Re: Machine hard lock

Re: Machine hard lock

[Martin Tessun]

hostmaster-uTGiPWHMmZuzZXS1Dc/lvw@public.gmane.org wrote:
> The combination of the latest devicescape stack ( 051221 )
> and latest dscape branch makes my Powerbook hard lock after 
> rmmod the kernel modules and stopping wpa_supplicant.
> 
> Basicly i do the following before it hard locks :
> 
> rmmod bcm43xx
> killall wpa_supplicant 2>/dev/null
> rmmod rate_control
> rmmod 80211

At least you get them inserted ;) When I try to modprobe rate_control I 
get the following:

Dec 22 15:27:48 kirk kernel: 80211: module license 'unspecified' taints 
kernel.
---> This line
Dec 22 15:27:48 kirk kernel: 80211: ieee80211_tx_packet_data is bigger 
than the skb->cb (48 > 40)
---> This line
Dec 22 15:27:48 kirk kernel: rate_control: Unknown symbol 
ieee80211_rate_control_unregister
Dec 22 15:27:48 kirk kernel: rate_control: Unknown symbol sta_info_release
Dec 22 15:27:48 kirk kernel: rate_control: Unknown symbol 
ieee80211_rate_control_register
Dec 22 15:27:48 kirk kernel: rate_control: Unknown symbol sta_info_get

Most probably caused by my 64-bit-kernel. I have to dig into that.

Nevertheless, I would address this sort of hangup to jbenc and probably 
jgarzik on linux-kernel (or better netdev).

Regards,
Martin

Re: Machine hard lock

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 192 bytes --]

On Thu, 2005-12-22 at 17:51 +0100, Martin Tessun wrote:

> At least you get them inserted ;) When I try to modprobe rate_control I 
> get the following:

insmod 80211 first.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

Re: Machine hard lock

[Martin Tessun]

Johannes Berg wrote:
> On Thu, 2005-12-22 at 17:51 +0100, Martin Tessun wrote:
> 
>> At least you get them inserted ;) When I try to modprobe rate_control I 
>> get the following:
> 
> insmod 80211 first.
> 
> johannes

No, this ist not the problem.
When inserting 80211 I get the same message;

kernel: 80211: ieee80211_tx_packet_data is bigger than the skb->cb (48 > 
40)

And *this* is the real problem. The two structs don't match somehow.

Regards,
Martin

Re: Machine hard lock

[Michael Buesch]

[-- Attachment #1: Type: text/plain, Size: 672 bytes --]

On Thursday 22 December 2005 18:49, Martin Tessun wrote:
> Johannes Berg wrote:
> > On Thu, 2005-12-22 at 17:51 +0100, Martin Tessun wrote:
> > 
> >> At least you get them inserted ;) When I try to modprobe rate_control I 
> >> get the following:
> > 
> > insmod 80211 first.
> > 
> > johannes
> 
> No, this ist not the problem.
> When inserting 80211 I get the same message;
> 
> kernel: 80211: ieee80211_tx_packet_data is bigger than the skb->cb (48 > 
> 40)
> 
> And *this* is the real problem. The two structs don't match somehow.

You're on a 64bit machine.
Run a 32bit kernel, or wait a few weeks until this is fixed.

-- 
Greetings Michael.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Machine hard lock

[Arnaldo Carvalho de Melo]

On 12/22/05, Michael Buesch <mbuesch-KuiJ5kEpwI6ELgA04lAiVw@public.gmane.org> wrote:
> On Thursday 22 December 2005 18:49, Martin Tessun wrote:
> > Johannes Berg wrote:
> > > On Thu, 2005-12-22 at 17:51 +0100, Martin Tessun wrote:
> > >
> > >> At least you get them inserted ;) When I try to modprobe rate_control I
> > >> get the following:
> > >
> > > insmod 80211 first.
> > >
> > > johannes
> >
> > No, this ist not the problem.
> > When inserting 80211 I get the same message;
> >
> > kernel: 80211: ieee80211_tx_packet_data is bigger than the skb->cb (48 >
> > 40)
> >
> > And *this* is the real problem. The two structs don't match somehow.
>
> You're on a 64bit machine.
> Run a 32bit kernel, or wait a few weeks until this is fixed.

Or temporarily increase the size of struct sk_buff::cb from [40] to
[48] if he wants to test other features present in this kernel release
he is trying to compile.

- Arnaldo

Re: Machine hard lock

[Martin Tessun]

Arnaldo Carvalho de Melo wrote:
> On 12/22/05, Michael Buesch <mbuesch-KuiJ5kEpwI6ELgA04lAiVw@public.gmane.org> wrote:
>> On Thursday 22 December 2005 18:49, Martin Tessun wrote:
>>> Johannes Berg wrote:
>>>> On Thu, 2005-12-22 at 17:51 +0100, Martin Tessun wrote:
>>>>
>>>>> At least you get them inserted ;) When I try to modprobe rate_control I
>>>>> get the following:
>>>> insmod 80211 first.
>>>>
>>>> johannes
>>> No, this ist not the problem.
>>> When inserting 80211 I get the same message;
>>>
>>> kernel: 80211: ieee80211_tx_packet_data is bigger than the skb->cb (48 >
>>> 40)
>>>
>>> And *this* is the real problem. The two structs don't match somehow.
>> You're on a 64bit machine.
>> Run a 32bit kernel, or wait a few weeks until this is fixed.
> 
> Or temporarily increase the size of struct sk_buff::cb from [40] to
> [48] if he wants to test other features present in this kernel release
> he is trying to compile.

Great. now it works ;) I can do some more tests now...
(Sometimes it is helpful to read the docs of tcp.h ;)

Thanks for that workaround.

Regards,
Martin