IDE IRQ probe brokenness.

IDE IRQ probe brokenness.

[David Woodhouse]

I have on my desk a machine where all PCI interrupts are routed to 
IRQ 0.

The IDE code doesn't seem very happy with it -- it seems to think that
hwif->irq == 0 means that no IRQ has been set. It should be using -1 for
that instead.

This error is in both 2.4 and 2.5.

-- 
dwmw2

Re: IDE IRQ probe brokenness.

[Alan Cox]

On Maw, 2003-06-10 at 09:33, David Woodhouse wrote:
> I have on my desk a machine where all PCI interrupts are routed to 
> IRQ 0.
> 
> The IDE code doesn't seem very happy with it -- it seems to think that
> hwif->irq == 0 means that no IRQ has been set. It should be using -1 for
> that instead.
> 
> This error is in both 2.4 and 2.5.

It isn't clear that -1 is safe either. One thing I would suggest for
your latest weird embedded beastie is that you tweak the IRQ assignment
so IRQ 0 disappears or becomes IRQ 63 or something. IDE does have this
problem, but so does USB and a lot of other code.

Basically we need a "NOT_AN_IRQ" value defined either globally or per
port

Alan

Re: IDE IRQ probe brokenness.

[Russell King]

On Tue, Jun 10, 2003 at 05:42:41PM +0100, Alan Cox wrote:
> Basically we need a "NOT_AN_IRQ" value defined either globally or per
> port

All it needs is for someone to add the definition for NO_IRQ to all
architecture header files.  ARM has an established use of NO_IRQ
already with currently 34 uses:

/*
 * Use this value to indicate lack of interrupt
 * capability
 */
#ifndef NO_IRQ
#define NO_IRQ  ((unsigned int)(-1))
#endif

I'd prefer not to do another mindless time wasteful change just because
someone decides to use a different name.

-- 
Russell King (rmk@arm.linux.org.uk)                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html