[lkml]   [2009]   [Jun]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH 9/9] x86/apic: support moorestown interrupt subsystem
On Mon, 29 Jun 2009, Feng Tang wrote:

> On Sat, 27 Jun 2009 01:18:50 +0800
> "Pan, Jacob jun" <> wrote:
> > >> -#ifdef CONFIG_ACPI
> > >> +#if defined(CONFIG_ACPI) || defined(CONFIG_SFI)
> > >
> > >Please add a new helper non-interactive Kconfig symbol instead of
> > >increasing the #ifdef jungle.
> > >
> > [[JPAN]] I agreed, maybe this should be part of the SFI patch or
> > already have similar plans?
> Yes, Ingo has given similar comments for SFI code. I thought about 2
> methods for this:
> 1. these "#ifdef...#endif" covers three functions:
> extern int io_apic_get_unique_id(int ioapic, int apic_id);
> extern int io_apic_get_version(int ioapic);
> extern int io_apic_get_redir_entries(int ioapic);
> how about just completely removing the "#ifdef...#endif", as ACPI/SFI
> codes are heavily used on x86 platforms, and it may bring one hundred
> additional bytes for None-ACPI/SFI platforms
> 2. create a new "CONFIG_ACPI_SFI" option, using this new option to cover
> this case, and let "ACPI"/"SFI" option select it in Kconfig files
> any comments? thanks

I vote #2 -- remove the #ifdefs.
All of these routines are both small and __init,
and the only build that would notice the extra bytes in .text
is the x86 IOAPIC ACPI=n build, which is uncommon today
and becoming more uncommon over time.

The SFI patch series doesn't actually depend on these routines,
the IO-APIC patch depends on them. So this change
should be in the IO-APIC series.

-Len Brown, Intel Open Source Technology Center

 \ /
  Last update: 2009-06-29 18:15    [W:0.034 / U:21.248 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site