Re: [GIT PULL] Early boot SLAB for 2.6.31

[Ingo Molnar <mingo@elte.hu>]

* Pekka Enberg <penberg@cs.helsinki.fi> wrote:

> Hi Ingo,
>
> Ingo Molnar wrote:
>>>>> What kind of conflicts are there against -tip? The diffstat  
>>>>> suggests it's mostly in-SLAB code, right? There shouldnt be much 
>>>>> to conflict, except kmemcheck - which has more or less trivial 
>>>>> callbacks there.
>>>> The conflicting bits are the patches that remove bootmem allocator 
>>>> uses in arch/x86 and kernel/sched.c.
>>> Give me an hour and i'll get some minimal testing done.
>>
>> This tree doesnt conflict (not even with kmecheck) - and the older  
>> bits you sent against the scheduler and against x86 doesnt apply  
>> anymore - but they do look scary.
>
> Btw, yeah, it doesn't conflict because I dropped the problematic patches  
> and did the bootmem fallback instead.
>
> But now you know why I tried to push all this to -tip. Your tree 
> is moving so fast that it's difficult to generate patches that 
> apply to both, -tip and mainline, in this particular area :-).

Hey, i'd agree normally, but the scheduler tree was very quiet in 
this cycle, for a change :-)

The main "problem" here really is the multi-tree impact of such 
broad changes. Those are best kept in a tree like -mm, which goes on 
top of all other trees and is thus basically the only tree that can 
do tree-wide changes.

Anyway, if you rebase to latest -git it should be fine - Linus 
pulled the scheduler and x86 bits.

	Ingo