On Fri, 8 Mar 2002, Martin J. Bligh wrote:
> >> If you applied an SGI patch that makes the zonelist contain all the zones
> >> of your machine, then the zonelist should not be NULL.
> >> If you allocate memory with gfp_mask & GFP_ZONEMASK == GFP_NORMAL from a
> >> HIGHMEM only node, then the first entry on the corresponding zonelist
> >> should be the first NORMAL zone on some other node.
> >> Am I missing something here ?
> >
> > You're missing the fact that I'm missing the SGI patch ;-)
Oh, I see. I was missing something then...;-)
>
> I should have also mentioned that:
>
> 1) I shouldn't need the SGI patch, though it might help performance.
Why shouldn't you need it ? It is NUMA generic, and totally arch
independent.
And it actually helps performance. I also allows the kernel to have a
single memory allocation path. I think it is cleaner than calling _alloc_pages()
from numa.c
> 2) The kernel panics without my fix, and runs fine with it.
I hope so :-)
But your fix is at the same time useless and harmless for UMA machines.
OTOH, the SGI patch doesn't modify __alloc_pages(). I think I'm a little
too picky here...
Cheers,
Samuel.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Mar 15 2002 - 22:00:10 EST