The fstat call can take a long time to finish when running over
NFS. Add a version of it that runs in the thread pool.
Adapt one of its users, raw_co_get_allocated_file size to use the new
version. That function is called via QMP under the qemu_global_mutex
so it has a large chance of blocking VCPU threads in case it takes too
long to finish.
Link: https://lore.kernel.org/r/20240409145917.6780-1-farosas@suse.de
Reviewed-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Signed-off-by: João Silva <jsilva@suse.de>
References: bsc#1211000
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Dario Faggioli <dfaggioli@suse.com>
Convert the remaining functions to make the QMP commands query-block
and query-named-block-nodes run in their entirety in a coroutine. With
this, any yield from those commands will return all the way back to
the main loop. This releases the BQL and the main loop and avoids
having the QMP command block another more important task from running.
Both commands need to be converted at once because hmp_info_block
calls both and it needs to be moved to a coroutine as well.
Now the wrapper for bdrv_co_get_allocated_file_size() can be made not
mixed and the wrapper for bdrv_co_block_device_info() can be removed.
Link: https://lore.kernel.org/r/20240409145917.6780-1-farosas@suse.de
Signed-off-by: Lin Ma <lma@suse.com>
References: bsc#1211000
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Dario Faggioli <dfaggioli@suse.com>
We're converting callers of bdrv_co_get_allocated_file_size() to run
in coroutines because that function will be made asynchronous when
called (indirectly) from the QMP dispatcher.
This function is a candidate because it calls bdrv_query_image_info()
-> bdrv_co_do_query_node_info() -> bdrv_co_get_allocated_file_size().
It is safe to turn this is a coroutine because the code it calls is
made up of either simple accessors and string manipulation functions
[1] or it has already been determined to be safe [2].
1) bdrv_refresh_filename(), bdrv_is_read_only(),
blk_enable_write_cache(), bdrv_cow_bs(), blk_get_public(),
throttle_group_get_name(), bdrv_write_threshold_get(),
bdrv_query_dirty_bitmaps(), throttle_group_get_config(),
bdrv_filter_or_cow_bs(), bdrv_skip_implicit_filters()
2) bdrv_co_do_query_node_info() (see previous commits);
This was the only caller of bdrv_query_image_info(), so we can remove
the wrapper for that function now.
Link: https://lore.kernel.org/r/20240409145917.6780-1-farosas@suse.de
References: bsc#1211000
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Dario Faggioli <dfaggioli@suse.com>
This function is a caller of bdrv_do_query_node_info(), which have
been converted to a coroutine. Convert this function as well so we're
closer from having the whole qmp_query_block as a single coroutine.
Also remove the wrapper for bdrv_co_do_query_node_info() now that all
its callers are converted.
Link: https://lore.kernel.org/r/20240409145917.6780-1-farosas@suse.de
References: bsc#1211000
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Dario Faggioli <dfaggioli@suse.com>
We're converting callers of bdrv_co_get_allocated_file_size() to run
in coroutines because that function will be made asynchronous when
called (indirectly) from the QMP dispatcher.
This function is a candidate because it calls bdrv_do_query_node_info(),
which in turn calls bdrv_co_get_allocated_file_size().
All the functions called from bdrv_do_query_node_info() onwards are
coroutine-safe, either have a coroutine version themselves[1] or are
mostly simple code/string manipulation[2].
1) bdrv_co_getlength(), bdrv_co_get_allocated_file_size(),
bdrv_co_get_info();
2) bdrv_refresh_filename(), bdrv_get_format_name(),
bdrv_get_full_backing_filename(), bdrv_query_snapshot_info_list(),
bdrv_get_specific_info();
Link: https://lore.kernel.org/r/20240409145917.6780-1-farosas@suse.de
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
References: bsc#1211000
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Dario Faggioli <dfaggioli@suse.com>
Move this function into a coroutine so we can convert the whole
qmp_query_block command into a coroutine in the next patches.
Placing the entire command in a coroutine allow us to yield all the
way back to the main loop, releasing the BQL and unblocking the main
loop.
