dri-devel Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: dri-devel@lists.freedesktop.org
Subject: Re: [PATCH v4 22/22] drm: omapdrm: Perform initialization/cleanup at probe/remove time
Date: Fri, 16 Dec 2016 15:54:21 +0200	[thread overview]
Message-ID: <2379089.6fJ5i85Q6b@avalon> (raw)
In-Reply-To: <0791e4c6-657e-da9e-54ad-021b0617c4d5@ti.com>

Hi Tomi,

On Friday 16 Dec 2016 14:44:26 Tomi Valkeinen wrote:
> On 14/12/16 02:27, Laurent Pinchart wrote:
> > The drm driver .load() operation is prone to race conditions as it
> > initializes the driver after registering the device nodes. Its usage is
> > deprecated, inline it in the probe function and call drm_dev_alloc() and
> > drm_dev_register() explicitly.
> > 
> > For consistency inline the .unload() handler in the remove function as
> > well.
> > 
> > Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> > ---
> > Changes since v1:
> > 
> > - Call drm_kms_helper_poll_fini() before omap_fbdev_free() in the remove
> > 
> >   handler.
> > 
> > - Keep storing drm_device in the platform device private data.
> > ---
> > 
> >  drivers/gpu/drm/omapdrm/omap_connector.c |   2 -
> >  drivers/gpu/drm/omapdrm/omap_drv.c       | 212 +++++++++++++-------------
> >  2 files changed, 107 insertions(+), 107 deletions(-)

[snip]

> > diff --git a/drivers/gpu/drm/omapdrm/omap_drv.c
> > b/drivers/gpu/drm/omapdrm/omap_drv.c index 0a2d461d62cf..a3b37823b271
> > 100644
> > --- a/drivers/gpu/drm/omapdrm/omap_drv.c
> > +++ b/drivers/gpu/drm/omapdrm/omap_drv.c

[snip]

> > @@ -837,30 +741,128 @@ static struct drm_driver omap_drm_driver = {
> >  	.patchlevel = DRIVER_PATCHLEVEL,
> >  };
> > 
> > -static int pdev_probe(struct platform_device *device)
> > +static int pdev_probe(struct platform_device *pdev)
> >  {
> > -	int r;
> > +	struct omap_drm_platform_data *pdata = pdev->dev.platform_data;
> > +	struct omap_drm_private *priv;
> > +	struct drm_device *ddev;
> > +	unsigned int i;
> > +	int ret;
> > +
> > +	DBG("%s", pdev->name);
> > 
> >  	if (omapdss_is_initialized() == false)
> >  		return -EPROBE_DEFER;
> >  	
> >  	omap_crtc_pre_init();
> > 
> > -	r = omap_connect_dssdevs();
> > -	if (r) {
> > +	ret = omap_connect_dssdevs();
> > +	if (ret) {
> >  		omap_crtc_pre_uninit();
> > -		return r;
> > +		goto err_crtc_uninit;
> 
> This calls omap_crtc_pre_uninit here and in the err handler.

Oops, will fix.

