These functions are not used outside of block/io.c, there is no reason
why they should be globally available.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Callers can now pass a pointer to an integer that bdrv_drain_invoke()
(and its recursive callees) will increment for every
bdrv_drain_invoke_entry() operation they schedule.
bdrv_drain_invoke_entry() in turn will decrement it once it has invoked
BlockDriver.bdrv_co_drain_end().
We use atomic operations to access the pointee, because the
bdrv_do_drained_end() caller may wish to end drained sections for
multiple nodes in different AioContexts (bdrv_drain_all_end() does, for
example).
This is the first step to moving the polling for BdrvCoDrainData.done to
become true out of bdrv_drain_invoke() and into the root drained_end
function.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Commit 5cb2737e92 laid out why
bdrv_do_drained_end() must decrement the quiesce_counter after
bdrv_drain_invoke(). It did not give a very good reason why it has to
happen after bdrv_parent_drained_end(), instead only claiming symmetry
to bdrv_do_drained_begin().
It turns out that delaying it for so long is wrong.
Situation: We have an active commit job (i.e. a mirror job) from top to
base for the following graph:
filter
|
[file]
|
v
top --[backing]--> base
Now the VM is closed, which results in the job being cancelled and a
bdrv_drain_all() happening pretty much simultaneously.
Beginning the drain means the job is paused once whenever one of its
nodes is quiesced. This is reversed when the drain ends.
With how the code currently is, after base's drain ends (which means
that it will have unpaused the job once), its quiesce_counter remains at
1 while it goes to undrain its parents (bdrv_parent_drained_end()). For
some reason or another, undraining filter causes the job to be kicked
and enter mirror_exit_common(), where it proceeds to invoke
block_job_remove_all_bdrv().
Now base will be detached from the job. Because its quiesce_counter is
still 1, it will unpause the job once more. So in total, undraining
base will unpause the job twice. Eventually, this will lead to the
job's pause_count going negative -- well, it would, were there not an
assertion against this, which crashes qemu.
The general problem is that if in bdrv_parent_drained_end() we undrain
parent A, and then undrain parent B, which then leads to A detaching the
child, bdrv_replace_child_noperm() will undrain A as if we had not done
so yet; that is, one time too many.
It follows that we cannot decrement the quiesce_counter after invoking
bdrv_parent_drained_end().
Unfortunately, decrementing it before bdrv_parent_drained_end() would be
wrong, too. Imagine the above situation in reverse: Undraining A leads
to B detaching the child. If we had already decremented the
quiesce_counter by that point, bdrv_replace_child_noperm() would undrain
B one time too little; because it expects bdrv_parent_drained_end() to
issue this undrain. But bdrv_parent_drained_end() won't do that,
because B is no longer a parent.
Therefore, we have to do something else. This patch opts for
introducing a second quiesce_counter that counts how many times a
child's parent has been quiesced (though c->role->drained_*). With
that, bdrv_replace_child_noperm() just has to undrain the parent exactly
that many times when removing a child, and it will always be right.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This patch is used in the 'block/stream: introduce a bottom node'
that is following. Instead of the base node, the caller may pass
the node that has the base as its backing image to the function
bdrv_is_allocated_above() with a new parameter include_base = true
and get rid of the dependency on the base that may change during
commit/stream parallel jobs. Now, if the specified base is not
found in the backing image chain, the QEMU will abort.
Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Message-id: 1559152576-281803-2-git-send-email-andrey.shinkevich@virtuozzo.com
[mreitz: Squashed in the following as a rebase on conflicting patches:]
Message-id: e3cf99ae-62e9-8b6e-5a06-d3c8b9363b85@redhat.com
Signed-off-by: Max Reitz <mreitz@redhat.com>
This fixes at least one overflow in qcow2_process_discards, which
passes 64bit region length to bdrv_pdiscard where bytes (or sectors in
the past) parameter is int since its introduction in 0b919fae.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
When ending a drained section, bdrv_do_drained_end() currently first
decrements the quiesce_counter, and only then actually ends the drain.
The bdrv_drain_invoke(bs, false) call may cause graph changes. Say the
graph change involves replacing an existing BB's ("blk") BDS
(blk_bs(blk)) by @bs. Let us introducing the following values:
- bs_oqc = old_quiesce_counter
(so bs->quiesce_counter == bs_oqc - 1)
- obs_qc = blk_bs(blk)->quiesce_counter (before bdrv_drain_invoke())
Let us assume there is no blk_pread_unthrottled() involved, so
blk->quiesce_counter == obs_qc (before bdrv_drain_invoke()).
Now replacing blk_bs(blk) by @bs will reduce blk->quiesce_counter by
obs_qc (making it 0) and increase it by bs_oqc-1 (making it bs_oqc-1).
bdrv_drain_invoke() returns and we invoke bdrv_parent_drained_end().
This will decrement blk->quiesce_counter by one, so it would be -1 --
were there not an assertion against that in blk_root_drained_end().
We therefore have to keep the quiesce_counter up at least until
bdrv_drain_invoke() returns, so that bdrv_parent_drained_end() does the
right thing for the parents @bs got during bdrv_drain_invoke().
But let us delay it even further, namely until bdrv_parent_drained_end()
returns, because then it mirrors bdrv_do_drained_begin(): There, we
first increment the quiesce_counter, then begin draining the parents,
and then call bdrv_drain_invoke(). It makes sense to let
bdrv_do_drained_end() unravel this exactly in reverse.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
drv_co_block_status digs bs->file for additional, more accurate search
for hole inside region, reported as DATA by bs since 5daa74a6eb.
This accuracy is not free: assume we have qcow2 disk. Actually, qcow2
knows, where are holes and where is data. But every block_status
request calls lseek additionally. Assume a big disk, full of
data, in any iterative copying block job (or img convert) we'll call
lseek(HOLE) on every iteration, and each of these lseeks will have to
iterate through all metadata up to the end of file. It's obviously
ineffective behavior. And for many scenarios we don't need this lseek
at all.
However, lseek is needed when we have metadata-preallocated image.
So, let's detect metadata-preallocation case and don't dig qcow2's
protocol file in other cases.
The idea is to compare allocation size in POV of filesystem with
allocations size in POV of Qcow2 (by refcounts). If allocation in fs is
significantly lower, consider it as metadata-preallocation case.
102 iotest changed, as our detector can't detect shrinked file as
metadata-preallocation, which don't seem to be wrong, as with metadata
preallocation we always have valid file length.
Two other iotests have a slight change in their QMP output sequence:
Active 'block-commit' returns earlier because the job coroutine yields
earlier on a blocking operation. This operation is loading the refcount
blocks in qcow2_detect_metadata_preallocation().
Suggested-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
There are a few places in which we turn a number of bytes into sectors
in order to compare the result against BDRV_REQUEST_MAX_SECTORS
instead of using BDRV_REQUEST_MAX_BYTES directly.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
No one is using these functions anymore, all callers have switched to
the byte-based bdrv_pread() and bdrv_pwrite()
Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
On a file system used by the customer, fallocate() returns an error
if the block is not properly aligned. So, bdrv_co_pwrite_zeroes()
fails. We can handle that case the same way as it is done for the
unsupported cases, namely, call to bdrv_driver_pwritev() that writes
zeroes to an image for the unaligned chunk of the block.
Suggested-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 1554474244-553661-1-git-send-email-andrey.shinkevich@virtuozzo.com
Message-Id: <1554474244-553661-1-git-send-email-andrey.shinkevich@virtuozzo.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
For qemu-img convert, we want an operation that zeroes out the whole
image if this can be done efficiently, but that returns an error
otherwise so we don't write explicit zeroes and immediately overwrite
them with the real data, potentially doubling the amount of data to be
written.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Eric Blake <eblake@redhat.com>
There is only a single caller of bdrv_make_zero(), which is qemu-img
convert. If the function fails, we just fall back to a different method
of zeroing out blocks on the target image. There is no good reason to
print error messages on stderr when the higher level operation will
actually succeed.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Eric Blake <eblake@redhat.com>
Use new qemu_iovec_init_buf() instead of
qemu_iovec_init_external( ... , 1), which simplifies the code.
While being here, use qemu_try_blockalign0 as well.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20190218140926.333779-3-vsementsov@virtuozzo.com
Message-Id: <20190218140926.333779-3-vsementsov@virtuozzo.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
In the block layer, synchronous APIs are often implemented by creating a
coroutine that calls the asynchronous coroutine-based implementation and
then waiting for completion with BDRV_POLL_WHILE().
For this to work with iothreads (more specifically, when the synchronous
API is called in a thread that is not the home thread of the block
device, so that the coroutine will run in a different thread), we must
make sure to call aio_wait_kick() at the end of the operation. Many
places are missing this, so that BDRV_POLL_WHILE() keeps hanging even if
the condition has long become false.
Note that bdrv_dec_in_flight() involves an aio_wait_kick() call. This
corresponds to the BDRV_POLL_WHILE() in the drain functions, but it is
generally not enough for most other operations because they haven't set
the return value in the coroutine entry stub yet. To avoid race
conditions there, we need to kick after setting the return value.
The race window is small enough that the problem doesn't usually surface
in the common path. However, it does surface and causes easily
reproducible hangs if the operation can return early before even calling
bdrv_inc/dec_in_flight, which many of them do (trivial error or no-op
success paths).
The bug in bdrv_truncate(), bdrv_check() and bdrv_invalidate_cache() is
slightly different: These functions even neglected to schedule the
coroutine in the home thread of the node. This avoids the hang, but is
obviously wrong, too. Fix those to schedule the coroutine in the right
AioContext in addition to adding aio_wait_kick() calls.
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
When draining a block node, we recurse to its parent and for subtree
drains also to its children. A single AIO_WAIT_WHILE() is then used to
wait for bdrv_drain_poll() to become true, which depends on all of the
nodes we recursed to. However, if the respective child or parent becomes
quiescent and calls bdrv_wakeup(), only the AioWait of the child/parent
is checked, while AIO_WAIT_WHILE() depends on the AioWait of the
original node.
Fix this by using a single AioWait for all callers of AIO_WAIT_WHILE().
This may mean that the draining thread gets a few more unnecessary
wakeups because an unrelated operation got completed, but we already
wake it up when something _could_ have changed rather than only if it
has certainly changed.
Apart from that, drain is a slow path anyway. In theory it would be
possible to use wakeups more selectively and still correctly, but the
gains are likely not worth the additional complexity. In fact, this
patch is a nice simplification for some places in the code.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
bdrv_drain_poll_top_level() was buggy because it didn't release the
AioContext lock of the node to be drained before calling aio_poll().
This way, callbacks called by aio_poll() would possibly take the lock a
second time and run into a deadlock with a nested AIO_WAIT_WHILE() call.
However, it turns out that the aio_poll() call isn't actually needed any
more. It was introduced in commit 91af091f92, which is effectively
reverted by this patch. The cases it was supposed to fix are now covered
by bdrv_drain_poll(), which waits for block jobs to reach a quiescent
state.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
bdrv_do_drained_begin/end() assume that they are called with the
AioContext lock of bs held. If we call drain functions from a coroutine
with the AioContext lock held, we yield and schedule a BH to move out of
coroutine context. This means that the lock for the home context of the
coroutine is released and must be re-acquired in the bottom half.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Truncation is the last to convert from open coded req handling to
reusing helpers. This time the permission check in prepare has to adapt
to the new caller: it checks a different permission bit, and doesn't
trigger the before write notifier.
Also, truncation should always trigger a bs->total_sectors update and in
turn call parent resize_cb. Update the condition in finish helper, too.
It's intended to do a duplicated bs->read_only check before calling
bdrv_co_write_req_prepare() so that we can be more informative with the
error message, as bdrv_co_write_req_prepare() doesn't have Error
parameter.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If we are growing the image and potentially using preallocation for the
new area, we need to make sure that no write requests are made to the
"preallocated" area which is [@old_size, @offset), not
[@offset, offset * 2 - @old_size).
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This brings the request handling logic inline with write and discard,
fixing write_gen, resize_cb, dirty bitmaps and image size refreshing.
