Re: [PATCH 1/3] KVM: x86 emulator: Fix task switch privilege checks

[Gleb Natapov <gleb@redhat.com>]

On Fri, Jan 27, 2012 at 04:02:30PM +0100, Kevin Wolf wrote:
> Am 27.01.2012 15:17, schrieb Joerg Roedel:
> > On Fri, Jan 27, 2012 at 02:55:12PM +0100, Kevin Wolf wrote:
> >> Am 27.01.2012 14:34, schrieb Joerg Roedel:
> >>> On Fri, Jan 27, 2012 at 01:58:38PM +0100, Kevin Wolf wrote:
> >>>> Am 25.01.2012 17:00, schrieb Joerg Roedel:
> >>>
> >>>> I just retried. I use kvm-kmod and kvm.git with HEAD at ff92e9b5 plus
> >>>> the tree patches of this series plus a printk to output exit_int_info in
> >>>> task_switch_intercept(). I ran taskswitch2 from kvm-unittests and got
> >>>> two failures and my VM86 unit test which hung when trying to return from
> >>>> VM86. I also ran the kernel that made me aware of the issue initially.
> >>>> All debug messages show exit_int_info = 0.
> >>>
> >>> Okay, you are testing on a K8 which has exactly this bug. As I just
> >>> found out it is documented as erratum 701. The good news is that this
> >>> only happens on K8 and Fam11h, any later AMD processor doesn't have this
> >>> bug.
> >>
> >> Meh. Unless you give me a newer processor, this doesn't really help
> >> me... Doesn't look like there's any way to get a workaround, is there? I
> >> guess I'll have to hack it locally and possibly break other guests with
> >> the hacked module.
> > 
> > No, unfortunatly there is no workaround for this problem. How do you
> > plan to hack around it?
> 
> I know that my guest only uses iret and exceptions for task switches, so
> I think in my case I can assume that any TASK_SWITCH_CALL is really a
> TASK_SWITCH_GATE and I don't have to skip an instruction.
> 
You still need to know what exception caused task switch. Some of them
require you to skip an instruction.

> Not quite upstreamable, obviously.
> 
> Kevin

--
			Gleb.