> > +	}
> > +
> > +	/* Allocate and initialize the driver private structure. */
> > +	priv = kzalloc(sizeof(*priv), GFP_KERNEL);
> > +	if (!priv) {
> > +		ret = -ENOMEM;
> > +		goto err_disconnect_dssdevs;
> > +	}
> > +
> > +	priv->omaprev = pdata->omaprev;
> > +	priv->wq = alloc_ordered_workqueue("omapdrm", 0);
> > +
> > +	init_waitqueue_head(&priv->commit.wait);
> > +	spin_lock_init(&priv->commit.lock);
> > +	spin_lock_init(&priv->list_lock);
> > +	INIT_LIST_HEAD(&priv->obj_list);
> > +
> > +	/* Allocate and initialize the DRM device. */
> > +	ddev = drm_dev_alloc(&omap_drm_driver, &pdev->dev);
> > +	if (IS_ERR(ddev)) {
> > +		ret = PTR_ERR(ddev);
> > +		goto err_free_priv;
> > +	}
> > +
> > +	ddev->dev_private = priv;
> > +	platform_set_drvdata(pdev, ddev);
> > +
> > +	omap_gem_init(ddev);
> > +
> > +	ret = omap_modeset_init(ddev);
> > +	if (ret) {
> > +		dev_err(&pdev->dev, "omap_modeset_init failed: ret=%d\n", 
ret);
> > +		goto err_free_drm_dev;
> > +	}
> > +
> > +	/* Initialize vblank handling, start with all CRTCs disabled. */
> > +	ret = drm_vblank_init(ddev, priv->num_crtcs);
> > +	if (ret) {
> > +		dev_err(&pdev->dev, "could not init vblank\n");
> > +		goto err_cleanup_modeset;
> >  	}
> > 
> > -	DBG("%s", device->name);
> > -	return drm_platform_init(&omap_drm_driver, device);
> > +	for (i = 0; i < priv->num_crtcs; i++)
> > +		drm_crtc_vblank_off(priv->crtcs[i]);
> > +
> > +	priv->fbdev = omap_fbdev_init(ddev);
> > +
> > +	drm_kms_helper_poll_init(ddev);
> > +
> > +	/*
> > +	 * Register the DRM device with the core and the connectors with
> > +	 * sysfs.
> > +	 */
> > +	ret = drm_dev_register(ddev, 0);
> > +	if (ret)
> > +		goto err_cleanup_helpers;
> > +
> > +	return 0;
> > +
> > +err_cleanup_helpers:
> > +	drm_kms_helper_poll_fini(ddev);
> > +	if (priv->fbdev)
> > +		omap_fbdev_free(ddev);
> > +	drm_vblank_cleanup(ddev);
> > +err_cleanup_modeset:
> > +	drm_mode_config_cleanup(ddev);
> > +	omap_drm_irq_uninstall(ddev);
> > +err_free_drm_dev:
> > +	omap_gem_deinit(ddev);
> > +	drm_dev_unref(ddev);
> > +err_free_priv:
> > +	destroy_workqueue(priv->wq);
> > +	kfree(priv);
> > +err_disconnect_dssdevs:
> > +	omap_disconnect_dssdevs();
> > +err_crtc_uninit:
> > +	omap_crtc_pre_uninit();
> > +	return ret;
> >  }
> > 
> > -static int pdev_remove(struct platform_device *device)
> > +static int pdev_remove(struct platform_device *pdev)
> >  {
> > +	struct drm_device *ddev = platform_get_drvdata(pdev);
> > +	struct omap_drm_private *priv = ddev->dev_private;
> > +
> >  	DBG("");
> > 
> > -	drm_put_dev(platform_get_drvdata(device));
> > +	drm_dev_unregister(ddev);
> > +
> > +	drm_kms_helper_poll_fini(ddev);
> > +
> > +	if (priv->fbdev)
> > +		omap_fbdev_free(ddev);
> > +
> > +	drm_mode_config_cleanup(ddev);
> > +
> > +	omap_drm_irq_uninstall(ddev);
> > +	omap_gem_deinit(ddev);
> > +
> > +	drm_dev_unref(ddev);
> > +
> > +	destroy_workqueue(priv->wq);
> > +	kfree(priv);
> > 
> >  	omap_disconnect_dssdevs();
> >  	omap_crtc_pre_uninit();
> 
> The old code calls drm_vblank_cleanup(), and the probe's error handling
> calls that, but not remove. Is that correct?

Now that drm_vblank_cleanup() is called by drm_dev_release() we can drop the 
call from the probe error path. It used to be needed as the function was 
called from drm_dev_unregister().

