public inbox for linux-media@vger.kernel.org
 help / color / mirror / Atom feed
From: Nicolas Dufresne <nicolas.dufresne@collabora.com>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Sakari Ailus <sakari.ailus@linux.intel.com>,
	Mauro Carvalho Chehab	 <mchehab@kernel.org>,
	Hans Verkuil <hverkuil@xs4all.nl>,
	Tiffany Lin	 <tiffany.lin@mediatek.com>,
	Andrew-CT Chen <andrew-ct.chen@mediatek.com>,
	 Yunfei Dong <yunfei.dong@mediatek.com>,
	Matthias Brugger <matthias.bgg@gmail.com>,
	AngeloGioacchino Del Regno
	<angelogioacchino.delregno@collabora.com>,
	linux-kernel@vger.kernel.org,
	 linux-arm-kernel@lists.infradead.org,
	linux-mediatek@lists.infradead.org,  kernel@collabora.com,
	linux-media@vger.kernel.org,
	Sebastian Fricke	 <sebastian.fricke@collabora.com>
Subject: Re: [PATCH v2 1/5] media: mc: add manual request completion
Date: Thu, 10 Apr 2025 16:31:42 -0400	[thread overview]
Message-ID: <bb1f8d1581169bdf2e04be67883ee3ae03488680.camel@collabora.com> (raw)
In-Reply-To: <20250410175010.GF27870@pendragon.ideasonboard.com>

Replying on the code comment, sorry for missing some bit earlier ...

