Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[Jean Tourrilhes]

On Mon, Sep 17, 2007 at 04:24:18AM +0900, YOSHIFUJI Hideaki / ?$B5HF#1QL@ wrote:
> Hello.
> 
> With latest git tree, bcm43xx driver does not work.
> By bisect, I've found the commit 4cf92a3c is the first "bad" commit.
> 
>     [PATCH] softmac: Fix ESSID problem
> 
>     Victor Porton reported that the SoftMAC layer had random problem when setting the ESSID :
>     http://bugzilla.kernel.org/show_bug.cgi?id=8686 After investigation, it turned out to be
>     worse, the SoftMAC layer is left in an inconsistent state. The fix is pretty trivial.
>     
>     Signed-off-by: Jean Tourrilhes <jt@hpl.hp.com>
>     Acked-by: Michael Buesch <mb@bu3sch.df>
>     Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
>     Signed-off-by: John W. Linville <linville@tuxdriver.com>
> 
> After reverting this commit, the driver starts working again.
> 
> Regards,
> 
> -- 
> YOSHIFUJI Hideaki @ USAGI Project  <yoshfuji@linux-ipv6.org>
> GPG-FP  : 9022 65EB 1ECF 3AD1 0BDF  80D8 4807 F894 E062 0EEA

	Hi,

	Thanks for taking the time to track down this problem and
narrow it so well. This helps a lot to have such good bug reports.
	I'm surprised by your report, it does not make sense that it
would behave this way. This means that I don't understand the driver
as well I was believing. Or maybe your initialisation sequence is
peculiar. I will defer to the driver maintainers.
	Thanks again...

	Jean

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[Larry Finger]

Jean Tourrilhes wrote:
> On Mon, Sep 17, 2007 at 04:24:18AM +0900, YOSHIFUJI Hideaki / ?$B5HF#1QL@ wrote:
>> Hello.
>>
>> With latest git tree, bcm43xx driver does not work.
>> By bisect, I've found the commit 4cf92a3c is the first "bad" commit.
>>
>>     [PATCH] softmac: Fix ESSID problem
>>
>>     Victor Porton reported that the SoftMAC layer had random problem when setting the ESSID :
>>     http://bugzilla.kernel.org/show_bug.cgi?id=8686 After investigation, it turned out to be
>>     worse, the SoftMAC layer is left in an inconsistent state. The fix is pretty trivial.
>>     
>>     Signed-off-by: Jean Tourrilhes <jt@hpl.hp.com>
>>     Acked-by: Michael Buesch <mb@bu3sch.df>
>>     Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
>>     Signed-off-by: John W. Linville <linville@tuxdriver.com>
>>
>> After reverting this commit, the driver starts working again.

Thank you for this thorough, if surprising, bug report.

I would like a few more details about your configuration so that I might be able to reproduce the
problem.

When you say "the latest git tree", do you mean Linus's mainstream tree?

Is your wireless device being controlled by NetworkManager, or by ifconfig up/down?

Which distro do you use?

Which BCM43xx device do you have?

Thanks,

Larry

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[Larry Finger]

John,

Yoshifuji Hideaki reported that commit 4cf92a3c broke the bcm43xx driver. I was able to duplicate
the problem with WEP encryption and ifconfig control of the device. The problem does not happen with
WPA or when using NetworkManager with WEP.

This patch was supposed to be a fix for the bug reported at
http://bugzilla.kernel.org/show_bug.cgi?id=8686; however, it does not.

This commit should be reverted before 2.6.23 is released.

Thanks,

Larry

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[Jean Tourrilhes]

On Mon, Sep 17, 2007 at 11:33:56PM -0500, Larry Finger wrote:
> John,
> 
> Yoshifuji Hideaki reported that commit 4cf92a3c broke the bcm43xx driver. I was able to duplicate
> the problem with WEP encryption and ifconfig control of the device. The problem does not happen with
> WPA or when using NetworkManager with WEP.
> 
> This patch was supposed to be a fix for the bug reported at
> http://bugzilla.kernel.org/show_bug.cgi?id=8686; however, it does not.
> 
> This commit should be reverted before 2.6.23 is released.
> 
> Thanks,
> 
> Larry

	Larry,

	Could you be more explicit ? Reverting the patch will just
bring back the old bug, and the old code was obviously wrong. I don't
like the idea of trading one bug for another bug. It looks to me like
nobody knows what's exactly happening in the driver and we are just
trying random fixes and see what breaks.
	Could you figure out what's exactly happening and make a
proper fix ?
	Regards,

	Jean

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[Michael Buesch]

On Tuesday 18 September 2007 19:42:02 Jean Tourrilhes wrote:
> On Mon, Sep 17, 2007 at 11:33:56PM -0500, Larry Finger wrote:
> > John,
> > 
> > Yoshifuji Hideaki reported that commit 4cf92a3c broke the bcm43xx driver. I was able to duplicate
> > the problem with WEP encryption and ifconfig control of the device. The problem does not happen with
> > WPA or when using NetworkManager with WEP.
> > 
> > This patch was supposed to be a fix for the bug reported at
> > http://bugzilla.kernel.org/show_bug.cgi?id=8686; however, it does not.
> > 
> > This commit should be reverted before 2.6.23 is released.
> > 
> > Thanks,
> > 
> > Larry
> 
> 	Larry,
> 
> 	Could you be more explicit ? Reverting the patch will just
> bring back the old bug, and the old code was obviously wrong. I don't
> like the idea of trading one bug for another bug. It looks to me like
> nobody knows what's exactly happening in the driver and we are just
> trying random fixes and see what breaks.
> 	Could you figure out what's exactly happening and make a
> proper fix ?

