[alsa-devel] [PATCH 0/2] ASoC: Refcount fixes

Hi
Nothing new, just yesterday's refcount fix for-2.6.38 and another minor for-2.6.39 rebased on top of for-2.6.39 and sent together.

Commit f6c2ed5 "ASoC: Fix the device references to codec and platform drivers" moved codec driver refcount increments from soc_bind_dai_link into soc_probe_codec.
However, the commit didn't remove try_module_get from soc_probe_aux_dev so the auxiliary device reference counts are incremented twice as the soc_probe_codec is called from soc_probe_aux_dev too.
Signed-off-by: Jarkko Nikula jhnikula@gmail.com Cc: Vinod Koul vinod.koul@intel.com Cc: Harsha Priya priya.harsha@intel.com --- sound/soc/soc-core.c | 3 --- 1 files changed, 0 insertions(+), 3 deletions(-)
diff --git a/sound/soc/soc-core.c b/sound/soc/soc-core.c index 8af47f7..e9f81c5 100644 --- a/sound/soc/soc-core.c +++ b/sound/soc/soc-core.c @@ -1670,9 +1670,6 @@ static int soc_probe_aux_dev(struct snd_soc_card *card, int num) goto out;
found: - if (!try_module_get(codec->dev->driver->owner)) - return -ENODEV; - ret = soc_probe_codec(card, codec); if (ret < 0) return ret;

On Thu, Jan 27, 2011 at 04:24:21PM +0200, Jarkko Nikula wrote:
Commit f6c2ed5 "ASoC: Fix the device references to codec and platform drivers" moved codec driver refcount increments from soc_bind_dai_link into soc_probe_codec.
However, the commit didn't remove try_module_get from soc_probe_aux_dev so the auxiliary device reference counts are incremented twice as the soc_probe_codec is called from soc_probe_aux_dev too.
This is already applied for 2.6.38.

Commit f6c2ed5 "ASoC: Fix the device references to codec and platform drivers" moved codec and platform driver refcount increments from soc_bind_dai_link to more appropriate places.
Adjust a little them so that refcounts are incremented before executing the driver probe functions.
Signed-off-by: Jarkko Nikula jhnikula@gmail.com Cc: Vinod Koul vinod.koul@intel.com Cc: Harsha Priya priya.harsha@intel.com --- I don't know any regression with this so probably 2.6.39 material. --- sound/soc/soc-core.c | 21 +++++++++++++-------- 1 files changed, 13 insertions(+), 8 deletions(-)
diff --git a/sound/soc/soc-core.c b/sound/soc/soc-core.c index e9f81c5..9dfbb8f 100644 --- a/sound/soc/soc-core.c +++ b/sound/soc/soc-core.c @@ -1412,26 +1412,31 @@ static int soc_probe_codec(struct snd_soc_card *card, codec->dapm.card = card; soc_set_name_prefix(card, codec);
+ if (!try_module_get(codec->dev->driver->owner)) + return -ENODEV; + if (codec->driver->probe) { ret = codec->driver->probe(codec); if (ret < 0) { dev_err(codec->dev, "asoc: failed to probe CODEC %s: %d\n", codec->name, ret); - return ret; + goto err_probe; } }
soc_init_codec_debugfs(codec);
/* mark codec as probed and add to card codec list */ - if (!try_module_get(codec->dev->driver->owner)) - return -ENODEV; - codec->probed = 1; list_add(&codec->card_list, &card->codec_dev_list); list_add(&codec->dapm.list, &card->dapm_list);
+ return 0; + +err_probe: + module_put(codec->dev->driver->owner); + return ret; }
@@ -1549,19 +1554,19 @@ static int soc_probe_dai_link(struct snd_soc_card *card, int num)
/* probe the platform */ if (!platform->probed) { + if (!try_module_get(platform->dev->driver->owner)) + return -ENODEV; + if (platform->driver->probe) { ret = platform->driver->probe(platform); if (ret < 0) { printk(KERN_ERR "asoc: failed to probe platform %s\n", platform->name); + module_put(platform->dev->driver->owner); return ret; } } /* mark platform as probed and add to card platform list */ - - if (!try_module_get(platform->dev->driver->owner)) - return -ENODEV; - platform->probed = 1; list_add(&platform->card_list, &card->platform_dev_list); }

On Thu, Jan 27, 2011 at 04:24:22PM +0200, Jarkko Nikula wrote:
Commit f6c2ed5 "ASoC: Fix the device references to codec and platform drivers" moved codec and platform driver refcount increments from soc_bind_dai_link to more appropriate places.
Adjust a little them so that refcounts are incremented before executing the driver probe functions.
Applied, thanks.

On Thu, 2011-01-27 at 16:24 +0200, Jarkko Nikula wrote:
Hi
Nothing new, just yesterday's refcount fix for-2.6.38 and another minor for-2.6.39 rebased on top of for-2.6.39 and sent together.
All
Acked-by: Liam Girdwood lrg@slimlogic.co.uk
participants (3)
-
Jarkko Nikula
-
Liam Girdwood
-
Mark Brown