[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1458219853.24844.10.camel@intel.com>
Date: Thu, 17 Mar 2016 13:04:15 +0000
From: "Pan, Harry" <harry.pan@...el.com>
To: "lars@...afoo.de" <lars@...afoo.de>,
"broonie@...nel.org" <broonie@...nel.org>
CC: "lgirdwood@...il.com" <lgirdwood@...il.com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"gs0622@...il.com" <gs0622@...il.com>,
"alsa-devel@...a-project.org" <alsa-devel@...a-project.org>,
"tiwai@...e.com" <tiwai@...e.com>,
"perex@...ex.cz" <perex@...ex.cz>
Subject: Re: [PATCH] ASoC: dapm: Do not traverse widget hooks to
snd-soc-dummy
> I'd say as a quick fix for stable check that card is not NULL in
> dapm_widget_show_component(). And as a longterm fix get rid of
> dapm_widget
> file. Nobody should hopefully use it anymore with debugfs being
> available as
> the far better alternative.
>
> - Lars
Well, that was my original approach while I realized problem
is caused by de-referencing a null pointer to its member of
'widget' list.
i.e.
https://chromium-review.googlesource.com/#/c/331285/3/sound/soc/soc-dap
m.c
+ if (WARN_ON(!codec->component.card))
+ return 0;
+
Additional remark is that I was working on chromium-os v3.18 kernel,
so that the interface and argument are sort of different than what
I then cherry-pick'ed then sent, which was based on latest 4.5.
Then I was stuck at here couple days keeping hunting root cause,
Turned out I found that commit skipped probing dummy component.
Long story short.
-Harry
Powered by blists - more mailing lists