When the whole conversion is completed, we'll be able to avoid a
priority inversion that happens when a QMP command calls a slow
(buggy) system call and blocks the vcpu thread from doing mmio due to
contention on the BQL.
About coroutine safety:
Most callees have coroutine versions themselves and thus are safe to
call in a coroutine. The remaining ones:
- bdrv_refresh_filename, bdrv_get_full_backing_filename: String
manipulation, nothing that would be unsafe for use in coroutines;
- bdrv_get_format_name: Just accesses a field;
- bdrv_get_specific_info, bdrv_query_snapshot_info_list: No locks or
anything that would poll or block.
(using a mixed wrapper for now, but after all callers are converted,
this can become a coroutine exclusively)
Link: https://lore.kernel.org/r/20240409145917.6780-1-farosas@suse.de
References: bsc#1211000
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Dario Faggioli <dfaggioli@suse.com>
There is a small window at the end of block device migration when
devices are being re-activated. This includes a resetting of some
fields of BDRVQcow2State at qcow2_co_invalidate_cache(). A concurrent
QMP query-block command can call qcow2_get_specific_info() during this
window and see the cleared values, which leads to an assert:
qcow2_get_specific_info: Assertion `false' failed
This is the same issue as Gitlab #1933, which has already been
resolved[1], but there the fix applied only to non-coroutine
commands. Once we move query-block to a coroutine the problem will
manifest again.
Add an operation blocker to the invalidation function to block the
query info path during this window.
Instead of failing query-block, which would be disruptive to users,
use the blocker to know when to reschedule the coroutine back into the
iohandler so it doesn't run while the BDRVQcow2State is inconsistent.
To avoid failing query-block when all block operations are blocked,
unblock the INFO operation at various places. This preserves the prior
situations where query-block used to work.
1 - https://gitlab.com/qemu-project/qemu/-/issues/1933
Link: https://lore.kernel.org/all/87bk6trl9i.fsf@suse.de/
Link: https://lore.kernel.org/r/20240409145917.6780-1-farosas@suse.de
References: bsc#1221812
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Dario Faggioli <dfaggioli@suse.com>
Some callers of this function are about to be converted to run in
coroutines, so allow it to be executed both inside and outside a
coroutine while we convert all the callers.
This will be reverted once all callers of bdrv_do_query_node_info run
in a coroutine.
Link: https://lore.kernel.org/r/20240409145917.6780-1-farosas@suse.de
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
References: bsc#1211000
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Dario Faggioli <dfaggioli@suse.com>
While using SCSI passthrough, Following scenario makes qemu doesn't
realized the capacity change of remote scsi target:
1. online resize the scsi target.
2. issue 'rescan-scsi-bus.sh -s ...' in host.
3. issue 'rescan-scsi-bus.sh -s ...' in vm.
In above scenario I used to experienced errors while accessing the
additional disk space in vm. I think the reasonable operations should
be:
1. online resize the scsi target.
2. issue 'rescan-scsi-bus.sh -s ...' in host.
3. issue 'block_resize' via qmp to notify qemu.
4. issue 'rescan-scsi-bus.sh -s ...' in vm.
The errors disappear once I notify qemu by block_resize via qmp.
So this patch replaces the number of logical blocks of READ CAPACITY
response from scsi target by qemu's bs->total_sectors. If the user in
vm wants to access the additional disk space, The administrator of
host must notify qemu once resizeing the scsi target.
Bonus is that domblkinfo of libvirt can reflect the consistent capacity
information between host and vm in case of missing block_resize in qemu.
E.g:
...
<disk type='block' device='lun'>
<driver name='qemu' type='raw'/>
<source dev='/dev/sdc' index='1'/>
<backingStore/>
<target dev='sda' bus='scsi'/>
<alias name='scsi0-0-0-0'/>
<address type='drive' controller='0' bus='0' target='0' unit='0'/>
</disk>
...
Before:
1. online resize the scsi target.
2. host:~ # rescan-scsi-bus.sh -s /dev/sdc
3. guest:~ # rescan-scsi-bus.sh -s /dev/sda
4 host:~ # virsh domblkinfo --domain $DOMAIN --human --device sda
Capacity: 4.000 GiB
Allocation: 0.000 B
Physical: 8.000 GiB
5. guest:~ # lsblk /dev/sda
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 8G 0 disk
└─sda1 8:1 0 2G 0 part
After:
1. online resize the scsi target.