The last of these issues broke iotest case 222, which is now fixed.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reuse the new bdrv_co_write_req_prepare/finish helpers. The variation
here is that discard requests don't affect bs->wr_highest_offset, and it
cannot extend the image.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Two problems exist when a write request that enlarges the image (i.e.
write beyond EOF) finishes:
1) parent is not notified about size change;
2) dirty bitmap is not resized although we try to set the dirty bits;
Fix them just like how bdrv_co_truncate works.
Reported-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
As a mechanical refactoring patch, this is the first step towards
unified and more correct write code paths. This is helpful because
multiple BlockDriverState fields need to be updated after modifying
image data, and it's hard to maintain in multiple places such as copy
offload, discard and truncate.
Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This matches the types used for bytes in the rest parts of block layer.
In the case of bdrv_co_truncate, new_bytes can be the image size which
probably doesn't fit in a 32 bit int.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Other I/O functions are already using a BdrvChild pointer in the API, so
make discard do the same. It makes it possible to initiate the same
permission checks before doing I/O, and much easier to share the
helper functions for this, which will be added and used by write,
truncate and copy range paths.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
A few trace points that can help reveal what is happening in a copy
offloading I/O path.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Serialized writes should be used in copy-on-write of backup(sync=none)
for image fleecing scheme.
We need to change an assert in bdrv_aligned_pwritev, added in
28de2dcd88. The assert may fail now, because call to
wait_serialising_requests here may become first call to it for this
request with serializing flag set. It occurs if the request is aligned
(otherwise, we should already set serializing flag before calling
bdrv_aligned_pwritev and correspondingly waited for all intersecting
requests). However, for aligned requests, we should not care about
outdating of previously read data, as there no such data. Therefore,
let's just update an assert to not care about aligned requests.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Pass read flags and write flags separately. This is needed to handle
coming BDRV_REQ_NO_SERIALISING clearly in following patches.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Here two things are fixed:
1. Architecture
On each recursion step, we go to the child of src or dst, only for one
of them. So, it's wrong to create tracked requests for both on each
step. It leads to tracked requests duplication.
2. Wait for serializing requests on write path independently of
BDRV_REQ_NO_SERIALISING
Before commit 9ded4a0114 "backup: Use copy offloading",
BDRV_REQ_NO_SERIALISING was used for only one case: read in
copy-on-write operation during backup. Also, the flag was handled only
on read path (in bdrv_co_preadv and bdrv_aligned_preadv).
After 9ded4a0114, flag is used for not waiting serializing operations
on backup target (in same case of copy-on-write operation). This
behavior change is unsubstantiated and potentially dangerous, let's
drop it and add additional asserts and documentation.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If the virtual disk size isn't aligned to full clusters,
bdrv_co_do_copy_on_readv() may get pnum == 0 before having the full
cluster completed, which will let it run into an assertion failure:
qemu-io: block/io.c:1203: bdrv_co_do_copy_on_readv: Assertion `skip_bytes < pnum' failed.
Check for EOF, assert that we read at least as much as the read request
originally wanted to have (which is true at EOF because otherwise
bdrv_check_byte_request() would already have returned an error) and
return success early even though we couldn't copy the full cluster.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Commit dcf94a23b1 ('block: Don't poll in parent drain callbacks')
removed polling in bdrv_child_cb_drained_begin() on the grounds that the
original bdrv_drain() already will poll and BdrvChildRole.drained_begin
calls must not cause graph changes (and therefore must not call
aio_poll() or the recursion through the graph will break.
This reasoning is correct for calls through bdrv_do_drained_begin().
However, BdrvChildRole.drained_begin is also called when a node that is
already in a drained section (i.e. bdrv_do_drained_begin() has already
returned and therefore can't poll any more) is attached to a new parent.
In this case, we must explicitly poll to have all requests completed
before the drained new child can be attached to the parent.
In bdrv_replace_child_noperm(), we know that we're not inside the
recursion of bdrv_do_drained_begin() because graph changes are not
allowed there, and bdrv_replace_child_noperm() is a graph change. The
call of BdrvChildRole.drained_begin() must therefore be followed by a
BDRV_POLL_WHILE() that waits for the completion of requests.
Reported-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This semantics is needed by drive-backup so implement it before using
this API there.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-id: 20180703023758.14422-3-famz@redhat.com
Signed-off-by: Jeff Cody <jcody@redhat.com>
src may be NULL if BDRV_REQ_ZERO_WRITE flag is set, in this case only
check dst and dst->bs. This bug was introduced when moving in the
request tracking code from bdrv_co_copy_range, in 37aec7d75e.
This especially fixes the possible segfault when initializing src_bs
with a NULL src.
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-id: 20180703023758.14422-2-famz@redhat.com
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. Now that all callers of vectored I/O have been converted
to use our preferred byte-based bdrv_co_p{read,write}v(), we can
delete the unused bdrv_co_{read,write}v().
Furthermore, this gets rid of the signature difference between the
public bdrv_co_writev() and the callback .bdrv_co_writev (the
latter still exists, because some drivers still need more work
before they are fully byte-based).
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
in_flight and tracked requests need to be tracked in every layer during
recursion. For now the only user is qemu-img convert where overlapping
requests and IOThreads don't exist, therefore this change doesn't make
much difference form user point of view, but it is incorrect as part of
the API. Fix it.
Reported-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
When growing an image, block drivers (especially protocol drivers) may
initialise the newly added area. I/O requests to the same area need to
wait for this initialisation to be completed so that data writes don't
get overwritten and reads don't read uninitialised data.
To avoid overhead in the fast I/O path by adding new locking in the
protocol drivers and to restrict the impact to requests that actually
touch the new area, reuse the existing tracked request infrastructure in
block/io.c and mark all discard requests as serialising.
With this change, it is safe for protocol drivers to make
.bdrv_co_truncate actually asynchronous.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
This moves the bdrv_truncate() implementation from block.c to block/io.c
so it can have access to the tracked requests infrastructure.
This involves making refresh_total_sectors() public (in block_int.h).
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
bdrv_drain_all_*() used bdrv_next() to iterate over all root nodes and
did a subtree drain for each of them. This works fine as long as the
graph is static, but sadly, reality looks different.
If the graph changes so that root nodes are added or removed, we would
have to compensate for this. bdrv_next() returns each root node only
once even if it's the root node for multiple BlockBackends or for a
monitor-owned block driver tree, which would only complicate things.
The much easier and more obviously correct way is to fundamentally
change the way the functions work: Iterate over all BlockDriverStates,
no matter who owns them, and drain them individually. Compensation is
only necessary when a new BDS is created inside a drain_all section.
Removal of a BDS doesn't require any action because it's gone afterwards
anyway.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
In the future, bdrv_drained_all_begin/end() will drain all invidiual
nodes separately rather than whole subtrees. This means that we don't
want to propagate the drain to all parents any more: If the parent is a
BDS, it will already be drained separately. Recursing to all parents is
unnecessary work and would make it an O(n²) operation.
Prepare the drain function for the changed drain_all by adding an
ignore_bds_parents parameter to the internal implementation that
prevents the propagation of the drain to BDS parents. We still (have to)
propagate it to non-BDS parents like BlockBackends or Jobs because those
are not drained separately.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Before we can introduce a single polling loop for all nodes in
bdrv_drain_all_begin(), we must make sure to run it outside of coroutine
context like we already do for bdrv_do_drained_begin().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We cannot allow aio_poll() in bdrv_drain_invoke(begin=true) until we're
done with propagating the drain through the graph and are doing the
single final BDRV_POLL_WHILE().
Just schedule the coroutine with the callback and increase bs->in_flight
to make sure that the polling phase will wait for it.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_do_drained_begin() is only safe if we have a single
BDRV_POLL_WHILE() after quiescing all affected nodes. We cannot allow
that parent callbacks introduce a nested polling loop that could cause
graph changes while we're traversing the graph.
Split off bdrv_do_drained_begin_quiesce(), which only quiesces a single
node without waiting for its requests to complete. These requests will
be waited for in the BDRV_POLL_WHILE() call down the call chain.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Anything can happen inside BDRV_POLL_WHILE(), including graph
changes that may interfere with its callers (e.g. child list iteration
in recursive callers of bdrv_do_drained_begin).
Switch to a single BDRV_POLL_WHILE() call for the whole subtree at the
end of bdrv_do_drained_begin() to avoid such effects. The recursion
happens now inside the loop condition. As the graph can only change
between bdrv_drain_poll() calls, but not inside of it, doing the
recursion here is safe.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
For bdrv_drain(), recursively waiting for child node requests is
pointless because we didn't quiesce their parents, so new requests could
come in anyway. Letting the function work only on a single node makes it
more consistent.
For subtree drains and drain_all, we already have the recursion in
bdrv_do_drained_begin(), so the extra recursion doesn't add anything
either.
Remove the useless code.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
We already requested that block jobs be paused in .bdrv_drained_begin,
but no guarantee was made that the job was actually inactive at the
point where bdrv_drained_begin() returned.
This introduces a new callback BdrvChildRole.bdrv_drained_poll() and
uses it to make bdrv_drain_poll() consider block jobs using the node to
be drained.
For the test case to work as expected, we have to switch from
block_job_sleep_ns() to qemu_co_sleep_ns() so that the test job is even
considered active and must be waited for when draining the node.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Commit 91af091f92 added an additional aio_poll() to BDRV_POLL_WHILE()
in order to make sure that all pending BHs are executed on drain. This
was the wrong place to make the fix, as it is useless overhead for all
other users of the macro and unnecessarily complicates the mechanism.
This patch effectively reverts said commit (the context has changed a
bit and the code has moved to AIO_WAIT_WHILE()) and instead polls in the
loop condition for drain.
The effect is probably hard to measure in any real-world use case
because actual I/O will dominate, but if I run only the initialisation
part of 'qemu-img convert' where it calls bdrv_block_status() for the
whole image to find out how much data there is copy, this phase actually
needs only roughly half the time after this patch.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
All involved nodes are already idle, we called bdrv_do_drain_begin() on
them.
The comment in the code suggested that this was not correct because the
completion of a request on one node could spawn a new request on a
different node (which might have been drained before, so we wouldn't
drain the new request). In reality, new requests to different nodes
aren't spawned out of nothing, but only in the context of a parent
request, and they aren't submitted to random nodes, but only to child
nodes. As long as we still poll for the completion of the parent request
(which we do), draining each root node separately is good enough.
Remove the additional polling code from bdrv_drain_all_begin() and
replace it with an assertion that all nodes are already idle after we
drained them separately.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
All callers pass false for the 'recursive' parameter now. Remove it.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
bdrv_do_drain_begin/end() implement already everything that
bdrv_drain_all_begin/end() need and currently still do manually: Disable
external events, call parent drain callbacks, call block driver
callbacks.
It also does two more things:
The first is incrementing bs->quiesce_counter. bdrv_drain_all() already
stood out in the test case by behaving different from the other drain
variants. Adding this is not only safe, but in fact a bug fix.
The second is calling bdrv_drain_recurse(). We already do that later in
the same function in a loop, so basically doing an early first iteration
doesn't hurt.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
As long as nobody keeps the other I/O thread from working, there is no
reason why bdrv_drain() wouldn't work with cross-AioContext events. The
key is that the root request we're waiting for is in the AioContext
we're polling (which it always is for bdrv_drain()) so that aio_poll()
is woken up in the end.
Add a test case that shows that it works. Remove the comment in
bdrv_drain() that claims otherwise.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Introduce the bdrv_co_copy_range() API for copy offloading. Block
drivers implementing this API support efficient copy operations that
avoid reading each block from the source device and writing it to the
destination devices. Examples of copy offload primitives are SCSI
EXTENDED COPY and Linux copy_file_range(2).
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20180601092648.24614-2-famz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Message-id: 20180421132929.21610-5-mreitz@redhat.com
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
This flag signifies that a write request will not change the visible
disk content. With this flag set, it is sufficient to have the
BLK_PERM_WRITE_UNCHANGED permission instead of BLK_PERM_WRITE.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Message-id: 20180421132929.21610-4-mreitz@redhat.com
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
We have too many driver callback interfaces; simplify the mess
somewhat by merging the flags parameter of .bdrv_co_writev_flags()
into .bdrv_co_writev(). Note that as long as a driver doesn't set
.supported_write_flags, the flags argument will be 0 and behavior is
identical. Also note that the public function bdrv_co_writev() still
lacks a flags argument; so the driver signature is thus intentionally
slightly different. But that's not the end of the world, nor the first
time that the driver interface differs slightly from the public
interface.
