Re: [PATCH] device property: Fix usecount for of_graph_get_port_parent()

From: Rob Herring
Date: Thu Jul 27 2017 - 18:19:04 EST


On Thu, Jul 27, 2017 at 4:44 AM, Tony Lindgren <tony@xxxxxxxxxxx> wrote:
> Fix inconsistent use of of_graph_get_port_parent() where
> asoc_simple_card_parse_graph_dai() does of_node_get() before
> calling it while other callers do not. We can fix this by
> not trashing the node passed to of_graph_get_port_parent().
>
> Let's make sure the users have correct refcounts and remove
> related incorrect of_node_put() calls for of_for_each_phandle
> as that's done by of_phandle_iterator_next().
>
> Let's fix both issues with a single patch to avoid kobject
> refcounts getting messed up more if two patches are merged
> separately.
>
> Otherwise strange issues can happen caused by memory corruption
> caused by too many kobject_del() calls such as:
>
> BUG: sleeping function called from invalid context at
> kernel/locking/mutex.c:747
> ...
> (___might_sleep)
> (__mutex_lock)
> (mutex_lock_nested)
> (kernfs_remove)
> (kobject_del)
> (kobject_put)
> (of_get_next_parent)
> (of_graph_get_port_parent)
> (asoc_simple_card_parse_graph_dai [snd_soc_simple_card_utils])
> (asoc_graph_card_probe [snd_soc_audio_graph_card])
>
> Fixes: 0ef472a973eb ("of_graph: add of_graph_get_port_parent()")
> Fixes: 2692c1c63c29 ("ASoC: add audio-graph-card support")
> Fixes: 1689333f8311 ("ASoC: simple-card-utils: add asoc_simple_card_parse_graph_dai()")
> Cc: Mark Brown <broonie@xxxxxxxxxx>
> Cc: Takashi Iwai <tiwai@xxxxxxxx>
> Cc: Kuninori Morimoto <kuninori.morimoto.gx@xxxxxxxxxxx>
> Cc: alsa-devel@xxxxxxxxxxxxxxxx
> Signed-off-by: Tony Lindgren <tony@xxxxxxxxxxx>
> ---
> drivers/of/property.c | 9 +++++++++
> sound/soc/generic/audio-graph-card.c | 5 +----
> sound/soc/generic/simple-card-utils.c | 5 -----
> 3 files changed, 10 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/of/property.c b/drivers/of/property.c
> --- a/drivers/of/property.c
> +++ b/drivers/of/property.c
> @@ -708,6 +708,15 @@ struct device_node *of_graph_get_port_parent(struct device_node *node)
> {
> unsigned int depth;
>
> + if (!node)
> + return NULL;
> +
> + /*
> + * Preserve usecount for passed in node as of_get_next_parent()
> + * will do of_node_put() on it.
> + */
> + of_node_get(node);

I think this messes up of_graph_get_remote_port_parent(). First it
calls of_graph_get_remote_endpoint which returns the endpoint node
with ref count incremented. Then you are incrementing it again here.

> +
> /* Walk 3 levels up only if there is 'ports' node. */
> for (depth = 3; depth && node; depth--) {
> node = of_get_next_parent(node);
> diff --git a/sound/soc/generic/audio-graph-card.c b/sound/soc/generic/audio-graph-card.c
> --- a/sound/soc/generic/audio-graph-card.c
> +++ b/sound/soc/generic/audio-graph-card.c
> @@ -224,7 +224,6 @@ static int asoc_graph_card_parse_of(struct graph_card_data *priv)
>
> of_for_each_phandle(&it, rc, node, "dais", NULL, 0) {
> ret = asoc_graph_card_dai_link_of(it.node, priv, idx++);
> - of_node_put(it.node);
> if (ret < 0)
> return ret;

I think you need a put here.

Rob