Le jeudi 10 avril 2025 à 20:50 +0300, Laurent Pinchart a écrit :
> Hi Nicolas,
> 
> Thank you for the patch.
> 
> On Thu, Apr 10, 2025 at 11:39:56AM -0400, Nicolas Dufresne wrote:
> > From: Hans Verkuil <hverkuil@xs4all.nl>
> > 
> > By default when the last request object is completed, the whole
> > request completes as well.
> > 
> > But sometimes you want to manually complete a request in a driver,
> > so add a manual complete mode for this.
> 
> I didn't immediately understand this was about delaying completion of
> the request. It would be nice to make that more explicit in the commit
> message and in the documentation of
> media_request_mark_manual_completion(). A sample use case would also
> help.
> 
> > In req_queue the driver marks the request for manual completion by
> > calling media_request_mark_manual_completion, and when the driver
> > wants to manually complete the request it calls
> > media_request_manual_complete().
> > 
> > Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
> > Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
> > ---
> >  drivers/media/mc/mc-request.c | 38 ++++++++++++++++++++++++++++++++++++--
> >  include/media/media-request.h | 36 +++++++++++++++++++++++++++++++++++-
> >  2 files changed, 71 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/media/mc/mc-request.c b/drivers/media/mc/mc-request.c
> > index 5edfc2791ce7c7485def5db675bbf53ee223d837..398d0806d1d274eb8c454fc5c37b77476abe1e74 100644
> > --- a/drivers/media/mc/mc-request.c
> > +++ b/drivers/media/mc/mc-request.c
> > @@ -54,6 +54,7 @@ static void media_request_clean(struct media_request *req)
> >  	req->access_count = 0;
> >  	WARN_ON(req->num_incomplete_objects);
> >  	req->num_incomplete_objects = 0;
> > +	req->manual_completion = false;
> >  	wake_up_interruptible_all(&req->poll_wait);
> >  }
> >  
> > @@ -313,6 +314,7 @@ int media_request_alloc(struct media_device *mdev, int *alloc_fd)
> >  	req->mdev = mdev;
> >  	req->state = MEDIA_REQUEST_STATE_IDLE;
> >  	req->num_incomplete_objects = 0;
> > +	req->manual_completion = false;
> >  	kref_init(&req->kref);
> >  	INIT_LIST_HEAD(&req->objects);
> >  	spin_lock_init(&req->lock);
> > @@ -459,7 +461,7 @@ void media_request_object_unbind(struct media_request_object *obj)
> >  
> >  	req->num_incomplete_objects--;
> >  	if (req->state == MEDIA_REQUEST_STATE_QUEUED &&
> > -	    !req->num_incomplete_objects) {
> > +	    !req->num_incomplete_objects && !req->manual_completion) {
> >  		req->state = MEDIA_REQUEST_STATE_COMPLETE;
> >  		completed = true;
> >  		wake_up_interruptible_all(&req->poll_wait);
> > @@ -488,7 +490,7 @@ void media_request_object_complete(struct media_request_object *obj)
> >  	    WARN_ON(req->state != MEDIA_REQUEST_STATE_QUEUED))
> >  		goto unlock;
> >  
> > -	if (!--req->num_incomplete_objects) {
> > +	if (!--req->num_incomplete_objects && !req->manual_completion) {
> >  		req->state = MEDIA_REQUEST_STATE_COMPLETE;
> >  		wake_up_interruptible_all(&req->poll_wait);
> >  		completed = true;
> > @@ -499,3 +501,35 @@ void media_request_object_complete(struct media_request_object *obj)
> >  		media_request_put(req);
> >  }
> >  EXPORT_SYMBOL_GPL(media_request_object_complete);
> > +
> > +void media_request_manual_complete(struct media_request *req)
> > +{
> > +	unsigned long flags;
> > +	bool completed = false;
> > +
> > +	if (WARN_ON(!req))
> > +		return;
> > +	if (WARN_ON(!req->manual_completion))
> > +		return;
> > +
> > +	spin_lock_irqsave(&req->lock, flags);
> > +	if (WARN_ON(req->state != MEDIA_REQUEST_STATE_QUEUED))
> > +		goto unlock;
> > +
> > +	req->manual_completion = false;
> > +	/*
> > +	 * It is expected that all other objects in this request are
> > +	 * completed when this function is called. WARN if that is
> > +	 * not the case.
> > +	 */
> > +	if (!WARN_ON(req->num_incomplete_objects)) {
> > +		req->state = MEDIA_REQUEST_STATE_COMPLETE;
> > +		wake_up_interruptible_all(&req->poll_wait);
> > +		completed = true;
> > +	}
> > +unlock:
> > +	spin_unlock_irqrestore(&req->lock, flags);
> > +	if (completed)
> > +		media_request_put(req);
> > +}
> > +EXPORT_SYMBOL_GPL(media_request_manual_complete);
> > diff --git a/include/media/media-request.h b/include/media/media-request.h
> > index d4ac557678a78372222704400c8c96cf3150b9d9..645d18907be7148ca50dcc9248ff06bd8ccdf953 100644
> > --- a/include/media/media-request.h
> > +++ b/include/media/media-request.h
> > @@ -56,6 +56,10 @@ struct media_request_object;
> >   * @access_count: count the number of request accesses that are in progress
> >   * @objects: List of @struct media_request_object request objects
> >   * @num_incomplete_objects: The number of incomplete objects in the request
> > + * @manual_completion: if true, then the request won't be marked as completed
> > + * when @num_incomplete_objects reaches 0. Call media_request_manual_complete()
> > + * to set this field to false and complete the request
> 
> I'd drop "set this field to false and " here.

I agree, it sounds like an invitation to write it directly, we don't
really want this. If Hans is fine I'll drop it.

> 
> > + * if @num_incomplete_objects == 0.
> 
>  * after @num_incomplete_objects reaches 0.
> 
> >   * @poll_wait: Wait queue for poll
> >   * @lock: Serializes access to this struct
> >   */
> > @@ -68,6 +72,7 @@ struct media_request {
> >  	unsigned int access_count;
> >  	struct list_head objects;
> >  	unsigned int num_incomplete_objects;
> > +	bool manual_completion;
> >  	wait_queue_head_t poll_wait;
> >  	spinlock_t lock;
> >  };
> > @@ -218,6 +223,35 @@ media_request_get_by_fd(struct media_device *mdev, int request_fd);
> >  int media_request_alloc(struct media_device *mdev,
> >  			int *alloc_fd);
> >  
> > +/**
> > + * media_request_mark_manual_completion - Set manual_completion to true
> > + *
> > + * @req: The request
> > + *
> > + * Mark that the request has to be manually completed by calling
> > + * media_request_manual_complete().
> > + *
> > + * This function should be called in the req_queue callback.
> 
> s/should/shall/ unless it's not a hard requirement. Any way to catch
> incorrect call patterns ?

I think we should be more strict, I can edit to shall.

About ways to check, perhaps I can add a lockdep_assert_held() on the
mdev->req_queue_mutex along with checking that the state is QUEUED. The
state check would catch trying to do so in req_validate().

