Skip to contentSkip to conversation

Add capture support to compress API

PATCH In review Richard Fitzgerald 6 participants 20 messages Apr 18, 2013 – Apr 22, 2013 2 messages not in the archive

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover Add capture support to compress API

This chain of patches does some clearing up and add capture support to the compressed API.

Missing is support for capture streams that don't use the copy callback. Also I have included two patches at the end of the chain that remove the buffer pointers and just use the cumulative values however as our setup uses the copy callback these are untested past build so should be taken with caution.

Thanks,
Charles

Thess patches apply to:
git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git
branch "for-next"

based on top of commit
126825e ALSA: snd-usb: add quirks handler for DSD streams

- RichF

Charles Keepax (7):
ALSA: compress_core: Update calc_avail to use cumulative values
ALSA: compress_core: Calculate avail correctly for capture streams
ALSA: compress_core: Deconstify copy callback buffer
ALSA: compress_core: Add support for capture streams
ASoC: soc-compress: Deduce stream direction
ALSA: compress_core: Remove unused hw_pointer
ALSA: compress_core: Rework writes to use cumulative values

 include/sound/compress_driver.h |    4 +-
 sound/core/compress_offload.c   |  117 +++++++++++++++++++++++++-------------
 sound/soc/soc-compress.c        |   11 +++-
 3 files changed, 87 insertions(+), 45 deletions(-)
Signature · 1 line
1.7.2.5
#1 lore raw

PATCH 1/7 ALSA: compress_core: Update calc_avail to use cumulative values

The app_pointer is managed locally by the compress core for memory mapped DSPs but for DSPs that are not memory mapped this would have to be manually updated from within the DSP driver itself, which is hardly very idiomatic.

This patch switches to using the cumulative values to calculate the available buffer space because these are already gracefully passed out of the DSP driver to the compress core and otherwise should be functionally equivalent.

---
 sound/core/compress_offload.c |   23 +++++------------------
 1 files changed, 5 insertions(+), 18 deletions(-)
sound/core/compress_offload.c +5 -18
index c84abc8..27bd81a 100644
@@ -160,8 +160,6 @@ static int snd_compr_update_tstamp(struct snd_compr_stream *stream,
 static size_t snd_compr_calc_avail(struct snd_compr_stream *stream,
 		struct snd_compr_avail *avail)
 {
-	long avail_calc; /*this needs to be signed variable */
-
 	memset(avail, 0, sizeof(*avail));
 	snd_compr_update_tstamp(stream, &avail->tstamp);
 	/* Still need to return avail even if tstamp can't be filled in */
@@ -184,22 +182,11 @@ static size_t snd_compr_calc_avail(struct snd_compr_stream *stream,
 		return stream->runtime->buffer_size;
 	}
 
-	/* FIXME: this routine isn't consistent, in one test we use
-	 * cumulative values and in the other byte offsets. Do we
-	 * really need the byte offsets if the cumulative values have
-	 * been updated? In the PCM interface app_ptr and hw_ptr are
-	 * already cumulative */
-
-	avail_calc = stream->runtime->buffer_size -
-		(stream->runtime->app_pointer - stream->runtime->hw_pointer);
-	pr_debug("calc avail as %ld, app_ptr %lld, hw+ptr %lld\n", avail_calc,
-			stream->runtime->app_pointer,
-			stream->runtime->hw_pointer);
-	if (avail_calc >= stream->runtime->buffer_size)
-		avail_calc -= stream->runtime->buffer_size;
-	pr_debug("ret avail as %ld\n", avail_calc);
-	avail->avail = avail_calc;
-	return avail_calc;
+	avail->avail = stream->runtime->buffer_size -
+		(stream->runtime->total_bytes_available -
+		 stream->runtime->total_bytes_transferred);
+	pr_debug("ret avail as %lld\n", avail->avail);
+	return avail->avail;
 }
 
 static inline size_t snd_compr_get_avail(struct snd_compr_stream *stream)
Signature · 6 lines
1.7.2.5

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#2 lore raw

PATCH 2/7 ALSA: compress_core: Calculate avail correctly for capture streams

In reply to Richard Fitzgerald
---
 sound/core/compress_offload.c |   29 ++++++++++++++++++-----------
 1 files changed, 18 insertions(+), 11 deletions(-)
sound/core/compress_offload.c +18 -11
index 27bd81a..1f69863 100644
@@ -153,7 +153,10 @@ static int snd_compr_update_tstamp(struct snd_compr_stream *stream,
 	pr_debug("dsp consumed till %d total %d bytes\n",
 		tstamp->byte_offset, tstamp->copied_total);
 	stream->runtime->hw_pointer = tstamp->byte_offset;
-	stream->runtime->total_bytes_transferred = tstamp->copied_total;
+	if (stream->direction == SND_COMPRESS_PLAYBACK)
+		stream->runtime->total_bytes_transferred = tstamp->copied_total;
+	else
+		stream->runtime->total_bytes_available = tstamp->copied_total;
 	return 0;
 }
 