Indeed. The patch does fix a bug. So reverting it is not really
an option. I'm not sure how that patch can introduce such breakage, though.
It must be some special ordering of wext calls that trigger this.
It might be that there's another bug similiar to this one breaking
the state machine.
Can you monitor if all needed steps are properly done (assoc, etc...)
to get a WEP connection? If not, where does it fail?

-- 
Greetings Michael.

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[Larry Finger]

Jean Tourrilhes wrote:
> 
> 	Could you be more explicit ? Reverting the patch will just
> bring back the old bug, and the old code was obviously wrong. I don't
> like the idea of trading one bug for another bug. It looks to me like
> nobody knows what's exactly happening in the driver and we are just
> trying random fixes and see what breaks.
> 	Could you figure out what's exactly happening and make a
> proper fix ?
> 	Regards,

It took me nearly two days, but I finally found the correct fix for the bug that made it impossible
to set the essid. The resulting patch has been sent to John.

Larry

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[Jean Tourrilhes]

On Tue, Sep 18, 2007 at 10:03:03PM -0500, Larry Finger wrote:
> Jean Tourrilhes wrote:
> > 
> > 	Could you be more explicit ? Reverting the patch will just
> > bring back the old bug, and the old code was obviously wrong. I don't
> > like the idea of trading one bug for another bug. It looks to me like
> > nobody knows what's exactly happening in the driver and we are just
> > trying random fixes and see what breaks.
> > 	Could you figure out what's exactly happening and make a
> > proper fix ?
> > 	Regards,
> 
> It took me nearly two days, but I finally found the correct fix for the bug that made it impossible
> to set the essid. The resulting patch has been sent to John.
> 
> Larry

	Yes, the fix looked more extensive than I though, and I guess
you also fixed other corner cases in the process. Thanks for your
work, I think you did the right thing...
	Regards,

	Jean

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[YOSHIFUJI Hideaki / 吉藤英明]

Hello.

What is the status of this issue?
2.6.23-rc8 does not help me.

Regression is not right even if the change does
"fix" some aspects.   Please apply the "right" fix
or revert it before 2.6.23.

Thank you.

In article <20070919163237.GA7914@bougret.hpl.hp.com> (at Wed, 19 Sep 2007 09:32:37 -0700), Jean Tourrilhes <jt@hpl.hp.com> says:

> On Tue, Sep 18, 2007 at 10:03:03PM -0500, Larry Finger wrote:
> > Jean Tourrilhes wrote:
> > > 
> > > 	Could you be more explicit ? Reverting the patch will just
> > > bring back the old bug, and the old code was obviously wrong. I don't
> > > like the idea of trading one bug for another bug. It looks to me like
> > > nobody knows what's exactly happening in the driver and we are just
> > > trying random fixes and see what breaks.
> > > 	Could you figure out what's exactly happening and make a
> > > proper fix ?
> > > 	Regards,
> > 
> > It took me nearly two days, but I finally found the correct fix for the bug that made it impossible
> > to set the essid. The resulting patch has been sent to John.
> > 
> > Larry
> 
> 	Yes, the fix looked more extensive than I though, and I guess
> you also fixed other corner cases in the process. Thanks for your
> work, I think you did the right thing...
> 	Regards,
> 
> 	Jean
> 

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[Larry Finger]

YOSHIFUJI Hideaki / =E5=90=89=E8=97=A4=E8=8B=B1=E6=98=8E wrote:
> Hello.
>=20
> What is the status of this issue?
> 2.6.23-rc8 does not help me.
>=20
> Regression is not right even if the change does
> "fix" some aspects.   Please apply the "right" fix
> or revert it before 2.6.23.

The right fix is in the chain. I sent it to John Linville who has sent =
it to Jeff Garzik. I
certainly hope it makes it before 2.6.23, as it is a serious regression=
=2E

Do you have the patch that fixes it? If not, please let me know and I'l=
l send it to you.

Larry

-
To unsubscribe from this list: send the line "unsubscribe linux-wireles=
s" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[YOSHIFUJI Hideaki / 吉藤英明]

In article <46F87B27.5040106@lwfinger.net> (at Mon, 24 Sep 2007 22:06:15 -0500), Larry Finger <larry.finger@lwfinger.net> says:

> The right fix is in the chain. I sent it to John Linville who has sent it to Jeff Garzik. I
> certainly hope it makes it before 2.6.23, as it is a serious regression.
> 
> Do you have the patch that fixes it? If not, please let me know and I'll send it to you.

No, I don't have it in my hand.  Please sent it to me. Thank you.

--yoshfuji

Re: 2.6.23-rc regression: bcm43xx does not work after commit 4cf92a3c

[YOSHIFUJI Hideaki / 吉藤英明]

In article <20070925.122951.98475715.yoshfuji@linux-ipv6.org> (at Tue, 25 Sep 2007 12:29:51 +0900 (JST)), YOSHIFUJI Hideaki / 吉藤英明 <yoshfuji@linux-ipv6.org> says:

> In article <46F87B27.5040106@lwfinger.net> (at Mon, 24 Sep 2007 22:06:15 -0500), Larry Finger <larry.finger@lwfinger.net> says:
> 
> > The right fix is in the chain. I sent it to John Linville who has sent it to Jeff Garzik. I
> > certainly hope it makes it before 2.6.23, as it is a serious regression.
> > 
> > Do you have the patch that fixes it? If not, please let me know and I'll send it to you.
> 
> No, I don't have it in my hand.  Please sent it to me. Thank you.

Okay, I confirm that current tree does not have this issue
any longer.  Thank you!

--yoshfuji