Skip to contentSkip to conversation

ALSA: core: Allow drivers to set R/W wait time.

PATCH Superseded Liam Girdwood 2 participants 3 messages Jul 3, 2018 – Jul 4, 2018 Sound

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 ALSA: core: Allow drivers to set R/W wait time.

Currently ALSA core blocks userspace for about 10 seconds for PCM R/W IO. This needs to be configurable for modern hardware like DSPs where no pointer update in milliseconds can indicate terminal DSP errors.

Add a substream variable to set the wait time in ms. This allows userspace and drivers to recover more quickly from terminal DSP errors.

---

Changes since V1 :-
o Remove API method and alow drivers to set directly.
o Validate time is driver supplied times.

V2 :-
o Max wait calc now in ms.
o checkpatch.

 include/sound/pcm.h  |  1 +
 sound/core/pcm_lib.c | 12 +++++++++---
 2 files changed, 10 insertions(+), 3 deletions(-)
include/sound/pcm.h +1 -0
index e054c583d3b3..fcdf358a25f0 100644
@@ -462,6 +462,7 @@ struct snd_pcm_substream {
         /* -- timer section -- */
 	struct snd_timer *timer;		/* timer */
 	unsigned timer_running: 1;	/* time is running */
+	long wait_time;	/* time in ms for R/W to wait for avail */
 	/* -- next substream -- */
 	struct snd_pcm_substream *next;
 	/* -- linked substreams -- */
sound/core/pcm_lib.c +9 -3
index 44b5ae833082..82a41e4d7170 100644
@@ -1832,12 +1832,18 @@ static int wait_for_avail(struct snd_pcm_substream *substream,
 	if (runtime->no_period_wakeup)
 		wait_time = MAX_SCHEDULE_TIMEOUT;
 	else {
-		wait_time = 10;
+		/* use wait time from substream if available */
+		if (substream->wait_time)
+			wait_time = substream->wait_time;
+		else
+			wait_time = 10 * 1000; /* 10 secs */
+
 		if (runtime->rate) {
-			long t = runtime->period_size * 2 / runtime->rate;
+			long t = runtime->period_size * 2 /
+				 (runtime->rate / 1000);
 			wait_time = max(t, wait_time);
 		}
-		wait_time = msecs_to_jiffies(wait_time * 1000);
+		wait_time = msecs_to_jiffies(wait_time);
 	}
 
 	for (;;) {
Signature · 1 line
2.17.1
#1 lore raw
Quoted 48 lines · Liam Girdwood

Currently ALSA core blocks userspace for about 10 seconds for PCM R/W IO. This needs to be configurable for modern hardware like DSPs where no pointer update in milliseconds can indicate terminal DSP errors.

Add a substream variable to set the wait time in ms. This allows userspace and drivers to recover more quickly from terminal DSP errors.

---

Changes since V1 :-
o Remove API method and alow drivers to set directly.
o Validate time is driver supplied times.

V2 :-
o Max wait calc now in ms.
o checkpatch.

 include/sound/pcm.h  |  1 +
 sound/core/pcm_lib.c | 12 +++++++++---
 2 files changed, 10 insertions(+), 3 deletions(-)
include/sound/pcm.h +1 -0
index e054c583d3b3..fcdf358a25f0 100644
@@ -462,6 +462,7 @@ struct snd_pcm_substream {
         /* -- timer section -- */
 	struct snd_timer *timer;		/* timer */
 	unsigned timer_running: 1;	/* time is running */
+	long wait_time;	/* time in ms for R/W to wait for avail */
 	/* -- next substream -- */
 	struct snd_pcm_substream *next;
 	/* -- linked substreams -- */
sound/core/pcm_lib.c +4 -1
index 44b5ae833082..82a41e4d7170 100644
@@ -1832,12 +1832,18 @@ static int wait_for_avail(struct snd_pcm_substream *substream,
 	if (runtime->no_period_wakeup)
 		wait_time = MAX_SCHEDULE_TIMEOUT;
 	else {
-		wait_time = 10;
+		/* use wait time from substream if available */
+		if (substream->wait_time)
+			wait_time = substream->wait_time;
+		else

sound/core/pcm_lib.c static int wait_for_avail(struct snd_pcm_substream *substream, · line 1846

