From: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
To: Shuah Khan <shuahkh@osg.samsung.com>
Cc: tiwai@suse.com, clemens@ladisch.de, hans.verkuil@cisco.com,
laurent.pinchart@ideasonboard.com, sakari.ailus@linux.intel.com,
javier@osg.samsung.com, pawel@osciak.com,
m.szyprowski@samsung.com, kyungmin.park@samsung.com,
perex@perex.cz, arnd@arndb.de, dan.carpenter@oracle.com,
tvboxspy@gmail.com, crope@iki.fi, ruchandani.tina@gmail.com,
corbet@lwn.net, chehabrafael@gmail.com, k.kozlowski@samsung.com,
stefanr@s5r6.in-berlin.de, inki.dae@samsung.com,
jh1009.sung@samsung.com, elfring@users.sourceforge.net,
prabhakar.csengg@gmail.com, sw0312.kim@samsung.com,
p.zabel@pengutronix.de, ricardo.ribalda@gmail.com,
labbott@fedoraproject.org, pierre-louis.bossart@linux.intel.com,
ricard.wanderlof@axis.com, julian@jusst.de,
takamichiho@gmail.com, dominic.sacre@gmx.de, misterpib@gmail.com,
daniel@zonque.org, gtmkramer@xs4all.nl, normalperson@yhbt.net,
joe@oampo.co.uk, linuxbugs@vittgam.net, johan@oljud.se,
klock.android@gmail.com, nenggun.kim@samsung.com,
j.anaszewski@samsung.com, geliangtang@163.com,
linux-kernel@vger.kernel.org, linux-media@vger.kernel.org,
linux-api@vger.kernel.org, alsa-devel@alsa-project.org
Subject: Re: [PATCH v2 22/22] media: Ensure media device unregister is done only once
Date: Thu, 4 Feb 2016 08:40:44 -0200 [thread overview]
Message-ID: <20160204084044.234f1867@recife.lan> (raw)
In-Reply-To: <9c22d4395f92102051383110cae9de09494d7257.1454557589.git.shuahkh@osg.samsung.com>
Em Wed, 03 Feb 2016 21:03:54 -0700
Shuah Khan <shuahkh@osg.samsung.com> escreveu:
> media_device_unregister() checks if the media device
> is registered or not as the first step. However, the
> MEDIA_FLAG_REGISTERED bit doesn't get cleared until
> the end leaving a large window for two drivers to
> attempt media device unregister.
>
> The above leads to general protection faults when
> device is removed.
>
> Fix the problem with two phase media device unregister.
> Add a new interface media_devnode_start_unregister()
> to clear the MEDIA_FLAG_REGISTERED bit. Change
> media_device_unregister() call this interface to mark
> the start of unregister. This will ensure that media
> device unregister is done only once.
Hmm... it sounds simpler to use a kref at device register
and when other drivers need instantiate media_dev. We can
then do a kref_put() at unregister, and run the actual
unregister code and media_ctl kfree only when kref is
decremented on all drivers.
Regards,
Mauro
>
> Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
> ---
> drivers/media/media-device.c | 12 ++++++------
> drivers/media/media-devnode.c | 15 ++++++++++-----
> include/media/media-devnode.h | 17 +++++++++++++++++
> 3 files changed, 33 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/media/media-device.c b/drivers/media/media-device.c
> index 1f5d67e..584d46e 100644
> --- a/drivers/media/media-device.c
> +++ b/drivers/media/media-device.c
> @@ -747,17 +747,17 @@ void media_device_unregister(struct media_device *mdev)
> struct media_entity *next;
> struct media_interface *intf, *tmp_intf;
> struct media_entity_notify *notify, *nextp;
> + int ret;
>
> if (mdev == NULL)
> return;
>
> - spin_lock(&mdev->lock);
> -
> - /* Check if mdev was ever registered at all */
> - if (!media_devnode_is_registered(&mdev->devnode)) {
> - spin_unlock(&mdev->lock);
> + /* Start unregister - continue if necessary */
> + ret = media_devnode_start_unregister(&mdev->devnode);
> + if (ret)
> return;
> - }
> +
> + spin_lock(&mdev->lock);
>
> /* Remove all entities from the media device */
> list_for_each_entry_safe(entity, next, &mdev->entities, graph_obj.list)
> diff --git a/drivers/media/media-devnode.c b/drivers/media/media-devnode.c
> index 29409f4..c27f9e7 100644
> --- a/drivers/media/media-devnode.c
> +++ b/drivers/media/media-devnode.c
> @@ -272,15 +272,20 @@ error:
> return ret;
> }
>
> -void media_devnode_unregister(struct media_devnode *mdev)
> +int __must_check media_devnode_start_unregister(struct media_devnode *mdev)
> {
> - /* Check if mdev was ever registered at all */
> - if (!media_devnode_is_registered(mdev))
> - return;
> -
> mutex_lock(&media_devnode_lock);
> + if (!media_devnode_is_registered(mdev)) {
> + mutex_unlock(&media_devnode_lock);
> + return -EINVAL;
> + }
> clear_bit(MEDIA_FLAG_REGISTERED, &mdev->flags);
> mutex_unlock(&media_devnode_lock);
> + return 0;
> +}
> +
> +void media_devnode_unregister(struct media_devnode *mdev)
> +{
> device_unregister(&mdev->dev);
> }
>
> diff --git a/include/media/media-devnode.h b/include/media/media-devnode.h
> index fe42f08..6f08677 100644
> --- a/include/media/media-devnode.h
> +++ b/include/media/media-devnode.h
> @@ -120,6 +120,23 @@ int __must_check media_devnode_register(struct media_devnode *mdev,
> struct module *owner);
>
> /**
> + * media_devnode_start_unregister - start unregister on a media device node
> + * @mdev: the device node to start unregister
> + *
> + * This clears the MEDIA_FLAG_REGISTERED bit to indicate that unregister
> + * is in progress.
> + *
> + * This function can safely be called if the device node has never been
> + * registered or has already been unregistered.
> + *
> + * Zero is returned on success.
> + *
> + * -EINVAL is returned if the device node has never been
> + * registered or has already been unregistered.
> + */
> +int __must_check media_devnode_start_unregister(struct media_devnode *mdev);
> +
> +/**
> * media_devnode_unregister - unregister a media device node
> * @mdev: the device node to unregister
> *
next prev parent reply other threads:[~2016-02-04 10:41 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-02-04 4:03 [PATCH v2 00/22] Sharing media resources across ALSA and au0828 drivers Shuah Khan
2016-02-04 4:03 ` [PATCH v2 01/22] uapi/media.h: Declare interface types for ALSA Shuah Khan
2016-02-04 8:38 ` Mauro Carvalho Chehab
2016-02-04 4:03 ` [PATCH v2 02/22] media: Add ALSA Media Controller function entities Shuah Khan
2016-02-04 8:40 ` Mauro Carvalho Chehab
2016-02-04 4:03 ` [PATCH v2 03/22] media: Media Controller register/unregister entity_notify API Shuah Khan
2016-02-04 9:03 ` Mauro Carvalho Chehab
2016-02-04 4:03 ` [PATCH v2 04/22] media: Media Controller enable/disable source handler API Shuah Khan
2016-02-04 4:03 ` [PATCH v2 05/22] media: Media Controller export non locking __media_entity_setup_link() Shuah Khan
2016-02-04 4:03 ` [PATCH v2 06/22] media: Media Controller non-locking __media_entity_pipeline_start/stop() Shuah Khan
2016-02-04 9:18 ` Mauro Carvalho Chehab
2016-02-04 15:00 ` Shuah Khan
2016-02-04 4:03 ` [PATCH v2 07/22] media: v4l-core add enable/disable source common interfaces Shuah Khan
2016-02-04 6:29 ` kbuild test robot
2016-02-04 9:25 ` Mauro Carvalho Chehab
2016-02-04 4:03 ` [PATCH v2 08/22] media: Move au8522_media_pads enum to au8522.h from au8522_priv.h Shuah Khan
2016-02-04 4:03 ` [PATCH v2 09/22] media: au8522 change to create MC pad for ALSA Audio Out Shuah Khan
2016-02-04 4:03 ` [PATCH v2 10/22] media: Change v4l-core to check if source is free Shuah Khan