@@ -164,12 +167,9 @@ static size_t snd_compr_calc_avail(struct snd_compr_stream *stream,
 	snd_compr_update_tstamp(stream, &avail->tstamp);
 	/* Still need to return avail even if tstamp can't be filled in */
 
-	/* FIXME: This needs to be different for capture stream,
-	   available is # of compressed data, for playback it's
-	   remainder of buffer */
-
 	if (stream->runtime->total_bytes_available == 0 &&
-			stream->runtime->state == SNDRV_PCM_STATE_SETUP) {
+			stream->runtime->state == SNDRV_PCM_STATE_SETUP &&
+			stream->direction == SND_COMPRESS_PLAYBACK) {
 		pr_debug("detected init and someone forgot to do a write\n");
 		return stream->runtime->buffer_size;
 	}
@@ -178,13 +178,20 @@ static size_t snd_compr_calc_avail(struct snd_compr_stream *stream,
 			stream->runtime->total_bytes_transferred);
 	if (stream->runtime->total_bytes_available ==
 				stream->runtime->total_bytes_transferred) {
-		pr_debug("both pointers are same, returning full avail\n");
-		return stream->runtime->buffer_size;
+		if (stream->direction == SND_COMPRESS_PLAYBACK) {
+			pr_debug("both pointers are same, returning full avail\n");
+			return stream->runtime->buffer_size;
+		} else {
+			pr_debug("both pointers are same, returning no avail\n");
+			return 0;
+		}
 	}
 
-	avail->avail = stream->runtime->buffer_size -
-		(stream->runtime->total_bytes_available -
-		 stream->runtime->total_bytes_transferred);
+	avail->avail = stream->runtime->total_bytes_available -
+			stream->runtime->total_bytes_transferred;
+	if (stream->direction == SND_COMPRESS_PLAYBACK)
+		avail->avail = stream->runtime->buffer_size - avail->avail;
+
 	pr_debug("ret avail as %lld\n", avail->avail);
 	return avail->avail;
 }
Signature · 6 lines
1.7.2.5

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#3 lore raw

PATCH 3/7 ALSA: compress_core: Deconstify copy callback buffer

In reply to Richard Fitzgerald

The buffer passed to the copy callback should not be const because the copy callback can be used for capture and playback.

---
 include/sound/compress_driver.h |    2 +-
 sound/core/compress_offload.c   |    8 +++++---
 sound/soc/soc-compress.c        |    2 +-
 3 files changed, 7 insertions(+), 5 deletions(-)
include/sound/compress_driver.h +1 -1
index ff6c741..db8273a 100644
@@ -121,7 +121,7 @@ struct snd_compr_ops {
 	int (*trigger)(struct snd_compr_stream *stream, int cmd);
 	int (*pointer)(struct snd_compr_stream *stream,
 			struct snd_compr_tstamp *tstamp);
-	int (*copy)(struct snd_compr_stream *stream, const char __user *buf,
+	int (*copy)(struct snd_compr_stream *stream, char __user *buf,
 		       size_t count);
 	int (*mmap)(struct snd_compr_stream *stream,
 			struct vm_area_struct *vma);
sound/core/compress_offload.c +5 -3
index 1f69863..52ca4cc 100644
@@ -272,10 +272,12 @@ static ssize_t snd_compr_write(struct file *f, const char __user *buf,
 	if (avail > count)
 		avail = count;
 
-	if (stream->ops->copy)
-		retval = stream->ops->copy(stream, buf, avail);
-	else
+	if (stream->ops->copy) {
+		char __user* cbuf = (char __user*)buf;
+		retval = stream->ops->copy(stream, cbuf, avail);
+	} else {
 		retval = snd_compr_write_data(stream, buf, avail);
+	}
 	if (retval > 0)
 		stream->runtime->total_bytes_available += retval;
 
sound/soc/soc-compress.c +1 -1
index 29093a3..da83e56 100644
@@ -315,7 +315,7 @@ static int soc_compr_pointer(struct snd_compr_stream *cstream,
 }
 
 static int soc_compr_copy(struct snd_compr_stream *cstream,
-			  const char __user *buf, size_t count)
+			  char __user *buf, size_t count)
 {
 	struct snd_soc_pcm_runtime *rtd = cstream->private_data;
 	struct snd_soc_platform *platform = rtd->platform;
Signature · 6 lines
1.7.2.5

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#4 lore raw

PATCH 4/7 ALSA: compress_core: Add support for capture streams

In reply to Richard Fitzgerald
---
 sound/core/compress_offload.c |   43 +++++++++++++++++++++++++++++++++++++++-
 1 files changed, 41 insertions(+), 2 deletions(-)
sound/core/compress_offload.c +41 -2
index 52ca4cc..52a2765 100644
@@ -296,7 +296,41 @@ static ssize_t snd_compr_write(struct file *f, const char __user *buf,
 static ssize_t snd_compr_read(struct file *f, char __user *buf,
 		size_t count, loff_t *offset)
 {
-	return -ENXIO;
+	struct snd_compr_file *data = f->private_data;
+	struct snd_compr_stream *stream;
+	size_t avail;
+	int retval;
+
+	if (snd_BUG_ON(!data))
+		return -EFAULT;
+
+	stream = &data->stream;
+	mutex_lock(&stream->device->lock);
+
+	/* read is allowed when stream is running */
+	if (stream->runtime->state != SNDRV_PCM_STATE_RUNNING) {
+		retval = -EBADFD;
+		goto out;
+	}
+
+	avail = snd_compr_get_avail(stream);
+	pr_debug("avail returned %ld\n", (unsigned long)avail);
+	/* calculate how much we can read from buffer */
+	if (avail > count)
+		avail = count;
+
+	if (stream->ops->copy) {
+		retval = stream->ops->copy(stream, buf, avail);
+	} else {
+		retval = -ENXIO;
+		goto out;
+	}
+	if (retval > 0)
+		stream->runtime->total_bytes_transferred += retval;
+
+out:
+	mutex_unlock(&stream->device->lock);
+	return retval;
 }
 
 static int snd_compr_mmap(struct file *f, struct vm_area_struct *vma)
@@ -481,9 +515,14 @@ snd_compr_set_params(struct snd_compr_stream *stream, unsigned long arg)
 		retval = stream->ops->set_params(stream, params);
 		if (retval)
 			goto out;
-		stream->runtime->state = SNDRV_PCM_STATE_SETUP;
+
 		stream->metadata_set = false;
 		stream->next_track = false;
+
+		if (stream->direction == SND_COMPRESS_PLAYBACK)
+			stream->runtime->state = SNDRV_PCM_STATE_SETUP;
+		else
+			stream->runtime->state = SNDRV_PCM_STATE_PREPARED;
 	} else {
 		return -EPERM;
 	}
Signature · 6 lines
1.7.2.5

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#5 lore raw

PATCH 5/7 ASoC: soc-compress: Deduce stream direction

In reply to Richard Fitzgerald

Previously we just hard coded all streams as playback streams, this patch checks the DAI to see if it is a capture or playback stream. It is worth noting that at this time only unidirectional streams are supported.

---
 sound/soc/soc-compress.c |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)
sound/soc/soc-compress.c +8 -1
index da83e56..3853f7e 100644
@@ -384,7 +384,14 @@ int soc_new_compress(struct snd_soc_pcm_runtime *rtd, int num)
 	/* check client and interface hw capabilities */
 	snprintf(new_name, sizeof(new_name), "%s %s-%d",
 			rtd->dai_link->stream_name, codec_dai->name, num);
-	direction = SND_COMPRESS_PLAYBACK;
+
+	if (codec_dai->driver->playback.channels_min)
+		direction = SND_COMPRESS_PLAYBACK;
+	else if (codec_dai->driver->capture.channels_min)
+		direction = SND_COMPRESS_CAPTURE;
+	else
+		return -EINVAL;
+
 	compr = kzalloc(sizeof(*compr), GFP_KERNEL);
 	if (compr == NULL) {
 		snd_printk(KERN_ERR "Cannot allocate compr\n");
Signature · 6 lines
1.7.2.5

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#6 lore raw

PATCH 6/7 ALSA: compress_core: Remove unused hw_pointer

In reply to Richard Fitzgerald

Only tested as far as build.

---
 include/sound/compress_driver.h |    1 -
 sound/core/compress_offload.c   |    2 --
 2 files changed, 0 insertions(+), 3 deletions(-)
include/sound/compress_driver.h +0 -1
index db8273a..2d7de96 100644
@@ -56,7 +56,6 @@ struct snd_compr_runtime {
 	u64 buffer_size;
 	u32 fragment_size;
 	u32 fragments;
-	u64 hw_pointer;
 	u64 app_pointer;
 	u64 total_bytes_available;
 	u64 total_bytes_transferred;
sound/core/compress_offload.c +0 -2
index 52a2765..36d7688 100644
@@ -152,7 +152,6 @@ static int snd_compr_update_tstamp(struct snd_compr_stream *stream,
 	stream->ops->pointer(stream, tstamp);
 	pr_debug("dsp consumed till %d total %d bytes\n",
 		tstamp->byte_offset, tstamp->copied_total);
-	stream->runtime->hw_pointer = tstamp->byte_offset;
 	if (stream->direction == SND_COMPRESS_PLAYBACK)
 		stream->runtime->total_bytes_transferred = tstamp->copied_total;
 	else
@@ -657,7 +656,6 @@ static int snd_compr_stop(struct snd_compr_stream *stream)
 	if (!retval) {
 		stream->runtime->state = SNDRV_PCM_STATE_SETUP;
 		wake_up(&stream->runtime->sleep);
-		stream->runtime->hw_pointer = 0;
 		stream->runtime->app_pointer = 0;
 		stream->runtime->total_bytes_available = 0;
 		stream->runtime->total_bytes_transferred = 0;
Signature · 6 lines
1.7.2.5

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#7 lore raw

PATCH 7/7 ALSA: compress_core: Rework writes to use cumulative values

In reply to Richard Fitzgerald

This patch reworks the writes to use cumulative values thus making the app_pointer unecessary and removing it.

Only tested as far as build.

---
 include/sound/compress_driver.h |    1 -
 sound/core/compress_offload.c   |   18 +++++++++++-------
 2 files changed, 11 insertions(+), 8 deletions(-)
include/sound/compress_driver.h +0 -1
index 2d7de96..9031a26 100644
@@ -56,7 +56,6 @@ struct snd_compr_runtime {
 	u64 buffer_size;
 	u32 fragment_size;
 	u32 fragments;
-	u64 app_pointer;
 	u64 total_bytes_available;
 	u64 total_bytes_transferred;
 	wait_queue_head_t sleep;
sound/core/compress_offload.c +11 -7
index 36d7688..7941ace 100644
@@ -28,11 +28,13 @@
 #include <linux/file.h>
 #include <linux/fs.h>
 #include <linux/list.h>
+#include <linux/math64.h>
 #include <linux/mm.h>
 #include <linux/mutex.h>
 #include <linux/poll.h>
 #include <linux/slab.h>
 #include <linux/sched.h>
+#include <linux/types.h>
 #include <linux/uio.h>
 #include <linux/uaccess.h>
 #include <linux/module.h>
@@ -223,21 +225,24 @@ static int snd_compr_write_data(struct snd_compr_stream *stream,
 	void *dstn;
 	size_t copy;
 	struct snd_compr_runtime *runtime = stream->runtime;
+	/* 64-bit Modulus */
+	u64 app_pointer = div64_u64(runtime->total_bytes_available,
+				    runtime->buffer_size);
+	app_pointer = runtime->total_bytes_available -
+		      (app_pointer * runtime->buffer_size);
 
-	dstn = runtime->buffer + runtime->app_pointer;
+	dstn = runtime->buffer + app_pointer;
 	pr_debug("copying %ld at %lld\n",
-			(unsigned long)count, runtime->app_pointer);
-	if (count < runtime->buffer_size - runtime->app_pointer) {
+			(unsigned long)count, app_pointer);
+	if (count < runtime->buffer_size - app_pointer) {
 		if (copy_from_user(dstn, buf, count))
 			return -EFAULT;
-		runtime->app_pointer += count;
 	} else {
-		copy = runtime->buffer_size - runtime->app_pointer;
+		copy = runtime->buffer_size - app_pointer;
 		if (copy_from_user(dstn, buf, copy))
 			return -EFAULT;
 		if (copy_from_user(runtime->buffer, buf + copy, count - copy))
 			return -EFAULT;
-		runtime->app_pointer = count - copy;
 	}
 	/* if DSP cares, let it know data has been written */
 	if (stream->ops->ack)
@@ -656,7 +661,6 @@ static int snd_compr_stop(struct snd_compr_stream *stream)
 	if (!retval) {
 		stream->runtime->state = SNDRV_PCM_STATE_SETUP;
 		wake_up(&stream->runtime->sleep);
-		stream->runtime->app_pointer = 0;
 		stream->runtime->total_bytes_available = 0;
 		stream->runtime->total_bytes_transferred = 0;
 	}
Signature · 6 lines
1.7.2.5

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#8 lore raw
In reply to Richard Fitzgerald

REMOVE THIS ADDRESS FROM THIS MAILING LIST PLEASE

Quoted 41 lines · Richard Fitzgerald

This chain of patches does some clearing up and add capture support to the compressed API.

Missing is support for capture streams that don't use the copy callback. Also I have included two patches at the end of the chain that remove the buffer pointers and just use the cumulative values however as our setup uses the copy callback these are untested past build so should be taken with caution.

