Instead of just checking once in exactly this order if there are
dependendies, non-COW clusters and new allocation, this starts looping
around these. This way we can, for example, gather non-COW clusters after
new allocations as long as the host cluster offsets stay contiguous.
Once handle_dependencies() is extended so that COW areas of in-flight
allocations can be overwritten, this allows to continue with gathering
other clusters (we wouldn't be able to do that without this change
because we would have missed a possible second dependency in one of the
next clusters).
This means that in the typical sequential write case, we can combine the
COW overwrite of one cluster with the allocation of the next cluster as
soon as something like Delayed COW gets actually implemented. It is only
by avoiding splitting requests this way that Delayed COW actually starts
improving performance noticably.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This patch is mainly to separate the indentation change from the
semantic changes. All that really changes here is that everything moves
into a while loop, all 'goto done' become 'break' and at the end of the
loop a new 'break is inserted.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Instead of expecting a single l2meta, have a list of them. This allows
to still have a single I/O request for the guest data, even though
multiple l2meta may be needed in order to describe both a COW overwrite
and a new cluster allocation (typical sequential write case).
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This gets rid of the nb_clusters and keep_clusters and the associated
complicated calculations. Just advance the number of bytes that have
been processed and everything is fine.
This patch advances the variables even after the last operation even
though they aren't used any more afterwards to make things look more
uniform. A later patch will turn the whole thing into a loop and then
it actually starts making sense.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This makes handle_alloc() and handle_copied() return byte-granularity
host offsets instead of returning always the cluster start. This is
required so that qcow2_alloc_cluster_offset() can stop aligning
everything to cluster boundaries.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Look only for clusters that start at a given physical offset.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Now *bytes is used to return the length of the area that can be written
to without performing an allocation or COW.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
handle_copied() uses its bytes parameter now to determine how many
clusters it should try to find.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Things can be simplified a bit now. No semantic changes.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The interface works completely on a byte granularity now and duplicated
parameters are removed.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
handle_alloc() is now called with the offset at which the actual new
allocation starts instead of the offset at which the whole write request
starts, part of which may already be processed.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
We already communicate the same information in *bytes.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This moves some code that prepares the allocation of new clusters to
where the actual allocation happens. This is the minimum required to be
able to move it to a separate function in the next patch.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This is a more precise description of what really constitutes a
dependency. The behaviour doesn't change at this point because the COW
area of the old request is still aligned to cluster boundaries and
therefore an overlap is detected wheneven the requests touch any part of
the same cluster.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The old code detected an overlapping allocation even when the
allocations didn't actually overlap, but were only adjacent.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Handling overlapping allocations isn't just a detail of cluster
allocation. It is rather one of three ways to get the host cluster
offset for a write request:
1. If a request overlaps an in-flight allocations, the cluster offset
can be taken from there (this is what handle_dependencies will evolve
into) or the request must just wait until the allocation has
completed. Accessing the L2 is not valid in this case, it has
outdated information.
2. Outside overlapping areas, check the clusters that can be written to
as they are, with no COW involved.
3. If a COW is required, allocate new clusters
Changing the code to reflect this doesn't change the behaviour because
overlaps cannot exist for clusters that are kept in step 2. It does
however make it easier for later patches to work on clusters that belong
to an allocation that is still in flight.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The unlock wakes up the next coroutine, but the currently running
coroutine will lock it again before it yields, so this doesn't make a
lot of sense.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This should be based on the virtual disk size, not on the size of the
image.
Interesting observation: With some VM state stored in the image file,
percentages higher than 100% are possible, even though snapshots
themselves are ignored. This is a qcow2 bug to be fixed another day: The
VM state should be discarded in the active L2 tables after completing
the snapshot creation.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The new parameter is unused yet.
This part was missing in commit 787e4a8500.
Cc: Kevin Wolf <kwolf@redhat.com>
Cc: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Commit 787e4a85 [block: Add options QDict to bdrv_file_open() prototypes] didn't
update rbd.c accordingly.
Cc: Kevin Wolf <kwolf@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Liu Yuan <tailai.ly@taobao.com>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
A file name may only specified if no host or socket path is specified.
The latter two may not appear at the same time either.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
The URL method already takes care to apply the default port when none is
specfied. Directly specifying driver-specific options required the port
number until now. Allow leaving it out and apply the default.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
In order to achieve this, the .bdrv_probe callbacks of all drivers must
cope with this. The DMG driver is the only one that bases its decision
on the filename and it needs to be changed.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
If a driver needs structured data and not just a string, it can provide
a .bdrv_parse_filename callback now that parses the command line string
into separate options. Keeping this separate from .bdrv_open_filename
ensures that the preferred way of directly specifying the options always
works as well if parsing the string works.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
The existing parsers for the file name now parse everything into the
bdrv_open() options QDict. Instead of using these parsers, you can now
directly specify the options on the command line, like this:
qemu-system-x86_64 -drive file=nbd:,file.port=1234,file.host=::1
Clearly the file=... part could use further improvement, but it's a
start.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
The NBD block supports an URL syntax, for which a URL parser returns
separate hostname and port fields. It also supports the traditional qemu
syntax encoded in a filename. Until now, after parsing the URL to get
each piece of information, a new string is built to be fed to socket
functions.
Instead of building a string in the URL case that is immediately parsed
again, parse the string in both cases and use the QemuOpts interface to
qemu-sockets.c.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Need to pass an options QDict to qcow2_open() now. This fixes a segfault
on the migration target with qcow2.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Sheepdog (neither quorum nor unsafe mode) will refuse to serve IO requests when
number of alive nodes is less than that of copies specified by users. This will
return 0x19 to QEMU client which currently doesn't recognize it.