2. host:~ # rescan-scsi-bus.sh -s /dev/sdc
3. guest:~ # rescan-scsi-bus.sh -s /dev/sda
4 host:~ # virsh domblkinfo --domain $DOMAIN --human --device sda
Capacity: 4.000 GiB
Allocation: 0.000 B
Physical: 8.000 GiB
5. guest:~ # lsblk /dev/sda
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 4G 0 disk
└─sda1 8:1 0 2G 0 part
References: [SUSE-JIRA] (SLE-20965)
Signed-off-by: Lin Ma <lma@suse.com>
Currently the QemuLockCnt data structure and associated functions are
in the include/qemu/thread.h header. Move them to their own
qemu/lockcnt.h. The main reason for doing this is that it means we
can autogenerate the documentation comments into the docs/devel
documentation.
The copyright/author in the new header is drawn from lockcnt.c,
since the header changes were added in the same commit as
lockcnt.c; since neither thread.h nor lockcnt.c state an explicit
license, the standard default of GPL-2-or-later applies.
We include the new header (and the .c file, which was accidentally
omitted previously) in the "RCU" part of MAINTAINERS, since that
is where the lockcnt.rst documentation is categorized.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20240816132212.3602106-7-peter.maydell@linaro.org
Indicate that 'MDTS and Size Limits Exclude Metadata (MEM)' in the
Controller Attributes (CTRATT) I/O Command Set Independent Identify
Controller Data Structure.
Signed-off-by: Arun Kumar <arun.kka@samsung.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
[k.jensen: updated commit message]
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
In the context of backup fleecing, discarding the source will not work
when the fleecing image has a larger granularity than the one used for
block-copy operations (can happen if the backup target has smaller
cluster size), because cbw_co_pdiscard_snapshot() will align down the
discard requests and thus effectively ignore then.
To make @discard-source work in such a scenario, allow specifying the
minimum cluster size used for block-copy operations and thus in
particular also the granularity for discard requests to the source.
The type 'size' (corresponding to uint64_t in C) is used in QAPI to
rule out negative inputs and for consistency with already existing
@cluster-size parameters. Since block_copy_calculate_cluster_size()
uses int64_t for its result, a check that the input is not too large
is added in block_copy_state_new() before calling it. The calculation
in block_copy_calculate_cluster_size() is done in the target int64_t
type.
Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Acked-by: Markus Armbruster <armbru@redhat.com> (QAPI schema)
Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Message-Id: <20240711120915.310243-2-f.ebner@proxmox.com>
[vsementsov: switch version to 9.2 in QAPI doc]
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Minor bugs and errors related to ufs-test are resolved. Some
permissions and code implementations that are not synchronized
with the ufs spec are edited.
Signed-off-by: Yoochan Jeong <yc01.jeong@samsung.com>
Reviewed-by: Jeuk Kim <jeuk20.kim@samsung.com>
Signed-off-by: Jeuk Kim <jeuk20.kim@samsung.com>
Allowing an unlimited number of clients to any web service is a recipe
for a rudimentary denial of service attack: the client merely needs to
open lots of sockets without closing them, until qemu no longer has
any more fds available to allocate.
For qemu-nbd, we default to allowing only 1 connection unless more are
explicitly asked for (-e or --shared); this was historically picked as
a nice default (without an explicit -t, a non-persistent qemu-nbd goes
away after a client disconnects, without needing any additional
follow-up commands), and we are not going to change that interface now
(besides, someday we want to point people towards qemu-storage-daemon
instead of qemu-nbd).
But for qemu proper, and the newer qemu-storage-daemon, the QMP
nbd-server-start command has historically had a default of unlimited
number of connections, in part because unlike qemu-nbd it is
inherently persistent until nbd-server-stop. Allowing multiple client
sockets is particularly useful for clients that can take advantage of
MULTI_CONN (creating parallel sockets to increase throughput),
although known clients that do so (such as libnbd's nbdcopy) typically
use only 8 or 16 connections (the benefits of scaling diminish once
more sockets are competing for kernel attention). Picking a number
large enough for typical use cases, but not unlimited, makes it
slightly harder for a malicious client to perform a denial of service
merely by opening lots of connections withot progressing through the
handshake.
