public inbox for linux-media@vger.kernel.org
 help / color / mirror / Atom feed
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: Hans Verkuil <hverkuil@xs4all.nl>
Cc: linux-media@vger.kernel.org, m.szyprowski@samsung.com,
	pawel@osciak.com, awalls@md.metrocast.net,
	Hans Verkuil <hans.verkuil@cisco.com>
Subject: Re: [RFC PATCH 1/8] vb2: push the mmap semaphore down to __buf_prepare()
Date: Thu, 21 Nov 2013 20:04:45 +0100	[thread overview]
Message-ID: <6539252.6X3kkSkupS@avalon> (raw)
In-Reply-To: <1385047326-23099-2-git-send-email-hverkuil@xs4all.nl>

Hi Hans,

Thank you for the patch.

On Thursday 21 November 2013 16:21:59 Hans Verkuil wrote:
> From: Hans Verkuil <hans.verkuil@cisco.com>
> 
> Rather than taking the mmap semaphore at a relatively high-level function,
> push it down to the place where it is really needed.
> 
> It was placed in vb2_queue_or_prepare_buf() to prevent racing with other
> vb2 calls, however, I see no way that any race can happen.

What about the following scenario ? Both QBUF calls are performed on the same 
buffer.

	CPU 0							CPU 1
	-------------------------------------------------------------------------
	QBUF								QBUF
		locks the queue mutex				waits for the queue mutex
	vb2_qbuf
	vb2_queue_or_prepare_buf
	__vb2_qbuf
		checks vb->state, calls
	__buf_prepare
	call_qop(q, wait_prepare, q);
		unlocks the queue mutex
										locks the queue mutex
									vb2_qbuf
									vb2_queue_or_prepare_buf
									__vb2_qbuf
										checks vb->state, calls
									__buf_prepare
									call_qop(q, wait_prepare, q);
										unlocks the queue mutex

									queue the buffer, set buffer
									 state to queue

	queue the buffer, set buffer
	 state to queue


We would thus end up queueing the buffer twice. The vb->state check needs to 
be performed after the brief release of the queue mutex.

> Moving it down offers opportunities to simplify the code.
> 
> Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
> ---
>  drivers/media/v4l2-core/videobuf2-core.c | 74 +++++++++++-----------------
>  1 file changed, 30 insertions(+), 44 deletions(-)
> 
> diff --git a/drivers/media/v4l2-core/videobuf2-core.c
> b/drivers/media/v4l2-core/videobuf2-core.c index 91412d4..d2b2efb 100644
> --- a/drivers/media/v4l2-core/videobuf2-core.c
> +++ b/drivers/media/v4l2-core/videobuf2-core.c
> @@ -1205,6 +1205,7 @@ static void __enqueue_in_driver(struct vb2_buffer *vb)
> static int __buf_prepare(struct vb2_buffer *vb, const struct v4l2_buffer
> *b) {
>  	struct vb2_queue *q = vb->vb2_queue;
> +	struct rw_semaphore *mmap_sem;
>  	int ret;
> 
>  	ret = __verify_length(vb, b);
> @@ -1219,7 +1220,25 @@ static int __buf_prepare(struct vb2_buffer *vb, const
> struct v4l2_buffer *b) ret = __qbuf_mmap(vb, b);
>  		break;
>  	case V4L2_MEMORY_USERPTR:
> +		/*
> +		 * In case of user pointer buffers vb2 allocators need to get direct
> +		 * access to userspace pages. This requires getting the mmap 
semaphore
> +		 * for read access in the current process structure. The same 
semaphore
> +		 * is taken before calling mmap operation, while both 
qbuf/prepare_buf
> +		 * and mmap are called by the driver or v4l2 core with the driver's 
lock
> +		 * held. To avoid an AB-BA deadlock (mmap_sem then driver's lock in 
mmap
> +		 * and driver's lock then mmap_sem in qbuf/prepare_buf) the videobuf2
> +		 * core releases the driver's lock, takes mmap_sem and then takes the
> +		 * driver's lock again.
> +		 */
> +		mmap_sem = &current->mm->mmap_sem;
> +		call_qop(q, wait_prepare, q);
> +		down_read(mmap_sem);
> +		call_qop(q, wait_finish, q);
> +
>  		ret = __qbuf_userptr(vb, b);
> +
> +		up_read(mmap_sem);
>  		break;
>  	case V4L2_MEMORY_DMABUF:
>  		ret = __qbuf_dmabuf(vb, b);
> @@ -1245,80 +1264,47 @@ static int vb2_queue_or_prepare_buf(struct vb2_queue
> *q, struct v4l2_buffer *b, struct v4l2_buffer *,
>  						   struct vb2_buffer *))
>  {
> -	struct rw_semaphore *mmap_sem = NULL;
>  	struct vb2_buffer *vb;
>  	int ret;
> 
> -	/*
> -	 * In case of user pointer buffers vb2 allocators need to get direct
> -	 * access to userspace pages. This requires getting the mmap semaphore
> -	 * for read access in the current process structure. The same semaphore
> -	 * is taken before calling mmap operation, while both qbuf/prepare_buf
> -	 * and mmap are called by the driver or v4l2 core with the driver's lock
> -	 * held. To avoid an AB-BA deadlock (mmap_sem then driver's lock in mmap
> -	 * and driver's lock then mmap_sem in qbuf/prepare_buf) the videobuf2
> -	 * core releases the driver's lock, takes mmap_sem and then takes the
> -	 * driver's lock again.
> -	 *
> -	 * To avoid racing with other vb2 calls, which might be called after
> -	 * releasing the driver's lock, this operation is performed at the
> -	 * beginning of qbuf/prepare_buf processing. This way the queue status
> -	 * is consistent after getting the driver's lock back.
> -	 */
> -	if (q->memory == V4L2_MEMORY_USERPTR) {
> -		mmap_sem = &current->mm->mmap_sem;
> -		call_qop(q, wait_prepare, q);
> -		down_read(mmap_sem);
> -		call_qop(q, wait_finish, q);
> -	}
> -
>  	if (q->fileio) {
>  		dprintk(1, "%s(): file io in progress\n", opname);
> -		ret = -EBUSY;
> -		goto unlock;
> +		return -EBUSY;
>  	}
> 
>  	if (b->type != q->type) {
>  		dprintk(1, "%s(): invalid buffer type\n", opname);
> -		ret = -EINVAL;
> -		goto unlock;
> +		return -EINVAL;
>  	}
> 
>  	if (b->index >= q->num_buffers) {
>  		dprintk(1, "%s(): buffer index out of range\n", opname);
> -		ret = -EINVAL;
> -		goto unlock;
> +		return -EINVAL;
>  	}
> 
>  	vb = q->bufs[b->index];
>  	if (NULL == vb) {
>  		/* Should never happen */
>  		dprintk(1, "%s(): buffer is NULL\n", opname);
> -		ret = -EINVAL;
> -		goto unlock;
> +		return -EINVAL;
>  	}
> 
>  	if (b->memory != q->memory) {
>  		dprintk(1, "%s(): invalid memory type\n", opname);
> -		ret = -EINVAL;
> -		goto unlock;
> +		return -EINVAL;
>  	}
> 
>  	ret = __verify_planes_array(vb, b);
>  	if (ret)
> -		goto unlock;
> +		return ret;
> 
>  	ret = handler(q, b, vb);
> -	if (ret)
> -		goto unlock;
> -
> -	/* Fill buffer information for the userspace */
> -	__fill_v4l2_buffer(vb, b);
> +	if (!ret) {
> +		/* Fill buffer information for the userspace */
> +		__fill_v4l2_buffer(vb, b);
> 
> -	dprintk(1, "%s() of buffer %d succeeded\n", opname, vb->v4l2_buf.index);
> -unlock:
> -	if (mmap_sem)
> -		up_read(mmap_sem);
> +		dprintk(1, "%s() of buffer %d succeeded\n", opname, vb-
>v4l2_buf.index);
> +	}
>  	return ret;
>  }
-- 
Regards,

