intel-gfx.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: dri-devel@lists.freedesktop.org
Cc: Alex Deucher <alexander.deucher@amd.com>,
	Daniel Vetter <daniel.vetter@ffwll.ch>,
	Intel Graphics Development <intel-gfx@lists.freedesktop.org>
Subject: Re: [PATCH] drm: Clean up pending events in the core
Date: Mon, 11 Jan 2016 01:48:23 +0200	[thread overview]
Message-ID: <3634528.gCZFVPLAH7@avalon> (raw)
In-Reply-To: <1452346126-20091-1-git-send-email-daniel.vetter@ffwll.ch>

Hi Daniel,

Thank you for the patch.

On Saturday 09 January 2016 14:28:46 Daniel Vetter wrote:
> There's really no reason to not do so, instead of replicating this
> for every use-case and every driver. Now we can't just nuke the events,
> since that would still mean that all drm_event users would need to know
> when that has happened, since calling e.g. drm_send_event isn't allowed
> any more. Instead just unlink them from the file, and detect this case
> and handle it appropriately in all functions.
> 
> v2: Adjust existing kerneldoc too.
> 
> Cc: Alex Deucher <alexander.deucher@amd.com>
> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> ---
>  drivers/gpu/drm/drm_fops.c | 35 ++++++++++++++++++++++++++---------
>  include/drm/drmP.h         |  2 ++
>  2 files changed, 28 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_fops.c b/drivers/gpu/drm/drm_fops.c
> index d85af1b2a238..d32b24c74e08 100644
> --- a/drivers/gpu/drm/drm_fops.c
> +++ b/drivers/gpu/drm/drm_fops.c
> @@ -264,6 +264,7 @@ static int drm_open_helper(struct file *filp, struct
> drm_minor *minor) INIT_LIST_HEAD(&priv->fbs);
>  	mutex_init(&priv->fbs_lock);
>  	INIT_LIST_HEAD(&priv->blobs);
> +	INIT_LIST_HEAD(&priv->pending_event_list);
>  	INIT_LIST_HEAD(&priv->event_list);
>  	init_waitqueue_head(&priv->event_wait);
>  	priv->event_space = 4096; /* set aside 4k for event buffer */
> @@ -353,18 +354,16 @@ static void drm_events_release(struct drm_file
> *file_priv) {
>  	struct drm_device *dev = file_priv->minor->dev;
>  	struct drm_pending_event *e, *et;
> -	struct drm_pending_vblank_event *v, *vt;
>  	unsigned long flags;
> 
>  	spin_lock_irqsave(&dev->event_lock, flags);
> 
> -	/* Remove pending flips */
> -	list_for_each_entry_safe(v, vt, &dev->vblank_event_list, base.link)
> -		if (v->base.file_priv == file_priv) {
> -			list_del(&v->base.link);
> -			drm_vblank_put(dev, v->pipe);
> -			v->base.destroy(&v->base);
> -		}

Where does this code go ?

> +	/* Unlink pending events */
> +	list_for_each_entry_safe(e, et, &file_priv->pending_event_list,
> +				 pending_link) {
> +		list_del(&e->pending_link);
> +		e->file_priv = NULL;
> +	}
> 
>  	/* Remove unconsumed events */
>  	list_for_each_entry_safe(e, et, &file_priv->event_list, link) {
> @@ -712,6 +711,7 @@ int drm_event_reserve_init(struct drm_device *dev,
>  	file_priv->event_space -= e->length;
> 
>  	p->event = e;
> +	list_add(&p->pending_link, &file_priv->pending_event_list);
>  	p->file_priv = file_priv;
> 
>  	/* we *could* pass this in as arg, but everyone uses kfree: */
> @@ -736,7 +736,10 @@ void drm_event_cancel_free(struct drm_device *dev,
>  {
>  	unsigned long flags;
>  	spin_lock_irqsave(&dev->event_lock, flags);
> -	p->file_priv->event_space += p->event->length;
> +	if (p->file_priv) {
> +		p->file_priv->event_space += p->event->length;
> +		list_del(&p->pending_link);
> +	}
>  	spin_unlock_irqrestore(&dev->event_lock, flags);
>  	p->destroy(p);
>  }
> @@ -750,11 +753,21 @@ EXPORT_SYMBOL(drm_event_cancel_free);
>   * This function sends the event @e, initialized with
> drm_event_reserve_init(),
>   * to its associated userspace DRM file. Callers must already hold
>   * dev->event_lock, see drm_send_event() for the unlocked version.
> + *
> + * Note that the core will take care of unlinking and disarming events when
> the
> + * corresponding DRM file is closed. Drivers need to worry about that and

