On 1 July 2012 17:40, Russell King - ARM Linux [off-list ref] wrote:

On Sun, Jul 01, 2012 at 05:28:16PM +0530, Jassi Brar wrote:

quoted

On 1 July 2012 16:29, Russell King - ARM Linux [off-list ref] wrote:

quoted

On Sun, Jul 01, 2012 at 12:44:01PM +0200, Alessandro Rubini wrote:

quoted

How should I address the problem? (original code, published on
sourceforge was simply replicating a number of amba drivers into pci
drivers, but I don't think massive code duplication is ever sensible,
thus I preferred to reuse the existing drivers).

I think the answer is... those primecell drivers need fixing in some way.
Re-defining CS, DS and ES in drivers is rather silly given that they're
x86 segment register names - so if PL330 can be fixed to make its names
more specific, that sorts it out.

I am OK prefixing the regs with PL330_ or somesuch.

The CS, DS and ES regs were named so as to match exactly the
terminology of the PL330 manual. So apparently even the ARM Ltd didn't
imagine ARM and X86 galaxies colliding so soon :)

Whatever it says in documentation is neither here nor there.  We're humans,
we can re-interpret, and we are capable of adding prefixes to names when
they're stupidly chosen.

Notice how I added MMCI_ and MCI_ to the register definitions for PL180
for example.  I always do that kind of thing as a rule to avoid these kinds
of problems right from the outset, and I don't understand why others don't
do the same.  We've been through soo many "this symbol clashes with
something else in the kernel tree" now that we really should be doing this
automatically, and not waiting for the clash to happen.

I fully agree with your point and IIRC I always add some prefix to
definitions in header files.
Private defines in a .c file, without redundant prefixes, sounded like
safe to me at the time, but perhaps I was wrong.

Thanks.