ASoC: soc-dapm.c: add comment for kzalloc()/kfree() on snd_soc_dai_link_event_pre_pmu()

snd_soc_dai_link_event_pre_pmu() is using kzalloc()/kfree() for params.
It looks we don't need to use these, but are necessary.
But, it is difficult to know why it is necessary without any comments.
This patch adds the reasons via comment.

Link: https://lore.kernel.org/all/Yxc2wzbZsSVZNf8Y@sirena.org.uk/
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/878rmubn47.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>

diff --git a/sound/soc/soc-dapm.c b/sound/soc/soc-dapm.c
index 76615d5..fc2f75e 100644 (file)
--- a/sound/soc/soc-dapm.c
+++ b/sound/soc/soc-dapm.c
@@ -3842,6 +3842,15 @@ snd_soc_dai_link_event_pre_pmu(struct snd_soc_dapm_widget *w,
        unsigned int fmt;
        int ret = 0;
 
+       /*
+        * NOTE
+        *
+        * snd_pcm_hw_params is quite large (608 bytes on arm64) and is
+        * starting to get a bit excessive for allocation on the stack,
+        * especially when you're building with some of the KASAN type
+        * stuff that increases stack usage.
+        * So, we use kzalloc()/kfree() for params in this function.
+        */
        params = kzalloc(sizeof(*params), GFP_KERNEL);
        if (!params)
                return -ENOMEM;
@@ -3939,7 +3948,9 @@ snd_soc_dai_link_event_pre_pmu(struct snd_soc_dapm_widget *w,
        runtime->rate = params_rate(params);
 
 out:
+       /* see above NOTE */
        kfree(params);
+
        return ret;
 }