Ideally, we should be rewriting all of these drivers to use modern
byte-based interfaces. But that's a more invasive patch to write
and audit, compared to the simplification done here.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. Now that all drivers with aio callbacks are using the
byte-based interfaces, we can remove the sector-based versions.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. Add new sector-based aio callbacks for read and write,
to match the fact that bdrv_aio_pdiscard is already byte-based.
Ideally, drivers should be converted to use coroutine callbacks
rather than aio; but that is not quite as trivial (and if we were
to do that conversion, the null-aio driver would disappear), so for
the short term, converting the signature but keeping things with
aio is easier. However, we CAN declare that a driver that uses
the byte-based aio interfaces now defaults to byte-based
operations, and must explicitly provide a refresh_limits override
to stick with larger alignments (making the alignment issues more
obvious directly in the drivers touched in the next few patches).
Once all drivers are converted, the sector-based aio callbacks will
be removed; in the meantime, a FIXME comment is added due to a
slight inefficiency that will be touched up as part of that later
cleanup.
Simplify some instances of 'bs->drv' into 'drv' while touching this,
since the local variable already exists to reduce typing.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
qemu_aio_coroutine_enter() is (indirectly) called recursively when
processing co_queue_wakeup. This can lead to stack exhaustion.
This patch rewrites co_queue_wakeup in an iterative fashion (instead of
recursive) with bounded memory usage to prevent stack exhaustion.
qemu_co_queue_run_restart() is inlined into qemu_aio_coroutine_enter()
and the qemu_coroutine_enter() call is turned into a loop to avoid
recursion.
There is one change that is worth mentioning: Previously, when
coroutine A queued coroutine B, qemu_co_queue_run_restart() entered
coroutine B from coroutine A. If A was terminating then it would still
stay alive until B yielded. After this patch B is entered by A's parent
so that a A can be deleted immediately if it is terminating.
It is safe to make this change since B could never interact with A if it
was terminating anyway.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20180322152834.12656-3-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
BlockDriverState has the BDRV_POLL_WHILE() macro to wait on event loop
activity while a condition evaluates to true. This is used to implement
synchronous operations where it acts as a condvar between the IOThread
running the operation and the main loop waiting for the operation. It
can also be called from the thread that owns the AioContext and in that
case it's just a nested event loop.
BlockBackend needs this behavior but doesn't always have a
BlockDriverState it can use. This patch extracts BDRV_POLL_WHILE() into
the AioWait abstraction, which can be used with AioContext and isn't
tied to BlockDriverState anymore.
This feature could be built directly into AioContext but then all users
would kick the event loop even if they signal different conditions.
Imagine an AioContext with many BlockDriverStates, each time a request
completes any waiter would wake up and re-check their condition. It's
nicer to keep a separate AioWait object for each condition instead.
Please see "block/aio-wait.h" for details on the API.
The name AIO_WAIT_WHILE() avoids the confusion between AIO_POLL_WHILE()
and AioContext polling.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The normal bdrv_co_pwritev() use is either
- BDRV_REQ_ZERO_WRITE clear and iovector provided
- BDRV_REQ_ZERO_WRITE set and iovector == NULL
while
- the flag clear and iovector == NULL is an assertion failure
in bdrv_co_do_zero_pwritev()
- the flag set and iovector provided is in fact allowed
(the flag prevails and zeroes are written)
However the alignment logic does not support the latter case so the padding
areas get overwritten with zeroes.
Currently, general functions like bdrv_rw_co() do provide iovector
regardless of flags. So, keep it supported and use bdrv_co_do_zero_pwritev()
alignment for it which also makes the code a bit more obvious anyway.
Signed-off-by: Anton Nefedov <anton.nefedov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. Now that all drivers have been updated to provide the
byte-based .bdrv_co_block_status(), we can delete the sector-based
interface.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. Update the generic helpers, and all passthrough clients
(blkdebug, commit, mirror, throttle) accordingly.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. Now that the block layer exposes byte-based allocation,
it's time to tackle the drivers. Add a new callback that operates
on as small as byte boundaries. Subsequent patches will then update
individual drivers, then finally remove .bdrv_co_get_block_status().
The new code also passes through the 'want_zero' hint, which will
allow subsequent patches to further optimize callers that only care
about how much of the image is allocated (want_zero is false),
rather than full details about runs of zeroes and which offsets the
allocation actually maps to (want_zero is true). As part of this
effort, fix another part of the documentation: the claim in commit
4c41cb4 that BDRV_BLOCK_ALLOCATED is short for 'DATA || ZERO' is a
lie at the block layer (see commit e88ae2264), even though it is
how the bit is computed from the driver layer. After all, there
are intentionally cases where we return ZERO but not ALLOCATED at
the block layer, when we know that a read sees zero because the
backing file is too short. Note that the driver interface is thus
slightly different than the public interface with regards to which
bits will be set, and what guarantees are provided on input.
We also add an assertion that any driver using the new callback will
make progress (the only time pnum will be 0 is if the block layer
already handled an out-of-bounds request, or if there is an error);
the old driver interface did not provide this guarantee, which
could lead to some inf-loops in drastic corner-case failures.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Allow block driver to map and unmap a buffer for later I/O, as a performance
hint.
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20180116060901.17413-5-famz@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
We need to remember how many of the drain sections in which a node is
were recursive (i.e. subtree drain rather than node drain), so that they
can be correctly applied when children are added or removed during the
drained section.
With this change, it is safe to modify the graph even inside a
bdrv_subtree_drained_begin/end() section.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_drained_begin() waits for the completion of requests in the whole
subtree, but it only actually keeps its immediate bs parameter quiesced
until bdrv_drained_end().
Add a version that keeps the whole subtree drained. As of this commit,
graph changes cannot be allowed during a subtree drained section, but
this will be fixed soon.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This is in preparation for subtree drains, i.e. drained sections that
affect not only a single node, but recursively all child nodes, too.
Calling the parent callbacks for drain is pointless when we just came
from that parent node recursively and leads to multiple increases of
bs->quiesce_counter in a single drain call. Don't do it.
In order for this to work correctly, the parent callback must be called
for every bdrv_drain_begin/end() call, not only for the outermost one:
If we have a node N with two parents A and B, recursive draining of A
should cause the quiesce_counter of B to increase because its child N is
drained independently of B. If now B is recursively drained, too, A must
increase its quiesce_counter because N is drained independently of A
only now, even if N is going from quiesce_counter 1 to 2.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_do_drained_begin() restricts the call of parent callbacks and
aio_disable_external() to the outermost drain section, but the block
driver callbacks are always called. bdrv_do_drained_end() must match
this behaviour, otherwise nodes stay drained even if begin/end calls
were balanced.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Block jobs are already paused using the BdrvChildRole drain callbacks,
so we don't need an additional block_job_pause_all() call.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_drained_begin() doesn't increase bs->quiesce_counter recursively
and also doesn't notify other parent nodes of children, which both means
that the child nodes are not actually drained, and bdrv_drained_begin()
is providing useful functionality only on a single node.
To keep things consistent, we also shouldn't call the block driver
callbacks recursively.
A proper recursive drain version that provides an actually working
drained section for child nodes will be introduced later.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Drain requests are propagated to child nodes, parent nodes and directly
to the AioContext. The order in which this happened was different
between all combinations of drain/drain_all and begin/end.
The correct order is to keep children only drained when their parents
are also drained. This means that at the start of a drained section, the
AioContext needs to be drained first, the parents second and only then
the children. The correct order for the end of a drained section is the
opposite.
This patch changes the three other functions to follow the example of
bdrv_drained_begin(), which is the only one that got it right.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
The device is drained, so there is no point in waiting for requests at
the end of the drained section. Remove the bdrv_drain_recurse() calls
there.
The bdrv_drain_recurse() calls were introduced in commit 481cad48e5
in order to call the .bdrv_co_drain_end() driver callback. This is now
done by a separate bdrv_drain_invoke() call.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Now that the bdrv_drain_invoke() calls are pulled up to the callers of
bdrv_drain_recurse(), the 'begin' parameter isn't needed any more.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
bdrv_drain_all_begin() used to call the .bdrv_co_drain_begin() driver
callback inside its polling loop. This means that how many times it got
called for each node depended on long it had to poll the event loop.
This is obviously not right and results in nodes that stay drained even
after bdrv_drain_all_end(), which calls .bdrv_co_drain_begin() once per
node.
Fix bdrv_drain_all_begin() to call the callback only once, too.
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
This change separates bdrv_drain_invoke(), which calls the BlockDriver
drain callbacks, from bdrv_drain_recurse(). Instead, the function
performs its own recursion now.
One reason for this is that bdrv_drain_recurse() can be called multiple
times by bdrv_drain_all_begin(), but the callbacks may only be called
once. The separation is necessary to fix this bug.
The other reason is that we intend to go to a model where we call all
driver callbacks first, and only then start polling. This is not fully
achieved yet with this patch, as bdrv_drain_invoke() contains a
BDRV_POLL_WHILE() loop for the block driver callbacks, which can still
call callbacks for any unrelated event. It's a step in this direction
anyway.
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
The .drained_begin/end callbacks can (directly or indirectly via
aio_poll()) cause block nodes to be removed or the current BdrvChild to
point to a different child node.
Use QLIST_FOREACH_SAFE() to make sure we don't access invalid
BlockDriverStates or accidentally continue iterating the parents of the
new child node instead of the node we actually came from.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Tested-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We currently do not guard everywhere against a NULL bs->drv where we
should be doing so. Most of the places fixed here just do not care
about that case at all.
Some care implicitly, e.g. through a prior function call to
bdrv_getlength() which would always fail for an ejected BDS. Add an
assert there to make it more obvious.
Other places seem to care, but do so insufficiently: Freeing clusters in
a qcow2 image is an error-free operation, but it may leave the image in
an unusable state anyway. Giving qcow2_free_clusters() an error code is
not really viable, it is much easier to note that bs->drv may be NULL
even after a successful driver call. This concerns bdrv_co_flush(), and
the way the check is added to bdrv_co_pdiscard() (in every iteration
instead of only once).
Finally, some places employ at least an assert(bs->drv); somewhere, that
may be reasonable (such as in the reopen code), but in
bdrv_has_zero_init(), it is definitely not. Returning 0 there in case
of an ejected BDS saves us much headache instead.
Reported-by: R. Nageswara Sastry <nasastry@in.ibm.com>
Buglink: https://bugs.launchpad.net/qemu/+bug/1728660
Signed-off-by: Max Reitz <mreitz@redhat.com>
Message-id: 20171110203111.7666-4-mreitz@redhat.com
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Now that bdrv_is_allocated accepts non-aligned inputs, we can
remove the TODO added in commit d6a644bb.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Any device that has request_alignment greater than 512 should be
unable to report status at a finer granularity; it may also be
simpler for such devices to be guaranteed that the block layer
has rounded things out to the granularity boundary (the way the
block layer already rounds all other I/O out). Besides, getting
the code correct for super-sector alignment also benefits us
for the fact that our public interface now has byte granularity,
even though none of our drivers have byte-level callbacks.
Add an assertion in blkdebug that proves that the block layer
never requests status of unaligned sections, similar to what it
does on other requests (while still keeping the generic helper
in place for when future patches add a throttle driver). Note
that iotest 177 already covers this (it would fail if you use
just the blkdebug.c hunk without the io.c changes). Meanwhile,
we can drop assertions in callers that no longer have to pass
in sector-aligned addresses.
There is a mid-function scope added for 'count' and 'longret',
for a couple of reasons: first, an upcoming patch will add an
'if' statement that checks whether a driver has an old- or
new-style callback, and can conveniently use the same scope for
less indentation churn at that time. Second, since we are
trying to get rid of sector-based computations, wrapping things
in a scope makes it easier to group and see what will be
deleted in a final cleanup patch once all drivers have been
converted to the new-style callback.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. In the common case, allocation is unlikely to ever use
values that are not naturally sector-aligned, but it is possible
that byte-based values will let us be more precise about allocation
at the end of an unaligned file that can do byte-based access.
Changing the name of the function from bdrv_get_block_status_above()
to bdrv_block_status_above() ensures that the compiler enforces that
all callers are updated. Likewise, since it a byte interface allows
an offset mapping that might not be sector aligned, split the mapping
out of the return value and into a pass-by-reference parameter. For
now, the io.c layer still assert()s that all uses are sector-aligned,
but that can be relaxed when a later patch implements byte-based
block status in the drivers.
For the most part this patch is just the addition of scaling at the
callers followed by inverse scaling at bdrv_block_status(), plus
updates for the new split return interface. But some code,
particularly bdrv_block_status(), gets a lot simpler because it no
longer has to mess with sectors. Likewise, mirror code no longer
computes s->granularity >> BDRV_SECTOR_BITS, and can therefore drop
an assertion about alignment because the loop no longer depends on
alignment (never mind that we don't really have a driver that
reports sub-sector alignments, so it's not really possible to test
the effect of sub-sector mirroring). Fix a neighboring assertion to
use is_power_of_2 while there.
For ease of review, bdrv_get_block_status() was tackled separately.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually converting to byte-based interfaces, as they are
easier to reason about than sector-based. Convert another internal
type (no semantic change), and rename it to match the corresponding
public function rename.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually converting to byte-based interfaces, as they are
easier to reason about than sector-based. Convert another internal
function (no semantic change).
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually converting to byte-based interfaces, as they are
easier to reason about than sector-based. Convert another internal
type (no semantic change), and rename it to match the corresponding
public function rename.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually converting to byte-based interfaces, as they are
easier to reason about than sector-based. Convert another internal
function (no semantic change); and as with its public counterpart,
rename to bdrv_co_block_status() and split the offset return, to
make the compiler enforce that we catch all uses. For now, we
assert that callers and the return value still use aligned data,
but ultimately, this will be the function where we hand off to a
byte-based driver callback, and will eventually need to add logic
to ensure we round calls according to the driver's
request_alignment then touch up the result handed back to the
caller, to start permitting a caller to pass unaligned offsets.
Note that we are now prepared to accepts 'bytes' larger than INT_MAX;
this is okay as long as we clamp things internally before violating
any 32-bit limits, and makes no difference to how a client will
use the information (clients looping over the entire file must
already be prepared for consecutive calls to return the same status,
as drivers are already free to return shorter-than-maximal status
due to any other convenient split points, such as when the L2 table
crosses cluster boundaries in qcow2).
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. In the common case, allocation is unlikely to ever use
values that are not naturally sector-aligned, but it is possible
that byte-based values will let us be more precise about allocation
at the end of an unaligned file that can do byte-based access.
Changing the name of the function from bdrv_get_block_status() to
bdrv_block_status() ensures that the compiler enforces that all
callers are updated. For now, the io.c layer still assert()s that
all callers are sector-aligned, but that can be relaxed when a later
patch implements byte-based block status in the drivers.
There was an inherent limitation in returning the offset via the
return value: we only have room for BDRV_BLOCK_OFFSET_MASK bits, which
means an offset can only be mapped for sector-aligned queries (or,
if we declare that non-aligned input is at the same relative position
modulo 512 of the answer), so the new interface also changes things to
return the offset via output through a parameter by reference rather
than mashed into the return value. We'll have some glue code that
munges between the two styles until we finish converting all uses.
For the most part this patch is just the addition of scaling at the
callers followed by inverse scaling at bdrv_block_status(), coupled
with the tweak in calling convention. But some code, particularly
bdrv_is_allocated(), gets a lot simpler because it no longer has to
mess with sectors.
For ease of review, bdrv_get_block_status_above() will be tackled
separately.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually converting to byte-based interfaces, as they are
easier to reason about than sector-based. Change the internal
loop iteration of zeroing a device to track by bytes instead of
sectors (although we are still guaranteed that we iterate by steps
that are sector-aligned).
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
In the process of converting sector-based interfaces to bytes,
I'm finding it easier to represent a byte count as a 64-bit
integer at the block layer (even if we are internally capped
by SIZE_MAX or even INT_MAX for individual transactions, it's
still nicer to not have to worry about truncation/overflow
issues on as many variables). Update the signature of
bdrv_round_to_clusters() to uniformly use int64_t, matching
the signature already chosen for bdrv_is_allocated and the
fact that off_t is also a signed type, then adjust clients
according to the required fallout (even where the result could
now exceed 32 bits, no client is directly assigning the result
into a 32-bit value without breaking things into a loop first).
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Not all callers care about which BDS owns the mapping for a given
range of the file, or where the zeroes lie within that mapping. In
particular, bdrv_is_allocated() cares more about finding the
largest run of allocated data from the guest perspective, whether
or not that data is consecutive from the host perspective, and
whether or not the data reads as zero. Therefore, doing subsequent
refinements such as checking how much of the format-layer
allocation also satisfies BDRV_BLOCK_ZERO at the protocol layer is
wasted work - in the best case, it just costs extra CPU cycles
during a single bdrv_is_allocated(), but in the worst case, it
results in a smaller *pnum, and forces callers to iterate through
more status probes when visiting the entire file for even more
extra CPU cycles.
This patch only optimizes the block layer (no behavior change when
want_zero is true, but skip unnecessary effort when it is false).
Then when subsequent patches tweak the driver callback to be
byte-based, we can also pass this hint through to the driver.
Tweak BdrvCoGetBlockStatusData to declare arguments in parameter
order, rather than mixing things up (minimizing padding is not
necessary here).
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Not all callers care about which BDS owns the mapping for a given
range of the file. This patch merely simplifies the callers by
consolidating the logic in the common call point, while guaranteeing
a non-NULL file to all the driver callbacks, for no semantic change.
The only caller that does not care about pnum is bdrv_is_allocated,
as invoked by vvfat; we can likewise add assertions that the rest
of the stack does not have to worry about a NULL pnum.
Furthermore, this will also set the stage for a future cleanup: when
a caller does not care about which BDS owns an offset, it would be
nice to allow the driver to optimize things to not have to return
BDRV_BLOCK_OFFSET_VALID in the first place. In the case of fragmented
allocation (for example, it's fairly easy to create a qcow2 image
where consecutive guest addresses are not at consecutive host
addresses), the current contract requires bdrv_get_block_status()
to clamp *pnum to the limit where host addresses are no longer
consecutive, but allowing a NULL file means that *pnum could be
set to the full length of known-allocated data.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
BlockDriverState has a bdrv_co_drain() callback but no equivalent for
the end of the drain. The throttle driver (block/throttle.c) needs a way
to mark the end of the drain in order to toggle io_limits_disabled
correctly, thus bdrv_co_drain_end is needed.
Signed-off-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Improve our braindead copy-on-read implementation. Pre-patch,
we have multiple issues:
- we create a bounce buffer and perform a write for the entire
request, even if the active image already has 99% of the
clusters occupied, and really only needs to copy-on-read the
remaining 1% of the clusters
- our bounce buffer was as large as the read request, and can
needlessly exhaust our memory by using double the memory of
the request size (the original request plus our bounce buffer),
rather than a capped maximum overhead beyond the original
- if a driver has a max_transfer limit, we are bypassing the
normal code in bdrv_aligned_preadv() that fragments to that
limit, and instead attempt to read the entire buffer from the
driver in one go, which some drivers may assert on
- a client can request a large request of nearly 2G such that
rounding the request out to cluster boundaries results in a
byte count larger than 2G. While this cannot exceed 32 bits,
it DOES have some follow-on problems:
-- the call to bdrv_driver_pread() can assert for exceeding
BDRV_REQUEST_MAX_BYTES, if the driver is old and lacks
.bdrv_co_preadv
-- if the buffer is all zeroes, the subsequent call to
bdrv_co_do_pwrite_zeroes is a no-op due to a negative size,
which means we did not actually copy on read
Fix all of these issues by breaking up the action into a loop,
where each iteration is capped to sane limits. Also, querying
the allocation status allows us to optimize: when data is
already present in the active layer, we don't need to bounce.
Note that the code has a telling comment that copy-on-read
should probably be a filter driver rather than a bolt-on hack
in io.c; but that remains a task for another day.
CC: qemu-stable@nongnu.org
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Make it possible to inject errors on writes performed during a
read operation due to copy-on-read semantics.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Handle a 0-length block status request up front, with a uniform
return value claiming the area is not allocated.
Most callers don't pass a length of 0 to bdrv_get_block_status()
and friends; but it definitely happens with a 0-length read when
copy-on-read is enabled. While we could audit all callers to
ensure that they never make a 0-length request, and then assert
that fact, it was just as easy to fix things to always report
success (as long as the callers are careful to not go into an
infinite loop). However, we had inconsistent behavior on whether
the status is reported as allocated or defers to the backing
layer, depending on what callbacks the driver implements, and
possibly wasting quite a few CPU cycles to get to that answer.
Consistently reporting unallocated up front doesn't really hurt
anything, and makes it easier both for callers (0-length requests
now have well-defined behavior) and for drivers (drivers don't
have to deal with 0-length requests).
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Both callers already had bytes available, but were scaling to
sectors. Move the scaling to internal code. In the case of
bdrv_aligned_pwritev(), we are now passing the exact offset
rather than a rounded sector-aligned value, but that's okay
as long as dirty bitmap widens start/bytes to granularity
boundaries.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_co_get_block_status_from_file() and
bdrv_co_get_block_status_from_backing() set *file to bs->file and
bs->backing respectively, so that bdrv_co_get_block_status() can recurse
to them. Future block drivers won't have to duplicate code to implement
this.
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
There are trace probes in bdrv_co_readv|writev, however, the
block drivers are being gradually moved over to using the
bdrv_co_preadv|pwritev functions instead. As a result some
block drivers miss the current probes. Move the probes
into bdrv_co_preadv|pwritev instead, so that they are triggered
by more (all?) I/O code paths.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170804105036.11879-1-berrange@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
We've had a shadowed 'ret' variable, which risks returning the wrong
value, introduced in commit b9c64947.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20170710150559.30163-1-den@openvz.org
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Eric Blake <eblake@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This will let the callback take a CoMutex in the next patch.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20170629132749.997-8-pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
It will be needed in following commits for persistent bitmaps.
If bitmap is loaded from read-only storage (and we can't mark it
"in use" in this storage) corresponding BdrvDirtyBitmap should be
read-only.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-id: 20170628120530.31251-11-vsementsov@virtuozzo.com
Signed-off-by: Max Reitz <mreitz@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. In the common case, allocation is unlikely to ever use
values that are not naturally sector-aligned, but it is possible
that byte-based values will let us be more precise about allocation
at the end of an unaligned file that can do byte-based access.
Changing the signature of the function to use int64_t *pnum ensures
that the compiler enforces that all callers are updated. For now,
the io.c layer still assert()s that all callers are sector-aligned,
but that can be relaxed when a later patch implements byte-based
block status. Therefore, for the most part this patch is just the
addition of scaling at the callers followed by inverse scaling at
bdrv_is_allocated(). But some code, particularly stream_run(),
gets a lot simpler because it no longer has to mess with sectors.
Leave comments where we can further simplify by switching to
byte-based iterations, once later patches eliminate the need for
sector-aligned operations.
For ease of review, bdrv_is_allocated() was tackled separately.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_is_allocated_above() was relying on intermediate->total_sectors,
which is a field that can have stale contents depending on the value
of intermediate->has_variable_length. An audit shows that we are safe
(we were first calling through bdrv_co_get_block_status() which in
turn calls bdrv_nb_sectors() and therefore just refreshed the current
length), but it's nicer to favor our accessor functions to avoid having
to repeat such an audit, even if it means refresh_total_sectors() is
called more frequently.
Suggested-by: John Snow <jsnow@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We are gradually moving away from sector-based interfaces, towards
byte-based. In the common case, allocation is unlikely to ever use
values that are not naturally sector-aligned, but it is possible
that byte-based values will let us be more precise about allocation
at the end of an unaligned file that can do byte-based access.
Changing the signature of the function to use int64_t *pnum ensures
that the compiler enforces that all callers are updated. For now,
the io.c layer still assert()s that all callers are sector-aligned
on input and that *pnum is sector-aligned on return to the caller,
but that can be relaxed when a later patch implements byte-based
block status. Therefore, this code adds usages like
DIV_ROUND_UP(,BDRV_SECTOR_SIZE) to callers that still want aligned
values, where the call might reasonbly give non-aligned results
in the future; on the other hand, no rounding is needed for callers
that should just continue to work with byte alignment.