Quoted 10 lines
sound/core/pcm_lib.c +5 -2
+			wait_time = 10 * 1000; /* 10 secs */
+
 		if (runtime->rate) {
-			long t = runtime->period_size * 2 / runtime->rate;
+			long t = runtime->period_size * 2 /
+				 (runtime->rate / 1000);
 			wait_time = max(t, wait_time);
 		}
-		wait_time = msecs_to_jiffies(wait_time * 1000);
+		wait_time = msecs_to_jiffies(wait_time);

I can take the patch as is, but would you like to keep runtime->rate max check also in the case with substream->wait_time set?

Also, substream->wait_time isn't cleared at each open, right? The driver can do it if wants, and the driver can set statically at creation time if wants, too.

thanks,

Takashi

#2 lore raw
Quoted 56 lines · Takashi Iwai
Quoted 54 lines · Liam Girdwood

Currently ALSA core blocks userspace for about 10 seconds for PCM R/W IO. This needs to be configurable for modern hardware like DSPs where no pointer update in milliseconds can indicate terminal DSP errors.

Add a substream variable to set the wait time in ms. This allows userspace and drivers to recover more quickly from terminal DSP errors.

---

Changes since V1 :-
o Remove API method and alow drivers to set directly.
o Validate time is driver supplied times.

V2 :-
o Max wait calc now in ms.
o checkpatch.

 include/sound/pcm.h  |  1 +
 sound/core/pcm_lib.c | 12 +++++++++---
 2 files changed, 10 insertions(+), 3 deletions(-)
include/sound/pcm.h +1 -0
index e054c583d3b3..fcdf358a25f0 100644
@@ -462,6 +462,7 @@ struct snd_pcm_substream {
         /* -- timer section -- */
 	struct snd_timer *timer;		/* timer */
 	unsigned timer_running: 1;	/* time is running */
+	long wait_time;	/* time in ms for R/W to wait for avail */
 	/* -- next substream -- */
 	struct snd_pcm_substream *next;
 	/* -- linked substreams -- */
sound/core/pcm_lib.c +7 -2
index 44b5ae833082..82a41e4d7170 100644
@@ -1832,12 +1832,18 @@ static int wait_for_avail(struct snd_pcm_substream
 *substream,
 	if (runtime->no_period_wakeup)
 		wait_time = MAX_SCHEDULE_TIMEOUT;
 	else {
-		wait_time = 10;
+		/* use wait time from substream if available */
+		if (substream->wait_time)
+			wait_time = substream->wait_time;
+		else
+			wait_time = 10 * 1000; /* 10 secs */
+
 		if (runtime->rate) {
-			long t = runtime->period_size * 2 / runtime->rate;
+			long t = runtime->period_size * 2 /
Quoted 9 lines

sound/core/pcm_lib.c static int wait_for_avail(struct snd_pcm_substream *substream, · line 1846

Quoted 5 lines
sound/core/pcm_lib.c +2 -1
+				 (runtime->rate / 1000);
 			wait_time = max(t, wait_time);
 		}
-		wait_time = msecs_to_jiffies(wait_time * 1000);
+		wait_time = msecs_to_jiffies(wait_time);

I can take the patch as is, but would you like to keep runtime->rate max check also in the case with substream->wait_time set?

Good point, I was trying to keep the same constraints but it does seem like a good idea if the driver is setting wait_time then it should be set correctly for any hw_params. This also gives options to set wait time lower that two periods too if desired.

Quoted 3 lines

Also, substream->wait_time isn't cleared at each open, right? The driver can do it if wants, and the driver can set statically at creation time if wants, too.

SOF currently does this at creation time, but I can update this to be more flexible in hw_params.

I'll send a V4.

Thanks

Liam

Quoted 5 lines

thanks,

Takashi