Messages in this thread | | | Date | Mon, 27 Oct 2008 17:37:47 +0900 | Subject | Re: [PATCH 2/2] x86: replace BIO_VMERGE_BOUNDARY with BIOVEC_PHYS_MERGEABLE | From | FUJITA Tomonori <> |
| |
On Mon, 27 Oct 2008 19:21:41 +1100 Jeremy Fitzhardinge <jeremy@goop.org> wrote:
> FUJITA Tomonori wrote: > > The block layer always done the physical merge if possible. We don't > > provide any kernel parameter to disable it. > > > > The iommu_bio_merge parameter had been used to enable the virtual > > merge. As I wrote, the virtual merge feature was completely > > removed. Effectively, the iommu_bio_merge parameter is meaningless > > now. > > > > Under Xen, pages which appear to be pseudo-physically adjacent are not > necessarily really physically adjacent. We need to hook > BIOVEC_PHYS_MERGEABLE to prevent the bio layer from inappropriately > merging requests across non-contiguous page boundaries.
I'm not familiar with what Xen does but why can't Xen just override BIOVEC_PHYS_MERGEABLE?
Why does Xen need to hook BIOVEC_PHYS_MERGEABLE to the iommu_bio_merge parameter (as this patch does)? BIOVEC_PHYS_MERGEABLE and the iommu_bio_merge parameter are not related at all.
| |