linux-kernel - Re: Why do we need reset_control_get

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]

Message-ID: <1469703408.12835.30.camel@pengutronix.de>
Date:	Thu, 28 Jul 2016 12:56:48 +0200
From:	Philipp Zabel <p.zabel@...gutronix.de>
To:	Arnd Bergmann <arnd@...db.de>
Cc:	Masahiro Yamada <yamada.masahiro@...ionext.com>,
	Hans de Goede <hdegoede@...hat.com>,
	Lee Jones <lee.jones@...aro.org>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>,
	Axel Lin <axel.lin@...ics.com>,
	Maxime Ripard <maxime.ripard@...e-electrons.com>
Subject: Re: Why do we need reset_control_get_optional() ?

Am Donnerstag, den 28.07.2016, 12:09 +0200 schrieb Arnd Bergmann:
> On Thursday, July 28, 2016 11:43:00 AM CEST Philipp Zabel wrote:
> > > I want to deprecate _optional variants in the following steps:
> > > 
> > > [1] Add "depends on RESET_CONTROLLER" to drivers
> > >     for which reset_control is mandatory.
> > > 
> > >     We can find those driver easily by grepping
> > >     the reference to non-optional reset_control_get().
> > 
> > Since we have the stubs, the RESET_CONTROLLER dependency is only at
> > runtime, not at build time.
> > 
> > I think Arnd wanted to move this in the opposite direction and remove
> > the configurable RESET_CONTROLLER symbol. Maybe we should let all
> > drivers that currently request non-optional resets have:
> >         depends on (ARCH_HAS_)RESET_CONTROLLER || COMPILE_TEST
> > ?
> 
> There are various ways to improve the current situation.
> 
> I think it's important that a driver that has an optional
> reset line behaves in exactly the same way whether the reset
> subsystem is enabled or disabled when no reset line is
> provided for a machine.

Agreed.

> When a driver requires a reset line, we can either have a
> build-time failure when the reset subsystem is disabled
> (enforcing the Kconfig dependency),

That would have been the case before commit 5bcd0b7f3c56 ("reset: Add
(devm_)reset_control_get stub functions"). Now that we allow compiling
all drivers with the stubs, I suppose option B is the way to go:

> or cause a runtime failure if either there is no reset line or
> the subsystem is disabled.

> In my experimental patch, I make the _optional functions
> return NULL if no "resets" property is provided but return
> an error if there are reset lines but the subsystem is
> disabled, i.e. an optional reset must be used if it's in the
> DT, but can be ignored otherwise.

I have to admit I don't understand the purpose of that.
reset_controller_get_optional returns -ENOENT if there's no reset in the
DT, and all drivers act accordingly.

regards
Philipp