> 
> > + */
> > +static inline void
> > +media_request_mark_manual_completion(struct media_request *req)
> > +{
> > +	req->manual_completion = true;
> > +}
> > +
> > +/**
> > + * media_request_manual_complete - Set manual_completion to false
> 
> The main purpose of the function is to complete the request, not setting
> manual_completion to false.

Indeed, that's documenting an implementation detail.

> 
> > + *
> > + * @req: The request
> > + *
> > + * Set @manual_completion to false, and if @num_incomplete_objects
> > + * is 0, then mark the request as completed.
> > + *
> > + * If there are still incomplete objects in the request, then
> > + * WARN for that since that suggests a driver error.
> 
> If that's an error then I'd document it more explicitly, as the first
> sentence makes it sound that both cases are valid. Maybe

Its a programming error, I should rephrase this in the next version.

> 
>  * This function completes a request that was marked for manual completion by an
>  * earlier call to media_request_mark_manual_completion(). The request's
>  * @manual_completion flag is reset to false.
>  *
>  * All objects contained in the request must have been completed previously. It
>  * is an error to call this function otherwise. The request will not be
>  * completed in that case, and the function will WARN.
> 
> > + */
> > +void media_request_manual_complete(struct media_request *req);
> > +
> >  #else
> >  
> >  static inline void media_request_get(struct media_request *req)
> > @@ -336,7 +370,7 @@ void media_request_object_init(struct media_request_object *obj);
> >   * @req: The media request
> >   * @ops: The object ops for this object
> >   * @priv: A driver-specific priv pointer associated with this object
> > - * @is_buffer: Set to true if the object a buffer object.
> > + * @is_buffer: Set to true if the object is a buffer object.
> >   * @obj: The object
> >   *
> >   * Bind this object to the request and set the ops and priv values of

thanks a lot for the review,

-- 
Nicolas Dufresne
Principal Engineer at Collabora

  parent reply	other threads:[~2025-04-10 20:31 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-04-10 15:39 [PATCH v2 0/5] Add manual request completion to the MediaTek VCodec driver Nicolas Dufresne
2025-04-10 15:39 ` [PATCH v2 1/5] media: mc: add manual request completion Nicolas Dufresne
2025-04-10 17:50   ` Laurent Pinchart
2025-04-10 18:26     ` Nicolas Dufresne
2025-04-10 18:42       ` Hans Verkuil
2025-04-10 18:41     ` Nicolas Dufresne
2025-04-10 19:05       ` Laurent Pinchart
2025-04-10 19:58         ` Nicolas Dufresne
2025-04-10 20:31     ` Nicolas Dufresne [this message]
2025-04-11  6:31       ` Hans Verkuil
2025-04-10 15:39 ` [PATCH v2 2/5] media: vicodec: add support for manual completion Nicolas Dufresne
2025-04-10 15:39 ` [PATCH v2 3/5] media: mc: add debugfs node to keep track of requests Nicolas Dufresne
2025-04-10 15:39 ` [PATCH v2 4/5] media: vcodec: Implement manual request completion Nicolas Dufresne
2025-04-14  9:11   ` AngeloGioacchino Del Regno
2025-04-14 19:19     ` Nicolas Dufresne
2025-04-10 15:40 ` [PATCH v2 5/5] media: mtk-vcodec: Don't try to decode 422/444 VP9 Nicolas Dufresne
2025-04-14  9:11   ` AngeloGioacchino Del Regno

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=bb1f8d1581169bdf2e04be67883ee3ae03488680.camel@collabora.com \
    --to=nicolas.dufresne@collabora.com \
    --cc=andrew-ct.chen@mediatek.com \
    --cc=angelogioacchino.delregno@collabora.com \
    --cc=hverkuil@xs4all.nl \
    --cc=kernel@collabora.com \
    --cc=laurent.pinchart@ideasonboard.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-mediatek@lists.infradead.org \
    --cc=matthias.bgg@gmail.com \
    --cc=mchehab@kernel.org \
    --cc=sakari.ailus@linux.intel.com \
    --cc=sebastian.fricke@collabora.com \
    --cc=tiffany.lin@mediatek.com \
    --cc=yunfei.dong@mediatek.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