Re: [PATCH 1/2]: serial8250: Use native_io_delay on the x86

From: Alan Cox
Date: Wed Mar 17 2010 - 14:36:22 EST


On Wed, 17 Mar 2010 11:25:46 -0700
"H. Peter Anvin" <hpa@xxxxxxxxx> wrote:

> On 03/17/2010 05:30 AM, Simon Kagstrom wrote:
> > #ifdef __i386__
> > - outb(0xff, 0x080);
> > + native_io_delay();
> > #endif
>
> There is something a lot more weird about this. First of all, it's
> #ifdef'd out on all but __i386__ including x86-64; second, it looks like
> this is specific to synchronizing to the IER.
>
> I'm wondering if the right thing isn't to add a dummy write to the SCR
> register (or similar.)

You just need a write to something on the ISA bus which is 'safe' so that
you don't end up reading back what you wrote to an non-existant port as
some old chipsets will return the last ISA result when you do this rather
than 0xff.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/