Re: [PATCH] b43: Accessing the TSF via mac80211

[Michael Buesch <mb@bu3sch.de>]

On Sunday 25 January 2009 13:55:49 Michael Buesch wrote:
> On Sunday 25 January 2009 13:51:50 Michael Buesch wrote:
> > On Sunday 25 January 2009 13:41:51 Alina Friedrichsen wrote:
> > > This allows the mac80211 high level code to access the TSF. This is e.g. needed for BSSID merges in the IBSS mode.
> > 
> > Thanks, can you also remove our private debugfs hook?
> > 
> > > Signed-off-by: Alina Friedrichsen <x-alina@gmx.net>
> > > ---
> > > diff -urN wireless-testing.orig/drivers/net/wireless/b43/main.c wireless-testing/drivers/net/wireless/b43/main.c
> > > --- wireless-testing.orig/drivers/net/wireless/b43/main.c	2009-01-25 06:12:13.000000000 +0100
> > > +++ wireless-testing/drivers/net/wireless/b43/main.c	2009-01-25 13:09:35.000000000 +0100
> > > @@ -3177,6 +3177,25 @@
> > >  	return 0;
> > >  }
> > >  
> > > +static u64 b43_op_get_tsf(struct ieee80211_hw *hw)
> > > +{
> > > +	struct b43_wl *wl = hw_to_b43_wl(hw);
> > > +	struct b43_wldev *dev = wl->current_dev;
> > > +	u64 tsf;
> > > +
> > 
> > You must add a check and locking here.
> > 
> > 	if (dev && (b43_status(dev) >= B43_STAT_INITIALIZED)) {
> > 		mutex_lock(&wl->mutex);
> > 		spin_lock_irq(&wl->irq_lock);
> > 
> > > +	b43_tsf_read(dev, &tsf);
> > 
> > 		spin_unlock_irq(&wl->irq_lock);
> > 		mutex_unlock(&wl->mutex);
> > 	} else
> > 		tsf = 0; //FIXME what to do on error?
> > 
> > > +	return tsf;
> > > +}
> 
> Whoops, I got it wrong :D
> This is what it should look like:
> 
> static u64 b43_op_get_tsf(struct ieee80211_hw *hw)
> {
>        struct b43_wl *wl = hw_to_b43_wl(hw);
>        struct b43_wldev *dev = wl->current_dev;
>        u64 tsf;
> 
> 	mutex_lock(&wl->mutex);
> 	spin_lock_irq(&wl->irq_lock);
> 	if (dev && (b43_status(dev) >= B43_STAT_INITIALIZED))
> 		b43_tsf_read(dev, &tsf);
> 	else
> 		tsf = 0;//FIXME? what to do on error?
> 	spin_unlock_irq(&wl->irq_lock);
> 	mutex_unlock(&wl->mutex);
> 
>        return tsf;
> }
> 

Damn, I'm not awake, yet.
The   dev = wl->current_dev   also must be inside of the mutex.

-- 
Greetings, Michael.