,
Jonathan



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/2011015025.GB24704@elie



Bug#611946: Freezes apparently caused by ath9k

2011-02-22 Thread Tony Houghton
I've reassigned this bug to the latest unstable kernel because that's
affected too. I think the problem is caused by the ath9k driver because
any attempt to disable wireless or even disconnect usually causes
similar crashes.

I tried replacing network-manager with wicd but that crashed the system
when it connected instead of when disconnected.

2.6.32 from squeeze does not have this problem, but I've tried Ubuntu
Maverick which has kernel 2.6.35 and that exhibits similar symptoms. I
think I should report this upstream but I don't know the best way
to do that.



-- 
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/2011012031.7ae65200@toddler