[PATCH 2/2] I/O port access routines

[Luc Verhaegen]

On Thu, Nov 19, 2009 at 11:21:40AM +1100, Daniel Stone wrote:
> On Thu, Nov 19, 2009 at 12:37:10AM +0100, Luc Verhaegen wrote:
> > So how would we go about using any of this then? Do we get to have 
> > another huge API/ABI breakage like when libpciaccess was introduced just 
> > because the normal in/out routines should no longer be used?
> 
> It's not even an ABI break, let alone API.  If you read the patch,
> you'll note that new calls are being added, and nothing modified;
> drivers which choose to make use of these calls can make use of them.
> If you want to be compatible back to 6.9, you can ifdef it, or make a
> wrapper if you really care that much.
> 
> > Oh, and that's before i start wondering about gratuitous void pointers 
> > and goto.
> 
> There's absolutely nothing wrong with void pointers (what do you
> recommend instead?), and goto is used in one place, as an error path to
> unwind, which is a very common coding practice (cf. the kernel) which
> avoids an entire class of errors related to failing to free things on
> an error return, mainly because you have the same unwind code duplicated
> in 17 different places.

Well, that one goto is rather gratuitous.

But the main point of this mail was ignored. Why did existing xorg pci 
infrastructure have to be reinvented like that instead of adding a new 
backend and fixing up the bad patches? Why did RAC get thrown away like 
that? Why does this NIH have to keep on repeating itself?

Luc Verhaegen.