On Mon, Sep 10, 2007 at 05:22:35AM +0200, Tristan Gingold wrote:
> On Mon, Sep 10, 2007 at 11:31:23AM +0900, Isaku Yamahata wrote:
> > On Mon, Sep 10, 2007 at 09:32:42AM +0800, Zhang, Xing Z wrote:
> > > I think Isaku have the answer. Is it related to MCA?
> >
> > No. It's for xm dump-core.
> > When I wrote the patch c/s 3c8039aa5004,
> > I thought the area should also be dumped to the file.
> > The fact that it uses v->arch.privregs seconded.
> >
> > Yes, the guest shouldn't write to the area.
> > For that we can make it read-only by using ASSIGN_readonly.
> Yes, seems to be required. Can you take care of doing the patch ?
>
> > If we think it's the infomation leak (i.e. the guest shouln't read the
> > area),
> > another choice is not to dump the area.
> > With Tristan's get vcpu context patch the area isn't necessary
> > for the guest dump. Fortunatly the current crash utility doesn't look
> > the .xen_ia64_mmapped_regs section, the change won't break the
> > compatibility.
> >
> > My vote is the latter option. But I'm not sure the area should be dumped or
> > not.
> This is my preference too.
I'm willing to take care of it.
I'm not sure which you prefer. If we choose not to dump the area,
it isn't necessary to map the area into the guest pseudo physical address space.
So making it read-only doesn't make sense.
--
yamahata
_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel
|
Home