overriding ioops

overriding ioops

[Marc Singer]

I know that you me to do this differently that I did in 2.6.4.  I need
to override the hwif->OUTB (et, all) functions.  Do I implement my own
version of the default_hwif_ function?  

Re: overriding ioops

[Bartlomiej Zolnierkiewicz]

On Saturday 05 of June 2004 22:19, Marc Singer wrote:
> I know that you me to do this differently that I did in 2.6.4.  I need
> to override the hwif->OUTB (et, all) functions.  Do I implement my own
> version of the default_hwif_ function?

No, arch specific code shouldn't touch generic one.
Your host driver should take care of this.

If you need some changes in generic code please describe them
and I'll try to fix/enhance it.

[ I know about polling but this is really non trivial change as i.e.
  we have 4 versions of IDE PIO code (!) - I'm cleaning this mess now. ]

Thanks,
Bartlomiej

Re: overriding ioops

[Marc Singer]

On Sat, Jun 05, 2004 at 10:42:19PM +0200, Bartlomiej Zolnierkiewicz wrote:
> On Saturday 05 of June 2004 22:19, Marc Singer wrote:
> > I know that you me to do this differently that I did in 2.6.4.  I need
> > to override the hwif->OUTB (et, all) functions.  Do I implement my own
> > version of the default_hwif_ function?
> 
> No, arch specific code shouldn't touch generic one.
> Your host driver should take care of this.

I apologize for the redundant question.  After I send the first
message, I couldn't find it in my sent folder so I thought I might
have deleted it.

> 
> If you need some changes in generic code please describe them
> and I'll try to fix/enhance it.
> 
> [ I know about polling but this is really non trivial change as i.e.
>   we have 4 versions of IDE PIO code (!) - I'm cleaning this mess now. ]

OK.  I'll implement something for now and wait for you to get a handle
on it before submitting a patch.

> 
> Thanks,
> Bartlomiej

Re: overriding ioops

[Marc Singer]

On Sat, Jun 05, 2004 at 10:42:19PM +0200, Bartlomiej Zolnierkiewicz wrote:
> On Saturday 05 of June 2004 22:19, Marc Singer wrote:
> > I know that you me to do this differently that I did in 2.6.4.  I need
> > to override the hwif->OUTB (et, all) functions.  Do I implement my own
> > version of the default_hwif_ function?
> 
> No, arch specific code shouldn't touch generic one.
> Your host driver should take care of this.
> 
> If you need some changes in generic code please describe them
> and I'll try to fix/enhance it.
> 
> [ I know about polling but this is really non trivial change as i.e.
>   we have 4 versions of IDE PIO code (!) - I'm cleaning this mess now. ]

I wonder if we're talking about the same thing.  My platform does not
have DMA nor does it have an IRQ.  PIO, as I understand it, refers to
programmed IO and not to polling proper.  Do you intend to support
polling?

Re: overriding ioops

[Bartlomiej Zolnierkiewicz]

On Sunday 06 of June 2004 00:25, Marc Singer wrote:
> On Sat, Jun 05, 2004 at 10:42:19PM +0200, Bartlomiej Zolnierkiewicz wrote:
> > On Saturday 05 of June 2004 22:19, Marc Singer wrote:
> > > I know that you me to do this differently that I did in 2.6.4.  I need
> > > to override the hwif->OUTB (et, all) functions.  Do I implement my own
> > > version of the default_hwif_ function?
> >
> > No, arch specific code shouldn't touch generic one.
> > Your host driver should take care of this.
> >
> > If you need some changes in generic code please describe them
> > and I'll try to fix/enhance it.
> >
> > [ I know about polling but this is really non trivial change as i.e.
> >   we have 4 versions of IDE PIO code (!) - I'm cleaning this mess now. ]
>
> I wonder if we're talking about the same thing.  My platform does not
> have DMA nor does it have an IRQ.  PIO, as I understand it, refers to
> programmed IO and not to polling proper.  Do you intend to support
> polling?

I am speaking about PIO mode data transfers not PIO register access
and I intend to use polling instead of IRQs for them.

It requires some time as I need to clean existing mess first:
- task_*() and flagged_task_*() in ide-taskfile.c
- read_intr() and write_intr() in ide-disk.c
- etc.