s/need to/need not/ ?

> can
> + * call this function upon completion of the asynchrnous work

s/asynchrnous/asynchronous/

> unconditionally. */
>  void drm_send_event_locked(struct drm_device *dev, struct drm_pending_event
> *e) {
>  	assert_spin_locked(&dev->event_lock);
> 
> +	if (!e->file_priv) {

I don't think this could happen before this patch as e->file_priv is 
dereferenced below, and I don't see anything in this patch that makes the 
condition possible.

> +		e->destroy(e);
> +		return;
> +	}
> +
> +	list_del(&e->pending_link);
>  	list_add_tail(&e->link,
>  		      &e->file_priv->event_list);
>  	wake_up_interruptible(&e->file_priv->event_wait);
> @@ -769,6 +782,10 @@ EXPORT_SYMBOL(drm_send_event_locked);
>   * This function sends the event @e, initialized with
> drm_event_reserve_init(),
>   * to its associated userspace DRM file. This function acquires
> dev->event_lock,
>   * see drm_send_event_locked() for callers which already hold this lock.
> + *
> + * Note that the core will take care of unlinking and disarming events when
> the
> + * corresponding DRM file is closed. Drivers need to worry about that and

s/need to/need not/ ?

> can
> + * call this function upon completion of the asynchrnous work

s/asynchrnous/asynchronous/

> unconditionally. */
>  void drm_send_event(struct drm_device *dev, struct drm_pending_event *e)
>  {
> diff --git a/include/drm/drmP.h b/include/drm/drmP.h
> index ae73abf5c2cf..3d78a7406d54 100644
> --- a/include/drm/drmP.h
> +++ b/include/drm/drmP.h
> @@ -283,6 +283,7 @@ struct drm_ioctl_desc {
>  struct drm_pending_event {
>  	struct drm_event *event;
>  	struct list_head link;
> +	struct list_head pending_link;
>  	struct drm_file *file_priv;
>  	pid_t pid; /* pid of requester, no guarantee it's valid by the time
>  		      we deliver the event, for tracing only */
> @@ -346,6 +347,7 @@ struct drm_file {
>  	struct list_head blobs;
> 
>  	wait_queue_head_t event_wait;
> +	struct list_head pending_event_list;
>  	struct list_head event_list;
>  	int event_space;

-- 
Regards,

Laurent Pinchart

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

  reply	other threads:[~2016-01-10 23:48 UTC|newest]

