linux-media.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Philipp Zabel <p.zabel@pengutronix.de>
To: linux-media@vger.kernel.org
Cc: Mauro Carvalho Chehab <mchehab@kernel.org>,
	Hans Verkuil <hverkuil@xs4all.nl>,
	kernel@pengutronix.de
Subject: [PATCH 23/28] media: coda: lock capture queue wakeup against encoder stop command
Date: Tue, 18 Jun 2019 18:45:30 +0200	[thread overview]
Message-ID: <20190618164535.20162-24-p.zabel@pengutronix.de> (raw)
In-Reply-To: <20190618164535.20162-1-p.zabel@pengutronix.de>

Make sure that an encoder stop command running concurrently with an
encoder finish_run always either flags the last returned buffer or wakes
up the capture queue to signal the end of stream condition afterwards.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
---
 drivers/media/platform/coda/coda-bit.c    |  7 +++++++
 drivers/media/platform/coda/coda-common.c | 19 ++++++++++++++-----
 drivers/media/platform/coda/coda.h        |  6 ++++++
 3 files changed, 27 insertions(+), 5 deletions(-)

diff --git a/drivers/media/platform/coda/coda-bit.c b/drivers/media/platform/coda/coda-bit.c
index 7bcdfe8dcf3d..44085e3d43d5 100644
--- a/drivers/media/platform/coda/coda-bit.c
+++ b/drivers/media/platform/coda/coda-bit.c
@@ -1540,6 +1540,12 @@ static void coda_finish_encode(struct coda_ctx *ctx)
 	struct coda_dev *dev = ctx->dev;
 	u32 wr_ptr, start_ptr;
 
+	/*
+	 * Lock to make sure that an encoder stop command running in parallel
+	 * will either already have marked src_buf as last, or it will wake up
+	 * the capture queue after the buffers are returned.
+	 */
+	mutex_lock(&ctx->wakeup_mutex);
 	src_buf = v4l2_m2m_src_buf_remove(ctx->fh.m2m_ctx);
 	dst_buf = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx);
 
@@ -1580,6 +1586,7 @@ static void coda_finish_encode(struct coda_ctx *ctx)
 
 	dst_buf = v4l2_m2m_dst_buf_remove(ctx->fh.m2m_ctx);
 	coda_m2m_buf_done(ctx, dst_buf, VB2_BUF_STATE_DONE);
+	mutex_unlock(&ctx->wakeup_mutex);
 
 	ctx->gopcounter--;
 	if (ctx->gopcounter < 0)
diff --git a/drivers/media/platform/coda/coda-common.c b/drivers/media/platform/coda/coda-common.c
index 0d64ddc49494..829d1e565911 100644
--- a/drivers/media/platform/coda/coda-common.c
+++ b/drivers/media/platform/coda/coda-common.c
@@ -1034,19 +1034,27 @@ static int coda_encoder_cmd(struct file *file, void *fh,
 	if (ret < 0)
 		return ret;
 
+	mutex_lock(&ctx->wakeup_mutex);
 	buf = v4l2_m2m_last_src_buf(ctx->fh.m2m_ctx);
 	if (buf) {
+		/*
+		 * If the last output buffer is still on the queue, make sure
+		 * that decoder finish_run will see the last flag and report it
+		 * to userspace.
+		 */
 		buf->flags |= V4L2_BUF_FLAG_LAST;
 	} else {
 		/* Set the stream-end flag on this context */
 		ctx->bit_stream_param |= CODA_BIT_STREAM_END_FLAG;
 
-		flush_work(&ctx->pic_run_work);
-
-		/* If there is no buffer in flight, wake up */
-		if (!ctx->streamon_out || ctx->qsequence == ctx->osequence)
-			coda_wake_up_capture_queue(ctx);
+		/*
+		 * If the last output buffer has already been taken from the
+		 * queue, wake up the capture queue and signal end of stream
+		 * via the -EPIPE mechanism.
+		 */
+		coda_wake_up_capture_queue(ctx);
 	}
+	mutex_unlock(&ctx->wakeup_mutex);
 
 	return 0;
 }
@@ -2466,6 +2474,7 @@ static int coda_open(struct file *file)
 
 	mutex_init(&ctx->bitstream_mutex);
 	mutex_init(&ctx->buffer_mutex);
+	mutex_init(&ctx->wakeup_mutex);
 	INIT_LIST_HEAD(&ctx->buffer_meta_list);
 	spin_lock_init(&ctx->buffer_meta_lock);
 
