Continuous monitoring of wifi stats

Continuous monitoring of wifi stats

[John Goyette]

Hello all,

 

I am looking for some advice on a Libertas related issue.  I am having a problem with a recent feature we added to our embedded system that continuously monitors wifi statistics, i.e., signal strength.  We are using a Blackfin platform with a Marvell 8686 device in the G-SPI configuration.  Our kernel is based on a 2.6.28.6 release.  

 

The problem occurs after pressing Ctrl-C to terminate a user space program that is polling the wireless stats in a loop.  A number of libertas error messages appear indicating a failure to download command 0x0000.  I am pretty sure there is no command 0x0000, so I do not know how it got queued.  Here is a sample of the error messages from dmesg:

 

libertas: command 0x0000 timed out
libertas: requeueing command 0x0000 due to timeout (#1)
libertas: if_spi_host_to_card: invalid size requested: 0
libertas: DNLD_CMD: hw_host_to_card failed: -22
libertas: command 0x0000 timed out
libertas: requeueing command 0x0000 due to timeout (0000002 <http://schicksw/mantis/view.php?id=2> )
libertas: if_spi_host_to_card: invalid size requested: 0
libertas: DNLD_CMD: hw_host_to_card failed: -22
libertas: command 0x0000 timed out
libertas: requeueing command 0x0000 due to timeout (0000003 <http://schicksw/mantis/view.php?id=3> )
libertas: if_spi_host_to_card: invalid size requested: 0
libertas: DNLD_CMD: hw_host_to_card failed: -22
libertas: command 0x0000 timed out
libertas: Excessive timeouts submitting command 0x0000

 

 

This can be reproduced by enabling IEEE Power Savings mode and flooding the card with wireless stats requests by running something like the following from a terminal:

while true; do iwconfig eth1;done

 

Let this run for 30-60 secs, and then press Ctrl-C.  With IEEE power savings off, it may take several minutes, but does eventually give the same problem.  I have downloaded the compat-wireless-2.6.30 release and was able to cross-compile it for our kernel.  It helped a little in that it seemed to take longer before the error messaged occurred after pressing ctrl-c, but it did not completely resolve the problem.

 

Has anyone seen something similar?  Are there any more recent patches that may address this issue?

 

Thanks.

 

-John Goyette

Schick Technologies, Inc.

 

Re: Continuous monitoring of wifi stats

[Dan Williams]

On Wed, 2009-09-16 at 13:24 -0400, John Goyette wrote:
> Hello all,
> 
>  
> 
> I am looking for some advice on a Libertas related issue.  I am having a problem with a recent feature we added to our embedded system that continuously monitors wifi statistics, i.e., signal strength.  We are using a Blackfin platform with a Marvell 8686 device in the G-SPI configuration.  Our kernel is based on a 2.6.28.6 release.  

Any chance you can grab the driver from 2.6.31 and rebuild it for your
kernel?  There are a few fixes from earlier this summer for GSPI and
Power Saving mode from Andrey Yurovsky that may fix your issue.  I'm
thinking specifically of b3781c74373489fa325ce64efdf72f6c8567b783 and a
few of the other PS cleanups from July.

Dan

>  
> 
> The problem occurs after pressing Ctrl-C to terminate a user space program that is polling the wireless stats in a loop.  A number of libertas error messages appear indicating a failure to download command 0x0000.  I am pretty sure there is no command 0x0000, so I do not know how it got queued.  Here is a sample of the error messages from dmesg:
> 
>  
> 
> libertas: command 0x0000 timed out
> libertas: requeueing command 0x0000 due to timeout (#1)
> libertas: if_spi_host_to_card: invalid size requested: 0
> libertas: DNLD_CMD: hw_host_to_card failed: -22
> libertas: command 0x0000 timed out
> libertas: requeueing command 0x0000 due to timeout (0000002 <http://schicksw/mantis/view.php?id=2> )
> libertas: if_spi_host_to_card: invalid size requested: 0
> libertas: DNLD_CMD: hw_host_to_card failed: -22
> libertas: command 0x0000 timed out
> libertas: requeueing command 0x0000 due to timeout (0000003 <http://schicksw/mantis/view.php?id=3> )
> libertas: if_spi_host_to_card: invalid size requested: 0
> libertas: DNLD_CMD: hw_host_to_card failed: -22
> libertas: command 0x0000 timed out
> libertas: Excessive timeouts submitting command 0x0000
> 
>  
> 
> 
> 
> This can be reproduced by enabling IEEE Power Savings mode and flooding the card with wireless stats requests by running something like the following from a terminal:
> 
> while true; do iwconfig eth1;done
> 
>  
> 
> Let this run for 30-60 secs, and then press Ctrl-C.  With IEEE power savings off, it may take several minutes, but does eventually give the same problem.  I have downloaded the compat-wireless-2.6.30 release and was able to cross-compile it for our kernel.  It helped a little in that it seemed to take longer before the error messaged occurred after pressing ctrl-c, but it did not completely resolve the problem.
> 
>  
> 
> Has anyone seen something similar?  Are there any more recent patches that may address this issue?
> 
>  
> 
> Thanks.
> 
>  
> 
> -John Goyette
> 
> Schick Technologies, Inc.
> 
>  
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Continuous monitoring of wifi stats

