lkml.org 
[lkml]   [2004]   [May]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: arm-lh7a40x IDE support in 2.6.6
Date
On Friday 14 of May 2004 23:33, Marc Singer wrote:

> > > Listen. It is not my intention to be clever. All I want to do is get
> > > things working and to not break other people's code. I'm certain that
> > > most people are working with the same assumptions. Aside from some
> > > naive snippets I've see promulgated, the bulk of the kernel work I've
> > > see is sane given limited information. Certainly, once one
> > > understands a sybsystem completely then the quality rises. I hope
> > > you'll admit that the IDE code is overly intricate.
> >
> > With changes like this nobody will ever be able to understand
> > IDE subsystem completely. ;-)
>
> I get the feeling that you're blaming me and others for making the IDE
> code a mess. Might I suggest that this isn't a very productive tack?

Yep, sorry.

> The most helpful thing to do is to a) provide best-practice examples,
> and b) to include some documentation. I'm not talking about anything
> extensive, but statements like
>
> "All references to linux/ide.h must reside in the ide tree."
>
> Are pretty darn helpful.

OK, will do (this can take a while). Can I ask you to review it later?
I think your comments will be very valuable.

> > > > - you are setting IDE_NO_IRQ in ide_init_hwif_ports() which is used
> > > > in many places in generic IDE code - anybody wanting to understand
> > > > interactions with your code + generic code will have serious
> > > > problems (especially if knows _nothing_ about lpd7a40x)
> > >
> > > I don't know what you mean. I grep for that constant and found it
> > > nowhere except for ide-io.c and in my code. It doesn't take much to
> > > find the references.
> >
> > I'm talking about ide_init_hwif_ports() function.
>
> Most of the ARM arch's use it. Perhaps all of them need a good once
> over.

Since some time I have a patch killing <asm/arch-*/ide.h>. :)

> > > > > The OUTB breaks my interface because I don't really have byte-level
> > > > > access to the resgisters. So, is selectproc a pre-select procedure
> > > > > or should it be a substitute?
> > > >
> > > > pre-select but you can change it to be substitute if you need
> > > > (just remember to update all users if you decide to do this)
> > >
> > > I'll have to search the kernel to see what uses it. Maybe the better
> > > way would be to define a new select proc that *is* a substitute.
> >
> > Nope.
>
> So then we break anyone who is using the selectproc as a pre-select
> proc? I don't understand what you mean here. There are several
> drivers that use this function. How do you propose that we provide
> both types of behavior with one entry point?

You can add last line of SELECT_DRIVE() to all ->selectproc()
implementations and add 'else' to SELECT_DRIVE().

> > > of the core code is fine. It is that core code that is necessary to
> > > make the test possible.
> >
> > Stuff in arch-lh7a40x/ide.h is really a driver code but
> > abuses subsystem code instead - that's my complain.
>
> Right. We agree. I am talking about the core code. Not the ide
> code. The core is what supports the CPU.

Good.

Cheers,
Bartlomiej

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:03    [W:0.058 / U:1.052 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site