This change does not eliminate CVE-2024-7409 on its own, but reduces
the chance for fd exhaustion or unlimited memory usage as an attack
surface. On the other hand, by itself, it makes it more obvious that
with a finite limit, we have the problem of an unauthenticated client
holding 100 fds opened as a way to block out a legitimate client from
being able to connect; thus, later patches will further add timeouts
to reject clients that are not making progress.
This is an INTENTIONAL change in behavior, and will break any client
of nbd-server-start that was not passing an explicit max-connections
parameter, yet expects more than 100 simultaneous connections. We are
not aware of any such client (as stated above, most clients aware of
MULTI_CONN get by just fine on 8 or 16 connections, and probably cope
with later connections failing by relying on the earlier connections;
libvirt has not yet been passing max-connections, but generally
creates NBD servers with the intent for a single client for the sake
of live storage migration; meanwhile, the KubeSAN project anticipates
a large cluster sharing multiple clients [up to 8 per node, and up to
100 nodes in a cluster], but it currently uses qemu-nbd with an
explicit --shared=0 rather than qemu-storage-daemon with
nbd-server-start).
We considered using a deprecation period (declare that omitting
max-parameters is deprecated, and make it mandatory in 3 releases -
then we don't need to pick an arbitrary default); that has zero risk
of breaking any apps that accidentally depended on more than 100
connections, and where such breakage might not be noticed under unit
testing but only under the larger loads of production usage. But it
does not close the denial-of-service hole until far into the future,
and requires all apps to change to add the parameter even if 100 was
good enough. It also has a drawback that any app (like libvirt) that
is accidentally relying on an unlimited default should seriously
consider their own CVE now, at which point they are going to change to
pass explicit max-connections sooner than waiting for 3 qemu releases.
Finally, if our changed default breaks an app, that app can always
pass in an explicit max-parameters with a larger value.
It is also intentional that the HMP interface to nbd-server-start is
not changed to expose max-connections (any client needing to fine-tune
things should be using QMP).
Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-12-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[ericb: Expand commit message to summarize Dan's argument for why we
break corner-case back-compat behavior without a deprecation period]
Signed-off-by: Eric Blake <eblake@redhat.com>
Upcoming patches to fix a CVE need to track an opaque pointer passed
in by the owner of a client object, as well as request for a time
limit on how fast negotiation must complete. Prepare for that by
changing the signature of nbd_client_new() and adding an accessor to
get at the opaque pointer, although for now the two servers
(qemu-nbd.c and blockdev-nbd.c) do not change behavior even though
they pass in a new default timeout value.
Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20240807174943.771624-11-eblake@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
[eblake: s/LIMIT/MAX_SECS/ as suggested by Dan]
Signed-off-by: Eric Blake <eblake@redhat.com>
Upstream clang 18 (and backports to clang 17 in Fedora and RHEL)
implemented support for __attribute__((cleanup())) in its Thread Safety
Analysis, so we can now actually have a proper implementation of
WITH_GRAPH_RDLOCK_GUARD() that understands when we acquire and when we
release the lock.
-Wthread-safety is now only enabled if the compiler is new enough to
understand this pattern. In theory, we could have used some #ifdefs to
keep the existing basic checks on old compilers, but as long as someone
runs a newer compiler (and our CI does), we will catch locking problems,
so it's probably not worth keeping multiple implementations for this.
The implementation can't use g_autoptr any more because the glib macros
define wrapper functions that don't have the right TSA attributes, so
the compiler would complain about them. Just use the cleanup attribute
directly instead.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20240627181245.281403-3-kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Extend copy command to copy user data across different namespaces via
support for specifying a namespace for each source range
Signed-off-by: Arun Kumar <arun.kka@samsung.com>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Commit 73064edfb8 ("hw/nvme: flexible data placement emulation")
intorudced NVMe FDP feature to nvme-subsys and nvme-ctrl with a
single endurance group #1 supported. This means that controller should
return proper identify data to host with Identify Endurance Group List
(CNS 19h). But, yes, only just for the endurance group #1. This patch
allows host applications to ask for which endurance group is available
and utilize FDP through that endurance group.
