Re: Requesting for freeze exception for RMRR

From: Andrew Cooper <andrew.cooper3@citrix.com>
To: Wei Liu <wei.liu2@citrix.com>, Jan Beulich <JBeulich@suse.com>,
	Tiejun Chen <tiejun.chen@intel.com>
Cc: Kevin <kevin.tian@intel.com>,
	"ian.campbell@citrix.com" <ian.campbell@citrix.com>,
	George Dunlap <George.Dunlap@eu.citrix.com>,
	"ian.jackson@eu.citrix.com" <ian.jackson@eu.citrix.com>,
	Yong Y Wang <yong.y.wang@intel.com>,
	"xen-devel@lists.xen.org" <xen-devel@lists.xen.org>
Subject: Re: Requesting for freeze exception for RMRR
Date: Fri, 17 Jul 2015 16:11:25 +0100	[thread overview]
Message-ID: <55A91B1D.1050406@citrix.com> (raw)
In-Reply-To: <20150717140157.GM12455@zion.uk.xensource.com>

On 17/07/15 15:01, Wei Liu wrote:
> On Fri, Jul 17, 2015 at 02:43:05PM +0100, Jan Beulich wrote:
>>>>> On 17.07.15 at 15:21, <wei.liu2@citrix.com> wrote:
>>> The major concern seems to be around the PCI allocation algorithm. Jan
>>> has different opinion from George. George provided a simple solution
>>> that will not make things worse than before, while Jan prefers to get
>>> everything right.
>>>
>>> To be fair, the PCI allocation code in a bad state is not really
>>> contributor's fault.
>>>
>>> Jan also pointed out on IRC he thinks the proper logic he asked for is
>>> not very hard to implement.
>>>
>>> Given we either take George's route, which already seems to have a
>>> patch, or Jan's route, which he thinks shouldn't be too hard to
>>> implement, I'm inclined to say give this series another week (24th
>>> deadline still applied). Note that we've been working on this for ages,
>>> any delay is going to burn up more energy than necessary.
>>>
>>> Jan and George, if you disagree with what I say above, please reply.
>> My main disagreement here continues to be that we're talking
>> about a bug fix, and hence I don't view this as needing a freeze
>> exception in the first place (at least not at this point in time). Yes,
>> the bug fix involves adding code that looks like a new feature, but
>> that happens with bug fixes.
>>
> Fine then. I'm not going to argue feature vs bug fix at this stage.  The
> final resolution is still the same. Tiejun can continue working on this
> next week.

Sorry for being slow in my maintainership role with this series.  (I
have been busy with the migration v2 side of things).

I can appreciate Wei's position that, despite this being a bugfix, it
does exhibit itself as a new feature, and we don't want to be merging a
new feature beyond the hard feature freeze point.

The PCI allocation code is in a state, but it was in a similarly bad
state before.  I agree with Jan's point of the risk that these new
changes cause a regression in booting guests, although we can mitigate
that somewhat by testing.

I feel at this point that we shouldn't block the RMRR bugfix on also
fixing the PCI allocation algorithm (which was a pre-existing issue).

Therefore, I recommend that v9 gets respun to v10 to address the current
comments, and accepted.  Afterwards, the PCI allocation algorithm gets
worked on as a bugfix activity, to pro actively cater for the risk of
regression.

~Andrew