[PATCH2/4] hda: dynamic jack id

[PATCH2/4] hda: dynamic jack id

[Matthew Ranostay]

This patch duplicates the jack->id pointer with kstrdup() to prevent
scoping issues from calling autoprobing functions from the HDA section.

Signed-off-by: Matthew Ranostay <mranostay@embeddedalley.com>
---

diff --git a/sound/core/jack.c b/sound/core/jack.c
index 4978db1..284432f 100644
--- a/sound/core/jack.c
+++ b/sound/core/jack.c
@@ -34,6 +34,7 @@ static int snd_jack_dev_free(struct snd_device *device)
 	else
 		input_free_device(jack->input_dev);

+	kfree(jack->id);
 	kfree(jack);

 	return 0;
@@ -87,7 +88,7 @@ int snd_jack_new(struct snd_card *card, const char *id, int type,
 	if (jack == NULL)
 		return -ENOMEM;

-	jack->id = id;
+	jack->id = kstrdup(id, GFP_KERNEL);

 	jack->input_dev = input_allocate_device();
 	if (jack->input_dev == NULL) {

[PATCH2/4] hda: dynamic jack id

[Matthew Ranostay]

Allow jack->id to be kstrdup()'ed to prevent scope issues from
calling functions.

Signed-off-by: Matthew Ranostay <mranostay@embeddedalley.com>
---

diff --git a/sound/core/jack.c b/sound/core/jack.c
index 4978db1..284432f 100644
--- a/sound/core/jack.c
+++ b/sound/core/jack.c
@@ -34,6 +34,7 @@ static int snd_jack_dev_free(struct snd_device *device)
 	else
 		input_free_device(jack->input_dev);

+	kfree(jack->id);
 	kfree(jack);

 	return 0;
@@ -87,7 +88,7 @@ int snd_jack_new(struct snd_card *card, const char *id, int type,
 	if (jack == NULL)
 		return -ENOMEM;

-	jack->id = id;
+	jack->id = kstrdup(id, GFP_KERNEL);

 	jack->input_dev = input_allocate_device();
 	if (jack->input_dev == NULL) {

Re: [PATCH2/4] hda: dynamic jack id

[Mark Brown]

On Tue, Oct 21, 2008 at 01:26:48PM -0400, Matthew Ranostay wrote:
> Allow jack->id to be kstrdup()'ed to prevent scope issues from
> calling functions.

Your patch doesn't allow it to be duplicated - it does actually
duplicate it.  Do you mean to say that you want to duplicate the string
so that users can automatically generate the string without having to
manage the memory for it?

That should be OK for the embedded case, the size of the string isn't
such a big deal if you're able to handle a reasonable amount of audio
data, and it's probably going to be a common case with autoprobable
things like HDA.

I'd prefer a clearer changelog but:

Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>