The "driver" entry in the options QDict is now only missing if we're
opening an image with format probing.
We also catch cases now where both the drv argument and a "driver"
option is specified, e.g. by specifying -drive format=qcow2,driver=raw
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
The idea of bdrv_fill_options() is to convert every parameter for
opening images, in particular the filename and flags, to entries in the
options QDict.
This patch starts with moving the filename parsing and driver probing
part from bdrv_file_open() to the new function.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
'buf' is not used actually, so remove it and related snprintf() statement.
Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
In order to let event defines use existing types later, instead of
redefine new ones, some old type defines for spice and vnc are changed,
and BlockErrorAction is moved from block.h to qapi schema. Note that
BlockErrorAction is not merged with BlockdevOnError.
At this point, VncInfo is not made a child of VncBasicInfo, because
VncBasicInfo has mandatory fields where VncInfo makes them optional.
Signed-off-by: Wenchao Xia <wenchaoqemu@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
With virtio-blk dataplane, I/O errors might occur while QEMU is
not in the main I/O thread. However, it's invalid to call vm_stop
when we're neither in a VCPU thread nor in the main I/O thread,
even if we were to take the iothread mutex around it.
To avoid this problem, we can raise a request to the main I/O thread,
similar to what QEMU does when vm_stop is called from a CPU thread.
We know that bdrv_error_action is called from an AIO callback, and
the moment at which the callback will fire is not well-defined; it
depends on the moment at which the disk or OS finishes the operation,
which can happen at any time. Note that QEMU is certainly not in a CPU
thread and we do not need to call cpu_stop_current() like vm_stop() does.
However, we need to ensure that any action taken by management will
result in correct detection of the error _and_ a running VM. In particular:
- the event must be raised after the iostatus has been set, so that
"info block" will return an iostatus that matches the event.
- the VM must be stopped after the iostatus has been set, so that
"info block" will return an iostatus that matches the runstate.
The ordering between the STOP and BLOCK_IO_ERROR events is preserved;
BLOCK_IO_ERROR is documented to come first.
This makes bdrv_error_action() thread safe (assuming QMP events are,
which is attacked by a separate series).
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Now that all backend drivers are using QemuOpts, remove all
QEMUOptionParameter related codes.
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Chunyan Liu <cyliu@suse.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Change block layer to support both QemuOpts and QEMUOptionParameter.
After this patch, it will change backend drivers one by one. At the end,
QEMUOptionParameter will be removed and only QemuOpts is kept.
Signed-off-by: Dong Xu Wang <wdongxu@linux.vnet.ibm.com>
Signed-off-by: Chunyan Liu <cyliu@suse.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Block I/O throttling uses timers and currently always adds them to the
main loop. Throttling will break if bdrv_set_aio_context() is used to
move a BlockDriverState to a different AioContext.
This patch adds throttle_detach/attach_aio_context() interfaces so the
throttling timers and uses them to move timers to the new AioContext.
Note that bdrv_set_aio_context() already drains all requests so we're
sure no throttled requests are pending.
The test cases need to be updated since the throttle_init() interface
has changed.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Up until now all BlockDriverState instances have used the QEMU main loop
for fd handlers, timers, and BHs. This is not scalable on SMP guests
and hosts so we need to move to a model with multiple event loops on
different host CPUs.
bdrv_set_aio_context() assigns the AioContext event loop to use for a
particular BlockDriverState. It first detaches the entire
BlockDriverState graph from the current AioContext and then attaches to
the new AioContext.
This function will be used by virtio-blk data-plane to assign a
BlockDriverState to its IOThread AioContext. Make
bdrv_aio_set_context() public since data-plane should not include
block_int.h.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Modify bdrv_drain_all() to take into account that BlockDriverState
instances may be running in different AioContexts.
This patch changes the implementation of bdrv_drain_all() while
preserving the semantics. Previously kicking throttled requests and
checking for pending requests were done across all BlockDriverState
instances in sequence. Now we process each BlockDriverState in turn,
making sure to acquire and release its AioContext.
This prevents race conditions between the thread executing
bdrv_drain_all() and the thread running the AioContext.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
bdrv_close_all(), bdrv_commit_all(), bdrv_flush_all(),
bdrv_invalidate_cache_all(), and bdrv_clear_incoming_migration_all() are
called by main loop code and touch all BlockDriverState instances.
