Re: [PATCH 4/5] Centralise NO_IRQ definition

[Benjamin Herrenschmidt]

> The fact is, 0 _is_ special. Not just for hardware, but because 0 has a 
> magical meaning as "false" in the C language.

I don't agree, irq 0 has been a valid irq on a number of platforms for
ages (including your own G5, at least some of them have the SATA irq at
0 :) and this didn't cause any problem for most drivers. The few ones
that have done broken assumption have been easily fixed using NO_IRQ.

"Translating" it means some ugly translation work all over the place,
which means overhead in the interrupt path (ok, not that much but
still), plus finding some magic number to put 0 on, which makes things
much more complicated for archs that have interrupts sorted in nice
blocks of power of two, etc... a significant burden on arch/PIC code for
no good reason imho.

I hate arbitrary "translations" when they aren't strictly necessary.
It's common to have a constant for a "not valid" number in spaces where
"0" is a valid value, I don't think that "looking simpler" to dump
driver writers is worth it in this case.

Ben.


-
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/