Re: [PATCH] fix platform_rename_gsi related ia32 build breakage

[Len Brown <len.brown@intel.com>]

On Thu, 2004-11-11 at 17:30, Linus Torvalds wrote:
> 
> On Thu, 11 Nov 2004, Len Brown wrote:
> >
> > I used a function pointer here because the same kernel binary must
> be
> > able to run on an ES7000 or a non-ES7000, so the compile-time inline
> > idiom doesn't work.
> 
> Sure it does. Do something like this in a header file
> 
>         static inline int translate_irq_number(...)
>         {
>                 #ifdef CONFIG_ACPI_BOOT
>                         return fn_ptr_xxx();
>                 #else
>                         return irq;
>                 #endif
>         }

sure, we could add a wrapper for the wrapper,
but the ifdefs are already gone without doing this --
platform_rename_gsi is present with or without ACPI.

I suppose I could shrink the kernel by 4-bytes by compiling out the
function pointer when IO_APIC is not defined.  I'll be happy to optimize
for that case if you think it justifies the code; though if I were
optimizing for that case, this probably isn't where I'd start.

> which means that yes, it uses the function pointer when it is
> meaningful, but if there is no point, the code just goes away.
> 
> > If you read this far and have suggestions for a more descriptive
> name than platform_rename_gsi(), just let me know.
> 
> At _least_ write out what the hell "gsi" is.
> 
> TLA's are bad. "gsi" apparently isn't the Geological Survey of
> Ireland, but that's all I can tell from google.

The _gsi in platform_rename_gsi was consistent with the surrounding use
in the ACPI case.  I decided to re-use the same funtion for the MPS case
for simplicity, even though io_apic.c uses _irq.  If you like, I can add
a synonym using an inline for _irq, but I thought we were moving away
from using _irq, not towards it.

GSI = Global System Interrupt
IRQ = overused so much it means nothing at all

While this fact may not win you over as a fan, GSI is actually
consistent with the language in the ACPI spec.  I don't know if that is
where Bjorn came up with the name, but I do think it was a positive
change.  The reason is that "global" actually means something -- and it
still has the same semantics no matter if your machine passes around cpu
interrupt vectors or if it passes around pin numbers.

cheers,
-Len