Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Tue, 18 Jun 2019 07:16:28 +0200
Message-ID: <s5h4l4nbhhv.wl-tiwai@suse.de>
From:   Takashi Iwai <tiwai@suse.de>
To:     Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Cc:     alsa-devel@alsa-project.org,
        =?UTF-8?B?IkFtYWRldXN6IFPFgmF3acWEc2tp?= =?UTF-8?B?Ig==?= 
        <amadeuszx.slawinski@linux.intel.com>,
        Liam Girdwood <lgirdwood@gmail.com>,
        Cezary Rojewski <cezary.rojewski@intel.com>,
        Mark Brown <broonie@kernel.org>,
        Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>,
        Jie Yang <yang.jie@linux.intel.com>,
        linux-kernel@vger.kernel.org
Subject: Re: [alsa-devel] [PATCH v2 09/11] ASoC: Intel: hdac_hdmi: Set ops to NULL on remove
In-Reply-To: <1fd4af8e1b4bad5eda6f1e2f747b7988c74408fb.camel@linux.intel.com>
References: <20190617113644.25621-1-amadeuszx.slawinski@linux.intel.com>
        <20190617113644.25621-10-amadeuszx.slawinski@linux.intel.com>
        <75be86354032f4886cbaf7d430de2aa89eaab573.camel@linux.intel.com>
        <s5h7e9jc2s1.wl-tiwai@suse.de>
        <1fd4af8e1b4bad5eda6f1e2f747b7988c74408fb.camel@linux.intel.com>
User-Agent: Wanderlust/2.15.9 (Almost Unreal) SEMI/1.14.6 (Maruoka)
 FLIM/1.14.9 (=?UTF-8?B?R29qxY0=?=) APEL/10.8 Emacs/25.3
 (x86_64-suse-linux-gnu) MULE/6.0 (HANACHIRUSATO)
MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka")
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, 18 Jun 2019 06:19:15 +0200,
Ranjani Sridharan wrote:
> 
> On Mon, 2019-06-17 at 23:36 +0200, Takashi Iwai wrote:
> > On Mon, 17 Jun 2019 22:51:42 +0200,
> > Ranjani Sridharan wrote:
> > > 
> > > On Mon, 2019-06-17 at 13:36 +0200, Amadeusz Sławiński wrote:
> > > > When we unload Skylake driver we may end up calling
> > > > hdac_component_master_unbind(), it uses acomp->audio_ops, which
> > > > we
> > > > set
> > > > in hdmi_codec_probe(), so we need to set it to NULL in
> > > > hdmi_codec_remove(),
> > > > otherwise we will dereference no longer existing pointer.
> > > 
> > > Hi Amadeusz,
> > > 
> > > It looks like the audio_ops should be deleted
> > > snd_hdac_acomp_exit().
> > 
> > It's a different one.  The codec driver registers / de-registers the
> > notifier at its probe/remove, while the controller driver takes care
> > of snd_hdac_acomp_init().  snd_hdac_acomp_exit() is usually not
> > needed
> > to be called explicitly, as it's managed via devres.
> 
> Makes sense, thanks Takashi. But I am still wondering why we havent
> seen this issue with SOF yet. We run the module unload/reload stress
> test as well and havent seen this yet. 

Usually this isn't a problem as the controller is removed at first.
But if the codec is unbound at first by some reason, it can be a
problem without unregistering, I guess.


Takashi

> 
> Thanks,
> Ranjani
> > 
> > 
> > Takashi
> > 
> > > Also, this doesnt seem to be the case with when the SOF driver is
> > > removed.
> > > Could you please give a bit more context on what error you see when
> > > this happens?
> > > 
> > > Thanks,
> > > Ranjani
> > > > 
> > > > Signed-off-by: Amadeusz Sławiński <
> > > > amadeuszx.slawinski@linux.intel.com>
> > > > ---
> > > >  sound/soc/codecs/hdac_hdmi.c | 6 ++++++
> > > >  1 file changed, 6 insertions(+)
> > > > 
> > > > diff --git a/sound/soc/codecs/hdac_hdmi.c
> > > > b/sound/soc/codecs/hdac_hdmi.c
> > > > index 911bb6e2a1ac..9ee1bff548d8 100644
> > > > --- a/sound/soc/codecs/hdac_hdmi.c
> > > > +++ b/sound/soc/codecs/hdac_hdmi.c
> > > > @@ -1890,6 +1890,12 @@ static void hdmi_codec_remove(struct
> > > > snd_soc_component *component)
> > > >  {
> > > >  	struct hdac_hdmi_priv *hdmi =
> > > > snd_soc_component_get_drvdata(component);
> > > >  	struct hdac_device *hdev = hdmi->hdev;
> > > > +	int ret;
> > > > +
> > > > +	ret = snd_hdac_acomp_register_notifier(hdev->bus, NULL);
> > > > +	if (ret < 0)
> > > > +		dev_err(&hdev->dev, "notifier unregister failed: err:
> > > > %d\n",
> > > > +				ret);
> > > >  
> > > >  	pm_runtime_disable(&hdev->dev);
> > > >  }
> > > 
> > > 
>