Re: [PATCH 1/4] ath10k: Add support to configure pktlog filter

[Rajkumar Manoharan <rmanohar@qti.qualcomm.com>]

On Thu, Oct 02, 2014 at 02:37:13PM +0200, Michal Kazior wrote:
> On 2 October 2014 14:12, Rajkumar Manoharan <rmanohar@qti.qualcomm.com> wrote:
> > Add support to configure packet log filters (tx, rx, rate control)
> > via debugfs. To disable htt pktlog events set the filters to 0.
> [...]
> > +static ssize_t ath10k_write_pktlog_filter(struct file *file,
> > +                                         const char __user *ubuf,
> > +                                         size_t count, loff_t *ppos)
> > +{
> > +       struct ath10k *ar = file->private_data;
> > +       u32 filter;
> > +       int ret;
> > +
> > +       if (kstrtouint_from_user(ubuf, count, 0, &filter))
> > +               return -EINVAL;
> > +
> > +       mutex_lock(&ar->conf_mutex);
> > +       if (filter && (filter != ar->debug.pktlog_filter))
> > +               ret = ath10k_wmi_pdev_pktlog_enable(ar, filter);
> > +       else
> > +               ret = ath10k_wmi_pdev_pktlog_disable(ar);
> 
> You're not checking if ar->state is ON.
>
How about updating pktlog_filter alone and not sending WMI command when
state != ON. So that in the next drv_start pktlog will be either be
disabled or enabled based on filter value at debug_start.

> Even if I assume the above doesn't crash the driver/system this has
> strange semantics. On one hand the driver has to be after drv_start
> (e.g. at least one interface must be up) before it's possible to set
> it but on the other hand pktlog_filter is retained across
> drv_stop/start. I'd expect it to be possible to set the pktlog_filter
> regardless of driver state but maybe that's just me.
> 
So pktlog should be disabed at ath10k_debug_stop. isn't it?

-Rajkumar