ALSA: usb-audio: Improved lowlatency playback support

This is another attempt to improve further the handling of playback
stream in the low latency mode.  The latest workaround in commit
4267c5a ("ALSA: usb-audio: Work around for XRUN with low latency
playback") revealed that submitting URBs forcibly in advance may
trigger XRUN easily.  In the classical mode, this problem was avoided
by practically delaying the submission of the actual data with the
pre-submissions of silent data before triggering the stream start.
But that is exactly what we want to avoid.

Now, in this patch, instead of the previous workaround, we take a
similar approach as used in the implicit feedback mode.  The URBs are
queued at the PCM trigger start like before, but we check whether the
buffer has been already filled enough before each submission, and
stop queuing if the data overcomes the threshold.  The remaining URBs
are kept in the ready list, and they will be retrieved in the URB
complete callback of other (already queued) URBs.  In the complete
callback, we try to fill the data and submit as much as possible
again.  When there is no more available in-flight URBs that may handle
the pending data, we'll check in PCM ack callback and submit and
process URBs there in addition.  In this way, the amount of in-flight
URBs may vary dynamically and flexibly depending on the available data
without hitting XRUN.

The following things are changed to achieve the behavior above:

* The endpoint prepare callback is changed to return an error code;
  when there is no enough data available, it may return -EAGAIN.
  Currently only prepare_playback_urb() returns the error.

  The evaluation of the available data is a bit messy here; we can't
  check with snd_pcm_avail() at the point of prepare callback (as
  runtime->status->hwptr hasn't been updated yet), hence we manually
  estimate the appl_ptr and compare with the internal hwptr_done to
  calculate the available frames.

* snd_usb_endpoint_start() doesn't submit full URBs if the prepare
  callback returns -EAGAIN, and puts the remaining URBs to the ready
  list for the later submission.

* snd_complete_urb() treats the URBs in the low-latency mode similarly
  like the implicit feedback mode, and submissions are done in
  (now exported) snd_usb_queue_pending_output_urbs().

* snd_usb_queue_pending_output_urbs() again checks the error value
  from the prepare callback.  If it's -EAGAIN for the normal stream
  (i.e. not implicit feedback mode), we push it back to the ready list
  again.

* PCM ack callback is introduced for the playback stream, and it calls
  snd_usb_queue_pending_output_urbs() if there is no in-flight URB
  while the stream is running.  This corresponds to the case where the
  system needs the appl_ptr update for re-submitting a new URB.

* snd_usb_queue_pending_output_urbs() and the prepare EP callback
  receive in_stream_lock argument, which is a bool flag indicating the
  call path from PCM ack.  It's needed for avoiding the deadlock of
  snd_pcm_period_elapsed() calls.

* Set the new SNDRV_PCM_INFO_EXPLICIT_SYNC flag when the new
  low-latency mode is deployed.  This assures catching each applptr
  update even in the mmap mode.

Fixes: 4267c5a ("ALSA: usb-audio: Work around for XRUN with low latency playback")
Link: https://lore.kernel.org/r/20210929080844.11583-9-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>

Author: tiwai

sound/usb/card.h

@@ -74,8 +74,9 @@ struct snd_usb_endpoint {
 
 	atomic_t state;		/* running state */
 
-	void (*prepare_data_urb) (struct snd_usb_substream *subs,
-				  struct urb *urb);
+	int (*prepare_data_urb) (struct snd_usb_substream *subs,
+				 struct urb *urb,
+				 bool in_stream_lock);
 	void (*retire_data_urb) (struct snd_usb_substream *subs,
 				 struct urb *urb);
 
@@ -94,7 +95,6 @@ struct snd_usb_endpoint {
 	struct list_head ready_playback_urbs; /* playback URB FIFO for implicit fb */
 
 	unsigned int nurbs;		/* # urbs */
-	unsigned int nominal_queue_size; /* total buffer sizes in URBs */
 	unsigned long active_mask;	/* bitmask of active urbs */
 	unsigned long unlink_mask;	/* bitmask of unlinked urbs */
 	atomic_t submitted_urbs;	/* currently submitted urbs */

