Messages in this thread | | | Subject | Re: 2.6.10-rc2-mm4 | From | Jes Sorensen <> | Date | 02 Dec 2004 03:03:05 -0500 |
| |
>>>>> "William" == William Lee Irwin <wli@holomorphy.com> writes:
William> On Tue, 2004-11-30 at 10:21 -0800, Andrew Morton wrote: >>> For pages which have a physical address <4G. I assume this was >>> motivated by the lack of an IOMMU on ia32e?
William> On Tue, Nov 30, 2004 at 07:25:10PM +0100, Arjan van de Ven William> wrote: >> but there's the swiommu for those... so that can't be it >> realistically.... Is there code using the zone GFP mask yet ??
William> ZONE_NORMAL and ZONE_DMA are both too overloaded to handle William> the 4GB boundary. And it makes a lot of sense on more machine William> types than x86-64 (e.g. ia64, ia32 and others with 32-bit PCI William> but no zone representing it).
Hardware that doesn't require the broken 24 bit ISA zone already uses ZONE_DMA for all 32 bit memory - ie. ia64 does this. It works even on the broken-beyond-repair Intel boxes without an IOMMU.
Cheers, Jes - 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/
| |