This patch adds an error description when QEMU client receives it, other than
plainly printing 'Invalid error code'
Cc: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Cc: Kevin Wolf <kwolf@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Liu Yuan <tailai.ly@taobao.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Now that each AioContext has a ThreadPool and the main loop AioContext
can be fetched with bdrv_get_aio_context(), we can eliminate the concept
of a global thread pool from thread-pool.c.
The submit functions must take a ThreadPool* argument.
block/raw-posix.c and block/raw-win32.c use
aio_get_thread_pool(bdrv_get_aio_context(bs)) to fetch the main loop's
ThreadPool.
tests/test-thread-pool.c must be updated to reflect the new
thread_pool_submit() function prototypes.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
If an io_flush handler is not set, qemu_aio_wait doesn't invoke
callbacks.
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Using a blocking socket in the coroutine context reduces the chance of
switching to other work. This patch makes the sheepdog driver use a
non-blocking fd always.
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Otherwise, live migration of the top layer will miss zero clusters and
let the backing file show through. This also matches what is done in qed.
QCOW2_CLUSTER_ZERO clusters are invalid in v2 image files. Check this
directly in qcow2_get_cluster_offset instead of replicating the test
everywhere.
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
We already flush when the function completes. There is no need to flush
after every compressed cluster.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The update_cluster_refcount() function increments/decrements a cluster's
refcount and then returns the new refcount value.
There is no need to flush since both update_cluster_refcount() callers
already take care of this:
1. qcow2_alloc_bytes() calls update_cluster_refcount() when compressed
sectors will be appended to an existing cluster with enough free
space. qcow2_alloc_bytes() already flushes so there is no need to do
so in update_cluster_refcount().
2. qcow2_update_snapshot_refcount() sets a cache dependency on refcounts
if it needs to update L2 entries. It also flushes before completing.
Removing this flush significantly speeds up qcow2 snapshot creation:
$ qemu-img create -f qcow2 test.qcow2 -o size=50G,preallocation=metadata
$ time qemu-img snapshot -c new test.qcow2
Time drops from more than 3 minutes to under 1 second.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Users of qcow2_update_snapshot_refcount() do not flush consistently.
qcow2_snapshot_create() flushes but qcow2_snapshot_goto() and
qcow2_snapshot_delete() do not.
Solve this by moving the bdrv_flush() into
qcow2_update_snapshot_refcount().
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Compressed writes use qcow2_alloc_bytes() to allocate space with byte
granularity. The affected clusters' refcounts will be incremented but
we do not need to flush yet.
Set a L2 cache dependency on the refcount block cache, so that the
refcounts get written out before the L2 updates.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Since qcow2 metadata is cached we need to flush the caches, not just the
underlying file. Use bdrv_flush(bs) instead of bdrv_flush(bs->file).
Also add the error return path when bdrv_flush() fails and move the
flush after checking for qcow2_alloc_clusters() failure so that the
qcow2_alloc_clusters() error return value takes precedence.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
update_refcount() affects the refcount cache, it does not write to disk.
Therefore bdrv_flush(bs->file) does nothing. We need to flush the
refcount cache in order to write out the refcount updates!
While we're here also add error returns when qcow2_cache_flush() fails.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
qcow2 images now accept a boolean lazy_refcounts options. Use it like
this:
-drive file=test.qcow2,lazy_refcounts=on
If the option is specified on the command line, it overrides the default
specified by the qcow2 header flags that were set when creating the
image.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
It doesn't do anything yet except storing the options QDict in the
BlockDriverState.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
this patch adds iscsi_truncate which effectively allows for
online resizing of iscsi volumes. for this to work you have
to resize the volume on your storage and then call
block_resize command in qemu which will issue a
readcapacity16 to update the capacity.
v4:
- factor out complete readcapacity logic into a separate function
- handle capacity change check condition in readcapacity function
(this happens if the block_resize cmd is the first iscsi task
executed after a resize on the storage)
v3:
- remove switch statement in iscsi_open
- create separate patch for brdv_drain_all() in bdrv_truncate()
v2:
- add a general bdrv_drain_all() before bdrv_truncate() to avoid
in-flight AIOs while the device is truncated
- since no AIOs are in flight we can use a sync libiscsi call
to re-read the capacity
- factor out the readcapacity16 logic as it is redundant
to iscsi_open() and iscsi_truncate().
Signed-off-by: Peter Lieven <pl@kamp.de>
[allow any type of unit attention check condition in iscsi_readcapacity_sync(),
as in Message-ID: <51263A2A.6070304@dlhnet.de> - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
the storage might return a check condition status for various reasons.
(e.g. bus reset, capacity change, thin-provisioning info etc.)
currently all these informative status responses lead to an I/O error
which is populated to the guest. this patch introduces a retry mechanism
to avoid this.
Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This patch adds support for a unix domain socket for a connection
between qemu and local sheepdog server. You can use the unix domain
socket with the following syntax:
$ qemu sheepdog+unix:///<vdiname>?socket=<socket path>[#snapid]
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This uses the form "<host>:<port>" for the representation of the
sheepdog server to use inet_connect.
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The URI syntax is consistent with the NBD and Gluster syntax. The
syntax is
sheepdog[+tcp]://[host:port]/vdiname[#snapid|#tag]
Signed-off-by: MORITA Kazutaka <morita.kazutaka@lab.ntt.co.jp>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