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Minwoo Im <minwoo.im@samsung.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
bdrv_file_open and bdrv_open are completely equivalent, they are
never checked except to see which one to invoke. So merge them
into a single one.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Libaio defines IO_CMD_FDSYNC command to sync all outstanding
asynchronous I/O operations, by flushing out file data to the
disk storage. Enable linux-aio to submit such aio request.
When using aio=native without fdsync() support, QEMU creates
pthreads, and destroying these pthreads results in TLB flushes.
In a real-time guest environment, TLB flushes cause a latency
spike. This patch helps to avoid such spikes.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Prasad Pandit <pjp@fedoraproject.org>
Message-ID: <20240425070412.37248-1-ppandit@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The main loop has two AioContexts: qemu_aio_context and iohandler_ctx.
The main loop runs them both, but nested aio_poll() calls on
qemu_aio_context exclude iohandler_ctx.
Which one should qemu_get_current_aio_context() return when called from
the main loop? Document that it's always qemu_aio_context.
This has subtle effects on functions that use
qemu_get_current_aio_context(). For example, aio_co_reschedule_self()
does not work when moving from iohandler_ctx to qemu_aio_context because
qemu_get_current_aio_context() does not differentiate these two
AioContexts.
Document this in order to reduce the chance of future bugs.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20240506190622.56095-3-stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This patch adds support for MCQ defined in UFSHCI 4.0. This patch
utilized the legacy I/O codes as much as possible to support MCQ.
MCQ operation & runtime register is placed at 0x1000 offset of UFSHCI
register statically with no spare space among four registers (48B):
UfsMcqSqReg, UfsMcqSqIntReg, UfsMcqCqReg, UfsMcqCqIntReg
The maxinum number of queue is 32 as per spec, and the default
MAC(Multiple Active Commands) are 32 in the device.
Example:
-device ufs,serial=foo,id=ufs0,mcq=true,mcq-maxq=8
Signed-off-by: Minwoo Im <minwoo.im@samsung.com>
Reviewed-by: Jeuk Kim <jeuk20.kim@samsung.com>
Message-Id: <20240528023106.856777-3-minwoo.im@samsung.com>
Signed-off-by: Jeuk Kim <jeuk20.kim@samsung.com>
Add a parameter that enables discard-after-copy. That is mostly useful
in "push backup with fleecing" scheme, when source is snapshot-access
format driver node, based on copy-before-write filter snapshot-access
API:
[guest] [snapshot-access] ~~ blockdev-backup ~~> [backup target]
| |
| root | file
v v
[copy-before-write]
| |
| file | target
v v
[active disk] [temp.img]
In this case discard-after-copy does two things:
- discard data in temp.img to save disk space
- avoid further copy-before-write operation in discarded area
Note that we have to declare WRITE permission on source in
copy-before-write filter, for discard to work. Still we can't take it
unconditionally, as it will break normal backup from RO source. So, we
have to add a parameter and pass it thorough bdrv_open flags.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Fiona Ebner <f.ebner@proxmox.com>
Tested-by: Fiona Ebner <f.ebner@proxmox.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20240313152822.626493-5-vsementsov@yandex-team.ru>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Currently block_copy creates copy_bitmap in source node. But that is in
bad relation with .independent_close=true of copy-before-write filter:
source node may be detached and removed before .bdrv_close() handler
called, which should call block_copy_state_free(), which in turn should
remove copy_bitmap.
That's all not ideal: it would be better if internal bitmap of
block-copy object is not attached to any node. But that is not possible
now.
The simplest solution is just create copy_bitmap in filter node, where
anyway two other bitmaps are created.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Fiona Ebner <f.ebner@proxmox.com>
Tested-by: Fiona Ebner <f.ebner@proxmox.com>
Message-Id: <20240313152822.626493-4-vsementsov@yandex-team.ru>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
During drain, we do not care about virtqueue notifications, which is why
we remove the handlers on it. When removing those handlers, whether vq
notifications are enabled or not depends on whether we were in polling
mode or not; if not, they are enabled (by default); if so, they have
been disabled by the io_poll_start callback.
