linux-kernel - Re: [PATCH v8] clk: Add (devm_)clk_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] [day] [month] [year] [list]

Date:   Fri, 30 Nov 2018 10:55:48 -0800
From:   Stephen Boyd <sboyd@...nel.org>
To:     Phil Edworthy <phil.edworthy@...esas.com>,
        Russell King - ARM Linux <linux@...linux.org.uk>
Cc:     Michael Turquette <mturquette@...libre.com>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Geert Uytterhoeven <geert@...ux-m68k.org>,
        Uwe Kleine-König <u.kleine-koenig@...gutronix.de>,
        "linux-clk@...r.kernel.org" <linux-clk@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [PATCH v8] clk: Add (devm_)clk_get_optional() functions

Quoting Russell King - ARM Linux (2018-11-30 03:04:46)
> On Fri, Nov 30, 2018 at 10:25:37AM +0000, Phil Edworthy wrote:
> > On 30 November 2018 09:09 Stephen Boyd wrote:
> > > Quoting Phil Edworthy (2018-11-20 06:14:45)
> > > > This adds clk_get_optional() and devm_clk_get_optional() functions to
> > > > get optional clocks.
> > > > They behave the same as (devm_)clk_get except where there is no clock
> > > > producer. In this case, instead of returning -ENOENT, the function
> > > > returns NULL. This makes error checking simpler and allows
> > > > clk_prepare_enable, etc to be called on the returned reference without
> > > > additional checks.
> > > 
> > > Ok. I guess that works by virtue of how -ENOENT is returned by various
> > > functions that are called deeper in the clk_get() path? I'm cautiously
> > > optimistic. So cautious, we should probably add a comment to these optional
> > > functions that indicate they rely on the functions they call to return -ENOENT
> > > under the various conditions that make a clk optional.
> > Yes, it does indeed rely on how clk_get() is implemented.
> > Specifically, that if __of_clk_get_by_name() returns -EINVAL, the error is
> > superseded by clk_get_sys() returning -ENOENT.
> > As you say, a comment may help here.
> 
> Each time the question of the optional clk_get() stuff comes up, we go
> around the same discussions time and time again.  So far, each time
> has ended up flopping.
> 
> Yes, clk_get() can only ever return -ENOENT if it falls back to the
> non-DT methods, because it assumes that the clk tables are complete
> (it can do nothing else.)
> 
> I don't think it needs a comment because it's obvious from the code
> and also from the implementation point of view.

Ok. I would still suggest a comment on of_clk_get_by_name() that
indicates what types of return values happen in there. Otherwise it's
obvious from the code after reading 3 or 4 functions deep
(__of_clk_get_by_name -> __of_clk_get -> of_parse_phandle_with_args and
__of_clk_get_from_provider) that this is what happens.

> 
> > > > +static inline struct clk *clk_get_optional(struct device *dev, const
> > > > +char *id)
> > > 
> > > Any kernel doc for this function?
> > I took my cue from the surrounding functions, let me know if I have to add it.
> 
> I don't see you need to - this is an internal function by way of the
> "static inline" you have before it.  It's not an API function.

It's static inline in a header file. That is an API function as far as I
can tell.