Some BlockDriverState instances may be running in another AioContext.
Make sure to acquire the AioContext before closing the BlockDriverState.
This will protect against race conditions once virtio-blk data-plane is
using the BlockDriverState from another AioContext event loop.
Note that this patch does not convert bdrv_drain_all() yet since that
conversion is non-trivial.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Drop the assumption that we're using the main AioContext. Convert
qemu_aio_wait() to aio_poll() and qemu_bh_new() to aio_bh_new() so the
BlockDriverState AioContext is used.
Note there is still one qemu_aio_wait() left in bdrv_create() but we do
not have a BlockDriverState there and only main loop code invokes this
function.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Introduced in commit da557a. Spotted by Coverity.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The above bdrv_set_backing_hd already does this.
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This makes use of op_blocker and blocks all the operations except for
commit target, on each BlockDriverState->backing_hd.
The asserts for op_blocker in bdrv_swap are removed because with this
change, the target of block commit has at least the backing blocker of
its child, so the assertion is not true. Callers should do their check.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
We need to handle the coming backing_blocker properly, so don't open
code the assignment, instead, call bdrv_set_backing_hd to change
backing_hd.
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This is the common but non-trivial steps to assign or change the
backing_hd of BDS.
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This drops BlockDriverState.in_use with op_blockers:
- Call bdrv_op_block_all in place of bdrv_set_in_use(bs, 1).
- Call bdrv_op_unblock_all in place of bdrv_set_in_use(bs, 0).
- Check bdrv_op_is_blocked() in place of bdrv_in_use(bs).
The specific types are used, e.g. in place of starting block backup,
bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_BACKUP, ...).
There is one exception in block_job_create, where
bdrv_op_blocker_is_empty() is used, because we don't know the operation
type here. This doesn't matter because in a few commits away we will drop
the check and move it to callers that _do_ know the type.
- Check bdrv_op_blocker_is_empty() in place of assert(!bs->in_use).
Note: there is only bdrv_op_block_all and bdrv_op_unblock_all callers at
this moment. So although the checks are specific to op types, this
changes can still be seen as identical logic with previously with
in_use. The difference is error message are improved because of blocker
error info.
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
BlockDriverState.op_blockers is an array of lists with BLOCK_OP_TYPE_MAX
elements. Each list is a list of blockers of an operation type
(BlockOpType), that marks this BDS as currently blocked for a certain
type of operation with reason errors stored in the list. The rule of
usage is:
* BDS user who wants to take an operation should check if there's any
blocker of the type with bdrv_op_is_blocked().
* BDS user who wants to block certain types of operation, should call
bdrv_op_block (or bdrv_op_block_all to block all types of operations,
which is similar to the existing bdrv_set_in_use()).
* A blocker is only referenced by op_blockers, so the lifecycle is
managed by caller, and shouldn't be lost until unblock, so typically
a caller does these:
- Allocate a blocker with error_setg or similar, call bdrv_op_block()
to block some operations.
- Hold the blocker, do his job.
- Unblock operations that it blocked, with the same reason pointer
passed to bdrv_op_unblock().
- Release the blocker with error_free().
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
this patch tries to optimize zero write requests
by automatically using bdrv_write_zeroes if it is
supported by the format.
This significantly speeds up file system initialization and
should speed zero write test used to test backend storage
performance.
I ran the following 2 tests on my internal SSD with a
50G QCOW2 container and on an attached iSCSI storage.
a) mkfs.ext4 -E lazy_itable_init=0,lazy_journal_init=0 /dev/vdX
QCOW2 [off] [on] [unmap]
-----
runtime: 14secs 1.1secs 1.1secs
filesize: 937M 18M 18M
iSCSI [off] [on] [unmap]
----
runtime: 9.3s 0.9s 0.9s
b) dd if=/dev/zero of=/dev/vdX bs=1M oflag=direct
QCOW2 [off] [on] [unmap]
-----
runtime: 246secs 18secs 18secs
filesize: 51G 192K 192K
throughput: 203M/s 2.3G/s 2.3G/s
iSCSI* [off] [on] [unmap]
----
runtime: 8mins 45secs 33secs
throughput: 106M/s 1.2G/s 1.6G/s
allocated: 100% 100% 0%
* The storage was connected via an 1Gbit interface.
It seems to internally handle writing zeroes
via WRITESAME16 very fast.
Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If the filename given to bdrv_open() is prefixed with "json:", parse the
rest as a JSON object and merge the result into the options QDict. If
there are conflicts, the options QDict takes precedence.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_is_allocated() shouldn't return true for sectors that are
unallocated, but after the end of a short backing file, even though
such sectors are (correctly) marked as containing zeros.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
The immediately visible effect of this patch is that it fixes committing
a temporary snapshot to its backing file. Previously, it would fail with
a "permission denied" error because bdrv_inherited_flags() forced the
backing file to be read-only, ignoring the r/w reopen of bdrv_commit().
The bigger problem this revealed is that the original open flags must
actually only be applied to the temporary snapshot, and the original
image file must be treated as a backing file of the temporary snapshot
and get the right flags for that.
Reported-by: Jan Kiszka <jan.kiszka@web.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Use the same function as bdrv_open() for determining what the right
flags for bs->file are. Without doing this, a reopen means that
bs->file loses BDRV_O_CACHE_WB or BDRV_O_UNMAP if bs doesn't have it as
well.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
This reverts commit 3a389e7926. The commit
was wrong and what it tried to fix just works today without any change.
What the commit tried to fix:
When creating live snapshots, the new image file is opened with
BDRV_O_NO_BACKING because the whole backing chain is already opened.
It is then appended to the chain using bdrv_append(). The result of
this was that the image had a backing file, but BDRV_O_NO_BACKING
was still set. This is obviously inconsistent.
There used to be some places in qemu that closed and image and then
opened it again, with its old flags (a bdrv_open()/close() sequence
involves reopening the whole backing file chain, too). In this case
the BDRV_O_NO_BACKING flag meant that the backing chain wasn't
reopened and only the top layer was left.
(Most, but not all of these places are replaced by bdrv_reopen()
today, which doesn't touch the backing files at all.)
Other places that looked at bs->open_flags weren't interested in
BDRV_O_NO_BACKING, so no breakage there.
What it actually did:
The commit moved the BDRV_O_NO_BACKING away to the backing file.
Because the bdrv_open()/close() sequences only looked at the flags
of the top level BlockDriverState and used it for the whole chain,
the flag didn't hurt there any more. Obviously, it is still
inconsistent because the backing file may have another backing file,
but without practical impact.
At the same time, it swapped all other flags. This is practically
irrelevant as long as live snapshots only allow opening the new
layer with the same flags as the old top layer. It still doesn't
make any sense, and it is a time bomb that explodes as soon as the
flags can differ.
bdrv_append_temp_snapshot() is such a case: It adds the new flag
BDRV_O_TEMPORARY for the temporary snapshot. The swapping of commit
3a389e79 results in the following nonsensical configuration:
bs->open_flags: BDRV_O_TEMPORARY cleared
bs->file->open_flags: BDRV_O_TEMPORARY set
bs->backing_hd->open_flags: BDRV_O_TEMPORARY set
bs->backing_hd->file->open_flags: BDRV_O_TEMPORARY cleared
We're still lucky because the format layer ignores the flag and the
protocol layer happens to get the right value, but sooner or later
this is bound to go wrong...
What the right fix would have been:
Simply clear the BDRV_O_NO_BACKING flag when the BlockDriverState is
appended to an existing backing file chain, because now it does have
a backing file.
Commit 4ddc07ca already implemented this silently in bdrv_append(),
so we don't have to come up with a new fix.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Instead of having unlink() calls in the generic block layer, where we
aren't even guarateed to have a file name, move them to those block
drivers that are actually used and that always have a filename. Gets us
rid of some #ifdefs as well.
The patch also converts bs->is_temporary to a new BDRV_O_TEMPORARY open
flag so that it is inherited in the protocol layer and the raw-posix and
raw-win32 drivers can unlink the file.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Copy on Read makes sense on the format level where backing files are
implemented, but it's not required on the protocol level. While it
shouldn't actively break anything to have COR enabled on both layers,
needless serialisation and allocation checks may impact performance.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Instead of manipulation flags inline, move the derivation of the flags
of a backing file into a new function next to the existing functions
that derive flags for bs->file and for the block driver open function.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Instead of having bdrv_open_flags() as a function that creates flags for
several unrelated places and then adding open-coded flags on top, create
a new function that derives the flags for bs->file from the flags for bs.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Some block drivers have multiple BlockDriver instances with identical
format_name fields (e.g. gluster, nbd).
Both qemu-img and qemu will use bdrv_iterate_format() to list the
supported formats when a help option is invoked. As protocols and
formats may register multiple drivers, redundant listings of formats
occur (e.g., "Supported formats: ... gluster gluster gluster gluster ...
").
Since the list of driver formats will be small, this performs a simple
linear search on format_name, and ignores any duplicates.
The end result change is that the iterator will no longer receive
duplicate string names, nor will it receive NULL pointers.
Signed-off-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
error_is_set(&var) is the same as var != NULL, but it takes
whole-program analysis to figure that out. Unnecessarily hard for
optimizers, static checkers, and human readers. Commit 84d18f0 dumbed
it down to obvious, but a few more have crept in since, and
documentation was overlooked. Dumb these down, too.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
In 1.7.1 qcow2_create2 reopen the file for flushing without the BDRV_O_NO_BACKING
flags.
As a consequence the code would recursively open the whole backing chain.
These three stack arrays would pile up through the recursion and lead to a coroutine
stack overflow.
Convert these array to malloced buffers in order to streamline the coroutine
footprint.
Symptoms where freezes or segfaults on production machines while taking QMP externals
snapshots. The overflow disturbed coroutine switching.
[Resolved conflicts on qemu.git/master since the patch was against v1.7.1
--Stefan]
Signed-off-by: Benoit Canet <benoit.canet@gmail.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Since commit f298d071, block devices added with blockdev-add don't have
a QemuOpts around in dinfo->opts. Consequently, we can't rely any more
on QemuOpts catching duplicate IDs for block devices.
This patch adds a new check for duplicate IDs to bdrv_new(), and moves
the existing check that the ID isn't already taken for a node-name there
as well.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
This patch adds an errp parameter to bdrv_new() and updates all its
callers. The next patches will make use of this in order to check for
duplicate IDs. Most of the callers know that their ID is fine, so they
can simply assert that there is no error.
Behaviour doesn't change with this patch yet as bdrv_new() doesn't
actually assign errors to errp.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
bdrv_getlength could fail, check the return value before using it.
Return NULL and set errno if it fails. Callers are updated to handle
the error case.
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Insanely large requests could cause an integer overflow in
bdrv_rw_co() while converting sectors to bytes. This patch catches the
problem and returns an error (if we hadn't overflown the integer here,
bdrv_check_byte_request() would have rejected the request, so we're not
breaking anything that was supposed to work before).
We actually do have a test case that triggers behaviour where we
accidentally let such a request pass, so that it would return success,
but read 0 bytes instead of the requested 4 GB. It fails now like it
should.
If the vdi block driver wants to be able to deal with huge images, it
can't read the whole block bitmap at once into memory like it does
today, but needs to use a metadata cache like qcow2 does.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Commit 8f4754ed intended to protect against integer overflow bugs in
block drivers by making sure that a single request that is passed to
drivers is no longer than INT_MAX bytes.
However, meanwhile there are some callers that don't use that code path
any more but call bdrv_check_byte_request() directy, so let's add a
check there as well.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Since commit 9fd3171a, BDRV_O_SNAPSHOT uses an option QDict to specify
the originally requested image as the backing file of the newly created
temporary snapshot. This means that the filename is stored in
"file.filename", which is an option that is not parsed for protocol
names. Therefore things like -drive file=nbd:localhost:10809 were
broken because it looked for a local file with the literal name
'nbd:localhost:10809'.
This patch changes the way BDRV_O_SNAPSHOT works once again. We now open
the originally requested image as normal, and then do a similar
operation as for live snapshots to put the temporary snapshot on top.
This way, both driver specific options and parsed filenames work.
As a nice side effect, this results in code movement to factor
bdrv_append_temp_snapshot() out. This is a good preparation for moving
its call to drive_init() and friends eventually.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
When using the QDict option 'filename', it is supposed to be interpreted
literally. The code did correctly avoid guessing the protocol from any
string before the first colon, but it still called bdrv_parse_filename()
which would, for example, incorrectly remove a 'file:' prefix in the
raw-posix driver.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Limiting the size of a single request to INT_MAX not only fixes a
direct integer overflow in bdrv_check_request() (which would only
trigger bad behaviour with ridiculously huge images, as in close to
2^64 bytes), but can also prevent overflows in all block drivers.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
If it returns an error, the migrated VM will not be started, but qemu
exits with an error message.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Opening an encrypted image takes an additional step: setting the key.
