Re: [PATCH RFC 00/21] ALSA: pcm: add tracepoints for PCM params operation

From: Takashi Iwai <tiwai@suse.de>
To: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Cc: alsa-devel@alsa-project.org, lars@metafoo.de, clemens@ladisch.de
Subject: Re: [PATCH RFC 00/21] ALSA: pcm: add tracepoints for PCM params operation
Date: Mon, 15 May 2017 16:34:33 +0200	[thread overview]
Message-ID: <s5hvap289za.wl-tiwai@suse.de> (raw)
In-Reply-To: <bba04aee-5a88-ca35-3a05-41b23c738604@sakamocchi.jp>

On Mon, 15 May 2017 16:29:47 +0200,
Takashi Sakamoto wrote:
> 
> On May 15 2017 17:42, Takashi Iwai wrote:
> > On Sun, 14 May 2017 10:57:35 +0200,
> > Takashi Sakamoto wrote:
> >>
> >> Hi,
> >>
> >> In the last Audio Mini Conference held with Linux Plumber conference 2016[1], I mentioned about tracepoints for PCM params operation. This patchset is for the idea.
> >>
> >> In ALSA PCM interface, applications can get hardware capability by ioctl(2) with SNDRV_PCM_IOCTL_HW_REFINE/SNDRV_PCM_IOCT_HW_PARAMS in a shape of 'struct snd_pcm_hw_params'. In kernel side, relevant processing is somewhat complicated and developers sometimes have hard time to debug drivers for PCM constraints and rules.
> >>
> >> This patchset adds tracepoints for hw_params operations. When CONFIG_SND_DEBUG is enabled, you can see 'trace_hw_params_mask/trace_hw_params_interval' events of 'snd_pcm' subsystem. When applications execute ioctl(2) with SNDRV_PCM_IOCTL_HW_REFINE/SNDRV_PCM_IOCTL_HW_PARAMS, these events are probed. Developers can get how many PCM rules are added into runtime of PCM substream and which rule changed which parameters.
> >>
> >> This patchset also includes some improvements. The last three commits brings small changes to kernel/userspace interface for error handling.
> >>
> >> I'm happy to receive your comment for this patchset. For your information, low level application of SNDRV_PCM_IOCTL_HW_REFINE operation is available in my github repository[2].
> >
> > The patches look good through a quick glance.
> > The only concern I have is the function regression, since there are
> > lots of code rewrites.  How did you test?
> 
> Currently, I did four things:
> 
> 1. understand logic to process parameters, constraints and rules
> 2. add the tracepoints as early as the patchset
> 3. confirm that probed events include the same data commit by commit
> 4. do the above with refine-pcm-params.c and got valid results
> 
> For the above, I use ALSA Fireweorks/OXFW driver and supported
> devices, which I know correct behaviour.

Maybe we can have some test set using dummy driver, too?

Takashi