linux-kernel - Re: [PATCH 2/2] [media] exynos4-is: FIMC port parse should fail if there's no endpoint

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Tue, 22 Mar 2016 17:19:37 -0300
From:	Javier Martinez Canillas <javier@....samsung.com>
To:	Sylwester Nawrocki <s.nawrocki@...sung.com>
Cc:	linux-kernel@...r.kernel.org, Kukjin Kim <kgene@...nel.org>,
	Mauro Carvalho Chehab <mchehab@....samsung.com>,
	linux-samsung-soc@...r.kernel.org,
	Kyungmin Park <kyungmin.park@...sung.com>,
	Krzysztof Kozlowski <k.kozlowski@...sung.com>,
	linux-arm-kernel@...ts.infradead.org, linux-media@...r.kernel.org
Subject: Re: [PATCH 2/2] [media] exynos4-is: FIMC port parse should fail if
 there's no endpoint

Hello Sylwester,

On 03/11/2016 10:03 AM, Sylwester Nawrocki wrote:
> On 03/04/2016 09:20 PM, Javier Martinez Canillas wrote:
>> The fimc_md_parse_port_node() function return 0 if an endpoint node is
>> not found but according to Documentation/devicetree/bindings/graph.txt,
>> a port must always have at least one enpoint.
>>
>> So return an -EINVAL errno code to the caller instead, so it knows that
>> the port node parse failed due an invalid Device Tree description.
> 
> I don't think it is forbidden to have a port node in device tree
> containing no endpoint nodes. Empty port node means only that,
> for example, a subsystem has a port/bus for connecting external
> devices but nothing is actually connected to it.
> 
> In case of Exynos CSIS it might not be so useful to have an empty
> port node specified in some top level *.dtsi file and only
> the endpoints specified in a board specific dts file. Nevertheless,
> I wouldn't be saying in general a port node must always have some
> endpoint node defined.
> 

You are right, I asked Laurent and he confirms what you said that
it's possible to have ports with no endpoints. I still think the
DT binding docs could be more clear but that's a separate issue.

> I could apply this patch as it doesn't do any harm considering
> existing dts files in the kernel tree (arch/arm/boot/dts/
> exynos4412-trats2.dts), but the commit description would need to
> be changed.
> 

No worries, the current code is correct if endpoints are optional
and this patch is wrong so it should not be applied.
Best regards,
-- 
Javier Martinez Canillas
Open Source Group
Samsung Research America