Re: [PATCH V5 08/12] xen/vm_event: Check for VM_EVENT_FLAG_DUMMY only in Debug builds

From: Tamas K Lengyel <tamas.lengyel@zentific.com>
To: Andrew Cooper <andrew.cooper3@citrix.com>
Cc: "Tian, Kevin" <kevin.tian@intel.com>,
	"wei.liu2@citrix.com" <wei.liu2@citrix.com>,
	Ian Campbell <ian.campbell@citrix.com>,
	Stefano Stabellini <stefano.stabellini@eu.citrix.com>,
	Tim Deegan <tim@xen.org>, Steven Maresca <steve@zentific.com>,
	"xen-devel@lists.xen.org" <xen-devel@lists.xen.org>,
	Jan Beulich <jbeulich@suse.com>,
	"Dong, Eddie" <eddie.dong@intel.com>,
	Andres Lagar-Cavilla <andres@lagarcavilla.org>,
	Jun Nakajima <jun.nakajima@intel.com>,
	"rshriram@cs.ubc.ca" <rshriram@cs.ubc.ca>,
	Keir Fraser <keir@xen.org>,
	Daniel De Graaf <dgdegra@tycho.nsa.gov>,
	"yanghy@cn.fujitsu.com" <yanghy@cn.fujitsu.com>,
	Ian Jackson <ian.jackson@eu.citrix.com>
Subject: Re: [PATCH V5 08/12] xen/vm_event: Check for VM_EVENT_FLAG_DUMMY only in Debug builds
Date: Sat, 14 Feb 2015 00:02:35 +0100	[thread overview]
Message-ID: <CAErYnsgxVOLJJ=vEONSUzmq36Xvq35V4paWR5zZ5kRxy=9aovg@mail.gmail.com> (raw)
In-Reply-To: <54DE8208.30906@citrix.com>

On Sat, Feb 14, 2015 at 12:00 AM, Andrew Cooper
<andrew.cooper3@citrix.com> wrote:
> On 13/02/2015 22:53, Tamas K Lengyel wrote:
>> On Fri, Feb 13, 2015 at 11:48 PM, Tamas K Lengyel
>> <tamas.lengyel@zentific.com> wrote:
>>> On Fri, Feb 13, 2015 at 9:14 PM, Andrew Cooper
>>> <andrew.cooper3@citrix.com> wrote:
>>>> On 13/02/15 16:33, Tamas K Lengyel wrote:
>>>>> The flag is only used for debugging purposes, thus it should be only checked
>>>>> for in debug builds of Xen.
>>>>>
>>>>> Signed-off-by: Tamas K Lengyel <tamas.lengyel@zentific.com>
>>>> What is the purpose of the dummy flag?  I would have thought it would be
>>>> more dangerous to accidentally process a dummy response in a non-debug Xen.
>>>>
>>>> ~Andrew
>>> I honestly have no idea what was the real use-case for it. It is a way
>>> to signal to Xen to just remove the reponse from the ring without
>>> doing anything else with it so I figured it might be handy when
>>> debugging a toolstack.
>>>
>>> Tamas
>> As for processing it on a non-debug Xen: I think it's safe as the
>> response would just be handled according to the other flags that are
>> set in the response, like unpausing the vCPU that triggered the event.
>> IMHO its a wasteful check on production systems.
>
> If it is not useful then discard it completely (seeing as the entire
> system is being overhauled).  I can't see any purpose in having a method
> of telling Xen to ignore a request placed on the ring.  If the toolstack
> wishes Xen not to do anything then it should not have put a request on
> the ring in the first place.
>
> ~Andrew

Ack.

Tamas