Re: [PATCH v1] x86/hvm: Generic instruction re-execution mechanism for execute faults

From: "Jan Beulich" <JBeulich@suse.com>
To: aisaila@bitdefender.com
Cc: Andrei LUTAS <vlutas@bitdefender.com>,
	Tamas K Lengyel <tamas@tklengyel.com>,
	Wei Liu <wei.liu2@citrix.com>,
	Razvan Cojocaru <rcojocaru@bitdefender.com>,
	George Dunlap <George.Dunlap@eu.citrix.com>,
	Andrew Cooper <andrew.cooper3@citrix.com>,
	Mihai Dontu <mdontu@bitdefender.com>,
	Kevin Tian <kevin.tian@intel.com>,
	Jun Nakajima <jun.nakajima@intel.com>,
	xen-devel <xen-devel@lists.xenproject.org>,
	Anshul Makkar <anshul.makkar@citrix.com>,
	Roger Pau Monne <roger.pau@citrix.com>
Subject: Re: [PATCH v1] x86/hvm: Generic instruction re-execution mechanism for execute faults
Date: Thu, 22 Nov 2018 03:00:49 -0700	[thread overview]
Message-ID: <5BF67E5102000078001FEC44@prv1-mh.provo.novell.com> (raw)
In-Reply-To: <e5dcca5e-6f19-f2d6-05f1-8659ea1dd2bb@bitdefender.com>

>>> On 22.11.18 at 10:50, <aisaila@bitdefender.com> wrote:
> On 21.11.2018 20:55, Razvan Cojocaru wrote:
>>> +            if ( a == v )
>>> +                continue;
>>> +
>>> +            /* Pause, synced. */
>>> +            while ( !a->arch.in_host )
>> Why not use a->is_running as a way to know whether the vCPU is
>> running?
>> 
>> I think the logic of using vcpu_pause and expecting the running vcpu
>> to take a vmexit and thus set in_host is wrong because a vcpu that
>> wasn't running when vcpu_pause_nosync is called won't get scheduled
>> anymore, thus not taking a vmexit and this function will lockup.
> 
> We can resolve this by using while ( !vcpu_runnable(a) && 
> !a->arch.in_host ), if this is suitable.

Only if you can fully explain why the current infrastructure doesn't
work and can't be made work without the custom in_host flag.
From prior work I know things are subtle in some cases where one
would want to use ->is_running, but so far all issues could be
resolved without custom additions.

Jan

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel