Re: [Xen-devel] workaround for dom0 crash due to QEMU using O_DIRECT

[Alex Bligh <alex@xxxxxxxxxxx>]

--On 8 July 2013 15:18:53 -0400 Konrad Rzeszutek Wilk 
<konrad.wilk@xxxxxxxxxx> wrote:
> > speaking with Ian about the dom0 kernel crash caused by using O_DIRECT
> > in QEMU, we came up with a simple workaround that should turn the crash
> > into a data corruption problem (same as native).
> <chuckles> You should for fun also do 0xEE on the 'trade_page' whenever
> we update the PTE. That way we can detect the corruption as by default
> the trade_page would be 00.
Ha ha I've just read the patch. If I read right, what happens is there's
a fixed 'junk' page which gets mapped in whenever the granted page gets
mapped out.

Let's put aside the minor issue here that we've got a kernel patch
which doesn't actually fix the kernel's problem :-)

Have we not got a danger here that trade_page could end up written
to with VM A's data, and this could then find itself in VM B's disk?
Or do we know that every access by the kernel after withdrawal
of the grant is guaranteed to be a read? In which case making it read
only might be safer.

Also, our normal config has dom0 with completely fixed memory (no
ballooning) I believe. Is that something Diana needs to change when
testing this?

--
Alex Bligh

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel