Re: [Bugme-new] [Bug 10471] New: iwl3945 in 2.6.25 ignores state of hardware RF kill switch

Re: [Bugme-new] [Bug 10471] New: iwl3945 in 2.6.25 ignores state of hardware RF kill switch

[Andrew Morton]

(switched to email.  Please respond via emailed reply-to-all, not via the
bugzilla web interface).


On Thu, 17 Apr 2008 14:17:51 -0700 (PDT)
bugme-daemon@bugzilla.kernel.org wrote:

> http://bugzilla.kernel.org/show_bug.cgi?id=10471
> 
>            Summary: iwl3945 in 2.6.25 ignores state of hardware RF kill
>                     switch
>            Product: Drivers
>            Version: 2.5
>      KernelVersion: 2.6.25
>           Platform: All
>         OS/Version: Linux
>               Tree: Mainline
>             Status: NEW
>           Severity: normal
>           Priority: P1
>          Component: network-wireless
>         AssignedTo: drivers_network-wireless@kernel-bugs.osdl.org
>         ReportedBy: thoralf.dassler@gmail.com
> 
> 
> Latest working kernel version: 2.6.24.3 (possibly 2.6.24.4, but have not had
> it)
> Earliest failing kernel version: 2.6.25
> Distribution: Slackware 12
> Hardware Environment: Toshiba Satellite P200
> Software Environment: n/a
> Problem Description: With the 2.6.25 kernel, iwl3945 sees the state of the
> hardware RF kill switch, but ignores the state. This causes long boot delays
> because the adapter tries to make a connection when the switch is in the OFF
> position.
> 
> Steps to reproduce:
> boot up the system; none of the following showed with 2.6.24.3
> 
> 
> messages during boot:
> -----------------------------------
> SIOCSIFFLAGS: No such device
> [3 lines of irrelevant output on my iwl3945]
> SIOCSIFFLAGS: No such device
> 
> 
> dmesg:
> -----------
> ACPI: PCI Interrupt 0000:04:00.0[A] -> GSI 17 (level, low) -> IRQ 17
> PM: Writing back config space on device 0000:04:00.0 at offset 1 (was 100002,
> writing 100006)
> iwl3945: Radio disabled by HW RF Kill switch
> [message repeated 7 more times]
> 

Yeah, I got bitten by that too.  Apparently it's deliberate and other
wireless drivers will start to do the same thing soon.

It's totally dumb that the initscritps pause *at all* when the kill switch
is in "kill" mode.  Hopefully one day userspace will get fixed (I assume
the problem is in userspace).

Re: [Bugme-new] [Bug 10471] New: iwl3945 in 2.6.25 ignores state of hardware RF kill switch

[Andrew Morton]

On Fri, 18 Apr 2008 09:52:46 +0000 Thoralf Da__ler <thoralf.dassler@gmail.com> wrote:

> Yeah, it seems dumb that the init scripts pause, but I don't understand why this started with the 2.6.25 kernel, whereas it was fine before.

You have removed from Cc: the people who can explain it.  I restored them.

> Andrew Morton wrote:(switched to email.  Please respond via emailed reply-to-all, not via the
> bugzilla web interface).
> 
> 
> On Thu, 17 Apr 2008 14:17:51 -0700 (PDT)
> bugme-daemon@bugzilla.kernel.org wrote:
> 
>   http://bugzilla.kernel.org/show_bug.cgi?id=10471
> 

Re: [Bugme-new] [Bug 10471] New: iwl3945 in 2.6.25 ignores state of hardware RF kill switch

[drago01]

On Fri, Apr 18, 2008 at 11:01 AM, Andrew Morton
<akpm@linux-foundation.org> wrote:
> On Fri, 18 Apr 2008 09:52:46 +0000 Thoralf Da__ler <thoralf.dassler@gmail.com> wrote:
>
>  > Yeah, it seems dumb that the init scripts pause, but I don't understand why this started with the 2.6.25 kernel, whereas it was fine before.
>
>  You have removed from Cc: the people who can explain it.  I restored them.
>
>  > Andrew Morton wrote:(switched to email.  Please respond via emailed reply-to-all, not via the
>
> > bugzilla web interface).

iwl3945 was loading the ucode (firmware) during probe but then changed
to do this on ifup.
for 4965 this does not cause such problems because the rfkill state is
reported via an interrupt; while for 3965 its an ucode event...
no ucode means no ucode event ;) (but it should be able to read the
current state after an ifup)

RE: [Bugme-new] [Bug 10471] New: iwl3945 in 2.6.25 ignores state of hardware RF kill switch

[Chatre, Reinette]

First to address the subject of this bug, the driver does not ignore the
state of the hardware RF kill switch. More below.

On , Andrew Morton  wrote:

> On Thu, 17 Apr 2008 14:17:51 -0700 (PDT)
> bugme-daemon@bugzilla.kernel.org wrote:
> 

>> 
>> 
>> messages during boot:
>> -----------------------------------
>> SIOCSIFFLAGS: No such device
>> [3 lines of irrelevant output on my iwl3945]
>> SIOCSIFFLAGS: No such device

This is exactly the error returned by the driver when it detects that
the hardware RF kill switch is set. With several message like above in
your logs - could it be a user app that keeps trying to bring the
interface up even when it returns an error?


>> dmesg:
>> -----------
>> ACPI: PCI Interrupt 0000:04:00.0[A] -> GSI 17 (level, low) -> IRQ 17
>> PM: Writing back config space on device 0000:04:00.0 at offset 1
>> (was 100002, writing 100006) iwl3945: Radio disabled by HW RF Kill
>> switch [message repeated 7 more times]

It is right after printing this message that the driver returns
"ENODEV".

>From what I can tell the driver is doing the right thing. What is the
expected behavior in this scenario?

Reinette