Re: [MirageOS-devel] Memory exhaustion in Mirage

[Anil Madhavapeddy <anil@xxxxxxxxxx>]

On 24 Dec 2014, at 17:41, David Scott <scott.dj@xxxxxxxxx> wrote:

On Wed, Dec 24, 2014 at 4:58 PM, Anil Madhavapeddy <anil@xxxxxxxxxx> wrote:

Increasing the ring size requires patches to Xen's dom0 as well, and may not be entirely trivial due to the increase from a single page to multiple pages.  I know that these changes did happen for the blkback (and Thomas Leonard added indirect block descriptor support to our blkfront to take advantage), but I don't know if they are supported on dom0's netback (which is much more latency sensitive).

Dave, any idea of the status here?

It looks like dom0 netback in 3.18 still expects one grant ref for each ring:

http://lxr.free-electrons.com/source/drivers/net/xen-netback/netback.c#L1951

Some prototype patches were posted:

http://lists.xen.org/archives/html/xen-devel/2012-01/msg02661.html

but not merged yet

http://lists.xen.org/archives/html/xen-devel/2012-01/msg02709.html

Perhaps we should implement netback and experiment with this Mirage<->Mirage. At the moment we have blkback, blkfront, netfront but no netback. If we had a netback then you could make a Mirage switch which didn't have to bounce all traffic through dom0 (twice)

Doing this over netmap in dom0 might make sense... at the cost of some isolation, but we could assign a 1:1 mapping between netfront/netback which is sufficient for an openflow switch.

-anil

_______________________________________________
MirageOS-devel mailing list
MirageOS-devel@xxxxxxxxxxxxxxxxxxxx
http://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel