Re: [Qemu-devel] [PATCH v3 for-2.7 00/14] qapi: Allow blockdev-add for NBD

[Max Reitz <mreitz@redhat.com>]

[-- Attachment #1.1: Type: text/plain, Size: 2188 bytes --]

On 06.04.2016 20:28, Max Reitz wrote:
> Turns out NBD is not so simple to do if you do it right. Anyway, this
> series adds blockdev-add support for NBD clients.
> 
> Patch 1 adds a QDict function which I needed in later NBD patches for
> handling legacy options (move "host" to "address.data.host" etc.).
> 
> Patches 2, 3, 4, and 5 are minor patches with no functional relation to
> this series, other than later patches will touch the code they touch,
> too.
> 
> Patches 6 and 7 prepare the code for the addition of a new option
> prefix, which is "address.".
> 
> Patch 8 makes the NBD client accept a SocketAddress under the "address"
> option (or rather, a flattened SocketAddress QDict with its keys
> prefixed by "address."). The old options "host", "port", and "path" are
> supported as legacy options and translated to the respective
> SocketAddress representation.
> 
> Patch 9 drops usage of "host", "port", and "path" outside of
> nbd_has_filename_options_conflict(),
> nbd_process_legacy_socket_options(), and nbd_refresh_filename(), making
> those options nothing but legacy.
> 
> Patch 10, the goal of this series, is again not very complicated.
> 
> Patches 11, 12, and 13 are required for the iotest added in patch 16. It
> will invoke qemu-nbd, so patch 13 is required. Besides qemu-nbd, it will
> launch an NBD server VM concurrently to the client VM, which is why
> patch 14 is required. And finally, it will test whether we can add an
> NBD BDS by passing it a file descriptor, which patch 15 is needed for
> (so we use the socket_scm_helper to pass sockets to qemu).
> 
> Patch 14 then adds the iotest for NBD's blockdev-add interface.
> 
> 
> *** This series requires Daniel's qdict_crumple() function (from his
>     "Provide a QOM-based authorization API" series). ***
> 
> 
> v2:
> - Dropped patches 2 and 3; use Daniel's qdict_crumple() instead.
> - Patch 7: Not sure why the diff differs, seems functionally like the
>   same patch to me. Maybe git changed something about its default diff
>   algorithm.
> - Patch 8: Use qdict_crumple() instead of qdict_unflatten()
> - Patch 10: Rebase conflicts


Ping


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]