diff --git a/drivers/media/platform/coda/coda.h b/drivers/media/platform/coda/coda.h
index 6911c1c811ce..97845e58fb8b 100644
--- a/drivers/media/platform/coda/coda.h
+++ b/drivers/media/platform/coda/coda.h
@@ -258,6 +258,12 @@ struct coda_ctx {
 	bool				use_bit;
 	bool				use_vdoa;
 	struct vdoa_ctx			*vdoa;
+	/*
+	 * wakeup mutex used to serialize encoder stop command and finish_run,
+	 * ensures that finish_run always either flags the last returned buffer
+	 * or wakes up the capture queue to signal EOS afterwards.
+	 */
+	struct mutex			wakeup_mutex;
 };
 
 extern int coda_debug;
-- 
2.20.1


  parent reply	other threads:[~2019-06-18 16:45 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-06-18 16:45 [PATCH 00/28] media: coda: fixes and improvements Philipp Zabel
2019-06-18 16:45 ` [PATCH 01/28] media: coda: implement CMD_START to restart decoding Philipp Zabel
2019-06-18 16:45 ` [PATCH 02/28] media: coda: use mem2mem try_en/decoder_cmd helpers Philipp Zabel
2019-06-18 16:45 ` [PATCH 03/28] media: coda: fix mpeg2 sequence number handling Philipp Zabel
2019-06-18 16:45 ` [PATCH 04/28] media: coda: fix last buffer handling in V4L2_ENC_CMD_STOP Philipp Zabel
2019-06-18 16:45 ` [PATCH 05/28] media: coda: add coda_wake_up_capture_queue Philipp Zabel
2019-06-18 16:45 ` [PATCH 06/28] media: coda: fix V4L2_DEC_CMD_STOP when all buffers are already consumed Philipp Zabel
2019-06-18 16:45 ` [PATCH 07/28] media: coda: split decoder sequence initialization out of start decoding Philipp Zabel
2019-06-18 16:45 ` [PATCH 08/28] media: coda: add sequence initialization work Philipp Zabel
2019-06-18 16:45 ` [PATCH 09/28] media: coda: implement decoder source change event Philipp Zabel
2019-06-18 16:45 ` [PATCH 10/28] media: coda: integrate internal frame metadata into a structure Philipp Zabel
2019-06-18 16:45 ` [PATCH 11/28] media: coda: make coda_bitstream_queue more versatile Philipp Zabel
2019-06-18 16:45 ` [PATCH 12/28] media: coda: pad first buffer with repeated MPEG headers to fix sequence init Philipp Zabel
2019-06-18 16:45 ` [PATCH 13/28] media: coda: do not enforce 512-byte initial bitstream payload on CODA960 Philipp Zabel
2019-06-18 16:45 ` [PATCH 14/28] media: coda: flush bitstream ring buffer on decoder restart Philipp Zabel
2019-06-18 16:45 ` [PATCH 15/28] media: coda: increment sequence offset for the last returned frame Philipp Zabel
2019-06-18 16:45 ` [PATCH 16/28] media: coda: allow flagging last output buffer internally Philipp Zabel
2019-06-18 16:45 ` [PATCH 17/28] media: coda: mark the last output buffer on decoder stop command Philipp Zabel
2019-06-18 16:45 ` [PATCH 18/28] media: coda: only set the stream end flags if there are no more pending output buffers Philipp Zabel
2019-06-18 16:45 ` [PATCH 19/28] media: coda: mark the last output buffer on encoder stop command Philipp Zabel
2019-06-18 16:45 ` [PATCH 20/28] media: coda: retire coda_buf_is_end_of_stream Philipp Zabel
2019-06-18 16:45 ` [PATCH 21/28] media: coda: only wake up capture queue if no pending buffers to encode Philipp Zabel
2019-06-18 16:45 ` [PATCH 22/28] media: coda: flag the last encoded buffer Philipp Zabel
2019-06-18 16:45 ` Philipp Zabel [this message]
2019-06-18 16:45 ` [PATCH 24/28] media: coda: mark last pending buffer or last meta on decoder stop command Philipp Zabel
2019-06-18 16:45 ` [PATCH 25/28] media: coda: mark last returned frame Philipp Zabel
2019-06-18 16:45 ` [PATCH 26/28] media: coda: store device pointer in driver structure instead of pdev Philipp Zabel
2019-06-18 16:45 ` [PATCH 27/28] media: coda: add coda_slice_mode() function Philipp Zabel
2019-06-18 16:45 ` [PATCH 28/28] media: coda: encoder parameter change support Philipp Zabel

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190618164535.20162-24-p.zabel@pengutronix.de \
    --to=p.zabel@pengutronix.de \
    --cc=hverkuil@xs4all.nl \
    --cc=kernel@pengutronix.de \
    --cc=linux-media@vger.kernel.org \
    --cc=mchehab@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).