[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] SWIOMMU redundant copies?



Yes, the swiotlb check could be relaxed. Do we really do many mergeable
page-straddling I/Os in practice though? I would expect most block I/O to be
scattered by the page cache.

 -- Keir

On 14/3/08 15:50, "Stephen C. Tweedie" <sct@xxxxxxxxxx> wrote:

> So we'll create bios which span multiple pages, assuming that that's
> more efficient; but then we'll force them to be copied because they span
> multiple pages!
> 
> Have I missed something, or is this an oversight in the swiotlb.c patch
> above?
> 
> The reason I discovered this is that including the patch seems to be
> causing SWIOTLB overflows on some hardware.  Adding an extra test to
> swiotlb_map_sg() to catch sg segments which are truly machine-contig and
> don't need mapping, and avoid copying those, cures those symptoms for
> me, but I suspect there may be other issues lurking in this area.



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.