On 16/1/22 上午12:39, Eric Blake wrote:
> On 01/21/2016 04:22 AM, Rudy Zhang wrote:
>> Add hmp command for incremental backup in drive-backup.
>> It need a bitmap to backup data from drive-image to incremental image,
>> so before it need add bitmap for this device to track io.
>> Usage:
>> drive_backup [-n] [-f] device target [bitmap] [format]
>>
>> Signed-off-by: Rudy Zhang <rudyflyzh...@gmail.com>
>> ---
>> hmp-commands.hx | 5 +++--
>> hmp.c | 16 ++++++++++++++--
>> 2 files changed, 17 insertions(+), 4 deletions(-)
>>
>> diff --git a/hmp-commands.hx b/hmp-commands.hx
>> index bb52e4d..7378aaa 100644
>> --- a/hmp-commands.hx
>> +++ b/hmp-commands.hx
>> @@ -1180,12 +1180,13 @@ ETEXI
>>
>> {
>> .name = "drive_backup",
>> - .args_type = "reuse:-n,full:-f,device:B,target:s,format:s?",
>> - .params = "[-n] [-f] device target [format]",
>> + .args_type =
>> "reuse:-n,full:-f,device:B,target:s,bitmap:s?,format:s?",
>> + .params = "[-n] [-f] device target [bitmap] [format]",
> This is HMP, so it may not matter, but this is not backwards compatible.
> Scripts targetting the old style of passing a format will now have that
> format string interpreted as a bitmap name with no format. Better would
> be to stick [bitmap] at the end, not the middle.
But I have a question: If I don't want to input a 'format', only use 'bitmap',
it will let 'bitmap' as 'format', This problem how to do it.
>
>> @@ -1098,6 +1100,17 @@ void hmp_drive_backup(Monitor *mon, const QDict
>> *qdict)
>> return;
>> }
>>
>> + if (full && bitmap) {
>> + error_setg(&err, "Parameter 'bitmap' if conflict with '-f'");
> s/if conflict/conflicts/
oh，made a mistake.
>> + hmp_handle_error(mon, &err);
>> + return;
>> + } else if (full)
>> + sync = MIRROR_SYNC_MODE_FULL;
> Needs {}. Run your patch through scripts/checkpatch.pl, to flag this
> and other style violations.
I have checked these patches，but I ignored these warnings.