2016-02-04 4:03 ` [PATCH v2 11/22] media: dvb-frontend invoke enable/disable_source handlers Shuah Khan
2016-02-04 9:35 ` Mauro Carvalho Chehab
2016-02-04 14:58 ` Shuah Khan
2016-02-04 4:03 ` [PATCH v2 12/22] media: au0828 video remove au0828_enable_analog_tuner() Shuah Khan
2016-02-04 9:37 ` Mauro Carvalho Chehab
2016-02-04 14:49 ` Shuah Khan
2016-02-04 4:03 ` [PATCH v2 13/22] media: au0828 video change to use v4l_enable_media_source() Shuah Khan
2016-02-04 10:00 ` Mauro Carvalho Chehab
2016-02-04 4:03 ` [PATCH v2 14/22] media: au0828 change to use Managed Media Controller API Shuah Khan
2016-02-04 10:05 ` Mauro Carvalho Chehab
2016-02-04 10:08 ` Mauro Carvalho Chehab
2016-02-04 4:03 ` [PATCH v2 15/22] media: au0828 handle media_init and media_register window Shuah Khan
2016-02-04 4:03 ` [PATCH v2 16/22] media: au0828 create tuner to decoder link in disabled state Shuah Khan
2016-02-04 4:03 ` [PATCH v2 17/22] media: au0828 disable tuner to demod link Shuah Khan
2016-02-04 4:03 ` [PATCH v2 18/22] media: au0828 Use au8522_media_pads enum for pad defines Shuah Khan
2016-02-04 4:03 ` [PATCH v2 19/22] media: au0828-core register entity_notify hook Shuah Khan
2016-02-04 4:03 ` [PATCH v2 20/22] media: au0828 add enable, disable source handlers Shuah Khan
2016-02-04 10:26 ` Mauro Carvalho Chehab
2016-02-08 22:42 ` Shuah Khan
2016-02-08 23:00 ` Shuah Khan
2016-02-09 13:42 ` Mauro Carvalho Chehab
2016-02-09 10:51 ` Mauro Carvalho Chehab
2016-02-04 4:03 ` [PATCH v2 21/22] sound/usb: Use Media Controller API to share media resources Shuah Khan
2016-02-04 10:36 ` Mauro Carvalho Chehab
2016-02-04 10:46 ` Takashi Iwai
2016-02-07 11:25 ` kbuild test robot
2016-02-04 4:03 ` [PATCH v2 22/22] media: Ensure media device unregister is done only once Shuah Khan
2016-02-04 10:40 ` Mauro Carvalho Chehab [this message]
2016-02-04 10:54 ` [PATCH v2 00/22] Sharing media resources across ALSA and au0828 drivers Mauro Carvalho Chehab
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=20160204084044.234f1867@recife.lan \
--to=mchehab@osg.samsung.com \
--cc=alsa-devel@alsa-project.org \
--cc=arnd@arndb.de \
--cc=chehabrafael@gmail.com \
--cc=clemens@ladisch.de \
--cc=corbet@lwn.net \
--cc=crope@iki.fi \
--cc=dan.carpenter@oracle.com \
--cc=daniel@zonque.org \
--cc=dominic.sacre@gmx.de \
--cc=elfring@users.sourceforge.net \
--cc=geliangtang@163.com \
--cc=gtmkramer@xs4all.nl \
--cc=hans.verkuil@cisco.com \
--cc=inki.dae@samsung.com \
--cc=j.anaszewski@samsung.com \
--cc=javier@osg.samsung.com \
--cc=jh1009.sung@samsung.com \
--cc=joe@oampo.co.uk \
--cc=johan@oljud.se \
--cc=julian@jusst.de \
--cc=k.kozlowski@samsung.com \
--cc=klock.android@gmail.com \
--cc=kyungmin.park@samsung.com \
--cc=labbott@fedoraproject.org \
--cc=laurent.pinchart@ideasonboard.com \
--cc=linux-api@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.org \
--cc=linuxbugs@vittgam.net \
--cc=m.szyprowski@samsung.com \
--cc=misterpib@gmail.com \
--cc=nenggun.kim@samsung.com \
--cc=normalperson@yhbt.net \
--cc=p.zabel@pengutronix.de \
--cc=pawel@osciak.com \
--cc=perex@perex.cz \
--cc=pierre-louis.bossart@linux.intel.com \
--cc=prabhakar.csengg@gmail.com \
--cc=ricard.wanderlof@axis.com \
--cc=ricardo.ribalda@gmail.com \
--cc=ruchandani.tina@gmail.com \
--cc=sakari.ailus@linux.intel.com \
--cc=shuahkh@osg.samsung.com \
--cc=stefanr@s5r6.in-berlin.de \
--cc=sw0312.kim@samsung.com \
--cc=takamichiho@gmail.com \
--cc=tiwai@suse.com \
--cc=tvboxspy@gmail.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
Powered by JetHome