| Message ID | 20200213061147.29386-2-samuel@sholland.org (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | b6570fdb96edf45bcf71884bd2644bd73d348d1a |
| Headers | show |
| Series | simple-audio-card codec2codec support | expand |
On Thu 13 Feb 2020 at 07:11, Samuel Holland <samuel@sholland.org> wrote: > The PCM runtime was freed during PMU in the case that the event hook > encountered an error. However, it is also unconditionally freed during > PMD. Avoid a double-free by dropping the call to kfree in the PMU > hook. Oh ... Thanks for finding this. I thought that a widget which has failed PMU would not go through PMD, but It seems the return value dapm_seq_check_event is not checked. This brings another question/problem: A link which has failed in PMU, could try in PMD to hw_free/shutdown a dai which has not gone through startup/hw_params, right ? > > Fixes: a72706ed8208 ("ASoC: codec2codec: remove ephemeral variables") > Cc: stable@vger.kernel.org > Signed-off-by: Samuel Holland <samuel@sholland.org> > --- > sound/soc/soc-dapm.c | 3 --- > 1 file changed, 3 deletions(-) > > diff --git a/sound/soc/soc-dapm.c b/sound/soc/soc-dapm.c > index b6378f025836..935b5375ecc5 100644 > --- a/sound/soc/soc-dapm.c > +++ b/sound/soc/soc-dapm.c > @@ -3888,9 +3888,6 @@ snd_soc_dai_link_event_pre_pmu(struct snd_soc_dapm_widget *w, > runtime->rate = params_rate(params); > > out: > - if (ret < 0) > - kfree(runtime); > - > kfree(params); > return ret; > }
On Thu, Feb 13, 2020 at 09:37:18AM +0100, Jerome Brunet wrote: > This brings another question/problem: > A link which has failed in PMU, could try in PMD to hw_free/shutdown a > dai which has not gone through startup/hw_params, right ? I think so, yes.
On Thu 13 Feb 2020 at 12:37, Mark Brown <broonie@kernel.org> wrote: > On Thu, Feb 13, 2020 at 09:37:18AM +0100, Jerome Brunet wrote: > >> This brings another question/problem: >> A link which has failed in PMU, could try in PMD to hw_free/shutdown a >> dai which has not gone through startup/hw_params, right ? > > I think so, yes. Maybe this can be solved using the dai active counts which the codec-to-codec event is not updating. I'll try to come up with something.
diff --git a/sound/soc/soc-dapm.c b/sound/soc/soc-dapm.c index b6378f025836..935b5375ecc5 100644 --- a/sound/soc/soc-dapm.c +++ b/sound/soc/soc-dapm.c @@ -3888,9 +3888,6 @@ snd_soc_dai_link_event_pre_pmu(struct snd_soc_dapm_widget *w, runtime->rate = params_rate(params); out: - if (ret < 0) - kfree(runtime); - kfree(params); return ret; }
The PCM runtime was freed during PMU in the case that the event hook encountered an error. However, it is also unconditionally freed during PMD. Avoid a double-free by dropping the call to kfree in the PMU hook. Fixes: a72706ed8208 ("ASoC: codec2codec: remove ephemeral variables") Cc: stable@vger.kernel.org Signed-off-by: Samuel Holland <samuel@sholland.org> --- sound/soc/soc-dapm.c | 3 --- 1 file changed, 3 deletions(-)