Messages in this thread | | | Date | Fri, 26 Jun 2009 16:29:37 +0200 | From | Ingo Molnar <> | Subject | Re: [PATCH 3/9] x86/moorestown: add moorestown platform flags |
| |
* Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
> > The thing is, you are trying to defend a v1 patch-set here that > > is really indefensible: it's ugly and deficient in numerous > > smaller and larger details. I outlined numerous deficiencies > > already - and i'll review v2 too to see what else is there to > > fix. > > No I'm trying to understand what you actually want the thing to > look like.
It's a case by case thing and i pointed out a few specific directions in the review. The IO-APIC changes should probably go on top of Jeremy's IO-APIC driver-ization patches. They dont necessarily need their own IO-APIC driver (if the resulting line count increase is too much), but they should not wreck Jeremy's IO-APIC patches.
Bootup quirks that are small modificatons to existing PC initialization sequences should go into x86_quirks.
Timer related changes (the APB system timer) are mostly modular already by virtue of half of it being a clocksource and clockevents driver. The remaining bit of system timer handling should be abstracted out as a 'struct x86_system_timer' kind of structure, with ->init, ->timer_irq and ->shutdown functions.
[ Although it is beyond me why ABP was done - why wasnt HPET good enough? HPET can do per CPU clockevents too and it's just as off-chip (and hence fundamentally slow) as ABP. ]
Ingo
| |