linux-kernel - Re: [PATCH] pinctrl: bcm2835: make use of of_property_read_u32

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:	Wed, 3 Apr 2013 14:14:21 +0200
From:	Linus Walleij <linus.walleij@...aro.org>
To:	Tony Prisk <linux@...sktech.co.nz>
Cc:	Stephen Warren <swarren@...dotorg.org>,
	linux-rpi-kernel@...ts.infradead.org,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] pinctrl: bcm2835: make use of of_property_read_u32_index()

On Thu, Mar 28, 2013 at 7:18 AM, Tony Prisk <linux@...sktech.co.nz> wrote:
> On Wed, 2013-03-27 at 23:09 -0600, Stephen Warren wrote:
>> Use the new standard API of_property_read_u32_index() instead of open-
>> coding it.
>>
>> Signed-off-by: Stephen Warren <swarren@...dotorg.org>
>> ---
>> Note: This depends on the proposed patch "of: Add support for reading
>> a u32 from a multi-value property" by Tony Prisk.
>>
>> BTW, I realized why I didn't create that standard API, but wrote custom
>> prop_u32() instead; the code has already looked up the properties, and
>> validated their length, so prop_u32() can simply index into the property
>> data, whereas of_property_read_u32_index() needs to go search for the
>> property and re-validate it every time, so there's a bunch more overhead.
>> It also means duplicating the property name, although I could use a
>> define for that. I'm not entirely convinced that using this standard API
>> is a win in this case. LinusW, Tony, what do you think?
>> ---
>
> When I was writing the function I had a similar thought about the fact
> we need to work out the length first, which as you mentioned, requires
> all the prior work on the property anyway.
>
> I didn't bring it up, because I thought someone might say 'hey, you
> should add a function to get the count as well' :)
>
> In both the brcm and vt8500 use cases, we will have the issue of
> multiple lookups on the same property using 'generic' functions. Price
> we have to pay for generic code?

My take on this is that if you look in for example:
drivers/of/platform.c you find that for every node in the device
tree when it creates a platform device with a resource it will
for example call of_address_to_resource() twice. The same
for IRQs IIRC.

So basically this behaviour is inherent in all OF/DT code,
and no point trying to work around it.

I don't know if there is a way to fix the problem at the root,
like have the DT parser annotate nodes with the number of
resources of each type (like a decorated tree) so it can be
looked up quickly somehow. I'm not smart enough on DT for
such things...

Yours,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/