[PATCH] rt2800: do not enable tbtt unless we are in a beacon mode

[PATCH] rt2800: do not enable tbtt unless we are in a beacon mode

[Josef Bacik]

Please be gentle, I'm a fs developer and this is my first foray into drivers, as
I'm tired of building ralinks driver everytime I update my kernel.  Whenever I
load the rt2800pci driver my box bogs down, and a few printk's later I
discovered its because I was getting 10's of thousands of TBTT interrupts a
second.  I discovered this was because we were setting the beacon timer to 0,
which is apparently what TBTT keys off of.  It seems to me that we should only
be enabling TBTT when we are in a beacon transmitting mode, which from what I
can tell is in AD-HOC and other such modes where the mac80211 layer would have
given us a proper beacon_int to set the beacon timer to instead of 0.  So this
is my fix, only enable TBTT if our sync mode is for beacon.  This makes it so my
box doesn't die everytime I load the rt2800pci driver.  Let me know if this is
acceptable, I just learned all these terms about 15 minutes ago via wikipedia,
so I really am not familiar with how this stuff is supposed to work.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>

---
 drivers/net/wireless/rt2x00/rt2800lib.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/drivers/net/wireless/rt2x00/rt2800lib.c b/drivers/net/wireless/rt2x00/rt2800lib.c
index eb6d129..e90d6fd 100644
--- a/drivers/net/wireless/rt2x00/rt2800lib.c
+++ b/drivers/net/wireless/rt2x00/rt2800lib.c
@@ -560,7 +560,8 @@ void rt2800_config_intf(struct rt2x00_dev *rt2x00dev, struct rt2x00_intf *intf,
 		rt2800_register_read(rt2x00dev, BCN_TIME_CFG, &reg);
 		rt2x00_set_field32(&reg, BCN_TIME_CFG_TSF_TICKING, 1);
 		rt2x00_set_field32(&reg, BCN_TIME_CFG_TSF_SYNC, conf->sync);
-		rt2x00_set_field32(&reg, BCN_TIME_CFG_TBTT_ENABLE, 1);
+		rt2x00_set_field32(&reg, BCN_TIME_CFG_TBTT_ENABLE,
+				   (conf->sync == TSF_SYNC_BEACON));
 		rt2800_register_write(rt2x00dev, BCN_TIME_CFG, reg);
 	}
 
-- 
1.6.2.5

Re: [PATCH] rt2800: do not enable tbtt unless we are in a beacon mode

[Josef Bacik]

On Sun, Nov 15, 2009 at 9:33 PM, Josef Bacik <josef@redhat.com> wrote:
> Please be gentle, I'm a fs developer and this is my first foray into drivers, as
> I'm tired of building ralinks driver everytime I update my kernel.  Whenever I
> load the rt2800pci driver my box bogs down, and a few printk's later I
> discovered its because I was getting 10's of thousands of TBTT interrupts a
> second.  I discovered this was because we were setting the beacon timer to 0,
> which is apparently what TBTT keys off of.  It seems to me that we should only
> be enabling TBTT when we are in a beacon transmitting mode, which from what I
> can tell is in AD-HOC and other such modes where the mac80211 layer would have
> given us a proper beacon_int to set the beacon timer to instead of 0.  So this
> is my fix, only enable TBTT if our sync mode is for beacon.  This makes it so my
> box doesn't die everytime I load the rt2800pci driver.  Let me know if this is
> acceptable, I just learned all these terms about 15 minutes ago via wikipedia,
> so I really am not familiar with how this stuff is supposed to work.  Thanks,
>
> Signed-off-by: Josef Bacik <josef@redhat.com>
>
> ---
>  drivers/net/wireless/rt2x00/rt2800lib.c |    3 ++-
>  1 files changed, 2 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/net/wireless/rt2x00/rt2800lib.c b/drivers/net/wireless/rt2x00/rt2800lib.c
> index eb6d129..e90d6fd 100644
> --- a/drivers/net/wireless/rt2x00/rt2800lib.c
> +++ b/drivers/net/wireless/rt2x00/rt2800lib.c
> @@ -560,7 +560,8 @@ void rt2800_config_intf(struct rt2x00_dev *rt2x00dev, struct rt2x00_intf *intf,
>                rt2800_register_read(rt2x00dev, BCN_TIME_CFG, &reg);
>                rt2x00_set_field32(&reg, BCN_TIME_CFG_TSF_TICKING, 1);
>                rt2x00_set_field32(&reg, BCN_TIME_CFG_TSF_SYNC, conf->sync);
> -               rt2x00_set_field32(&reg, BCN_TIME_CFG_TBTT_ENABLE, 1);
> +               rt2x00_set_field32(&reg, BCN_TIME_CFG_TBTT_ENABLE,
> +                                  (conf->sync == TSF_SYNC_BEACON));
>                rt2800_register_write(rt2x00dev, BCN_TIME_CFG, reg);
>        }
>

Oh btw this is against the wireless-testing git tree, I forgot to
mention that.  Thanks,

Josef

Re: [PATCH] rt2800: do not enable tbtt unless we are in a beacon mode

[Luis R. Rodriguez]