[Andrey Yurovsky]

On Sun, Sep 20, 2009 at 8:10 AM, Dan Williams <dcbw@redhat.com> wrote:
> On Wed, 2009-09-16 at 13:24 -0400, John Goyette wrote:
>> Hello all,
>>
>>
>>
>> I am looking for some advice on a Libertas related issue.  I am having a problem with a recent feature we added to our embedded system that continuously monitors wifi statistics, i.e., signal strength.  We are using a Blackfin platform with a Marvell 8686 device in the G-SPI configuration.  Our kernel is based on a 2.6.28.6 release.
>
> Any chance you can grab the driver from 2.6.31 and rebuild it for your
> kernel?  There are a few fixes from earlier this summer for GSPI and
> Power Saving mode from Andrey Yurovsky that may fix your issue.  I'm
> thinking specifically of b3781c74373489fa325ce64efdf72f6c8567b783 and a
> few of the other PS cleanups from July.

I believe that John has the bug fixes in question -- John, can you
summarize your setup (which compat-wireless snapshot, etc)?

>> The problem occurs after pressing Ctrl-C to terminate a user space program that is polling the wireless stats in a loop.  A number of libertas error messages appear indicating a failure to download command 0x0000.  I am pretty sure there is no command 0x0000, so I do not know how it got queued.  Here is a sample of the error messages from dmesg:
>>
>>
>>
>> libertas: command 0x0000 timed out
>> libertas: requeueing command 0x0000 due to timeout (#1)
>> libertas: if_spi_host_to_card: invalid size requested: 0
>> libertas: DNLD_CMD: hw_host_to_card failed: -22
>> libertas: command 0x0000 timed out
>> libertas: requeueing command 0x0000 due to timeout (0000002 <http://schicksw/mantis/view.php?id=2> )
>> libertas: if_spi_host_to_card: invalid size requested: 0
>> libertas: DNLD_CMD: hw_host_to_card failed: -22
>> libertas: command 0x0000 timed out
>> libertas: requeueing command 0x0000 due to timeout (0000003 <http://schicksw/mantis/view.php?id=3> )
>> libertas: if_spi_host_to_card: invalid size requested: 0
>> libertas: DNLD_CMD: hw_host_to_card failed: -22
>> libertas: command 0x0000 timed out
>> libertas: Excessive timeouts submitting command 0x0000
>>
>>
>>
>>
>>
>> This can be reproduced by enabling IEEE Power Savings mode and flooding the card with wireless stats requests by running something like the following from a terminal:
>>
>> while true; do iwconfig eth1;done
>>
>>
>>
>> Let this run for 30-60 secs, and then press Ctrl-C.  With IEEE power savings off, it may take several minutes, but does eventually give the same problem.  I have downloaded the compat-wireless-2.6.30 release and was able to cross-compile it for our kernel.  It helped a little in that it seemed to take longer before the error messaged occurred after pressing ctrl-c, but it did not completely resolve the problem.
>>
>>
>>
>> Has anyone seen something similar?  Are there any more recent patches that may address this issue?
>>
>>
>>
>> Thanks.
>>
>>
>>
>> -John Goyette
>>
>> Schick Technologies, Inc.
>>
>>
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>