[alsa-devel] [PATCH 2/2] ALSA: hda - Fix another race in runtime PM refcounting
Takashi Iwai
tiwai at suse.de
Mon Apr 13 13:22:37 CEST 2015
Although some races in runtime PM refcount was fixed by the commit
[664c715573c2: ALSA: hda - Work around races of power up/down with
runtime PM], there is still a race in the following case:
CPU0: CPU1 :
runtime suspend:
codec->in_pm = 1
snd_hdac_power_up_pm():
pm_runtime_get_sync() skipped
suspend finished:
codec->in_pm = 0
snd_hdac_power_down_pm():
pm_runtime_put_*() is called!
For avoiding this situation, increment in_pm flag atomically when it's
non-zero, and decrement accordingly, to ensure that in_pm is set
consistently for the whole concurrent operations.
Also, since atomic_inc_not_zero() and atomic_dec_if_positive() are
lengthy inline functions, move snd_hdac_power_up_pm() and _down_pm()
to sound/hda/hdac_device.c as no inline functions.
Signed-off-by: Takashi Iwai <tiwai at suse.de>
---
include/sound/hdaudio.h | 32 ++++----------------------------
sound/hda/hdac_device.c | 30 ++++++++++++++++++++++++++++++
2 files changed, 34 insertions(+), 28 deletions(-)
diff --git a/include/sound/hdaudio.h b/include/sound/hdaudio.h
index 30446f17c6a6..2a8aa9dfb83d 100644
--- a/include/sound/hdaudio.h
+++ b/include/sound/hdaudio.h
@@ -139,39 +139,15 @@ static inline int snd_hdac_read_parm(struct hdac_device *codec, hda_nid_t nid,
#ifdef CONFIG_PM
void snd_hdac_power_up(struct hdac_device *codec);
void snd_hdac_power_down(struct hdac_device *codec);
+void snd_hdac_power_up_pm(struct hdac_device *codec);
+void snd_hdac_power_down_pm(struct hdac_device *codec);
#else
static inline void snd_hdac_power_up(struct hdac_device *codec) {}
static inline void snd_hdac_power_down(struct hdac_device *codec) {}
+static inline void snd_hdac_power_up_pm(struct hdac_device *codec) {}
+static inline void snd_hdac_power_down_pm(struct hdac_device *codec) {}
#endif
-/**
- * snd_hdac_power_up_pm - power up the codec
- * @codec: the codec object
- *
- * This function can be called in a recursive code path like init code
- * which may be called by PM suspend/resume again. OTOH, if a power-up
- * call must wake up the sleeper (e.g. in a kctl callback), use
- * snd_hdac_power_up() instead.
- */
-static inline void snd_hdac_power_up_pm(struct hdac_device *codec)
-{
- if (!atomic_read(&codec->in_pm))
- snd_hdac_power_up(codec);
-}
-
-/**
- * snd_hdac_power_down_pm - power down the codec
- * @codec: the codec object
- *
- * Like snd_hdac_power_up_pm(), this function is used in a recursive
- * code path like init code which may be called by PM suspend/resume again.
- */
-static inline void snd_hdac_power_down_pm(struct hdac_device *codec)
-{
- if (!atomic_read(&codec->in_pm))
- snd_hdac_power_down(codec);
-}
-
/*
* HD-audio codec base driver
*/
diff --git a/sound/hda/hdac_device.c b/sound/hda/hdac_device.c
index 92604bbcee5f..f75bf5622687 100644
--- a/sound/hda/hdac_device.c
+++ b/sound/hda/hdac_device.c
@@ -519,6 +519,36 @@ void snd_hdac_power_down(struct hdac_device *codec)
pm_runtime_put_autosuspend(dev);
}
EXPORT_SYMBOL_GPL(snd_hdac_power_down);
+
+/**
+ * snd_hdac_power_up_pm - power up the codec
+ * @codec: the codec object
+ *
+ * This function can be called in a recursive code path like init code
+ * which may be called by PM suspend/resume again. OTOH, if a power-up
+ * call must wake up the sleeper (e.g. in a kctl callback), use
+ * snd_hdac_power_up() instead.
+ */
+void snd_hdac_power_up_pm(struct hdac_device *codec)
+{
+ if (!atomic_inc_not_zero(&codec->in_pm))
+ snd_hdac_power_up(codec);
+}
+EXPORT_SYMBOL_GPL(snd_hdac_power_up_pm);
+
+/**
+ * snd_hdac_power_down_pm - power down the codec
+ * @codec: the codec object
+ *
+ * Like snd_hdac_power_up_pm(), this function is used in a recursive
+ * code path like init code which may be called by PM suspend/resume again.
+ */
+void snd_hdac_power_down_pm(struct hdac_device *codec)
+{
+ if (atomic_dec_if_positive(&codec->in_pm) < 0)
+ snd_hdac_power_down(codec);
+}
+EXPORT_SYMBOL_GPL(snd_hdac_power_down_pm);
#endif
/* codec vendor labels */
--
2.3.5
More information about the Alsa-devel
mailing list