Thread overview: 50+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-01-08 20:36 [PATCH 00/21] drm_event cleanup Daniel Vetter
2016-01-08 20:36 ` [PATCH 01/21] drm: kerneldoc for drm_fops.c Daniel Vetter
2016-01-08 20:36 ` [PATCH 02/21] drm: Add functions to setup/tear down drm_events Daniel Vetter
2016-01-08 21:10   ` Alex Deucher
2016-01-09 13:27   ` [PATCH] " Daniel Vetter
2016-01-08 20:36 ` [PATCH 03/21] drm/exynos: Use the new event init/free functions Daniel Vetter
2016-01-08 20:36 ` [PATCH 04/21] drm/vmwgfx: " Daniel Vetter
2016-01-08 20:36 ` [PATCH 05/21] drm: Create drm_send_event helpers Daniel Vetter
2016-01-09 13:28   ` [PATCH] " Daniel Vetter
2016-01-08 20:36 ` [PATCH 06/21] drm/fsl: Remove preclose hook Daniel Vetter
2016-01-08 20:36 ` [PATCH 07/21] drm/armada: Remove NULL open/pre/postclose hooks Daniel Vetter
2016-01-08 20:36 ` [PATCH 08/21] drm/gma500: Remove empty preclose hook Daniel Vetter
2016-01-08 20:36 ` [PATCH 09/21] drm: Clean up pending events in the core Daniel Vetter
2016-01-09 13:28   ` [PATCH] " Daniel Vetter
2016-01-10 23:48     ` Laurent Pinchart [this message]
2016-01-11 14:51       ` [Intel-gfx] " Daniel Stone
2016-01-14 18:49         ` Laurent Pinchart
2016-01-08 20:36 ` [PATCH 10/21] drm/i915: Nuke intel_modeset_preclose Daniel Vetter
2016-01-08 20:36 ` [PATCH 11/21] drm/atmel: Nuke preclose Daniel Vetter
2016-01-08 20:36 ` [PATCH 12/21] drm/exynos: Remove event cancelling from postclose Daniel Vetter
2016-01-08 20:36 ` [PATCH 13/21] drm/imx: Unconfuse preclose logic Daniel Vetter
2016-01-08 20:36 ` [PATCH 14/21] drm/msm: Nuke preclose hooks Daniel Vetter
2016-01-08 20:36 ` [PATCH 15/21] drm/omap: Nuke close hooks Daniel Vetter
2016-01-11  0:03   ` Laurent Pinchart
2016-01-11  7:10     ` Daniel Vetter
2016-01-08 20:36 ` [PATCH 16/21] drm/rcar: Nuke preclose hook Daniel Vetter
2016-01-11  0:01   ` Laurent Pinchart
2016-01-08 20:36 ` [PATCH 17/21] drm/shmob: " Daniel Vetter
2016-01-10 22:26   ` [PATCH 1/5] " Daniel Vetter
2016-01-10 22:26     ` [PATCH 2/5] drm/tegra: Stop cancelling page flip events Daniel Vetter
2016-01-10 22:26     ` [PATCH 3/5] drm/tilcdc: Nuke preclose hook Daniel Vetter
2016-01-10 22:26     ` [PATCH 4/5] drm/vc4: " Daniel Vetter
2016-01-10 22:26     ` [PATCH 5/5] drm/vmwgfx: " Daniel Vetter
2016-01-11 10:20       ` Thomas Hellstrom
2016-01-11  0:00     ` [PATCH 1/5] drm/shmob: " Laurent Pinchart
2016-01-08 20:36 ` [PATCH 18/21] drm/tegra: Stop cancelling page flip events Daniel Vetter
2016-01-08 20:36 ` [PATCH 19/21] drm/tilcdc: Nuke preclose hook Daniel Vetter
2016-01-08 20:36 ` [PATCH 20/21] drm/vc4: " Daniel Vetter
2016-01-08 20:36 ` [PATCH 21/21] drm/vmwgfx: " Daniel Vetter
2016-01-08 20:53   ` Thomas Hellstrom
2016-01-09 10:43     ` Daniel Vetter
2016-01-10 20:52       ` Thomas Hellstrom
2016-01-10 21:59         ` Daniel Vetter
2016-01-10 22:02   ` [PATCH] " Daniel Vetter
2016-01-10 22:17     ` Thomas Hellstrom
2016-01-10 22:22       ` Daniel Vetter
2016-01-08 21:47 ` [PATCH 00/21] drm_event cleanup Alex Deucher
2016-01-09 13:32 ` Daniel Vetter
2016-01-11 11:20 ` ✗ failure: Fi.CI.BAT Patchwork
2016-01-12 16:24   ` Daniel Vetter

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=3634528.gCZFVPLAH7@avalon \
    --to=laurent.pinchart@ideasonboard.com \
    --cc=alexander.deucher@amd.com \
    --cc=daniel.vetter@ffwll.ch \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-gfx@lists.freedesktop.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).