[PATCH 12/20] Drop FIQ dependency for GTA01 configuration.

[Werner Almesberger]

Jonas Bonn wrote:
> However, it does draw attention to the fact that a lot of code is
> being thrown in that isn't being tested across the permutations of
> build options that are available.

Yep, see the sub-thread on testing more permutations of the config
options. it's not only GTA01 vs. GTA02, but also things like turning
off a subsystem you don't need or that gets in the way of some
development work. (E.g., I hit this during my WLAN-SPI experiments,
when I simply wanted everything else that would printk things with
"mmc" or similar in it gone.)

> Anyway, I don't mean to rant... I agree with
> every word you write.  I just don't think it's worth ragging on this
> patch as it's just fixing up (poorly, I agree) what was already
> broken.  The style issues can be fixed and I will do so at first best
> opportunity.

Great, thanks ! By the way, that's also why it's important to set good
examples. You know when you're doing something that bends the rules and
plan to clean it up, but the next guy may just copy the bad example,
and if this isn't caught in the review, there goes the code rot.

> Anyway, thanks Werner, for the style tutorial... it is well written.
> I think we should post it on the wiki.

There is actually already pretty good material around. This specific
topic is covered in Documentation/SubmittingPatches, section 2, item
2 "#ifdefs are ugly".

Andi Kleen's OLS 2008 presentation "On submitting kernel patches" is
also worth a read. It's at page 253 of
http://www.linuxsymposium.org/2008/ols-2008-Proceedings-V1.pdf

- Werner