Thanks,
Charles

Thess patches apply to:
git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git
branch "for-next"

based on top of commit
126825e ALSA: snd-usb: add quirks handler for DSD streams

- RichF

Charles Keepax (7):
ALSA: compress_core: Update calc_avail to use cumulative values
ALSA: compress_core: Calculate avail correctly for capture streams
ALSA: compress_core: Deconstify copy callback buffer
ALSA: compress_core: Add support for capture streams
ASoC: soc-compress: Deduce stream direction
ALSA: compress_core: Remove unused hw_pointer
ALSA: compress_core: Rework writes to use cumulative values

 include/sound/compress_driver.h |    4 +-
 sound/core/compress_offload.c   |  117
+++++++++++++++++++++++++-------------
 sound/soc/soc-compress.c        |   11 +++-
 3 files changed, 87 insertions(+), 45 deletions(-)

--
1.7.2.5

#9 lore raw

At Thu, 18 Apr 2013 20:20:43 +1000,

Quoted 2 lines · Leanne Girdwood

REMOVE THIS ADDRESS FROM THIS MAILING LIST PLEASE

Added the right Girdwood to Cc.
Richard, you owe Liam for a wine he needs to bring tonight :)

Takashi

Quoted 44 lines
Quoted 41 lines · Richard Fitzgerald

This chain of patches does some clearing up and add capture support to the compressed API.

Missing is support for capture streams that don't use the copy callback. Also I have included two patches at the end of the chain that remove the buffer pointers and just use the cumulative values however as our setup uses the copy callback these are untested past build so should be taken with caution.

Thanks,
Charles

Thess patches apply to:
git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git
branch "for-next"

based on top of commit
126825e ALSA: snd-usb: add quirks handler for DSD streams

- RichF

Charles Keepax (7):
ALSA: compress_core: Update calc_avail to use cumulative values
ALSA: compress_core: Calculate avail correctly for capture streams
ALSA: compress_core: Deconstify copy callback buffer
ALSA: compress_core: Add support for capture streams
ASoC: soc-compress: Deduce stream direction
ALSA: compress_core: Remove unused hw_pointer
ALSA: compress_core: Rework writes to use cumulative values

 include/sound/compress_driver.h |    4 +-
 sound/core/compress_offload.c   |  117
+++++++++++++++++++++++++-------------
 sound/soc/soc-compress.c        |   11 +++-
 3 files changed, 87 insertions(+), 45 deletions(-)

--
1.7.2.5

#10 lore raw
In reply to Richard Fitzgerald

At Thu, 18 Apr 2013 10:58:41 +0100,

Quoted 14 lines · Richard Fitzgerald

This chain of patches does some clearing up and add capture support to the compressed API.

Missing is support for capture streams that don't use the copy callback. Also I have included two patches at the end of the chain that remove the buffer pointers and just use the cumulative values however as our setup uses the copy callback these are untested past build so should be taken with caution.

Thanks,
Charles

Quoted 6 lines

Thess patches apply to:
git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git
branch "for-next"

based on top of commit
126825e ALSA: snd-usb: add quirks handler for DSD streams

Vinod, Pierre, could you guys check these patches?

If these are OK, I can still merge for 3.10. Although it's already late for 3.10, there are little users of compress stuff, so the impact must be fairly small. If you guys aren't still happy, let's postpone for 3.11.

If we really merge it, don't you mind me merging soc-compress patch in my tree (directly to for-next branch), Mark?

thanks,

Takashi

Quoted 20 lines

- RichF

Charles Keepax (7):
ALSA: compress_core: Update calc_avail to use cumulative values
ALSA: compress_core: Calculate avail correctly for capture streams
ALSA: compress_core: Deconstify copy callback buffer
ALSA: compress_core: Add support for capture streams
ASoC: soc-compress: Deduce stream direction
ALSA: compress_core: Remove unused hw_pointer
ALSA: compress_core: Rework writes to use cumulative values

 include/sound/compress_driver.h |    4 +-
 sound/core/compress_offload.c   |  117 +++++++++++++++++++++++++-------------
 sound/soc/soc-compress.c        |   11 +++-
 3 files changed, 87 insertions(+), 45 deletions(-)
Signature · 1 line
1.7.2.5
#17 lore raw

Re: [PATCH 4/7] ALSA: compress_core: Add support for capture streams

In reply to a message not in the archive
Quoted 20 lines · Charles Keepax
---
 sound/core/compress_offload.c |   43 +++++++++++++++++++++++++++++++++++++++-
 1 files changed, 41 insertions(+), 2 deletions(-)
sound/core/compress_offload.c +5 -1
index 52ca4cc..52a2765 100644
@@ -296,7 +296,41 @@ static ssize_t snd_compr_write(struct file *f, const char __user *buf,
 static ssize_t snd_compr_read(struct file *f, char __user *buf,
 		size_t count, loff_t *offset)
 {
-	return -ENXIO;
+	struct snd_compr_file *data = f->private_data;
+	struct snd_compr_stream *stream;
+	size_t avail;
+	int retval;
+

sound/core/compress_offload.c static ssize_t snd_compr_write(struct file *f, const char __user *buf, · line 313

Quoted 10 lines
sound/core/compress_offload.c +10 -0
+	if (snd_BUG_ON(!data))
+		return -EFAULT;
+
+	stream = &data->stream;
+	mutex_lock(&stream->device->lock);
+
+	/* read is allowed when stream is running */
+	if (stream->runtime->state != SNDRV_PCM_STATE_RUNNING) {
+		retval = -EBADFD;
+		goto out;

minor one but shouldnt we allow read even when we have stoped or paused the stream. Stop is important as you might want to get all buffers after you have stopped.