On Mon, Nov 16, 2009 at 6:21 AM, Josef Bacik <josef@toxicpanda.com> wrote:
> On Sun, Nov 15, 2009 at 9:33 PM, Josef Bacik <josef@redhat.com> wrote:
>> Please be gentle, I'm a fs developer and this is my first foray into drivers, as
>> I'm tired of building ralinks driver everytime I update my kernel.  Whenever I
>> load the rt2800pci driver my box bogs down, and a few printk's later I
>> discovered its because I was getting 10's of thousands of TBTT interrupts a
>> second.  I discovered this was because we were setting the beacon timer to 0,
>> which is apparently what TBTT keys off of.  It seems to me that we should only
>> be enabling TBTT when we are in a beacon transmitting mode, which from what I
>> can tell is in AD-HOC and other such modes where the mac80211 layer would have
>> given us a proper beacon_int to set the beacon timer to instead of 0.  So this
>> is my fix, only enable TBTT if our sync mode is for beacon.  This makes it so my
>> box doesn't die everytime I load the rt2800pci driver.  Let me know if this is
>> acceptable, I just learned all these terms about 15 minutes ago via wikipedia,
>> so I really am not familiar with how this stuff is supposed to work.  Thanks,
>>
>> Signed-off-by: Josef Bacik <josef@redhat.com>
>>
>> ---
>>  drivers/net/wireless/rt2x00/rt2800lib.c |    3 ++-
>>  1 files changed, 2 insertions(+), 1 deletions(-)
>>
>> diff --git a/drivers/net/wireless/rt2x00/rt2800lib.c b/drivers/net/wireless/rt2x00/rt2800lib.c
>> index eb6d129..e90d6fd 100644
>> --- a/drivers/net/wireless/rt2x00/rt2800lib.c
>> +++ b/drivers/net/wireless/rt2x00/rt2800lib.c
>> @@ -560,7 +560,8 @@ void rt2800_config_intf(struct rt2x00_dev *rt2x00dev, struct rt2x00_intf *intf,
>>                rt2800_register_read(rt2x00dev, BCN_TIME_CFG, &reg);
>>                rt2x00_set_field32(&reg, BCN_TIME_CFG_TSF_TICKING, 1);
>>                rt2x00_set_field32(&reg, BCN_TIME_CFG_TSF_SYNC, conf->sync);
>> -               rt2x00_set_field32(&reg, BCN_TIME_CFG_TBTT_ENABLE, 1);
>> +               rt2x00_set_field32(&reg, BCN_TIME_CFG_TBTT_ENABLE,
>> +                                  (conf->sync == TSF_SYNC_BEACON));
>>                rt2800_register_write(rt2x00dev, BCN_TIME_CFG, reg);
>>        }
>>
>
> Oh btw this is against the wireless-testing git tree, I forgot to
> mention that.  Thanks,

This is assumed, if you want it in you should send to: linville though.

  Luis

Re: [PATCH] rt2800: do not enable tbtt unless we are in a beacon mode

[Ivo van Doorn]

[ Added john to CC list ]

On Monday 16 November 2009, Josef Bacik wrote:
> Please be gentle, I'm a fs developer and this is my first foray into drivers, as
> I'm tired of building ralinks driver everytime I update my kernel.  Whenever I
> load the rt2800pci driver my box bogs down, and a few printk's later I
> discovered its because I was getting 10's of thousands of TBTT interrupts a
> second.  I discovered this was because we were setting the beacon timer to 0,
> which is apparently what TBTT keys off of.  It seems to me that we should only
> be enabling TBTT when we are in a beacon transmitting mode, which from what I
> can tell is in AD-HOC and other such modes where the mac80211 layer would have
> given us a proper beacon_int to set the beacon timer to instead of 0.  So this
> is my fix, only enable TBTT if our sync mode is for beacon.  This makes it so my
> box doesn't die everytime I load the rt2800pci driver.  Let me know if this is
> acceptable, I just learned all these terms about 15 minutes ago via wikipedia,
> so I really am not familiar with how this stuff is supposed to work.  Thanks,
> 
> Signed-off-by: Josef Bacik <josef@redhat.com>

Good catch.

Acked-by: Ivo van Doorn <IvDoorn@gmail.com>

> ---
>  drivers/net/wireless/rt2x00/rt2800lib.c |    3 ++-
>  1 files changed, 2 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/net/wireless/rt2x00/rt2800lib.c b/drivers/net/wireless/rt2x00/rt2800lib.c
> index eb6d129..e90d6fd 100644
> --- a/drivers/net/wireless/rt2x00/rt2800lib.c
> +++ b/drivers/net/wireless/rt2x00/rt2800lib.c
> @@ -560,7 +560,8 @@ void rt2800_config_intf(struct rt2x00_dev *rt2x00dev, struct rt2x00_intf *intf,
>  		rt2800_register_read(rt2x00dev, BCN_TIME_CFG, &reg);
>  		rt2x00_set_field32(&reg, BCN_TIME_CFG_TSF_TICKING, 1);
>  		rt2x00_set_field32(&reg, BCN_TIME_CFG_TSF_SYNC, conf->sync);
> -		rt2x00_set_field32(&reg, BCN_TIME_CFG_TBTT_ENABLE, 1);
> +		rt2x00_set_field32(&reg, BCN_TIME_CFG_TBTT_ENABLE,
> +				   (conf->sync == TSF_SYNC_BEACON));
>  		rt2800_register_write(rt2x00dev, BCN_TIME_CFG, reg);
>  	}
>  

Re: [PATCH] rt2800: do not enable tbtt unless we are in a beacon mode

[Holger Schurig]

> Please be gentle, I'm a fs developer and this is my first foray
> into drivers, as I'm tired of building ralinks driver everytime
> I update my kernel.

We are (sometimes) gentle, but hey, even for FS linux code 
the "please be gentle" shouldn't be part of the commit message, 
but rather be in the comment part of the patch mail.

-- 
http://www.holgerschurig.de