-- 
Regards,

Laurent Pinchart

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

  reply	other threads:[~2016-12-16 13:53 UTC|newest]

Thread overview: 52+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-12-14  0:27 [PATCH v4 00/22] OMAP DRM fixes and improvements Laurent Pinchart
2016-12-14  0:27 ` [PATCH v4 01/22] drm: omapdrm: fb: Limit number of planes per framebuffer to two Laurent Pinchart
2016-12-14  0:27 ` [PATCH v4 02/22] drm: omapdrm: fb: Use format information provided by the DRM core Laurent Pinchart
2016-12-14 10:07   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 03/22] drm: omapdrm: fb: Simplify objects lookup when creating framebuffer Laurent Pinchart
2016-12-14  0:27 ` [PATCH v4 04/22] drm: omapdrm: fb: Simplify mode command checks " Laurent Pinchart
2016-12-14  0:27 ` [PATCH v4 05/22] drm: omapdrm: fb: Turn framebuffer creation error messages into debug Laurent Pinchart
2016-12-14  0:27 ` [PATCH v4 06/22] drm: omapdrm: Handle FIFO underflow IRQs internally Laurent Pinchart
2016-12-14 10:22   ` Tomi Valkeinen
2016-12-14 11:48     ` Laurent Pinchart
2016-12-14 13:13       ` Tomi Valkeinen
2016-12-15  9:02       ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 07/22] drm: omapdrm: Handle CRTC error IRQs directly Laurent Pinchart
2016-12-14  0:27 ` [PATCH v4 08/22] drm: omapdrm: Handle OCP error IRQ directly Laurent Pinchart
2016-12-14 10:24   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 09/22] drm: omapdrm: Replace DSS manager state check with omapdrm CRTC state Laurent Pinchart
2016-12-14 10:36   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 10/22] drm: omapdrm: Let the DRM core skip plane commit on inactive CRTCs Laurent Pinchart
2016-12-14 10:43   ` Tomi Valkeinen
2016-12-14 11:26     ` Laurent Pinchart
2016-12-14  0:27 ` [PATCH v4 11/22] drm: omapdrm: Check the CRTC software state at enable/disable time Laurent Pinchart
2016-12-14 14:54   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 12/22] drm: omapdrm: Prevent processing the same event multiple times Laurent Pinchart
2016-12-15 12:20   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 13/22] drm: omapdrm: Use a spinlock to protect the CRTC pending flag Laurent Pinchart
2016-12-14  0:27 ` [PATCH v4 14/22] drm: omapdrm: Keep vblank interrupt enabled while CRTC is active Laurent Pinchart
2016-12-15 12:52   ` Tomi Valkeinen
2016-12-15 14:51     ` Laurent Pinchart
2016-12-15 14:56       ` Tomi Valkeinen
2016-12-18  2:12   ` [PATCH v4.1 " Laurent Pinchart
2016-12-19  9:06     ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 15/22] drm: omapdrm: Don't expose the omap_irq_(un)register() functions Laurent Pinchart
2016-12-15 12:56   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 16/22] drm: omapdrm: Remove unused parameter from omap_drm_irq handler Laurent Pinchart
2016-12-15 12:57   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 17/22] drm: omapdrm: Don't call DISPC power handling in IRQ wait functions Laurent Pinchart
2016-12-15 13:00   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 18/22] drm: omapdrm: Inline the pipe2vbl function Laurent Pinchart
2016-12-14 10:25   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 19/22] drm: omapdrm: Simplify IRQ wait implementation Laurent Pinchart
2016-12-16 12:24   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 20/22] drm: omapdrm: Remove global variables Laurent Pinchart
2016-12-16 12:31   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 21/22] drm: omapdrm: Use sizeof(*var) instead of sizeof(type) for structures Laurent Pinchart
2016-12-15 13:02   ` Tomi Valkeinen
2016-12-14  0:27 ` [PATCH v4 22/22] drm: omapdrm: Perform initialization/cleanup at probe/remove time Laurent Pinchart
2016-12-16 12:44   ` Tomi Valkeinen
2016-12-16 13:54     ` Laurent Pinchart [this message]
2016-12-19  9:15   ` [PATCH v4.1 " Laurent Pinchart
2016-12-19  9:25     ` Tomi Valkeinen
2016-12-14  8:48 ` [PATCH v4 00/22] OMAP DRM fixes and improvements Tomi Valkeinen
2016-12-14 11:50   ` Laurent Pinchart

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=2379089.6fJ5i85Q6b@avalon \
    --to=laurent.pinchart@ideasonboard.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=tomi.valkeinen@ti.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