[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAJX1YtaHkNtz1B_mqMJGzW56LoREWFkCHXwr9GV8dVH4ZQkniA@mail.gmail.com>
Date: Tue, 30 May 2017 18:31:20 +0200
From: Gi-Oh Kim <gi-oh.kim@...fitbricks.com>
To: Paolo Bonzini <pbonzini@...hat.com>
Cc: Roman Penyaev <roman.penyaev@...fitbricks.com>,
Mikhail Sennikovskii <mikhail.sennikovskii@...fitbricks.com>,
Gleb Natapov <gleb@...nel.org>, kvm@...r.kernel.org,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
Andy Lutomirski <luto@...capital.net>
Subject: Re: [RFC] KVM: SVM: do not drop VMCB CPL to 0 if SS is not present
Hi,
I found a code to set unusable flag of segment register incorrectly.
https://lkml.org/lkml/2017/5/30/459
I guess above patch and current discussion could be related.
I guess following sequence could happen.
1. svm_get_segment() sets var->unusable of Stack Segment incorrectly
2. svm_set_segment() clears both of s->attrib and svm->vmcb->save.cpl.
Is it possible scenario?
On Tue, May 30, 2017 at 6:05 PM, Paolo Bonzini <pbonzini@...hat.com> wrote:
>
>
> On 30/05/2017 17:58, Roman Penyaev wrote:
>> Indeed, what is left is eventually take it from SS.RPL. J.
>
> Ahah! :) But I only suggested that in specific cases.
>
>> But jokes aside, with your last patch you seems fixed a race problem
>> when "CS.RPL is not equal to the CPL in the few instructions between
>> setting CR0.PE and reloading CS".
>
> Yes, exactly. The symptom was a crash (triple fault) when you kept
> interrupting with "info cpus" a guest that repeatedly went to protected
> mode and back to real mode.
>
>> We will have CPL in var->dpl, and it seems ok. All we need is not
>> to lose it on the way kernel->userspace->kernel.
>
> You're right. So what do you think of the other suggestion (svm.c
> doesn't clear attributes for unusable registers, QEMU only clears P for
> unusable registers)?
>
> Thanks,
>
> Paolo
--
Best regards,
Gi-Oh Kim
TEL: 0176 2697 8962
Powered by blists - more mailing lists