Re: [PATCH] clk: return proper ERR_PTR for clk_get when !HAVE_CLK

[Russell King - ARM Linux <linux@arm.linux.org.uk>]

On Mon, Feb 09, 2015 at 03:29:54PM +0100, Wolfram Sang wrote:
> ... which has to do with this. Historically, "clock-frequency" in DT
> means the i2c bus speed. This driver used the binding wrong to specify
> the IP core clock speed instead and also used a fixed I2C bus speed
> then. Now, when wanting to support a) flexible I2C bus speeds and b)
> clock information from standard DT clock bindings, we also need to
> ensure that we support the now deprecated and wrong binding as a
> fallback. And even after having learnt something about the clk API now,
> I still think the best fix is to replace all instances of 'if
> (!IS_ERR(dev->clk))' with (!IS_ERR_OR_NULL()). Because in our case, NULL
> is not success. We should skip any clk API interaction then and use the
> fallback mechanisms.

I disagree; I also utterly _hate_ IS_ERR_OR_NULL - this macro has been
responsible for a /lot/ of error handling bugs, and I wish the macro
would die.

The "test-for-zero-clock-rate" method I showed in my previous email
will work for you - and will avoid using this hateful macro while
still giving the behaviour you desire, and you won't be caring about
the detail of the clk API implementation either.

-- 
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.