linux-kernel - Re: [PATCH v4 1/2] clk: exynos5433: do not use CLK_IGNORE

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

Date:	Thu, 07 Jul 2016 17:20:35 +0200
From:	Sylwester Nawrocki <s.nawrocki@...sung.com>
To:	Andi Shyti <andi.shyti@...sung.com>
Cc:	Krzysztof Kozlowski <k.kozlowski@...sung.com>,
	Chanwoo Choi <cw00.choi@...sung.com>,
	Jaehoon Chung <jh80.chung@...sung.com>,
	Tomasz Figa <tomasz.figa@...il.com>,
	Michael Turquette <mturquette@...libre.com>,
	Stephen Boyd <sboyd@...eaurora.org>,
	Kukjin Kim <kgene@...nel.org>,
	linux-samsung-soc@...r.kernel.org, linux-clk@...r.kernel.org,
	linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
	Andi Shyti <andi@...zian.org>
Subject: Re: [PATCH v4 1/2] clk: exynos5433: do not use CLK_IGNORE_UNUSED for
 SPI clocks

On 07/07/2016 03:27 PM, Andi Shyti wrote:
>>> > > The CLK_IGNORE_UNUSED flag has to be avoided whenever possible.
>>> > > Use the CLK_IS_CRITICAL flag instead for critical SPI1 clocks,
>>> > > which enables the clock line during boot time.
>> > 
>> > I don't agree. Both flags should be avoided. Clk is critical does not
>> > solve the problem. It is just a better workaround for lack of proper
>> > clock consumers.
>> > 
>> > The IOCLK is not a critical clock. It can be disabled (e.g. when SoC
>> >  is used on a board without any SPI device connected).
>
> As we discussed offline there is no driver which is requesting
> this clock. We cannot ask to the spi driver to handle three
> clocks because the exynos5433 has its own peculiarities.
> 
> If we want this on the spi driver's side, we need to add a new
> compatible and check it everytime. To me it looks uglier than
> just keep it alive.

I took a closer look at what those IOCLK clocks exactly are and
unfortunately I must agree with Krzysztof.  These clock gates are
closely related to the IP blocks and to me proper approach is to
have them listed in DT and controlled by the SPI bus driver.

I checked and there is similar pattern for other IPs like I2S and
other SoCs, e.g. exynos7420.
Additionally parents of those IOCLK_SPI?_CLK clocks are currently
wrongly modelled as fixed rate clocks.  These clocks really don't
have a parent until some clock is fed externally to the SoC's I/O
pin.  But this issue could be addressed later.

I think it is not a big deal to add "samsung-exynos5433-spi"
compatible to the SPI driver along with a new variant data and
a flag like "has_cmu_ioclk" to indicate whether a third clock
should be handled or not. Presumably for now the ioclk clock can
just simply be enabled in probe(), this way it will be enabled
only for SPI controllers actually in use.