Re: workqueue cpu affinity

[Max Krasnyanskiy <maxk@qualcomm.com>]

Peter Zijlstra wrote:
> Sorry for being late,.. and I'm afraid most will have to wait a bit
> longer :-(
> 
> On Wed, 2008-06-11 at 12:02 -0700, Max Krasnyansky wrote:
>> Peter Zijlstra wrote:
> 
>>> Please don't take this too personal - I'm glad you're working on this.
>>> I'm just trying to see what we can generalize.
> 
>> Oh, no worries, I'm not taking this personally, except maybe the "most your
>> ideas suck" part which got me a little bit ;-). I'm definitely all for making
>> it suitable for more general usage.
>> This is actually first constructive criticism (except for the "most of your
>> ideas suck" part :). 
> 
> No, no, you understand me wrong (or I expressed myself wrong). Your
> ideas are good, its just the implementation / execution I have issues
> with.
> 
> Like with extending the isolation map, what didn't leave any room for
> hard-rt smp schedulers or multiple rt domains. Whereas the cpuset stuff
> does.
Yep. And I redid it completely and switched gears to fix/improve cpusets, 
genirq, etc.
Anyway, I think we're on the same page. Please look over the summary that I 
sent out and see if I missed anything.

Max