2021-11-18 17:34:01 +03:00
|
|
|
QEMU supports the NBD protocol, and has an internal NBD client (see
|
2018-06-09 18:17:58 +03:00
|
|
|
block/nbd.c), an internal NBD server (see blockdev-nbd.c), and an
|
|
|
|
external NBD server tool (see qemu-nbd.c). The common code is placed
|
|
|
|
in nbd/*.
|
|
|
|
|
|
|
|
The NBD protocol is specified here:
|
|
|
|
https://github.com/NetworkBlockDevice/nbd/blob/master/doc/proto.md
|
|
|
|
|
|
|
|
The following paragraphs describe some specific properties of NBD
|
2021-11-18 17:34:01 +03:00
|
|
|
protocol realization in QEMU.
|
2018-06-09 18:17:58 +03:00
|
|
|
|
|
|
|
= Metadata namespaces =
|
|
|
|
|
2021-11-18 17:34:01 +03:00
|
|
|
QEMU supports the "base:allocation" metadata context as defined in the
|
2018-06-09 18:17:58 +03:00
|
|
|
NBD protocol specification, and also defines an additional metadata
|
|
|
|
namespace "qemu".
|
|
|
|
|
|
|
|
== "qemu" namespace ==
|
|
|
|
|
2020-10-27 08:05:54 +03:00
|
|
|
The "qemu" namespace currently contains two available metadata context
|
|
|
|
types. The first is related to exposing the contents of a dirty
|
|
|
|
bitmap alongside the associated disk contents. That metadata context
|
|
|
|
is named with the following form:
|
2018-06-09 18:17:58 +03:00
|
|
|
|
|
|
|
qemu:dirty-bitmap:<dirty-bitmap-export-name>
|
|
|
|
|
|
|
|
Each dirty-bitmap metadata context defines only one flag for extents
|
|
|
|
in reply for NBD_CMD_BLOCK_STATUS:
|
|
|
|
|
2020-10-27 08:05:54 +03:00
|
|
|
bit 0: NBD_STATE_DIRTY, set when the extent is "dirty"
|
|
|
|
|
|
|
|
The second is related to exposing the source of various extents within
|
|
|
|
the image, with a single metadata context named:
|
|
|
|
|
|
|
|
qemu:allocation-depth
|
|
|
|
|
|
|
|
In the allocation depth context, the entire 32-bit value represents a
|
|
|
|
depth of which layer in a thin-provisioned backing chain provided the
|
|
|
|
data (0 for unallocated, 1 for the active layer, 2 for the first
|
|
|
|
backing layer, and so forth).
|
2018-06-09 18:17:58 +03:00
|
|
|
|
|
|
|
For NBD_OPT_LIST_META_CONTEXT the following queries are supported
|
2020-10-27 08:05:54 +03:00
|
|
|
in addition to the specific "qemu:allocation-depth" and
|
|
|
|
"qemu:dirty-bitmap:<dirty-bitmap-export-name>":
|
2018-06-09 18:17:58 +03:00
|
|
|
|
|
|
|
* "qemu:" - returns list of all available metadata contexts in the
|
|
|
|
namespace.
|
|
|
|
* "qemu:dirty-bitmap:" - returns list of all available dirty-bitmap
|
|
|
|
metadata contexts.
|
2018-12-15 16:53:04 +03:00
|
|
|
|
|
|
|
= Features by version =
|
|
|
|
|
|
|
|
The following list documents which qemu version first implemented
|
|
|
|
various features (both as a server exposing the feature, and as a
|
|
|
|
client taking advantage of the feature when present), to make it
|
|
|
|
easier to plan for cross-version interoperability. Note that in
|
|
|
|
several cases, the initial release containing a feature may require
|
|
|
|
additional patches from the corresponding stable branch to fix bugs in
|
|
|
|
the operation of that feature.
|
|
|
|
|
|
|
|
* 2.6: NBD_OPT_STARTTLS with TLS X.509 Certificates
|
|
|
|
* 2.8: NBD_CMD_WRITE_ZEROES
|
|
|
|
* 2.10: NBD_OPT_GO, NBD_INFO_BLOCK
|
|
|
|
* 2.11: NBD_OPT_STRUCTURED_REPLY
|
|
|
|
* 2.12: NBD_CMD_BLOCK_STATUS for "base:allocation"
|
|
|
|
* 3.0: NBD_OPT_STARTTLS with TLS Pre-Shared Keys (PSK),
|
|
|
|
NBD_CMD_BLOCK_STATUS for "qemu:dirty-bitmap:", NBD_CMD_CACHE
|
2020-09-17 10:50:22 +03:00
|
|
|
* 4.2: NBD_FLAG_CAN_MULTI_CONN for shareable read-only exports,
|
2019-08-23 17:37:23 +03:00
|
|
|
NBD_CMD_FLAG_FAST_ZERO
|
2020-10-27 08:05:54 +03:00
|
|
|
* 5.2: NBD_CMD_BLOCK_STATUS for "qemu:allocation-depth"
|
nbd/server: Allow MULTI_CONN for shared writable exports
According to the NBD spec, a server that advertises
NBD_FLAG_CAN_MULTI_CONN promises that multiple client connections will
not see any cache inconsistencies: when properly separated by a single
flush, actions performed by one client will be visible to another
client, regardless of which client did the flush.
We always satisfy these conditions in qemu - even when we support
multiple clients, ALL clients go through a single point of reference
into the block layer, with no local caching. The effect of one client
is instantly visible to the next client. Even if our backend were a
network device, we argue that any multi-path caching effects that
would cause inconsistencies in back-to-back actions not seeing the
effect of previous actions would be a bug in that backend, and not the
fault of caching in qemu. As such, it is safe to unconditionally
advertise CAN_MULTI_CONN for any qemu NBD server situation that
supports parallel clients.
Note, however, that we don't want to advertise CAN_MULTI_CONN when we
know that a second client cannot connect (for historical reasons,
qemu-nbd defaults to a single connection while nbd-server-add and QMP
commands default to unlimited connections; but we already have
existing means to let either style of NBD server creation alter those
defaults). This is visible by no longer advertising MULTI_CONN for
'qemu-nbd -r' without -e, as in the iotest nbd-qemu-allocation.
The harder part of this patch is setting up an iotest to demonstrate
behavior of multiple NBD clients to a single server. It might be
possible with parallel qemu-io processes, but I found it easier to do
in python with the help of libnbd, and help from Nir and Vladimir in
writing the test.
Signed-off-by: Eric Blake <eblake@redhat.com>
Suggested-by: Nir Soffer <nsoffer@redhat.com>
Suggested-by: Vladimir Sementsov-Ogievskiy <v.sementsov-og@mail.ru>
Message-Id: <20220512004924.417153-3-eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
2022-05-12 03:49:24 +03:00
|
|
|
* 7.1: NBD_FLAG_CAN_MULTI_CONN for shareable writable exports
|