Between open and the key set, the image must not be used.
We have some protection against accidental use in place: you can't
unpause a guest while we're missing keys. You can, however, hot-plug
block devices lacking keys into a running guest just fine, or insert
media lacking keys. In the latter case, notifying the guest of the
insert is delayed until the key is set, which may suffice to protect
at least some guests in common usage.
This patch makes the protection apply in more cases, in a rather
heavy-handed way: it doesn't let you open encrypted images unless
we're in a paused state.
It doesn't extend the protection to users other than the guest (block
jobs?). Use of runstate_check() from block.c is disgusting. Best I
can do right now.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
If the image file cannot be opened and was created as a temporary file,
it should be deleted; thus, in this case, we should jump to the
"unlink_and_fail" label and not just to "fail".
Reported-by: Benoît Canet <benoit@irqsave.net>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This patch keep the recursive way of doing things but simplify it by giving
two responsabilities to all block filters implementors.
They will need to do two things:
-Set the is_filter field of their block driver to true.
-Implement the bdrv_recurse_is_first_non_filter method of their block driver like
it is done on the Quorum block driver. (block/quorum.c)
[Paolo Bonzini <pbonzini@redhat.com> pointed out that this patch changes
the semantics of blkverify, which now recurses down both bs->file and
s->test_file.
-- Stefan]
Reported-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Currently, bdrv_debug_resume() requires every bs->drv in the BDS stack
to be NULL until a bs->drv with an implementation of bdrv_debug_resume()
is found. For a normal function, this would be fine, but this is a
function for debugging purposes and should therefore allow intermediate
BDS not to have a driver (i.e., be "ejected"). Otherwise, it is hard to
debug such situations.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
After migration has completed, we call bdrv_invalidate_cache() so that
drivers which cache some data drop their stale copy of the data and
reread it from the image file to get a new version of data that the
source modified while the migration was running.
Reloading metadata from the image file is useless, though, if the size
of the image file stays stale (this is a value that is cached for all
image formats in block.c). Reads from (meta)data after the old EOF
return only zeroes, causing image corruption.
We need to update bs->total_sectors in all layers that could potentially
have changed their size (i.e. backing files are not a concern - if they
are changed, we're in bigger trouble)
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Using an invalid option for a block device that is opened with
BDRV_O_PROTOCOL led to drv = NULL, and when trying to include the driver
name in the error message, qemu dereferenced it:
$ x86_64-softmmu/qemu-system-x86_64 -drive file=/tmp/test.qcow2,file.foo=bar
Segmentation fault (core dumped)
With this patch applied, the expected error message is printed:
$ x86_64-softmmu/qemu-system-x86_64 -drive file=/tmp/test.qcow2,file.foo=bar
qemu-system-x86_64: -drive file=/tmp/test.qcow2,file.foo=bar: could
not open disk image /tmp/test.qcow2: Block protocol 'file' doesn't
support the option 'foo'
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Currently, bdrv_file_open() always removes the "filename" option from
the options QDict after bdrv_parse_filename() has been (successfully)
called. However, for drivers with bdrv_needs_filename, it makes more
sense for bdrv_parse_filename() to overwrite the "filename" option and
for bdrv_file_open() to fetch the filename from there.
Since there currently are no drivers that implement
bdrv_parse_filename() and have bdrv_needs_filename set, this does not
change current behavior.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Moving only the node_name one field could lead to some inconsitencies where a
node_name was defined on a bs which was not registered in the graph node list.
bdrv_swap between a named node bs and a non named node bs would lead to this.
bdrv_make_anon would then crash because it would try to remove the bs from the
graph node list while it is not in it.
This patch remove named node bses from the graph node list before doing the swap
then insert them back.
Signed-off-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
For sg backends, bs->request_alignment is meaningless and may be 0.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
If TMPDIR is not specified, the default was to use /tmp for the working
copy of the block devices. Update this to /var/tmp instead, so systems
using tmp-on-tmpfs don't end up inadvertently using RAM for the block
device.
Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This option is now unnecessary since specifying BDRV_O_PROTOCOL as flag
will do exactly the same.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The fail and success paths of bdrv_file_open() may be further shortened
by reusing code already existent in bdrv_open(). This includes
bdrv_file_open() not taking the reference to options which allows the
removal of QDECREF(options) in that function.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The fail paths of bdrv_file_open() and bdrv_open() naturally exhibit
similarities, thus it is possible to reuse the one from bdrv_open() and
shorten the one in bdrv_file_open() accordingly.
Also, setting bs->options in bdrv_file_open() is not necessary if it is
already done in bdrv_open().
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Change bdrv_file_open() to take a simple pointer to an already existing
BDS instead of an indirect one. The BDS will be created in bdrv_open()
if necessary.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Remove the reference parameter and the related handling code from
bdrv_file_open(), since it exists in bdrv_open() now as well.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Add the bdrv_open() option BDRV_O_PROTOCOL which results in passing the
call to bdrv_file_open(). Additionally, make bdrv_file_open() static and
therefore bdrv_open() the only way to call it.
Consequently, all existing calls to bdrv_file_open() have to be adjusted
to use bdrv_open() with the BDRV_O_PROTOCOL flag instead.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Allow bdrv_open() to handle references to existing block devices just as
bdrv_file_open() is already capable of.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Make bdrv_open() take a pointer to a BDS pointer, similarly to
bdrv_file_open(). If a pointer to a NULL pointer is given, bdrv_open()
will create a new BDS with an empty name; if the BDS pointer is not
NULL, that existing BDS will be reused (in the same way as bdrv_open()
already did).
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Consider top level BlockDriverStates as well.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Tested-by: Benoit Canet <benoit@irqsave.net>
error_is_set(&var) is the same as var != NULL, but it takes
whole-program analysis to figure that out. Unnecessarily hard for
optimizers, static checkers, and human readers. Dumb it down to
obvious.
Gets rid of several dozen Coverity false positives.
Note that the obvious form is already used in many places.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Since we introduced node_name for named bs of the graph modify the opening by
reference to use it as a fallback.
This patch also enforce the separation of the device id and graph node
namespaces.
Signed-off-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The following patch will reuse bdrv_lookup_bs in order to open images by
references so the rules of usage of bdrv_lookup_bs must be relaxed a bit.
Signed-off-by: Benoit Canet <benoit@irqsave.net>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
On 32 bit hosts, size_t is too small for align as the bitmask
~(align - 1) will zero out the higher 32 bits of the offset.
While at it, change the local overlap_bytes variable to unsigned to
match the field in BdrvTrackedRequest.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
The behaviour of the ROUND_UP macro with negative numbers isn't obvious.
It happens to do the right thing in this please, but better avoid it.
Suggested-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
This adds assertions that the request that we actually end up passing to
the block driver (which includes RMW data and has therefore potentially
been rounded to alignment boundaries) is fully covered by the
overlap_{offset,size} fields of the associated BdrvTrackedRequest.
Suggested-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
The error path for a failure in one of the two bdrv_aligned_preadv()
calls leaked head_buf or tail_buf, respectively. This fixes the memory
leak.
Reported-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
This fixes a regression introduced in commit 2a05cbe42 ('block: Allow
block devices without files'):
$ qemu-system-x86_64 -drive driver=file
qemu-system-x86_64: block.c:892: bdrv_open_common: Assertion
`!drv->bdrv_needs_filename || filename != ((void *)0)' failed.
Now the respective check must be performed not only in bdrv_file_open(),
but also in bdrv_open().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Request sizes used to be rounded down to the next sector boundary,
allowing to bypass the I/O limit. Now all requests are accounted for
with their exact byte size.
Reported-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Instead of implementing the alignment adjustment here, use the now
existing functionality of bdrv_co_do_pwritev().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Instead of implementing the alignment adjustment here, use the now
existing functionality of bdrv_co_do_preadv().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
If a request calls wait_serialising_requests() and actually has to wait
in this function (i.e. a coroutine yield), other requests can run and
previously read data (like the head or tail buffer) could become
outdated. In this case, we would have to restart from the beginning to
read in the updated data.
However, we're lucky and don't actually need to do that: A request can
only wait in the first call of wait_serialising_requests() because we
mark it as serialising before that call, so any later requests would
wait. So as we don't wait in practice, we don't have to reload the data.
