Re: [Qemu-devel] [PATCH 1/2] Revert "migration: do not sent zero pages in bulk stage"

[Peter Lieven <lieven-lists@dlhnet.de>]

On 06.06.2013 14:16, Eric Blake wrote:
> ----- Original Message -----
>> Not sending zero pages breaks migration if a page is zero
>> at the source but not at the destination. This can e.g. happen
>> if different BIOS versions are used at source and destination.
>> It has also been reported that migration on pseries is completely
>> broken with this patch.
>>
>> This reverts commit f1c72795af573b24a7da5eb52375c9aba8a37972.
>>
>> +++ b/qapi-schema.json
>> @@ -499,9 +499,7 @@
>>   #
>>   # @total: total amount of bytes involved in the migration process
>>   #
>> -# @duplicate: number of duplicate (zero) pages (since 1.2)
>> -#
>> -# @skipped: number of skipped zero pages (since 1.5)
>> +# @duplicate: number of duplicate pages (since 1.2)
>>   #
>>   # @normal : number of normal pages (since 1.2)
>>   #
>> @@ -514,8 +512,8 @@
>>   ##
>>   { 'type': 'MigrationStats',
>>     'data': {'transferred': 'int', 'remaining': 'int', 'total': 'int' ,
>> -           'duplicate': 'int', 'skipped': 'int', 'normal': 'int',
>> -           'normal-bytes': 'int', 'dirty-pages-rate' : 'int' } }
>> +           'duplicate': 'int', 'normal': 'int', 'normal-bytes': 'int',
>> +           'dirty-pages-rate' : 'int' } }
> This hunk is questionable.  Removing something that we have previously sent
> over the wire may break clients that are expecting this field to exist.
> Rather than reverting the entire patch, you should consider keeping this
> field present in QMP, even if you now always populate it with 0.
>
You are right. I might still account zero pages in the bulk phase to give this
field a meaning. These pages are very likely not written on the destination except
for the cornercases. What do you think?

Peter