Re: unsigned long ioremap()?

[Jeff Garzik <jgarzik@mandrakesoft.com>]

Abramo Bagnara wrote:
> "David S. Miller" wrote:
> > There is a school of thought which believes that:
> >
> > struct xdev_regs {
> >         u32 reg1;
> >         u32 reg2;
> > };
> >
> >         val = readl(&regs->reg2);
> >
> > is cleaner than:
> >
> > #define REG1 0x00
> > #define REG2 0x04
> >
> >         val = readl(regs + REG2);

> The problem I see is that with the former solution nothing prevents from
> to do:
> 
>         regs->reg2 = 13;

Why should there be something to prevent that?

If a programmer does that to an ioremapped area, that is a bug.  Pure
and simple.

We do not need extra mechanisms simply to guard against programmers
doing the wrong thing all the time.


> That's indeed the reason to change ioremap prototype for 2.5.

Say what??

I have heard a good argument from rth about creating a pci_ioremap,
which takes a struct pci_dev argument.  But there is no reason to change
the ioremap prototype.

	Jeff


-- 
Jeff Garzik      | Game called on account of naked chick
Building 1024    |
MandrakeSoft     |