Quoted 42 lines
sound/core/compress_offload.c +26 -1
+	}
+
+	avail = snd_compr_get_avail(stream);
+	pr_debug("avail returned %ld\n", (unsigned long)avail);
+	/* calculate how much we can read from buffer */
+	if (avail > count)
+		avail = count;
+
+	if (stream->ops->copy) {
+		retval = stream->ops->copy(stream, buf, avail);
+	} else {
+		retval = -ENXIO;
+		goto out;
+	}
+	if (retval > 0)
+		stream->runtime->total_bytes_transferred += retval;
+
+out:
+	mutex_unlock(&stream->device->lock);
+	return retval;
 }
 
 static int snd_compr_mmap(struct file *f, struct vm_area_struct *vma)
@@ -481,9 +515,14 @@ snd_compr_set_params(struct snd_compr_stream *stream, unsigned long arg)
 		retval = stream->ops->set_params(stream, params);
 		if (retval)
 			goto out;
-		stream->runtime->state = SNDRV_PCM_STATE_SETUP;
+
 		stream->metadata_set = false;
 		stream->next_track = false;
+
+		if (stream->direction == SND_COMPRESS_PLAYBACK)
+			stream->runtime->state = SNDRV_PCM_STATE_SETUP;
+		else
+			stream->runtime->state = SNDRV_PCM_STATE_PREPARED;
 	} else {
 		return -EPERM;
 	}
Signature · 1 line
1.7.2.5
#21 lore raw

Re: [PATCH 5/7] ASoC: soc-compress: Deduce stream direction

In reply to a message not in the archive
Quoted 4 lines · Charles Keepax

Previously we just hard coded all streams as playback streams, this patch checks the DAI to see if it is a capture or playback stream. It is worth noting that at this time only unidirectional streams are supported.

well the intention was the compress device would be either playback or record. you should have two devices if you support both.

Quoted 30 lines
---
 sound/soc/soc-compress.c |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)
sound/soc/soc-compress.c +8 -1
index da83e56..3853f7e 100644
@@ -384,7 +384,14 @@ int soc_new_compress(struct snd_soc_pcm_runtime *rtd, int num)
 	/* check client and interface hw capabilities */
 	snprintf(new_name, sizeof(new_name), "%s %s-%d",
 			rtd->dai_link->stream_name, codec_dai->name, num);
-	direction = SND_COMPRESS_PLAYBACK;
+
+	if (codec_dai->driver->playback.channels_min)
+		direction = SND_COMPRESS_PLAYBACK;
+	else if (codec_dai->driver->capture.channels_min)
+		direction = SND_COMPRESS_CAPTURE;
+	else
+		return -EINVAL;
+
 	compr = kzalloc(sizeof(*compr), GFP_KERNEL);
 	if (compr == NULL) {
 		snd_printk(KERN_ERR "Cannot allocate compr\n");
Signature · 1 line
1.7.2.5
Acked-by
#11 lore raw
In reply to Takashi Iwai
Quoted 23 lines · Takashi Iwai

At Thu, 18 Apr 2013 10:58:41 +0100,

Quoted 14 lines · Richard Fitzgerald

This chain of patches does some clearing up and add capture support to the compressed API.

Missing is support for capture streams that don't use the copy callback. Also I have included two patches at the end of the chain that remove the buffer pointers and just use the cumulative values however as our setup uses the copy callback these are untested past build so should be taken with caution.

Thanks,
Charles

Quoted 6 lines

Thess patches apply to:
git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git
branch "for-next"

based on top of commit
126825e ALSA: snd-usb: add quirks handler for DSD streams

Quoted 6 lines

Vinod, Pierre, could you guys check these patches?

If these are OK, I can still merge for 3.10. Although it's already late for 3.10, there are little users of compress stuff, so the impact must be fairly small. If you guys aren't still happy, let's postpone for 3.11.

Sorry for delay and thrash on patches, my sound clone was bad...

