Re: Linux 2.4.20-rc4-ac1

[Adrian Bunk]

On Thu, Nov 28, 2002 at 08:15:02AM -0800, Martin J. Bligh wrote:

> > Martin, could you please review whether the patch below which does the
> > following is correct?
> > 
> > - kill the last occurence of CLUSTERED_APIC_NUMAQ
> > - only one definition of PCI_CONF1_ADDRESS is needed
> >   (#ifndef CONFIG_MULTIQUAD the BUS2LOCAL() has no effect)
> > - fix an #endif comment
> 
> I'm confused at to what made this break recently .... what release 
> did this break on? Either it was something very recent, everyone 
> else is asleep, or you're doing something wierd (like compling with
> MULTIQUAD on?)

Yes, I tried a compilation with MULTIQUAD on.

The "very recent" might be that I'm talking about -ac (look at the
Subject).

> CLUSTERED_APIC_NUMAQ is a part of the Summit patch introduction, 
> where John (cc'ed) splits clustered_apic_mode from a boolean into
> a switch (in 2.4 ... I did 2.5 differently). Maybe we merged half
> of a summit patch somehow, and didn't get the define?
>...

First of all:
I noticed this problem while doing a compile-only test, I have no real
knowledge of the code.

In 2.4.20-pre the following change was made in several places:
  clustered_apic_mode -> (clustered_apic_mode == CLUSTERED_APIC_NUMAQ)

In 2.4.20-pre-ac they were changed again, this time
  (clustered_apic_mode == CLUSTERED_APIC_NUMAQ) -> clustered_apic_logical

According to grep the occurence of CLUSTERED_APIC_NUMAQ my patch changes
was the last one _in the whole kernel sources of 2.4.20-rc4-ac1_.

It might be that my patch is wrong but there's undoubtable something
that needs to be fixed.

> M.

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed