| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Thu, 27 Apr 2017 23:13:50 +0100
From: Kieran Bingham <kieran.bingham@...asonboard.com>
To: Sakari Ailus <sakari.ailus@....fi>,
Kieran Bingham <kbingham@...nel.org>
Cc: laurent.pinchart@...asonboard.com, niklas.soderlund@...natech.se,
linux-media@...r.kernel.org, linux-renesas-soc@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/5] v4l2-subdev: Provide a port mapping for asynchronous
subdevs
Hi Sakari,
Thanks for taking a look
On 27/04/17 22:43, Sakari Ailus wrote:
> Hi Kieran,
>
> Could I ask you to rebase your patches on top of my V4L2 fwnode patches
> here?
>
> <URL:https://git.linuxtv.org/sailus/media_tree.git/log/?h=v4l2-acpi>
>
> It depends on the fwnode graph patches, merged here:
>
> <URL:https://git.linuxtv.org/sailus/media_tree.git/log/?h=v4l2-acpi-merge>
>
> I expect the fwnode graph patches in v4.12 so we'll have them in media-tree
> master soon.
>
> (I'm pushing these branches right now, it may take a while until it's really
> there.)
Sure, I'll merge those into my base.
> On Thu, Apr 27, 2017 at 07:26:00PM +0100, Kieran Bingham wrote:
>> From: Kieran Bingham <kieran.bingham+renesas@...asonboard.com>
>>
>> Devices such as the the ADV748x support multiple parallel stream routes
>> through a single chip. This leads towards needing to provide multiple
>> distinct entities and subdevs from a single device-tree node.
>>
>> To distinguish these separate outputs, the device-tree binding must
>> specify each endpoint link with a unique (to the device) non-zero port
>> number.
>>
>> This number allows async subdev registrations to identify the correct
>> subdevice to bind and link.
>>
>> Signed-off-by: Kieran Bingham <kieran.bingham+renesas@...asonboard.com>
>> ---
>> drivers/media/v4l2-core/v4l2-async.c | 7 +++++++
>> drivers/media/v4l2-core/v4l2-subdev.c | 1 +
>> include/media/v4l2-async.h | 1 +
>> include/media/v4l2-subdev.h | 2 ++
>> 4 files changed, 11 insertions(+)
>>
>> diff --git a/drivers/media/v4l2-core/v4l2-async.c b/drivers/media/v4l2-core/v4l2-async.c
>> index 1815e54e8a38..875e6ce646ec 100644
>> --- a/drivers/media/v4l2-core/v4l2-async.c
>> +++ b/drivers/media/v4l2-core/v4l2-async.c
>> @@ -42,6 +42,13 @@ static bool match_devname(struct v4l2_subdev *sd,
>>
>> static bool match_of(struct v4l2_subdev *sd, struct v4l2_async_subdev *asd)
>> {
>> + /*
>> + * If set, we must match the device tree port, with the subdev port.
>> + * This is a fast match, so do this first
>> + */
>> + if (sd->port && sd->port != asd->match.of.port)
>
> Zero is an entirely valid value for a port. I think it'd be good not to
> depend on non-zero port values for port matching.
Well then that pretty much dashes my chances on not parsing the DT in the ADV
driver.
>> + return -1;
>
> Any particular reason to return -1 from a function with bool return type?
Ahem, I clearly can't read ;-)
I think my mindset was thinking strcmp or something...
>> +
>> return !of_node_cmp(of_node_full_name(sd->of_node),
>> of_node_full_name(asd->match.of.node));
>> }
>> diff --git a/drivers/media/v4l2-core/v4l2-subdev.c b/drivers/media/v4l2-core/v4l2-subdev.c
>> index da78497ae5ed..67f816f90ac3 100644
>> --- a/drivers/media/v4l2-core/v4l2-subdev.c
>> +++ b/drivers/media/v4l2-core/v4l2-subdev.c
>> @@ -607,6 +607,7 @@ void v4l2_subdev_init(struct v4l2_subdev *sd, const struct v4l2_subdev_ops *ops)
>> sd->flags = 0;
>> sd->name[0] = '\0';
>> sd->grp_id = 0;
>> + sd->port = 0;
>> sd->dev_priv = NULL;
>> sd->host_priv = NULL;
>> #if defined(CONFIG_MEDIA_CONTROLLER)
>> diff --git a/include/media/v4l2-async.h b/include/media/v4l2-async.h
>> index 5b501309b6a7..2988960613ec 100644
>> --- a/include/media/v4l2-async.h
>> +++ b/include/media/v4l2-async.h
>> @@ -56,6 +56,7 @@ struct v4l2_async_subdev {
>> union {
>> struct {
>> const struct device_node *node;
>> + u32 port;
>
> What if instead of storing the device's OF node, you'd store the port node
> and used that for matching?
>
> Would that also solve the problem or do I miss something?
Actually - I was 'trying' to prevent having to parse the DT in the adv748x
driver if I didn't need to.
Once I have to parse the DT, then yes, I think storing the endpoint node is
probably the best thing to compare against.
And actually - you might have just solved my open question in the cover letter ...
I had got stuck in my mindset that if I were to use the endpoint 'leaf' node as
a comparator - that it would be 'instead' of the root node.
But actually - it could just be root-node + leaf-node to compare, which then
allows us the fallback of comparing just the root nodes if the leaf isn't set.
I'll respin with this either tomorrow or early next week.
>
>> } of;
>> struct {
>> const char *name;
>> diff --git a/include/media/v4l2-subdev.h b/include/media/v4l2-subdev.h
>> index 0ab1c5df6fac..1c1731b491e5 100644
>> --- a/include/media/v4l2-subdev.h
>> +++ b/include/media/v4l2-subdev.h
>> @@ -782,6 +782,7 @@ struct v4l2_subdev_platform_data {
>> * @ctrl_handler: The control handler of this subdev. May be NULL.
>> * @name: Name of the sub-device. Please notice that the name must be unique.
>> * @grp_id: can be used to group similar subdevs. Value is driver-specific
>> + * @port: driver-specific value to bind multiple subdevs with a single DT node.
>> * @dev_priv: pointer to private data
>> * @host_priv: pointer to private data used by the device where the subdev
>> * is attached.
>> @@ -814,6 +815,7 @@ struct v4l2_subdev {
>> struct v4l2_ctrl_handler *ctrl_handler;
>> char name[V4L2_SUBDEV_NAME_SIZE];
>> u32 grp_id;
>> + u32 port;
>> void *dev_priv;
>> void *host_priv;
>> struct video_device *devnode;
>
Regards
Kieran
Powered by blists - more mailing lists