Re: [Xen-devel] [PATCH v2] tools/tests/mem-sharing/memshrtool share-all test

[Andres Lagar-Cavilla <andres.lagarcavilla@xxxxxxxxx>]

On Apr 22, 2013, at 8:11 AM, George Dunlap <George.Dunlap@xxxxxxxxxxxxx> wrote:

> On 22/04/13 13:07, Ian Campbell wrote:
>> On Fri, 2013-03-22 at 19:25 +0000, Andres Lagar-Cavilla wrote:
>>> On Mar 21, 2013, at 8:17 AM, Tim Deegan <tim@xxxxxxx> wrote:
>>> 
>>>> At 09:34 -0400 on 18 Mar (1363599276), Tamas Lengyel wrote:
>>>>> Update memshrtool test program to allow sharing of all pages of two 
>>>>> domains
>>>>> with identical memory sizes. Currently the tool only allows sharing of
>>>>> specific pages. With this patch we can quickly share all pages between 
>>>>> clones
>>>>> and check how many pages were successfully deduplicated. The pages' 
>>>>> content
>>>>> are not checked, therefore this mode is only safe for clone domains.
>>>> Cc'ing Andres, who wrote the original tool.
>>>> 
>>>> Tim.
>>>> 
>>>>> v2: fix typo of source_info
>>>>> 
>>>>> Signed-off-by: Tamas Lengyel <tamas.lengyel@xxxxxxxxxxxx>
>>> Just a few minute comments.
>>> 
>>> The code in itself is correct as a first attempt. I am tempted to ack
>>> it on the basis of being a useful thing.
>> Did you conclude that you would ack it in the end or not?
>> 
>> WRT the freeze it seems this is new standalone functionality in a test
>> tool, which ought to be pretty safe. George CCd.
> 
> I think this is a "just barely" in terms of timing / risk / benefits 
> analysis, but:
> 
> Acked-by: George Dunlap <george.dunlap@xxxxxxxxxxxxx>
> 
It is clearly over the acceptable threshold in terms of a demo testing tool.

Acked-by: Andres Lagar-Cavilla <andres@xxxxxxxxxxxxxxxx>

Having said that, Tamas, can you add a blob of comment stating the expected 
operating conditions: both domains paused, sharing enabled, both domains able 
to stand on their own feet (vcpu contexts, device model, PV backends if any). 
Also clearly stating that this might corrupt domain memory if used carelessly. 
Just a few lines of prose in there.

Thanks (and sorry for dropping the ball for a bit)
Andres


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel