William Lee Irwin III <wli@holomorphy.com> writes:
> William Lee Irwin III <wli@holomorphy.com> writes:
> >> Seriously, just plop down the fresh zone type and all will be well.
> >> It's really incredibly easy.
>
> On Tue, Jan 28, 2003 at 12:28:04AM -0700, Eric W. Biederman wrote:
> > I will certainly take a look, tracing through that code can get a little
> > hairy.
>
> It can really be approached much more cavalierly than that. The only
> extant example aside from the original ZONE_DMA32 implementation I've
> seen is Simon Winwood's MPSS patch, which needed something on the order
> of 10 lines of code for a fresh zone type (for one arch).
>
> And most of the bulk of the ZONE_DMA32 implementation was stringing up
> the block layer to utilize it, not inserting the new zone type itself.
Primarily it appears that just another ZONE needs to be added, and then
free_area_init needs to be passed the proper parameters.
I still want to look closely at how the discontig mem case for NUMA is
setup. It is probably nothing to worry about but I want to make
certain it does not have any perverse behavior and also I want to be
certain I know how to setup a NUMA system properly, since I am looking
at the code anyway.
Eric
-
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 Jan 31 2003 - 22:00:19 EST