Re: [Qemu-devel] [PATCH 01/11] block: switch blk_write_compressed() to byte-based interface

[Pavel Butsykin <pbutsykin@virtuozzo.com>]

On 28.06.2016 13:53, Kevin Wolf wrote:
> Am 13.06.2016 um 16:23 hat Eric Blake geschrieben:
>> On 05/31/2016 03:15 AM, Denis V. Lunev wrote:
>>> From: Pavel Butsykin <pbutsykin@virtuozzo.com>
>>>
>>> This is a preparatory patch, which continues the general trend of the transition
>>> to the byte-based interfaces.
>>>
>>> Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
>>> Signed-off-by: Denis V. Lunev <den@openvz.org>
>>> CC: Jeff Cody <jcody@redhat.com>
>>> CC: Markus Armbruster <armbru@redhat.com>
>>> CC: Eric Blake <eblake@redhat.com>
>>> CC: John Snow <jsnow@redhat.com>
>>> CC: Stefan Hajnoczi <stefanha@redhat.com>
>>> CC: Kevin Wolf <kwolf@redhat.com>
>
>>> -int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
>>> -                          const uint8_t *buf, int nb_sectors)
>>> +int bdrv_pwrite_compressed(BlockDriverState *bs, int64_t offset,
>>> +                           const void *buf, int count)
>>>   {
>>>       BlockDriver *drv = bs->drv;
>>>       int ret;
>>> @@ -1791,14 +1791,15 @@ int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
>>>       if (!drv->bdrv_write_compressed) {
>>>           return -ENOTSUP;
>>>       }
>>> -    ret = bdrv_check_request(bs, sector_num, nb_sectors);
>>> +    ret = bdrv_check_byte_request(bs, offset, count);
>>>       if (ret < 0) {
>>>           return ret;
>>>       }
>>>
>>>       assert(QLIST_EMPTY(&bs->dirty_bitmaps));
>>>
>>> -    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
>>> +    return drv->bdrv_write_compressed(bs, offset >> BDRV_SECTOR_BITS, buf,
>>> +                                      count >> BDRV_SECTOR_BITS);
>>
>> If you are going to shift right, you need to first assert that offset
>> and count are aligned (and thus that our call to a sector interface
>> isn't going to operate on the wrong data).  See for example commit 166fe960.
>
> Yes, I would like to have these assertions at least.
>
> But I'm wondering what the point of converting the interface is when we
> don't intend to actually support sub-sector requests and the sector
> alignment is still required at the end of the series.

Because at the time of sending patches, the format drivers still had
the sector-based interfaces. In the end, the assertions are not
necessary, because the interface format driver now also byte-based.

> Kevin
>