Re: [Qemu-devel] [PATCH 1/2] block/raw-format: switch to BDRV_BLOCK_DATA with BDRV_BLOCK_RECURSE

From: Max Reitz <mreitz@redhat.com>
To: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>,
	Kevin Wolf <kwolf@redhat.com>
Cc: "qemu-devel@nongnu.org" <qemu-devel@nongnu.org>,
	"qemu-block@nongnu.org" <qemu-block@nongnu.org>,
	Denis Lunev <den@virtuozzo.com>
Subject: Re: [Qemu-devel] [PATCH 1/2] block/raw-format: switch to BDRV_BLOCK_DATA with BDRV_BLOCK_RECURSE
Date: Tue, 13 Aug 2019 17:03:57 +0200	[thread overview]
Message-ID: <d0b0fc4e-eb2f-796d-3413-366a6bb5aeca@redhat.com> (raw)
In-Reply-To: <6af67f42-b5ee-1ca9-6fb1-500fd993616a@virtuozzo.com>

[-- Attachment #1.1: Type: text/plain, Size: 2453 bytes --]

On 13.08.19 16:56, Vladimir Sementsov-Ogievskiy wrote:
> 13.08.2019 17:43, Max Reitz wrote:
>> On 13.08.19 13:04, Kevin Wolf wrote:
>>> Am 12.08.2019 um 20:11 hat Vladimir Sementsov-Ogievskiy geschrieben:
>>>> BDRV_BLOCK_RAW makes generic bdrv_co_block_status to fallthrough to
>>>> returned file. But is it correct behavior at all? If returned file
>>>> itself has a backing file, we may report as totally unallocated and
>>>> area which actually has data in bottom backing file.
>>>>
>>>> So, mirroring of qcow2 under raw-format is broken. Which is illustrated
>>>> by following commit with a test. Let's make raw-format behave more
>>>> correctly returning BDRV_BLOCK_DATA.
>>>>
>>>> Suggested-by: Max Reitz <mreitz@redhat.com>
>>>> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
>>>
>>> After some reading, I think I came to the conclusion that RAW is the
>>> correct thing to do. There is indeed a problem, but this patch is trying
>>> to fix it in the wrong place.
>>>
>>> In the case where the backing file contains some data, and we have a
>>> 'raw' node above the qcow2 overlay node, the content of the respective
>>> block is not defined by the queried backing file layer, so it is
>>> completely correct that bdrv_is_allocated() returns false,like it would
>>> if you queried the qcow2 layer directly.
>>
>> I disagree.  The queried backing file layer is the raw node.  As I said,
>> in my opinion raw nodes are not filter nodes, neither in behavior (they
>> have an offset option), nor in how they are generally used (as a format).
>>
>> The raw format does not support backing files.  Therefore, everything on
>> a raw node is allocated.
>>
> 
> Could you tell me at least, what means "allocated" ?
> 
> It's a term that describing a region somehow.. But how? Allocated where?
> In raw node, in its child or both? Am I right that if region allocated in
> one of non-cow children it is assumed to be allocated in parent too? Or what?
> 
> And it's unrelated to real disk allocation which (IMHO) directly shows that
> this a bad term.

It’s a term for COW backing chains.  If something is allocated on a
given node in a COW backing chain, it means it is either present in
exactly that node or in one of its storage children (in case the node is
a format node).  If it is not allocated, it is not, and read accesses
will be forwarded to the COW backing child.

Max

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]