Re: [Qemu-devel] [PATCH v5 30/42] qemu-img: Use child access functions

[Vladimir Sementsov-Ogievskiy]

19.06.2019 18:49, Max Reitz wrote:
> On 19.06.19 11:18, Vladimir Sementsov-Ogievskiy wrote:
>> 13.06.2019 1:09, Max Reitz wrote:
>>> This changes iotest 204's output, because blkdebug on top of a COW node
>>> used to make qemu-img map disregard the rest of the backing chain (the
>>> backing chain was broken by the filter).  With this patch, the
>>> allocation in the base image is reported correctly.
>>>
>>> Signed-off-by: Max Reitz <address@hidden>
>>> ---
>>>    qemu-img.c                 | 36 ++++++++++++++++++++----------------
>>>    tests/qemu-iotests/204.out |  1 +
>>>    2 files changed, 21 insertions(+), 16 deletions(-)
>>>
>>> diff --git a/qemu-img.c b/qemu-img.c
>>> index 07b6e2a808..7bfa6e5d40 100644
>>> --- a/qemu-img.c
>>> +++ b/qemu-img.c
>>> @@ -992,7 +992,7 @@ static int img_commit(int argc, char **argv)
>>>        if (!blk) {
>>>            return 1;
>>>        }
>>> -    bs = blk_bs(blk);
>>> +    bs = bdrv_skip_implicit_filters(blk_bs(blk));
>>
>> if filename is json, describing explicit filter over normal node, bs will be
>> explicit filter ...
>>
>>>    
>>>        qemu_progress_init(progress, 1.f);
>>>        qemu_progress_print(0.f, 100);
>>> @@ -1009,7 +1009,7 @@ static int img_commit(int argc, char **argv)
>>>            /* This is different from QMP, which by default uses the deepest 
>>> file in
>>>             * the backing chain (i.e., the very base); however, the 
>>> traditional
>>>             * behavior of qemu-img commit is using the immediate backing 
>>> file. */
>>> -        base_bs = backing_bs(bs);
>>> +        base_bs = bdrv_filtered_cow_bs(bs);
>>>            if (!base_bs) {
>>
>> and here we'll fail.
> 
> Right, will change to bdrv_backing_chain_next().
> 
>>>                error_setg(&local_err, "Image does not have a backing file");
>>>                goto done;
>>> @@ -1626,19 +1626,18 @@ static int 
>>> convert_iteration_sectors(ImgConvertState *s, int64_t sector_num)
>>>    
>>>        if (s->sector_next_status <= sector_num) {
>>>            int64_t count = n * BDRV_SECTOR_SIZE;
>>> +        BlockDriverState *src_bs = blk_bs(s->src[src_cur]);
>>> +        BlockDriverState *base;
>>>    
>>>            if (s->target_has_backing) {
>>> -
>>> -            ret = bdrv_block_status(blk_bs(s->src[src_cur]),
>>> -                                    (sector_num - src_cur_offset) *
>>> -                                    BDRV_SECTOR_SIZE,
>>> -                                    count, &count, NULL, NULL);
>>> +            base = bdrv_backing_chain_next(src_bs);
>>
>> As you described in another patch, will not we here get allocated in base as 
>> allocated, because of
>> counting filters above base?
> 
> Damn, yes.  So
> 
>      base = bdrv_filtered_cow_bs(bdrv_skip_rw_filters(src_bs));
> 
> I suppose.
> 
>> Hmm. I now think, why filters don't report everything as unallocated, would 
>> not it be more correct
>> than fallthrough to child?
> 
> I don’t know, actually.  Maybe, maybe not.
> 
>>>            } else {
>>> -            ret = bdrv_block_status_above(blk_bs(s->src[src_cur]), NULL,
>>> -                                          (sector_num - src_cur_offset) *
>>> -                                          BDRV_SECTOR_SIZE,
>>> -                                          count, &count, NULL, NULL);
>>> +            base = NULL;
>>>            }
>>> +        ret = bdrv_block_status_above(src_bs, base,
>>> +                                      (sector_num - src_cur_offset) *
>>> +                                      BDRV_SECTOR_SIZE,
>>> +                                      count, &count, NULL, NULL);
>>>            if (ret < 0) {
>>>                error_report("error while reading block status of sector %" 
>>> PRId64
>>>                             ": %s", sector_num, strerror(-ret));
> 
> [...]
> 
>>> @@ -2949,7 +2950,7 @@ static int img_map(int argc, char **argv)
>>>        if (!blk) {
>>>            return 1;
>>>        }
>>> -    bs = blk_bs(blk);
>>> +    bs = bdrv_skip_implicit_filters(blk_bs(blk));
>>
>> Hmm, another thought about implicit filters, how they could be here in 
>> qemu-img?
> 
> Hm, I don’t think they can.
> 
>> If implicit are only
>> job filters. Do you prepared it for implicit COR? But we discussed with 
>> Kevin that we'd better deprecate
>> copy-on-read option..
>>
>> So, if implicit filters are for compatibility, we'll have them only in 
>> block-jobs. So, seems no reason to support
>> them in qemu-img.
> 
> Seems reasonable, yes.
> 
>> Also, in block-jobs, we can abandon creating implicit filters above any 
>> filter nodes, as well as abandon creating any
>> filter nodes above implicit filters. This will still support old scenarios, 
>> but gives very simple and well defined scope
>> of using implicit filters and how to work with them. What do you think?
> 
> Hm, in what way would that make things simpler?
> 

This question was in my mind while I've finishing this paragraph) At least such 
restriction answer the question, where
should new filters be added: below or under implicit filters.. With such 
restriction we always can have only one implicit filter
over non-filter node, and above it should be explicit filter or non-filter 
node. But this need huge work to be done with small
benefit, so, forget it)


-- 
Best regards,
Vladimir