For the most part this patch is just the addition of scaling at the
callers followed by inverse scaling at bdrv_is_allocated(). But
some code, particularly bdrv_commit(), gets a lot simpler because it
no longer has to mess with sectors; also, it is now possible to pass
NULL if the caller does not care how much of the image is allocated
beyond the initial offset. Leave comments where we can further
simplify once a later patch eliminates the need for sector-aligned
requests through bdrv_is_allocated().
For ease of review, bdrv_is_allocated_above() will be tackled
separately.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Now that the last user [mirror_iteration()] has converted to using
bytes, we no longer need a function to round sectors to clusters.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We document that *file is valid if the return is not an error and
includes BDRV_BLOCK_OFFSET_VALID, but forgot to obey this contract
when a driver (such as blkdebug) lacks a callback. Messed up in
commit 67a0fd2 (v2.6), when we added the file parameter.
Enhance qemu-iotest 177 to cover this, using a sequence that would
print garbage or even SEGV, because it was dererefencing through
uninitialized memory. [The resulting test output shows that we
have less-than-ideal block status from the blkdebug driver, but
that's a separate fix coming up soon.]
Setting *file on all paths that return BDRV_BLOCK_OFFSET_VALID is
enough to fix the crash, but we can go one step further: always
setting *file, even on error, means that a broken caller that
blindly dereferences file without checking for error is now more
likely to get a reliable SEGV instead of randomly acting on garbage,
making it easier to diagnose such buggy callers. Adding an
assertion that file is set where expected doesn't hurt either.
CC: qemu-stable@nongnu.org
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
When we have a BDS with unallocated clusters, but asking the status
of its underlying bs->file or backing layer encounters an end-of-file
condition, we know that the rest of the unallocated area will read as
zeroes. However, pre-patch, this required two separate calls to
bdrv_get_block_status(), as the first call stops at the point where
the underlying file ends. Thanks to BDRV_BLOCK_EOF, we can now widen
the results of the primary status if the secondary status already
includes BDRV_BLOCK_ZERO.
In turn, this fixes a TODO mentioned in iotest 154, where we can now
see that all sectors in a partial cluster at the end of a file read
as zero when coupling the shorter backing file's status along with our
knowledge that the remaining sectors came from an unallocated cluster.
Also, note that the loop in bdrv_co_get_block_status_above() had an
inefficent exit: in cases where the active layer sets BDRV_BLOCK_ZERO
but does NOT set BDRV_BLOCK_ALLOCATED (namely, where we know we read
zeroes merely because our unallocated clusters lie beyond the backing
file's shorter length), we still ended up probing the backing layer
even though we already had a good answer.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20170505021500.19315-3-eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Just as the block layer already sets BDRV_BLOCK_ALLOCATED as a
shortcut for subsequent operations, there are also some optimizations
that are made easier if we can quickly tell that *pnum will advance
us to the end of a file, via a new BDRV_BLOCK_EOF which gets set
by the block layer.
This just plumbs up the new bit; subsequent patches will make use
of it.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20170505021500.19315-2-eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Change the 'int count' parameter in *pwrite_zeros, *pdiscard related
functions (and some others) to 'int bytes', as they both refer to bytes.
This helps with code legibility.
Signed-off-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
Message-id: 20170609101808.13506-1-el13635@mail.ntua.gr
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Calling aio_poll() directly may have been fine previously, but this is
the future, man! The difference between an aio_poll() loop and
BDRV_POLL_WHILE() is that BDRV_POLL_WHILE() releases the AioContext
around aio_poll().
This allows the IOThread to run fd handlers or BHs to complete the
request. Failure to release the AioContext causes deadlocks.
Using BDRV_POLL_WHILE() partially fixes a 'savevm' hang with -object
iothread.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Call bdrv_inc/dec_in_flight() for vmstate reads/writes. This seems
unnecessary at first glance because vmstate reads/writes are done
synchronously while the guest is stopped. But we need the bdrv_wakeup()
in bdrv_dec_in_flight() so the main loop sees request completion.
Besides, it's cleaner to count vmstate reads/writes like ordinary
read/write requests.
The bdrv_wakeup() partially fixes a 'savevm' hang with -object iothread.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Remove use of block_job_pause/resume from outside blockjob.c, thus
making them static. The new functions are used by the block layer,
so place them in blockjob_int.h.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Message-id: 20170508141310.8674-5-pbonzini@redhat.com
Signed-off-by: Jeff Cody <jcody@redhat.com>
Since we are already in coroutine context during the body of
bdrv_co_get_block_status(), we can shave off a few layers of
wrappers when recursing to query the protocol when a format driver
returned BDRV_BLOCK_RAW.
Note that we are already using the correct recursion later on in
the same function, when probing whether the protocol layer is sparse
in order to find out if we can add BDRV_BLOCK_ZERO to an existing
BDRV_BLOCK_DATA|BDRV_BLOCK_OFFSET_VALID.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170504173745.27414-1-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
tail_padding_bytes is calculated wrong. F.e. for
offset = 0
bytes = 2048
align = 512
we will have tail_padding_bytes = 512 which is definitely wrong. The patch
fixes that arithmetics.
Fortunately this problem is harmless, we will have 1 extra allocation and
free thus there is no need to put this into stable. The problem is here
from the very beginning.
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reported by Coverity. We already use bs in bdrv_inc_in_flight before
checking for NULL. It is unnecessary as all callers pass non-NULL bs, so
drop it.
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This reverts commit e3e0003a8f.
This commit was necessary for the 2.9 release because we were unable to
fix the underlying issue(s) in time. However, we will be for 2.10.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Acked-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The recursive bdrv_drain_recurse may run a block job completion BH that
drops nodes. The coming changes will make that more likely and use-after-free
would happen without this patch
Stash the bs pointer and use bdrv_ref/bdrv_unref in addition to
QLIST_FOREACH_SAFE to prevent such a case from happening.
Since bdrv_unref accesses global state that is not protected by the AioContext
lock, we cannot use bdrv_ref/bdrv_unref unconditionally. Fortunately the
protection is not needed in IOThread because only main loop can modify a graph
with the AioContext lock held.
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-Id: <20170418143044.12187-2-famz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Tested-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
In case of block migration, there may be writes to BlockBackends that do
not have the write permission taken. Before this issue is fixed (which
is not going to happen in 2.9), we therefore cannot assert that this is
the case.
Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Tested-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 20170411145050.31290-1-mreitz@redhat.com
Tested-by: Laurent Vivier <lvivier@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
bdrv_inc_in_flight and bdrv_dec_in_flight are mandatory for
BDRV_POLL_WHILE to work, even for the shortcut case where flush is
unnecessary. Move the if block to below bdrv_dec_in_flight, and BTW fix
the variable declaration position.
Signed-off-by: Fam Zheng <famz@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
BDRV_POLL_WHILE waits for the started I/O by releasing bs's ctx then polling
the main context, which relies on the yielded coroutine continuing on bs->ctx
before notifying qemu_aio_context with bdrv_wakeup().
Thus, using qemu_coroutine_enter to start I/O is wrong because if the coroutine
is entered from main loop, co->ctx will be qemu_aio_context, as a result of the
"release, poll, acquire" loop of BDRV_POLL_WHILE, race conditions happen when
both main thread and the iothread access the same BDS:
main loop iothread
-----------------------------------------------------------------------
blockdev_snapshot
aio_context_acquire(bs->ctx)
virtio_scsi_data_plane_handle_cmd
bdrv_drained_begin(bs->ctx)
bdrv_flush(bs)
bdrv_co_flush(bs) aio_context_acquire(bs->ctx).enter
...
qemu_coroutine_yield(co)
BDRV_POLL_WHILE()
aio_context_release(bs->ctx)
aio_context_acquire(bs->ctx).return
...
aio_co_wake(co)
aio_poll(qemu_aio_context) ...
co_schedule_bh_cb() ...
qemu_coroutine_enter(co) ...
/* (A) bdrv_co_flush(bs) /* (B) I/O on bs */
continues... */
aio_context_release(bs->ctx)
aio_context_acquire(bs->ctx)
Note that in above case, bdrv_drained_begin() doesn't do the "release,
poll, acquire" in BDRV_POLL_WHILE, because bs->in_flight == 0.
Fix this by using bdrv_coroutine_enter and enter coroutine in the right
context.
iotests 109 output is updated because the coroutine reenter flow during
mirror job complete is different (now through co_queue_wakeup, instead
of the unconditional qemu_coroutine_switch before), making the end job
len different.
Signed-off-by: Fam Zheng <famz@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
The assertion is currently failing. We can't require callers to have
write permissions when all they are doing is a read, so comment it out.
Add a FIXME comment in the code so that the check is re-enabled when
copy on read is refactored into its own filter driver.
Reported-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Richard W.M. Jones <rjones@redhat.com>
This fixes bdrv_co_get_block_status() for the bdrv_mirror_top block
driver, which must fall through to bs->backing instead of bs->file.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
This adds an assertion that ensures that the necessary resize permission
has been granted before bdrv_truncate() is called.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Acked-by: Fam Zheng <famz@redhat.com>
This adds assertions that ensure that the necessary write permissions
have been granted before someone attempts to write to a node.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
This is where we want to check the permissions, so we need to have the
BdrvChild around where they are stored.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
All that CoQueue needs in order to become thread-safe is help
from an external mutex. Add this to the API.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213181244.16297-6-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-16-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-15-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-13-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
AioContext is fairly self contained, the only dependency is QEMUTimer but
that in turn doesn't need anything else. So move them out of block-obj-y
to avoid introducing a dependency from io/ to block-obj-y.
main-loop and its dependency iohandler also need to be moved, because
later in this series io/ will call iohandler_get_aio_context.
[Changed copyright "the QEMU team" to "other QEMU contributors" as
suggested by Daniel Berrange and agreed by Paolo.
--Stefan]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213135235.12274-2-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
bdrv_io_plug and bdrv_io_unplug are only called (via their
BlockBackend equivalents) after starting asynchronous I/O.
bdrv_drain is not going to be called while they are running,
because---even if a coroutine runs for some reason---it will
only drain in the next iteration of the event loop through
bdrv_co_yield_to_drain.
So this mechanism is unnecessary, get rid of it.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20161129113334.605-1-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Discard is advisory, so rounding the requests to alignment
boundaries is never semantically wrong from the data that
the guest sees. But at least the Dell Equallogic iSCSI SANs
has an interesting property that its advertised discard
alignment is 15M, yet documents that discarding a sequence
of 1M slices will eventually result in the 15M page being
marked as discarded, and it is possible to observe which
pages have been discarded.
Between commits 9f1963b and b8d0a980, we converted the block
layer to a byte-based interface that ultimately ignores any
unaligned head or tail based on the driver's advertised
discard granularity, which means that qemu 2.7 refuses to
pass any discard request smaller than 15M down to the Dell
Equallogic hardware. This is a slight regression in behavior
compared to earlier qemu, where a guest executing discards
in power-of-2 chunks used to be able to get every page
discarded, but is now left with various pages still allocated
because the guest requests did not align with the hardware's
15M pages.
Since the SCSI specification says nothing about a minimum
discard granularity, and only documents the preferred
alignment, it is best if the block layer gives the driver
every bit of information about discard requests, rather than
rounding it to alignment boundaries early.
Rework the block layer discard algorithm to mirror the write
zero algorithm: always peel off any unaligned head or tail
and manage that in isolation, then do the bulk of the request
on an aligned boundary. The fallback when the driver returns
-ENOTSUP for an unaligned request is to silently ignore that
portion of the discard request; but for devices that can pass
the partial request all the way down to hardware, this can
result in the hardware coalescing requests and discarding
aligned pages after all.
Reported by: Peter Lieven <pl@kamp.de>
CC: qemu-stable@nongnu.org
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Commit 443668ca rewrote the write_zeroes logic to guarantee that
an unaligned request never crosses a cluster boundary. But
in the rewrite, the new code assumed that at most one iteration
would be needed to get to an alignment boundary.
However, it is easy to trigger an assertion failure: the Linux
kernel limits loopback devices to advertise a max_transfer of
only 64k. Any operation that requires falling back to writes
rather than more efficient zeroing must obey max_transfer during
that fallback, which means an unaligned head may require multiple
iterations of the write fallbacks before reaching the aligned
boundaries, when layering a format with clusters larger than 64k
atop the protocol of file access to a loopback device.
Test case:
$ qemu-img create -f qcow2 -o cluster_size=1M file 10M
$ losetup /dev/loop2 /path/to/file
$ qemu-io -f qcow2 /dev/loop2
qemu-io> w 7m 1k
qemu-io> w -z 8003584 2093056
In fairness to Denis (as the original listed author of the culprit
commit), the faulty logic for at most one iteration is probably all
my fault in reworking his idea. But the solution is to restore what
was in place prior to that commit: when dealing with an unaligned
head or tail, iterate as many times as necessary while fragmenting
the operation at max_transfer boundaries.
Reported-by: Ed Swierk <eswierk@skyportsystems.com>
CC: qemu-stable@nongnu.org
CC: Denis V. Lunev <den@openvz.org>
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Commit 3ff2f67a changed bdrv_co_flush() so that no flush is issues if
the image hasn't been dirtied since the last flush. This is not quite
correct: The condition should be that the image hasn't been dirtied
since the last _successful_ flush. This patch changes the logic
accordingly.
Without this fix, subsequent bdrv_co_flush() calls would return success
without actually doing anything even though the image is still dirty.
The difference is visible in some blkdebug test cases where error
messages incorrectly disappeared after commit 3ff2f67a.
Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: John Snow <jsnow@redhat.com>
Message-id: 1478300595-10090-1-git-send-email-kwolf@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
bdrv_drain_all() doesn't allow the caller to do anything after all
pending requests have been completed but before block jobs are
resumed.
This patch splits bdrv_drain_all() into _begin() and _end() for that
purpose. It also adds aio_{disable,enable}_external() calls to disable
external clients in the meantime.
An important restriction of this split is that no new block jobs or
BlockDriverStates can be created between the bdrv_drain_all_begin()
and bdrv_drain_all_end() calls. This is not a concern now because
we'll only be using this in bdrv_reopen_multiple(), but it must be
dealt with if we ever have other uses cases in the future.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
aio_poll is not thread safe; for example bdrv_drain can hang if
the last in-flight I/O operation is completed in the I/O thread after
the main thread has checked bs->in_flight.
The bug remains latent as long as all of it is called within
aio_context_acquire/aio_context_release, but this will change soon.
To fix this, if bdrv_drain is called from outside the I/O thread,
signal the main AioContext through a dummy bottom half. The event
loop then only runs in the I/O thread.
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <1477565348-5458-18-git-send-email-pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
We want the BDS event loop to run exclusively in the iothread that
owns the BDS's AioContext. This macro will provide the synchronization
between the two event loops; for now it just wraps the common idiom
of a while loop around aio_poll.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-Id: <1477565348-5458-8-git-send-email-pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
bdrv_requests_pending is checking children to also wait until internal
requests (such as metadata writes) have completed. However, checking
children is in general overkill. Children requests can be of two kinds:
- requests caused by an operation on bs, e.g. a bdrv_aio_write to bs
causing a write to bs->file->bs. In this case, the parent's in_flight
count will always be incremented by at least one for every request in
the child.
- asynchronous metadata writes or flushes. Such writes can be started
even if bs's in_flight count is zero, but not after the .bdrv_drain
callback has been invoked.
This patch therefore changes bdrv_drain to finish I/O in the parent
(after which the parent's in_flight will be locked to zero), call
bdrv_drain (after which the parent will not generate I/O on the child
anymore), and then wait for internal I/O in the children to complete.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-Id: <1477565348-5458-6-git-send-email-pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Unlike tracked_requests, this field also counts throttled requests,
and remains non-zero if an AIO operation needs a BH to be "really"
completed.
With this change, it is no longer necessary to have a dummy
BdrvTrackedRequest for requests that are never serialising, and
it is no longer necessary to poll the AioContext once after
bdrv_requests_pending(bs) returns false.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-Id: <1477565348-5458-5-git-send-email-pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
This allows drivers to implement ioctls in a coroutine-based way.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
All read/write functions already have a single coroutine-based function
on the BlockBackend level through which all requests go (no matter what
API style the external caller used) and which passes the requests down
to the block node level.
This patch exports a bdrv_co_ioctl() function and uses it to extend this
mode of operation to ioctls.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
This simplifies bottom half handlers by removing calls to qemu_bh_delete and
thus removing the need to stash the bottom half pointer in the opaque
datum.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Commit fe1a9cbc moved the flush_all routine from the bdrv layer to the
block-backend layer. In doing so, however, the semantics of the routine
changed slightly such that flush_all now used blk_flush instead of
bdrv_flush.
blk_flush can fail if the attached device model reports that it is not
"available," (i.e. the tray is open.) This changed the semantics of
flush_all such that it can now fail for e.g. open CDROM drives.
Reintroduce bdrv_flush_all to regain the old semantics without having to
alter the behavior of blk_flush or blk_flush_all, which are already
'doing the right thing.'
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Acked-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Previously was added the assert:
commit 1755da16e3
Author: Paolo Bonzini <pbonzini@redhat.com>
Date: Thu Oct 18 16:49:18 2012 +0200
block: introduce new dirty bitmap functionality
Now the compressed write is always in coroutine and setting the bits is
done after the write, so that we can return the dirty_bitmaps for the
compressed writes.
Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Jeff Cody <jcody@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: John Snow <jsnow@redhat.com>
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
There are no block drivers left that implement the old
.bdrv_write_compressed interface, so it can be removed. Also now we have
no need to use the bdrv_pwrite_compressed function and we can remove it
entirely.
Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Jeff Cody <jcody@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: John Snow <jsnow@redhat.com>
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
For bdrv_pwrite_compressed() it looks like most of the code creating
coroutine is duplicated in bdrv_prwv_co(). So we can just add a flag
(BDRV_REQ_WRITE_COMPRESSED) and use bdrv_prwv_co() as a generic one.
In the end we get coroutine oriented function for write compressed by using
bdrv_co_pwritev/blk_co_pwritev with BDRV_REQ_WRITE_COMPRESSED flag.
Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Jeff Cody <jcody@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: John Snow <jsnow@redhat.com>
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
CC: Jeff Cody <jcody@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: John Snow <jsnow@redhat.com>
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This is a preparatory patch, which continues the general trend of the
transition to the byte-based interfaces. bdrv_check_request() and
blk_check_request() are no longer used, thus we can remove them.
Signed-off-by: Pavel Butsykin <pbutsykin@virtuozzo.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Jeff Cody <jcody@redhat.com>
CC: Markus Armbruster <armbru@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: John Snow <jsnow@redhat.com>
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This patch reduce CPU usage of flush operations a bit. When we have one
flush completed we should kick only next operation. We should not start
all pending operations in the hope that they will go back to wait on
wait_queue.
Also there is a technical possibility that requests will get reordered
with the previous approach. After wakeup all requests are removed from
the wait queue. They become active and they are processed one-by-one
adding to the wait queue in the same order. Though new flush can arrive
while all requests are not put into the queue.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Tested-by: Evgeny Yakovlev <eyakovlev@virtuozzo.com>
Signed-off-by: Evgeny Yakovlev <eyakovlev@virtuozzo.com>
Message-id: 1471457214-3994-3-git-send-email-den@openvz.org
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Fam Zheng <famz@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The following commit
commit 3ff2f67a7c
Author: Evgeny Yakovlev <eyakovlev@virtuozzo.com>
Date: Mon Jul 18 22:39:52 2016 +0300
block: ignore flush requests when storage is clean
has introduced a regression.
There is a problem that it is still possible for 2 requests to execute
in non sequential fashion and sometimes this results in a deadlock
when bdrv_drain_one/all are called for BDS with such stalled requests.
1. Current flushed_gen and flush_started_gen is 1.
2. Request 1 enters bdrv_co_flush to with write_gen 1 (i.e. the same
as flushed_gen). It gets past flushed_gen != flush_started_gen and
sets flush_started_gen to 1 (again, the same it was before).
3. Request 1 yields somewhere before exiting bdrv_co_flush
4. Request 2 enters bdrv_co_flush with write_gen 2. It gets past
flushed_gen != flush_started_gen and sets flush_started_gen to 2.
5. Request 2 runs to completion and sets flushed_gen to 2
6. Request 1 is resumed, runs to completion and sets flushed_gen to 1.
However flush_started_gen is now 2.
From here on out flushed_gen is always != to flush_started_gen and all
further requests will wait on flush_queue. This change replaces
flush_started_gen with an explicitly tracked active flush request.
Signed-off-by: Evgeny Yakovlev <eyakovlev@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Message-id: 1471457214-3994-2-git-send-email-den@openvz.org
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Fam Zheng <famz@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Dell Equallogic iSCSI SANs have a very unusual advertised geometry:
$ iscsi-inq -e 1 -c $((0xb0)) iscsi://XXX/0
wsnz:0
maximum compare and write length:1
optimal transfer length granularity:0
maximum transfer length:0
optimal transfer length:0
maximum prefetch xdread xdwrite transfer length:0
maximum unmap lba count:30720
maximum unmap block descriptor count:2
optimal unmap granularity:30720
ugavalid:1
unmap granularity alignment:0
maximum write same length:30720
which says that both the maximum and the optimal discard size
is 15M. It is not immediately apparent if the device allows
discard requests not aligned to the optimal size, nor if it
allows discards at a finer granularity than the optimal size.
I tried to find details in the SCSI Commands Reference Manual
Rev. A on what valid values of maximum and optimal sizes are
permitted, but while that document mentions a "Block Limits
VPD Page", I couldn't actually find documentation of that page
or what values it would have, or if a SCSI device has an
advertisement of its minimal unmap granularity. So it is not
obvious to me whether the Dell Equallogic device is compliance
with the SCSI specification.
Fortunately, it is easy enough to support non-power-of-2 sizing,
even if it means we are less efficient than truly possible when
targetting that device (for example, it means that we refuse to
unmap anything that is not a multiple of 15M and aligned to a
15M boundary, even if the device truly does support a smaller
granularity where unmapping actually works).
Reported-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1469129688-22848-5-git-send-email-eblake@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Now that all drivers have a byte-based .bdrv_co_pdiscard(), we
no longer need to worry about the sector-based version. We can
also relax our minimum alignment to 1 for drivers that support it.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 1468624988-423-18-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
There's enough drivers with a sector-based callback that it will
be easier to switch one at a time. This patch adds a byte-based
callback, and then after all drivers are swapped, we'll drop the
sector-based callback.
[checkpatch doesn't like the space after coroutine_fn in
block_int.h, but it's consistent with the rest of the file]
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 1468624988-423-10-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Another step towards byte-based interfaces everywhere. Replace
the sector-based driver callback .bdrv_aio_discard() with a new
byte-based .bdrv_aio_pdiscard(). Only raw-posix and RBD drivers
are affected, so it was not worth splitting into multiple patches.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 1468624988-423-9-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Another step towards byte-based interfaces everywhere. Replace
the sector-based bdrv_aio_discard() with a new byte-based
bdrv_aio_pdiscard(), which silently ignores any unaligned head
or tail. Driver callbacks will be converted in followup patches.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-id: 1468624988-423-5-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
BlockRequest is the internal struct used by bdrv_aio_*. At the
moment, all such calls were sector-based, but we will eventually
convert to byte-based; start by changing the internal variables
to be byte-based. No change to behavior, although the read and
write code can now go byte-based through more of the stack.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-id: 1468624988-423-4-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Another step towards byte-based interfaces everywhere. Replace
the sector-based bdrv_discard() with a new byte-based
bdrv_pdiscard(), which silently ignores any unaligned head
or tail.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 1468624988-423-3-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Another step towards byte-based interfaces everywhere. Replace
the sector-based bdrv_co_discard() with a new byte-based
bdrv_co_pdiscard(), which silently ignores any unaligned head
or tail. Driver callbacks will be converted in followup patches.
By calculating the alignment outside of the loop, and clamping
the max discard to an aligned value, we can simplify the actions
done within the loop.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 1468624988-423-2-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Drivers should be able to rely on the block layer honoring the
max transfer length, rather than needing to return -EINVAL
(iscsi) or manually fragment things (nbd). We already fragment
write zeroes at the block layer; this patch adds the fragmentation
for normal writes, after requests have been aligned (fragmenting
before alignment would lead to multiple unaligned requests, rather
than just the head and tail).
When fragmenting a large request where FUA was requested, but
where we know that FUA is implemented by flushing all requests
rather than the given request, then we can still get by with
only one flush. Note, however, that we need a followup patch
to the raw format driver to avoid a regression in the number of
flushes actually issued.
The return value was previously nebulous on success (sometimes
zero, sometimes the length written); since we never have a short
write, and since fragmenting may store yet another positive
value in 'ret', change the function to always return 0 on success,
matching what we do in bdrv_aligned_preadv().
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-id: 1468607524-19021-4-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Drivers should be able to rely on the block layer honoring the
max transfer length, rather than needing to return -EINVAL
(iscsi) or manually fragment things (nbd). This patch adds
the fragmentation in the block layer, after requests have been
aligned (fragmenting before alignment would lead to multiple
unaligned requests, rather than just the head and tail).
The return value was previously nebulous on success on whether
it was zero or the length read; and fragmenting may introduce
yet other non-zero values if we use the last length read. But
as at least some callers are sloppy and expect only zero on
success, it is easiest to just guarantee 0.
[Fix uninitialized ret local variable in bdrv_aligned_preadv().
--Stefan]
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-id: 1468607524-19021-2-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Some guests (win2008 server for example) do a lot of unnecessary
flushing when underlying media has not changed. This adds additional
overhead on host when calling fsync/fdatasync.
This change introduces a write generation scheme in BlockDriverState.
Current write generation is checked against last flushed generation to
avoid unnessesary flushes.
The problem with excessive flushing was found by a performance test
which does parallel directory tree creation (from 2 processes).
Results improved from 0.424 loops/sec to 0.432 loops/sec.
Each loop creates 10^3 directories with 10 files in each.
This affected some blkdebug testcases that were expecting error logs from
failure-injected flushes which are now skipped entirely
(tests 026 071 089).
This also affects the performance of block jobs and thus BLOCK_JOB_READY
events for driver-mirror and active block-commit commands now arrives
faster, before QMP send successfully returns to caller (tests 141 144).
Signed-off-by: Evgeny Yakovlev <eyakovlev@virtuozzo.com>
Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1468870792-7411-5-git-send-email-den@openvz.org
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Fam Zheng <famz@redhat.com>
CC: John Snow <jsnow@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
In practice the entry argument is always known at creation time, and
it is confusing that sometimes qemu_coroutine_enter is used with a
non-NULL argument to re-enter a coroutine (this happens in
block/sheepdog.c and tests/test-coroutine.c). So pass the opaque value
at creation time, for consistency with e.g. aio_bh_new.
Mostly done with the following semantic patch:
@ entry1 @
expression entry, arg, co;
@@
- co = qemu_coroutine_create(entry);
+ co = qemu_coroutine_create(entry, arg);
...
- qemu_coroutine_enter(co, arg);
+ qemu_coroutine_enter(co);
@ entry2 @
expression entry, arg;
identifier co;
@@
- Coroutine *co = qemu_coroutine_create(entry);
+ Coroutine *co = qemu_coroutine_create(entry, arg);
...
- qemu_coroutine_enter(co, arg);
+ qemu_coroutine_enter(co);
@ entry3 @
expression entry, arg;
@@
- qemu_coroutine_enter(qemu_coroutine_create(entry), arg);
+ qemu_coroutine_enter(qemu_coroutine_create(entry, arg));
@ reentry @
expression co;
@@
- qemu_coroutine_enter(co, NULL);
+ qemu_coroutine_enter(co);
except for the aforementioned few places where the semantic patch
stumbled (as expected) and for test_co_queue, which would otherwise
produce an uninitialized variable warning.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This is the final patch for converting the common I/O path to take
a BdrvChild parameter instead of BlockDriverState.
The completion of this conversion means that all users that perform I/O
on an image need to actually hold a reference (in the form of BdrvChild,
possible as part of a BlockBackend) to that image. This also protects
against inconsistent use of BlockBackend vs. BlockDriverState functions
because direct use of a BlockDriverState isn't possible any more and
blk->root is private for block-backends.c.
In addition, we can now distinguish different users in the I/O path,
and the future op blockers work is going to add assertions based on
permissions stored in BdrvChild.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
It makes more sense to have ALL block size limit constraints
in the same struct. Improve the documentation while at it.
Simplify a couple of conditionals, now that we have audited and
documented that request_alignment is always non-zero.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
During bdrv_merge_limits(), we were computing initial limits
based on another BDS in two places. At first glance, the two
computations are not identical (one is doing straight copying,
the other is doing merging towards or away from zero) - but
when you realize that the first round is starting with all-0
memory, all of the merging happens to work. Factoring out the
merging makes it easier to track how two BDS limits are merged,
in case we have future reasons to merge in even more limits.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Sector-based limits are awkward to think about; in our on-going
quest to move to byte-based interfaces, convert max_discard and
discard_alignment. Rename them, using 'pdiscard' as an aid to
track which remaining discard interfaces need conversion, and so
that the compiler will help us catch the change in semantics
across any rebased code. The BlockLimits type is now completely
byte-based; and in iscsi.c, sector_limits_lun2qemu() is no
longer needed.
pdiscard_alignment is made unsigned (we use power-of-2 alignments
as bitmasks, where unsigned is easier to think about) while
leaving max_pdiscard signed (since we still have an 'int'
interface); this is comparable to what commit cf081fc did for
write zeroes limits. We may later want to make everything an
unsigned 64-bit limit - but that requires a bigger code audit.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Sector-based limits are awkward to think about; in our on-going
quest to move to byte-based interfaces, convert max_transfer_length
and opt_transfer_length. Rename them (dropping the _length suffix)
so that the compiler will help us catch the change in semantics
across any rebased code, and improve the documentation. Use unsigned
values, so that we don't have to worry about negative values and
so that bit-twiddling is easier; however, we are still constrained
by 2^31 of signed int in most APIs.
When a value comes from an external source (iscsi and raw-posix),
sanitize the results to ensure that opt_transfer is a power of 2.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We want to eventually stick request_alignment alongside other
BlockLimits, but first, we must ensure it is populated at the
same time as all other limits, rather than being a special case
that is set only when a block is first opened.
Now that all drivers have been updated to supply an override
of request_alignment during their .bdrv_refresh_limits(), as
needed, the block layer itself can defer setting the default
alignment until part of the overall bdrv_refresh_limits().
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
If the amount of data to read ends exactly on the total size
of the bs, then we were wasting time creating a local qiov
to read the data in preparation for what would normally be
appending zeroes beyond the end, even though this corner case
has nothing further to do.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We don't pass any flags on to drivers to handle. Tighten an
assert to explain why we pass 0 to bdrv_driver_preadv(), and add
some comments on things to be aware of if we want to turn on
per-BDS BDRV_REQ_FUA support during reads in the future. Also,
document that we may want to consider using unmap during
copy-on-read operations where the read is all zeroes.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
For symmetry with bdrv_aligned_preadv(), assert that the caller
really has aligned things properly. This requires adding an align
parameter, which is used now only in the new asserts, but will
come in handy in a later patch that adds auto-fragmentation to the
max transfer size, since that value need not always be a multiple
of the alignment, and therefore must be rounded down.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This is mandatory for correct backup creation. In the other case the
content under this area would be lost.
Dirty bits are set exactly like in bdrv_aligned_pwritev, i.e. they are set
even if notifier has returned a error.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Vladimir Sementsov-Ogievskiy<vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1466093381-6120-4-git-send-email-den@openvz.org
CC: Fam Zheng <famz@redhat.com>
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Actually we must set dirty bitmap dirty after we have written all our
zeroes for correct processing in drive mirror code. In the other case
we can face not zeroes in this area in mirror_iteration.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Vladimir Sementsov-Ogievskiy<vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1466093381-6120-3-git-send-email-den@openvz.org
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The request area is specified in bytes, not in sectors.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Vladimir Sementsov-Ogievskiy<vsementsov@virtuozzo.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 1466093381-6120-2-git-send-email-den@openvz.org
CC: Stefan Hajnoczi <stefanha@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Max Reitz <mreitz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
bdrv_drain_all() pauses all block jobs by using bdrv_next() to iterate
over all top-level BlockDriverStates. Therefore the code is unable to
find block jobs in other nodes.
This patch uses block_job_next() to iterate over all block jobs.
Signed-off-by: Alberto Garcia <berto@igalia.com>
Message-id: 55ee7d7d4a65c28aa1a1b28823897ef326f328e2.1464346103.git.berto@igalia.com
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
It is always true for open images now.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
This allows drivers to share code between normal I/O and vmstate
accesses.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
The return value of .bdrv_load/save_vmstate() can be any non-negative
number in case of success now. It used to be bytes/-errno.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
This brings it in line with .bdrv_save_vmstate().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
We already have a byte-based bdrv_pwritev(), but the read counterpart
was still missing. This commit adds it.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
If block drivers say that they can do an alignment < 512 bytes, let's
just suppose they mean it. raw-posix used to be an offender with respect
to this, but it can actually deal with byte-aligned requests now.
The default is still 512 bytes for any drivers that only implement
sector-based interfaces, but it is 1 now for drivers that implement
.bdrv_co_preadv.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
This patch makes bdrv_aligned_preadv() ready to accept byte-aligned
requests. Note that this doesn't mean that such requests are actually
made. The caller still ensures that all requests are aligned to at least
512 bytes.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
In a first step to convert the common I/O path to work on bytes rather
than sectors, this converts the copy-on-read logic that is used by
bdrv_aligned_preadv().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Add a new BDRV_REQ_MASK constant, and use it to make sure that
caller flags are always valid.
Tested with 'make check' and with qemu-iotests on both '-raw'
and '-qcow2'; the only failure turned up was fixed in the
previous commit.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Drivers that implement .bdrv_co_pwritev() get the flags passed as an
argument to said function, but we also unconditionally emulate the flags
anyway. We shouldn't do that.
Fix this by clearing all flags that the driver supports natively after
it returns from .bdrv_co_pwritev().
Fixes: 4df863f3 ('block: Make supported_write_flags a per-bds property')
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Now that all drivers have been converted to a byte interface,
we no longer need a sector interface.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Rename to bdrv_pwrite_zeroes() to let the compiler ensure we
cater to the updated semantics. Do the same for bdrv_co_write_zeroes().
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Update bdrv_co_do_write_zeroes() to be byte-based, and select
between the new byte-based bdrv_co_pwrite_zeroes() or the old
bdrv_co_write_zeroes(). The next patches will convert drivers,
then remove the old interface.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Another step towards removing sector-based interfaces: convert
the maximum write and minimum alignment values from sectors to
bytes. Rename the variables to let the compiler check that all
users are converted to the new semantics.
The maximum remains an int as long as BDRV_REQUEST_MAX_SECTORS
is constrained by INT_MAX (this means that we can't even
support a 2G write_zeroes, but just under it) - changing
operation lengths to unsigned or to 64-bits is a much bigger
audit, and debatable if we even want to do it (since at the
core, a 32-bit platform will still have ssize_t as its
underlying limit on write()).
Meanwhile, alignment is changed to 'uint32_t', since it makes no
sense to have an alignment larger than the maximum write, and
less painful to use an unsigned type with well-defined behavior
in bit operations than to have to worry about what happens if
a driver mistakenly supplies a negative alignment.
Add an assert that no one was trying to use sectors to get a
write zeroes larger than 2G, and therefore that a later conversion
to bytes won't be impacted by keeping the limit at 32 bits.
Signed-off-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We should split requests even if they are less than write_zeroes_alignment.
For example we can have the following request:
offset 62k
size 4k
write_zeroes_alignment 64k
The original code sent 1 request covering 2 qcow2 clusters, and resulted
in both clusters being allocated. But by splitting the request, we can
cater to the case where one of the two clusters can be zeroed as a
whole, for only 1 cluster allocated after the operation.
Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Eric Blake <eblake@redhat.com>
CC: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1463476543-3087-2-git-send-email-den@openvz.org>
[eblake: Avoid exceeding nb_sectors, hoist alignment checks out of
loop, and update testsuite to show that patch works]
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Similar to the "!drv || !drv->bdrv_aio_ioctl" case above, here it is
okay to set co.ret and return. As pointed out by Paolo, a BH will be
created as necessary by the caller (bdrv_co_maybe_schedule_bh).
Besides, as pointed out by Kevin, "data" was leaked before.
Reported-by: Kevin Wolf <kwolf@redhat.com>
Reported-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20160601015223.19277-1-famz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
I was thrown by the fact that the public type BlockRequest had
an anonymous union, but no obvious discriminator. Turns out
that the only client of the second branch of the union was code
internal to io.c, now that commit 91c6e4b killed public
multiwrite, so move it into io.c and improve the comments.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1463699150-19445-1-git-send-email-eblake@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
in a read-modify-write cycle a small request might cause
head and tail to fall into the same aligned block. Currently
QEMU reads the same block twice in this case which is
not necessary.
Signed-off-by: Peter Lieven <pl@kamp.de>
Message-id: 1464607873-28206-1-git-send-email-pl@kamp.de
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This changes the backup block job to use the job's BlockBackend for
performing its I/O. job->bs isn't used by the backup code any more
afterwards.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
This changes the streaming block job to use the job's BlockBackend for
performing the COR reads. job->bs isn't used by the streaming code any
more afterwards.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Until now, bdrv_drained_begin() used bdrv_drain() internally to drain
the queue. This is kind of backwards and caused quiescing code to be
duplicated because bdrv_drained_begin() had to ensure that no new
requests come in even after bdrv_drain() returns, whereas bdrv_drain()
had to have them because it could be called from other places.
Instead move the bdrv_drain() code to bdrv_drained_begin() and make
bdrv_drain() a simple wrapper around bdrv_drained_begin/end().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
The bdrv_next() users all leaked the BdrvNextIterator after completing
the iteration. Simply changing bdrv_next() to free the iterator before
returning NULL at the end of list doesn't work because some callers exit
the loop before looking at all BDSes.
This patch moves the BdrvNextIterator from the heap to the stack of
the caller and switches to a bdrv_first()/bdrv_next() interface for
initialising the iterator.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
We need to introduce a separate BdrvNextIterator struct that can keep
more state than just the current BDS in order to avoid using the bs->blk
pointer.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Since virtio-blk implements request merging itself these days, the only
remaining users are test cases for the function. That doesn't make the
function exactly useful any more.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Checking whether there are throttled requests requires going to the
associated BlockBackend, which we want to avoid.
All users of bdrv_requests_pending() in block/io.c already call
bdrv_parent_drained_begin() first, which restarts all throttled
requests, so no throttled requests can be left here and this is removal
of dead code.
The remaining users (assertions during graph manipulation in block.c)
don't care about requests that are still queued in the BlockBackend and
haven't been issued for a BlockDriverState yet.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
So far, bdrv_parent_drained_begin/end() was called for the duration of
the actual bdrv_drain() at the beginning of a drained section, but we
really should keep parents quiesced until the end of the drained
section.
This does not actually change behaviour at this point because the only
user of the .drained_begin/end BdrvChildRole callback is I/O throttling,
which already doesn't send any new requests after flushing its queue in
.drained_begin. The patch merely removes a trap for future users.
Reported-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
This removes the last part of I/O throttling from block/io.c and moves
it to the BlockBackend.
Instead of having knowledge about throttling inside io.c, we can call a
BdrvChild callback .drained_begin/end, which happens to drain the
throttled requests for BlockBackend parents.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
This patch changes where the throttling state is stored (used to be the
BlockDriverState, now it is the BlockBackend), but it doesn't actually
make it a BB level feature yet. For example, throttling is still
disabled when the BDS is detached from the BB.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
As a first step towards moving I/O throttling to the BlockBackend level,
this patch changes all pointers in struct ThrottleGroup from referencing
a BlockDriverState to referencing a BlockBackend.
This change is valid because we made sure that throttling can only be
enabled on BDSes which have a BB attached.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
The block layer has a couple of cases where it can lose
Force Unit Access semantics when writing a large block of
zeroes, such that the request returns before the zeroes
have been guaranteed to land on underlying media.
SCSI does not support FUA during WRITESAME(10/16); FUA is only
supported if it falls back to WRITE(10/16). But where the
underlying device is new enough to not need a fallback, it
means that any upper layer request with FUA semantics was
silently ignoring BDRV_REQ_FUA.
Conversely, NBD has situations where it can support FUA but not
ZERO_WRITE; when that happens, the generic block layer fallback
to bdrv_driver_pwritev() (or the older bdrv_co_writev() in qemu
2.6) was losing the FUA flag.
The problem of losing flags unrelated to ZERO_WRITE has been
latent in bdrv_co_do_write_zeroes() since commit aa7bfbff, but
back then, it did not matter because there was no FUA flag. It
became observable when commit 93f5e6d8 paved the way for flags
that can impact correctness, when we should have been using
bdrv_co_writev_flags() with modified flags. Compare to commit
9eeb6dd, which got flag manipulation right in
bdrv_co_do_zero_pwritev().
Symptoms: I tested with qemu-io with default writethrough cache
(which is supposed to use FUA semantics on every write), and
targetted an NBD client connected to a server that intentionally
did not advertise NBD_FLAG_SEND_FUA. When doing 'write 0 512',
the NBD client sent two operations (NBD_CMD_WRITE then
NBD_CMD_FLUSH) to get the fallback FUA semantics; but when doing
'write -z 0 512', the NBD client sent only NBD_CMD_WRITE.
The fix is do to a cleanup bdrv_co_flush() at the end of the
operation if any step in the middle relied on a BDS that does
not natively support FUA for that step (note that we don't
need to flush after every operation, if the operation is broken
into chunks based on bounce-buffer sizing). Each BDS gains a
new flag .supported_zero_flags, which parallels the use of
.supported_write_flags but only when accessing a zero write
operation (the flags MUST be different, because of SCSI having
different semantics based on WRITE vs. WRITESAME; and also
because BDRV_REQ_MAY_UNMAP only makes sense on zero writes).
Also fix some documentation to describe -ENOTSUP semantics,
particularly since iscsi depends on those semantics.
Down the road, we may want to add a driver where its
.bdrv_co_pwritev() honors all three of BDRV_REQ_FUA,
BDRV_REQ_ZERO_WRITE, and BDRV_REQ_MAY_UNMAP, and advertise
this via bs->supported_write_flags for blocks opened by that
driver; such a driver should NOT supply .bdrv_co_write_zeroes
nor .supported_zero_flags. But none of the drivers touched
in this patch want to do that (the act of writing zeroes is
different enough from normal writes to deserve a second
callback).
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Pre-patch, .supported_write_flags lives at the driver level, which
means we are blindly declaring that all block devices using a
given driver will either equally support FUA, or that we need a
fallback at the block layer. But there are drivers where FUA
support is a per-block decision: the NBD block driver is dependent
on the remote server advertising NBD_FLAG_SEND_FUA (and has
fallback code to duplicate the flush that the block layer would do
if NBD had not set .supported_write_flags); and the iscsi block
driver is dependent on the mode sense bits advertised by the
underlying device (and is currently silently ignoring FUA requests
if the underlying device does not support FUA).
The fix is to make supported flags as a per-BDS option, set during
.bdrv_open(). This patch moves the variable and fixes NBD and iscsi
to set it only conditionally; later patches will then further
simplify the NBD driver to quit duplicating work done at the block
layer, as well as tackle the fact that SCSI does not support FUA
semantics on WRITESAME(10/16) but only on WRITE(10/16).
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
There are no block drivers left that implement the old .bdrv_read/write
interface, so it can be removed now. This gets us rid of the
corresponding emulation functions, too.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Many parts of the block layer are already byte granularity. The block
driver interface, however, was still missing an interface that allows
making use of this. This patch introduces a new BlockDriver interface,
which is based on coroutines, vectored, has flags and uses a byte
granularity. This is now the preferred interface for new drivers.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
It used to be an internal helper function just for implementing
bdrv_co_do_readv/writev(), but now that it's a public interface, it
deserves a name without "do" in it.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Instead of registering emulation functions as .bdrv_co_writev, just
directly check whether the function is there or not, and use the AIO
interface if it isn't. This makes the read/write functions more
consistent with how things are done in other places (flush, discard,
etc.)
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
This is a function that simply calls into the block driver for doing a
write, providing the byte granularity interface we want to eventually
have everywhere, and using whatever interface that driver supports.
This one is a bit more interesting than the version for reads: It adds
support for .bdrv_co_writev_flags() everywhere, so that drivers
implementing this function can drop .bdrv_co_writev() now.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
This is a function that simply calls into the block driver for doing a
read, providing the byte granularity interface we want to eventually
have everywhere, and using whatever interface that driver supports.
For now, this is just a wrapper for calling bs->drv->bdrv_co_readv().
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Extract the handling of io_plug "depth" from linux-aio.c and let the
main bdrv_drain loop do nothing but wait on I/O.
Like the two newly introduced functions, bdrv_io_plug and bdrv_io_unplug
now operate on all children. The visit order is now symmetrical between
plug and unplug, making it possible for formats to implement plug/unplug.
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Extract the handling of throttling from bdrv_flush_io_queue. These
new functions will soon become BdrvChildRole callbacks, as they can
be generalized to "beginning of drain" and "end of drain".
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Do not call bdrv_drain_recurse twice in bdrv_co_drain. A small
tweak to the logic in Fam's patch, which is harmless since no
one implements bdrv_drain anyway. But better get it right.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We want to remove throttled_reqs from block/io.c. This is the easy
part---hide the handling of throttled_reqs during disable/enable of
throttling within throttle-groups.c.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The return value is unused and I am not sure why it would be useful.
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
We had to disable I/O throttling with synchronous requests because we
didn't use to run timers in nested event loops when the code was
introduced. This isn't true any more, and throttling works just fine
even when using the synchronous API.
The removed code is in fact dead code since commit a8823a3b ('block: Use
blk_co_pwritev() for blk_write()') because I/O throttling can only be
set on the top layer, but BlockBackend always uses the coroutine
interface now instead of using the sync API emulation in block.c.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <1458660792-3035-2-git-send-email-kwolf@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Using the nested aio_poll() in coroutine is a bad idea. This patch
replaces the aio_poll loop in bdrv_drain with a BH, if called in
coroutine.
For example, the bdrv_drain() in mirror.c can hang when a guest issued
request is pending on it in qemu_co_mutex_lock().
Mirror coroutine in this case has just finished a request, and the block
job is about to complete. It calls bdrv_drain() which waits for the
other coroutine to complete. The other coroutine is a scsi-disk request.
The deadlock happens when the latter is in turn pending on the former to
yield/terminate, in qemu_co_mutex_lock(). The state flow is as below
(assuming a qcow2 image):
mirror coroutine scsi-disk coroutine
-------------------------------------------------------------
do last write
qcow2:qemu_co_mutex_lock()
...
scsi disk read
tracked request begin
qcow2:qemu_co_mutex_lock.enter
qcow2:qemu_co_mutex_unlock()
bdrv_drain
while (has tracked request)
aio_poll()
In the scsi-disk coroutine, the qemu_co_mutex_lock() will never return
because the mirror coroutine is blocked in the aio_poll(blocking=true).
With this patch, the added qemu_coroutine_yield() allows the scsi-disk
coroutine to make progress as expected:
mirror coroutine scsi-disk coroutine
-------------------------------------------------------------
do last write
qcow2:qemu_co_mutex_lock()
...
scsi disk read
tracked request begin
qcow2:qemu_co_mutex_lock.enter
qcow2:qemu_co_mutex_unlock()
bdrv_drain.enter
> schedule BH
> qemu_coroutine_yield()
> qcow2:qemu_co_mutex_lock.return
> ...
tracked request end
...
(resumed from BH callback)
bdrv_drain.return
...
Reported-by: Laurent Vivier <lvivier@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Message-id: 1459855253-5378-2-git-send-email-famz@redhat.com
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