Laurent Pinchart


  reply	other threads:[~2013-11-21 19:04 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-11-21 15:21 [RFC PATCH 0/8] vb2: various cleanups and improvements Hans Verkuil
2013-11-21 15:21 ` [RFC PATCH 1/8] vb2: push the mmap semaphore down to __buf_prepare() Hans Verkuil
2013-11-21 19:04   ` Laurent Pinchart [this message]
2013-11-22  9:02     ` Hans Verkuil
2013-11-26 15:42       ` Laurent Pinchart
2013-11-27  7:12         ` Hans Verkuil
2013-11-27  8:37           ` Laurent Pinchart
2013-11-21 15:22 ` [RFC PATCH 2/8] vb2: simplify qbuf/prepare_buf by removing callback Hans Verkuil
2013-11-21 15:22 ` [RFC PATCH 3/8] vb2: remove the 'fileio = NULL' hack Hans Verkuil
2013-11-21 15:22 ` [RFC PATCH 4/8] vb2: retry start_streaming in case of insufficient buffers Hans Verkuil
2013-11-21 19:09   ` Laurent Pinchart
2013-11-22  8:43     ` Hans Verkuil
2013-11-26 15:46       ` Laurent Pinchart
2013-11-27  7:17         ` Hans Verkuil
2013-11-27 10:35           ` Laurent Pinchart
2013-11-21 15:22 ` [RFC PATCH 5/8] vb2: don't set index, don't start streaming for write() Hans Verkuil
2013-11-21 15:22 ` [RFC PATCH 6/8] vb2: return ENODATA in start_streaming in case of too few buffers Hans Verkuil
2013-11-21 15:22 ` [RFC PATCH 7/8] vb2: add thread support Hans Verkuil
2013-11-21 15:22 ` [RFC PATCH 8/8] vb2: Add videobuf2-dvb support Hans Verkuil

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=6539252.6X3kkSkupS@avalon \
    --to=laurent.pinchart@ideasonboard.com \
    --cc=awalls@md.metrocast.net \
    --cc=hans.verkuil@cisco.com \
    --cc=hverkuil@xs4all.nl \
    --cc=linux-media@vger.kernel.org \
    --cc=m.szyprowski@samsung.com \
    --cc=pawel@osciak.com \
    /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