I think I am okay with series approach. We need small fix for state check for read but that need not block this series

Quoted 4 lines

If we really merge it, don't you mind me merging soc-compress patch in my tree (directly to for-next branch), Mark?

i think that should be sensible :)

--
~Vinod

Quoted 26 lines

thanks,

Takashi

Quoted 20 lines

- RichF

Charles Keepax (7):
ALSA: compress_core: Update calc_avail to use cumulative values
ALSA: compress_core: Calculate avail correctly for capture streams
ALSA: compress_core: Deconstify copy callback buffer
ALSA: compress_core: Add support for capture streams
ASoC: soc-compress: Deduce stream direction
ALSA: compress_core: Remove unused hw_pointer
ALSA: compress_core: Rework writes to use cumulative values

 include/sound/compress_driver.h |    4 +-
 sound/core/compress_offload.c   |  117 +++++++++++++++++++++++++-------------
 sound/soc/soc-compress.c        |   11 +++-
 3 files changed, 87 insertions(+), 45 deletions(-)
Signature · 1 line
1.7.2.5
#12 lore raw
Quoted 6 lines · Vinod Koul
Quoted 5 lines · Takashi Iwai

At Thu, 18 Apr 2013 10:58:41 +0100,

Quoted 3 lines · Richard Fitzgerald

This chain of patches does some clearing up and add capture support to the compressed API.

Charles,

I am expecting capture support patches for tinycompress :)

Quoted 64 lines
Quoted 18 lines
Quoted 11 lines

Missing is support for capture streams that don't use the copy callback. Also I have included two patches at the end of the chain that remove the buffer pointers and just use the cumulative values however as our setup uses the copy callback these are untested past build so should be taken with caution.

Thanks,
Charles

Quoted 6 lines

Thess patches apply to:
git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git
branch "for-next"

based on top of commit
126825e ALSA: snd-usb: add quirks handler for DSD streams

Quoted 6 lines

Vinod, Pierre, could you guys check these patches?

If these are OK, I can still merge for 3.10. Although it's already late for 3.10, there are little users of compress stuff, so the impact must be fairly small. If you guys aren't still happy, let's postpone for 3.11.

Sorry for delay and thrash on patches, my sound clone was bad...

I think I am okay with series approach. We need small fix for state check for read but that need not block this series

Quoted 4 lines

If we really merge it, don't you mind me merging soc-compress patch in my tree (directly to for-next branch), Mark?

i think that should be sensible :)

--
~Vinod

Quoted 26 lines

thanks,

Takashi

Quoted 20 lines

- RichF

Charles Keepax (7):
ALSA: compress_core: Update calc_avail to use cumulative values
ALSA: compress_core: Calculate avail correctly for capture streams
ALSA: compress_core: Deconstify copy callback buffer
ALSA: compress_core: Add support for capture streams
ASoC: soc-compress: Deduce stream direction
ALSA: compress_core: Remove unused hw_pointer
ALSA: compress_core: Rework writes to use cumulative values

 include/sound/compress_driver.h |    4 +-
 sound/core/compress_offload.c   |  117 +++++++++++++++++++++++++-------------
 sound/soc/soc-compress.c        |   11 +++-
 3 files changed, 87 insertions(+), 45 deletions(-)
Signature · 1 line
1.7.2.5
#13 lore raw
Quoted 1 line

I am expecting capture support patches for tinycompress :)

Your expectations are correct. Charles has done them and I didn't notice that they hadn't been posted.
I think I need to rebase them first. Will do them soon as I get the time.

#18 lore raw

Re: [PATCH 4/7] ALSA: compress_core: Add support for capture streams

In reply to Vinod Koul
Quoted 14 lines
Quoted 14 lines

static ssize_t snd_compr_read(struct file *f, char __user *buf,

(quoted diff) +10 -1
 		size_t count, loff_t *offset)
 {
-	return -ENXIO;
+	struct snd_compr_file *data = f->private_data;
+	struct snd_compr_stream *stream;
+	size_t avail;
+	int retval;
+
+	if (snd_BUG_ON(!data))
+		return -EFAULT;
+
+	stream = &data->stream;
+	mutex_lock(&stream->device->lock);
Quoted 8 lines

Review

Quoted 5 lines
(quoted diff) +5 -0
+
+	/* read is allowed when stream is running */
+	if (stream->runtime->state != SNDRV_PCM_STATE_RUNNING) {
+		retval = -EBADFD;
+		goto out;

minor one but shouldnt we allow read even when we have stoped or paused the stream. Stop is important as you might want to get all buffers after you have stopped.

Yes, looks like a bug to me. Not sure that it's a blocker - new feature so nobody is going to be using this immediately.
As you can't capture sounds from the future(!) the DSP could only hold data up to the point you stopped/paused. But the DSP could have some audio packets queued that it hasn't sent to host yet and you still need to read them. You'd probably not have much queued anyway as it would be disturbing for user if they monitor recording through their device headphones and it has significant latency from the real world sounds?