sound/usb/endpoint.c

@@ -307,8 +307,9 @@ static void prepare_silent_urb(struct snd_usb_endpoint *ep,
 /*
  * Prepare a PLAYBACK urb for submission to the bus.
  */
-static void prepare_outbound_urb(struct snd_usb_endpoint *ep,
-				 struct snd_urb_ctx *ctx)
+static int prepare_outbound_urb(struct snd_usb_endpoint *ep,
+				struct snd_urb_ctx *ctx,
+				bool in_stream_lock)
 {
 	struct urb *urb = ctx->urb;
 	unsigned char *cp = urb->transfer_buffer;
@@ -320,9 +321,9 @@ static void prepare_outbound_urb(struct snd_usb_endpoint *ep,
 	case SND_USB_ENDPOINT_TYPE_DATA:
 		data_subs = READ_ONCE(ep->data_subs);
 		if (data_subs && ep->prepare_data_urb)
-			ep->prepare_data_urb(data_subs, urb);
-		else /* no data provider, so send silence */
-			prepare_silent_urb(ep, ctx);
+			return ep->prepare_data_urb(data_subs, urb, in_stream_lock);
+		/* no data provider, so send silence */
+		prepare_silent_urb(ep, ctx);
 		break;
 
 	case SND_USB_ENDPOINT_TYPE_SYNC:
@@ -351,13 +352,14 @@ static void prepare_outbound_urb(struct snd_usb_endpoint *ep,
 
 		break;
 	}
+	return 0;
 }
 
 /*
  * Prepare a CAPTURE or SYNC urb for submission to the bus.
  */
-static inline void prepare_inbound_urb(struct snd_usb_endpoint *ep,
-				       struct snd_urb_ctx *urb_ctx)
+static int prepare_inbound_urb(struct snd_usb_endpoint *ep,
+			       struct snd_urb_ctx *urb_ctx)
 {
 	int i, offs;
 	struct urb *urb = urb_ctx->urb;
@@ -382,6 +384,7 @@ static inline void prepare_inbound_urb(struct snd_usb_endpoint *ep,
 		urb->iso_frame_desc[0].offset = 0;
 		break;
 	}
+	return 0;
 }
 
 /* notify an error as XRUN to the assigned PCM data substream */
@@ -417,6 +420,16 @@ next_packet_fifo_dequeue(struct snd_usb_endpoint *ep)
 	return p;
 }
 
+static void push_back_to_ready_list(struct snd_usb_endpoint *ep,
+				    struct snd_urb_ctx *ctx)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&ep->lock, flags);
+	list_add_tail(&ctx->ready_list, &ep->ready_playback_urbs);
+	spin_unlock_irqrestore(&ep->lock, flags);
+}
+
 /*
  * Send output urbs that have been prepared previously. URBs are dequeued
  * from ep->ready_playback_urbs and in case there aren't any available
@@ -427,12 +440,14 @@ next_packet_fifo_dequeue(struct snd_usb_endpoint *ep)
  * is that host controllers don't guarantee the order in which they return
  * inbound and outbound packets to their submitters.
  *
- * This function is only used for implicit feedback endpoints. For endpoints
- * driven by dedicated sync endpoints, URBs are immediately re-submitted
- * from their completion handler.
+ * This function is used both for implicit feedback endpoints and in low-
+ * latency playback mode.
  */
-static void queue_pending_output_urbs(struct snd_usb_endpoint *ep)
+void snd_usb_queue_pending_output_urbs(struct snd_usb_endpoint *ep,
+				       bool in_stream_lock)
 {
+	bool implicit_fb = snd_usb_endpoint_implicit_feedback_sink(ep);
+
 	while (ep_state_running(ep)) {
 
 		unsigned long flags;
@@ -441,26 +456,39 @@ static void queue_pending_output_urbs(struct snd_usb_endpoint *ep)
 		int err, i;
 
 		spin_lock_irqsave(&ep->lock, flags);
-		if (ep->next_packet_queued > 0 &&
+		if ((!implicit_fb || ep->next_packet_queued > 0) &&
 		    !list_empty(&ep->ready_playback_urbs)) {
 			/* take URB out of FIFO */
 			ctx = list_first_entry(&ep->ready_playback_urbs,
 					       struct snd_urb_ctx, ready_list);
 			list_del_init(&ctx->ready_list);
-
-			packet = next_packet_fifo_dequeue(ep);
+			if (implicit_fb)
+				packet = next_packet_fifo_dequeue(ep);
 		}
 		spin_unlock_irqrestore(&ep->lock, flags);
 
 		if (ctx == NULL)
 			return;
 