This is an important assumption that may not be broken or data
corruption will happen. Document it with some assertions.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
This patch changes bdrv_co_do_pwritev() to actually be what its name
promises. If requests aren't properly aligned, it performs a RMW.
Requests touching the same block are serialised against the RMW request.
Further optimisation of this is possible by differentiating types of
requests (concurrent reads should actually be okay here).
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
We can only have a single wait_serialising_requests() call per request
because otherwise we can run into deadlocks where requests are waiting
for each other. The same is true when wait_serialising_requests() is not
at the very beginning of a request, so that other requests can be issued
between the start of the tracking and wait_serialising_requests().
Fix this by changing wait_serialising_requests() to ignore requests that
are already (directly or indirectly) waiting for the calling request.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Copy on Read wants to serialise with all requests touching the same
cluster, so wait_serialising_requests() rounded to cluster boundaries.
Other users like alignment RMW will have different requirements, though
(requests touching the same sector), so make it dynamic.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Change the API so that specific requests can be marked serialising. Only
these requests are checked for overlaps then.
This means that during a Copy on Read operation, not all requests
overlapping other requests are serialised any more, but only those that
actually overlap with the specific COR request.
Also remove COR from function and variable names because this
functionality can be useful in other contexts.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Odd file sizes could make bdrv_aligned_preadv() shorten the request in
non-aligned ways. Fix it by rounding to the required alignment instead
of 512 bytes.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Previously, it was not possible to use wait_for_overlapping_requests()
between tracked_request_begin()/end() because it would wait for itself.
Ignore the current request in the overlap check and run more of the
bdrv_co_do_preadv/pwritev code with a BdrvTrackedRequest present.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
This is going to become the bdrv_co_do_preadv() equivalent for writes.
In this patch, however, just a function taking byte offsets is created,
it doesn't align anything yet.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
First waiting for all COR requests to complete and calling the
throttling function afterwards means that the request could be delayed
and we still need to wait for the COR request even if it was issued only
after the throttled write request.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
This separates the part of bdrv_co_do_writev() that needs to happen
before the request is modified to match the backend alignment, and a
part that needs to be executed afterwards and passes the request to the
BlockDriver.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
Similar to bdrv_pread(), which aligns byte-aligned request to 512 byte
sectors, bdrv_co_do_preadv() takes a byte-aligned request and aligns it
to the alignment specified in bs->request_alignment.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
This separates the part of bdrv_co_do_readv() that needs to happen
before the request is modified to match the backend alignment, and a
part that needs to be executed afterwards and passes the request to the
BlockDriver.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Add a bs->request_alignment field that contains the required
offset/length alignment for I/O requests and fill it in the raw block
drivers. Use ioctls if possible, else see what alignment it takes for
O_DIRECT to succeed.
While at it, also expose the memory alignment requirements, which may be
(and in practice are) different from the disk alignment requirements.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
The alignment field is now set to the value that is promised to the
guest, rather than required by the host. The next patches will make
QEMU aware of the host-provided values, so make this clear.
The alignment is also not about memory buffers, but about the sectors on
the disk, change the documentation of the field.
At this point, the field is set by the device emulation, but completely
ignored by the block layer.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoit Canet <benoit@irqsave.net>
bs->buffer_alignment is set by the device emulation and contains the
logical block size of the guest device. This isn't something that the
block layer should know, and even less something to use for determining
the right alignment of buffers to be used for the host.
The new BlockLimits field opt_mem_alignment tells the qemu block layer
the optimal alignment to be used so that no bounce buffer must be used
in the driver.
This patch may change the buffer alignment from 4k to 512 for all
callers that used qemu_blockalign() with the top-level image format
BlockDriverState. The value was never propagated to other levels in the
tree, so in particular raw-posix never required anything else than 512.
While on disks with 4k sectors direct I/O requires a 4k alignment,
memory may still be okay when aligned to 512 byte boundaries. This is
what must have happened in practice, because otherwise this would
already have failed earlier. Therefore I don't expect regressions even
with this intermediate state. Later, raw-posix can implement the hook
and expose a different memory alignment requirement.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
For an O_DIRECT request to succeed, it's not only necessary that all
base addresses in the qiov are aligned, but also that each length in it
is aligned.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
When reopening with different flags, or when backing files disappear
from the chain, the limits may change. Make sure they get updated in
these cases.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Benoît Canet <benoit@irqsave.net>