#19 lore raw

Re: [PATCH 4/7] ALSA: compress_core: Add support for capture streams

In reply to Vinod Koul

KINDLY REMOVE THIS ADDRESS FROM YOUR MAILING LIST

Quoted 84 lines · Vinod Koul
Quoted 6 lines · Charles Keepax
---
 sound/core/compress_offload.c |   43

+++++++++++++++++++++++++++++++++++++++-

Review

Quoted 3 lines
 1 files changed, 41 insertions(+), 2 deletions(-)
(quoted diff) +0 -0

b/sound/core/compress_offload.c

sound/core/compress_offload.c static ssize_t snd_compr_write(struct file *f,

Quoted 4 lines

index 52ca4cc..52a2765 100644

sound/core/compress_offload.c +0 -0
@@ -296,7 +296,41 @@ static ssize_t snd_compr_write(struct file *f,

const char __user *buf,

Quoted 9 lines
sound/core/compress_offload.c +5 -1
 static ssize_t snd_compr_read(struct file *f, char __user *buf,
              size_t count, loff_t *offset)
 {
-     return -ENXIO;
+     struct snd_compr_file *data = f->private_data;
+     struct snd_compr_stream *stream;
+     size_t avail;
+     int retval;
+

sound/core/compress_offload.c static ssize_t snd_compr_write(struct file *f, · line 313

Quoted 10 lines
sound/core/compress_offload.c +10 -0
+     if (snd_BUG_ON(!data))
+             return -EFAULT;
+
+     stream = &data->stream;
+     mutex_lock(&stream->device->lock);
+
+     /* read is allowed when stream is running */
+     if (stream->runtime->state != SNDRV_PCM_STATE_RUNNING) {
+             retval = -EBADFD;
+             goto out;

minor one but shouldnt we allow read even when we have stoped or paused the stream. Stop is important as you might want to get all buffers after you have
stopped.

sound/core/compress_offload.c snd_compr_set_params(struct snd_compr_stream

Quoted 24 lines
sound/core/compress_offload.c +20 -0
+     }
+
+     avail = snd_compr_get_avail(stream);
+     pr_debug("avail returned %ld\n", (unsigned long)avail);
+     /* calculate how much we can read from buffer */
+     if (avail > count)
+             avail = count;
+
+     if (stream->ops->copy) {
+             retval = stream->ops->copy(stream, buf, avail);
+     } else {
+             retval = -ENXIO;
+             goto out;
+     }
+     if (retval > 0)
+             stream->runtime->total_bytes_transferred += retval;
+
+out:
+     mutex_unlock(&stream->device->lock);
+     return retval;
 }
 
 static int snd_compr_mmap(struct file *f, struct vm_area_struct *vma)
@@ -481,9 +515,14 @@ snd_compr_set_params(struct snd_compr_stream

*stream, unsigned long arg)

Quoted 18 lines
sound/core/compress_offload.c +6 -2
              retval = stream->ops->set_params(stream, params);
              if (retval)
                      goto out;
-             stream->runtime->state = SNDRV_PCM_STATE_SETUP;
+
              stream->metadata_set = false;
              stream->next_track = false;
+
+             if (stream->direction == SND_COMPRESS_PLAYBACK)
+                     stream->runtime->state = SNDRV_PCM_STATE_SETUP;
+             else
+                     stream->runtime->state = SNDRV_PCM_STATE_PREPARED;
      } else {
              return -EPERM;
      }
--

1.7.2.5

#14 lore raw
In reply to Vinod Koul

At Fri, 19 Apr 2013 19:00:58 +0530,

Quoted 36 lines · Vinod Koul
Quoted 23 lines · Takashi Iwai

At Thu, 18 Apr 2013 10:58:41 +0100,

Quoted 14 lines · Richard Fitzgerald

This chain of patches does some clearing up and add capture support to the compressed API.

Missing is support for capture streams that don't use the copy callback. Also I have included two patches at the end of the chain that remove the buffer pointers and just use the cumulative values however as our setup uses the copy callback these are untested past build so should be taken with caution.

Thanks,
Charles

Quoted 6 lines

Thess patches apply to:
git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git
branch "for-next"

based on top of commit
126825e ALSA: snd-usb: add quirks handler for DSD streams

Quoted 6 lines

Vinod, Pierre, could you guys check these patches?

If these are OK, I can still merge for 3.10. Although it's already late for 3.10, there are little users of compress stuff, so the impact must be fairly small. If you guys aren't still happy, let's postpone for 3.11.

Sorry for delay and thrash on patches, my sound clone was bad...

I think I am okay with series approach. We need small fix for state check for read but that need not block this series

Quoted 6 lines
Quoted 4 lines

If we really merge it, don't you mind me merging soc-compress patch in my tree (directly to for-next branch), Mark?

i think that should be sensible :)

OK, I merged the original patches now to for-next branch.

thanks,

Takashi

maintainer
#15 lore raw
In reply to Takashi Iwai
Quoted 2 lines · Takashi Iwai

If we really merge it, don't you mind me merging soc-compress patch in my tree (directly to for-next branch), Mark?

Yes, that's fine for me - for v3.11 I suspect it'd make sense to have it in my tree though as there'll probably be other stuff that depends on it.