Re: [PATCH 4/9 UPDATED-1] x86: Initialize 32bit logical apicid mappingearly during boot

From: Tejun Heo
Date: Thu Nov 18 2010 - 03:48:37 EST


Hello,

On 11/18/2010 09:42 AM, Ingo Molnar wrote:
> This is a very sensitive area of code that tends to blow up in nasty
> ways. So when we do changes here we want them super-finegrained. If
> you split it up into 30 reasonable patches - no problem at
> all. (here up to 5 would suffice i think)

I'll probably split it to three or four. That said, I don't really
think it would help bisection. But, anyways, no problem.

>> Yeah, what they do is ugly. It gets less uglier after the patchset.
>> I'll see if some of them can be dropped but I don't think putting them
>> inside ifdef'd inline functions necessarily improves things. It often
>> just makes things more difficult to follow.
>
> Can we remove them? Or does it make any sense on the 64-bit side?

Those are the parts where 32 and 64 actually differ. On 64, the
mapping between physical and logical are fixed (at least in the
generic arch code) which isn't true on 32, so it doesn't make much
sense for 64. For now, I would suggest leaving them ugly. At least
they're now properly localized after the patchset.

Thanks.

--
tejun
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/