Re: linux-next: manual merge of the rr_cpumask tree

[Mike Travis <travis@sgi.com>]

Rusty Russell wrote:
> On Tuesday 16 December 2008 02:57:09 Mike Travis wrote:
>> Rusty Russell wrote:
>>> On Monday 15 December 2008 16:39:45 Stephen Rothwell wrote:
>>>> Hi Rusty,
>>>>
>>>> Today's linux-next merge of the rr_cpumask tree got a conflict in
>>>> arch/x86/kernel/io_apic.c between commit
>>>> 3145e941fcfe2548fa2270afb1a05bab3a6bc418 ("x86, MSI: pass irq_cfg and
>>>> irq_desc") from the sparseirq tree and commit
>>>> 0de26520c7cabf36e1de090ea8092f011a6106ce ("cpumask: make irq_set_affinity
>>>> () take a const struct cpumask") from the rr_cpumask tree.
>>>>
>>>> There are lots of overlapping changes.  I fixed it up (see below) and can carry the
>>>> fix as necessary.
>>> I have performed this merge before.  The results are in
>>> 	git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-ingo.git
>>>
>>> Ingo, can you pull these into cpus4096?  It's just the cpumask/cpus4096 merge.
>>>
>>> Thanks,
>>> Rusty.
>> Hi Rusty,
>>
>> Did you update your tip/cpus4096 branch?
> 
> No.  As I said, I just did a merge of ingo's 4096 tree with the cpumask tree.
> He should pull that, and then the rest of the x86 patches can go on top.
> 
> Cheers,
> Rusty.

Hi Rusty,

Yes, I did finally figure that out.  Thanks for rebasing the part that Ingo needed
so maybe the rest can be pushed.  There was a problem when the 'for-ingo' tree was
merged as there were mixed up API calls in io_apic.c, but I fixed them in the first
x86 only patch.  I wasn't sure of what else to do.

All the x86 patches build and boot (and test! ;-) on x86_64.  My 32-bit box has gone
into a coma until I can figure out what's up (and the 64-bit box lost it's 32-bit
root partition.)  Since they are pretty much exactly as you had tested last week,
I not expecting any problems.

I'm working through the remainder of your cpumask patches now.

Thanks!
Mike