 		/* copy over the length information */
-		for (i = 0; i < packet->packets; i++)
-			ctx->packet_size[i] = packet->packet_size[i];
+		if (implicit_fb) {
+			for (i = 0; i < packet->packets; i++)
+				ctx->packet_size[i] = packet->packet_size[i];
+		}
 
 		/* call the data handler to fill in playback data */
-		prepare_outbound_urb(ep, ctx);
+		err = prepare_outbound_urb(ep, ctx, in_stream_lock);
+		/* can be stopped during prepare callback */
+		if (unlikely(!ep_state_running(ep)))
+			break;
+		if (err < 0) {
+			/* push back to ready list again for -EAGAIN */
+			if (err == -EAGAIN)
+				push_back_to_ready_list(ep, ctx);
+			else
+				notify_xrun(ep);
+			return;
+		}
 
 		err = usb_submit_urb(ctx->urb, GFP_ATOMIC);
 		if (err < 0) {
@@ -483,7 +511,6 @@ static void snd_complete_urb(struct urb *urb)
 {
 	struct snd_urb_ctx *ctx = urb->context;
 	struct snd_usb_endpoint *ep = ctx->ep;
-	unsigned long flags;
 	int err;
 
 	if (unlikely(urb->status == -ENOENT ||		/* unlinked */
@@ -504,17 +531,20 @@ static void snd_complete_urb(struct urb *urb)
 		if (unlikely(!ep_state_running(ep)))
 			goto exit_clear;
 
-		if (snd_usb_endpoint_implicit_feedback_sink(ep)) {
-			spin_lock_irqsave(&ep->lock, flags);
-			list_add_tail(&ctx->ready_list, &ep->ready_playback_urbs);
+		/* in low-latency and implicit-feedback modes, push back the
+		 * URB to ready list at first, then process as much as possible
+		 */
+		if (ep->lowlatency_playback ||
+		     snd_usb_endpoint_implicit_feedback_sink(ep)) {
+			push_back_to_ready_list(ep, ctx);
 			clear_bit(ctx->index, &ep->active_mask);
-			spin_unlock_irqrestore(&ep->lock, flags);
-			queue_pending_output_urbs(ep);
+			snd_usb_queue_pending_output_urbs(ep, false);
 			atomic_dec(&ep->submitted_urbs); /* decrement at last */
 			return;
 		}
 
-		prepare_outbound_urb(ep, ctx);
+		/* in non-lowlatency mode, no error handling for prepare */
+		prepare_outbound_urb(ep, ctx, false);
 		/* can be stopped during prepare callback */
 		if (unlikely(!ep_state_running(ep)))
 			goto exit_clear;
@@ -807,8 +837,9 @@ void snd_usb_endpoint_set_sync(struct snd_usb_audio *chip,
  * Pass NULL to deactivate each callback.
  */
 void snd_usb_endpoint_set_callback(struct snd_usb_endpoint *ep,
-				   void (*prepare)(struct snd_usb_substream *subs,
-						   struct urb *urb),
+				   int (*prepare)(struct snd_usb_substream *subs,
+						  struct urb *urb,
+						  bool in_stream_lock),
 				   void (*retire)(struct snd_usb_substream *subs,
 						  struct urb *urb),
 				   struct snd_usb_substream *data_subs)
@@ -1166,10 +1197,6 @@ static int data_ep_set_params(struct snd_usb_endpoint *ep)
 		INIT_LIST_HEAD(&u->ready_list);
 	}
 
-	/* total buffer bytes of all URBs plus the next queue;
-	 * referred in pcm.c
-	 */
-	ep->nominal_queue_size = maxsize * urb_packs * (ep->nurbs + 1);
 	return 0;
 
 out_of_memory:
@@ -1408,6 +1435,7 @@ int snd_usb_endpoint_get_clock_rate(struct snd_usb_audio *chip, int clock)
  */
 int snd_usb_endpoint_start(struct snd_usb_endpoint *ep)
 {
+	bool is_playback = usb_pipeout(ep->pipe);
 	int err;
 	unsigned int i;
 
@@ -1444,13 +1472,9 @@ int snd_usb_endpoint_start(struct snd_usb_endpoint *ep)
 
 	if (snd_usb_endpoint_implicit_feedback_sink(ep) &&
 	    !(ep->chip->quirk_flags & QUIRK_FLAG_PLAYBACK_FIRST)) {
-		for (i = 0; i < ep->nurbs; i++) {
-			struct snd_urb_ctx *ctx = ep->urb + i;
-			list_add_tail(&ctx->ready_list, &ep->ready_playback_urbs);
-		}
-
 		usb_audio_dbg(ep->chip, "No URB submission due to implicit fb sync\n");
-		return 0;
+		i = 0;
+		goto fill_rest;
 	}
 
 	for (i = 0; i < ep->nurbs; i++) {
@@ -1459,10 +1483,18 @@ int snd_usb_endpoint_start(struct snd_usb_endpoint *ep)
 		if (snd_BUG_ON(!urb))
 			goto __error;
 
-		if (usb_pipeout(ep->pipe)) {
-			prepare_outbound_urb(ep, urb->context);
-		} else {
-			prepare_inbound_urb(ep, urb->context);
+		if (is_playback)
+			err = prepare_outbound_urb(ep, urb->context, true);
+		else
+			err = prepare_inbound_urb(ep, urb->context);
+		if (err < 0) {
+			/* stop filling at applptr */
+			if (err == -EAGAIN)
+				break;
+			usb_audio_dbg(ep->chip,
+				      "EP 0x%x: failed to prepare urb: %d\n",
+				      ep->ep_num, err);
+			goto __error;
 		}
 
 		err = usb_submit_urb(urb, GFP_ATOMIC);
@@ -1476,8 +1508,22 @@ int snd_usb_endpoint_start(struct snd_usb_endpoint *ep)
 		atomic_inc(&ep->submitted_urbs);
 	}
 
+	if (!i) {
+		usb_audio_dbg(ep->chip, "XRUN at starting EP 0x%x\n",
+			      ep->ep_num);
+		goto __error;
+	}
+
 	usb_audio_dbg(ep->chip, "%d URBs submitted for EP 0x%x\n",
-		      ep->nurbs, ep->ep_num);
+		      i, ep->ep_num);
+
+ fill_rest:
+	/* put the remaining URBs to ready list */
+	if (is_playback) {
+		for (; i < ep->nurbs; i++)
+			push_back_to_ready_list(ep, ep->urb + i);
+	}
+
 	return 0;
 
 __error:
@@ -1629,7 +1675,7 @@ static void snd_usb_handle_sync_urb(struct snd_usb_endpoint *ep,
 		}
 
 		spin_unlock_irqrestore(&ep->lock, flags);
-		queue_pending_output_urbs(ep);
+		snd_usb_queue_pending_output_urbs(ep, false);
 
 		return;
 	}

sound/usb/endpoint.h

@@ -30,8 +30,9 @@ void snd_usb_endpoint_set_sync(struct snd_usb_audio *chip,
 			       struct snd_usb_endpoint *data_ep,
 			       struct snd_usb_endpoint *sync_ep);
 void snd_usb_endpoint_set_callback(struct snd_usb_endpoint *ep,
-				   void (*prepare)(struct snd_usb_substream *subs,
-						   struct urb *urb),
+				   int (*prepare)(struct snd_usb_substream *subs,
+						  struct urb *urb,
+						  bool in_stream_lock),
 				   void (*retire)(struct snd_usb_substream *subs,
 						  struct urb *urb),
 				   struct snd_usb_substream *data_subs);
@@ -48,5 +49,7 @@ int snd_usb_endpoint_implicit_feedback_sink(struct snd_usb_endpoint *ep);
 int snd_usb_endpoint_next_packet_size(struct snd_usb_endpoint *ep,
 				      struct snd_urb_ctx *ctx, int idx,
 				      unsigned int avail);
+void snd_usb_queue_pending_output_urbs(struct snd_usb_endpoint *ep,
+				       bool in_stream_lock);
 
 #endif /* __USBAUDIO_ENDPOINT_H */