Re: [Qemu-devel] [PATCH v4 0/4] block: Block driver callbacks fixes

[Manos Pitsidianakis <el13635@mail.ntua.gr>]

[-- Attachment #1: Type: text/plain, Size: 1537 bytes --]

On Wed, Jul 12, 2017 at 09:49:20AM +0200, Markus Armbruster wrote:
>Manos Pitsidianakis <el13635@mail.ntua.gr> writes:
>
>> This series makes implementing some of the bdrv_* callbacks easier for block
>> filters by passing requests to bs->file if bs->drv doesn't implement it instead
>> of failing, and adding default bdrv_co_get_block_status() implementations.
>>
>> This is based against Kevin Wolf's block branch, commit
>> da4bd74d2450ab72a7c26bbabb10c6a287dd043e
>
>Haven't seen BlockDriver member is_filter before.  Interesting.  It's
>documentation
>
>    /* set to true if the BlockDriver is a block filter */
>    bool is_filter;
>
>is seriously lacking.  What does it *mean* to be a block filter?  Which
>block layer facilities are affected, and how?

Currently it is only used in bdrv_recurse_is_first_non_filter. 

>Observation: driver "raw" is filter-like in the sense that all it does
>is pass along method arguments and results.  Can't say whether that
>makes it a filter in the sense of is_filter, because "the sense of
>is_filter" is nebulous to me :)

I'm not very acquainted with raw, so I can't really comment. But the 
drivers I'm working on, throttle and before write notifier have that 
exact semantic, ie they do something when IO is intercepted and pass 
everything to the BDSes below. There was a mini discussion about raw and 
filters in the previous version's thread.

I might add documentation to block_int.h in the future. When I first 
read it it felt nebulous to me too.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]