Because we do not care about those notifications after removing the
handlers, this is fine. However, we have to explicitly ensure they are
enabled when re-attaching the handlers, so we will resume receiving
notifications. We do this in virtio_queue_aio_attach_host_notifier*().
If such a function is called while we are in a polling section,
attaching the notifiers will then invoke the io_poll_start callback,
re-disabling notifications.
Because we will always miss virtqueue updates in the drained section, we
also need to poll the virtqueue once after attaching the notifiers.
Buglink: https://issues.redhat.com/browse/RHEL-3934
Signed-off-by: Hanna Czenczek <hreitz@redhat.com>
Message-ID: <20240202153158.788922-3-hreitz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Introduce a new flag 'backing-mask-protocol' for the block-stream QMP
command which instructs the internals to use 'raw' instead of the
protocol driver in case when a image is used without a dummy 'raw'
wrapper.
The flag is designed such that it can be always asserted by management
tools even when there isn't any update to backing files.
The flag will be used by libvirt so that the backing images still
reference the proper format even when libvirt will stop using the dummy
raw driver (raw driver with no other config). Libvirt needs this so that
the images stay compatible with older libvirt versions which didn't
expect that a protocol driver name can appear in the backing file format
field.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <bbee9a0a59748a8893289bf8249f568f0d587e62.1701796348.git.pkrempa@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Introduce a new flag 'backing-mask-protocol' for the block-commit QMP
command which instructs the internals to use 'raw' instead of the
protocol driver in case when a image is used without a dummy 'raw'
wrapper.
The flag is designed such that it can be always asserted by management
tools even when there isn't any update to backing files.
The flag will be used by libvirt so that the backing images still
reference the proper format even when libvirt will stop using the dummy
raw driver (raw driver with no other config). Libvirt needs this so that
the images stay compatible with older libvirt versions which didn't
expect that a protocol driver name can appear in the backing file format
field.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <2cb46e37093ce793ea1604abc8bbb90f4c8e434b.1701796348.git.pkrempa@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The LuringState typedef is defined twice, in include/block/raw-aio.h and
block/io_uring.c. Move it in include/block/aio.h, which is included
everywhere the typedef is needed, since include/block/aio.h already has
to define the forward reference to the struct.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The Big QEMU Lock (BQL) has many names and they are confusing. The
actual QemuMutex variable is called qemu_global_mutex but it's commonly
referred to as the BQL in discussions and some code comments. The
locking APIs, however, are called qemu_mutex_lock_iothread() and
qemu_mutex_unlock_iothread().
The "iothread" name is historic and comes from when the main thread was
split into into KVM vcpu threads and the "iothread" (now called the main
loop thread). I have contributed to the confusion myself by introducing
a separate --object iothread, a separate concept unrelated to the BQL.
The "iothread" name is no longer appropriate for the BQL. Rename the
locking APIs to:
- void bql_lock(void)
- void bql_unlock(void)
- bool bql_locked(void)
There are more APIs with "iothread" in their names. Subsequent patches
will rename them. There are also comments and documentation that will be
updated in later patches.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Paul Durrant <paul@xen.org>
Acked-by: Fabiano Rosas <farosas@suse.de>
Acked-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Acked-by: Peter Xu <peterx@redhat.com>
Acked-by: Eric Farman <farman@linux.ibm.com>
Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Acked-by: Hyman Huang <yong.huang@smartx.com>
Reviewed-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20240102153529.486531-2-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
aio_context_set_aio_params() doesn't use its undocumented
Error** argument. Remove it to simplify.
Note this removes a use of "unchecked Error**" in
iothread_set_aio_context_params().
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20231120171806.19361-1-philmd@linaro.org>
The AioContext lock no longer exists.
There is one noteworthy change:
- * More specifically, these functions use BDRV_POLL_WHILE(bs), which
- * requires the caller to be either in the main thread and hold
- * the BlockdriverState (bs) AioContext lock, or directly in the
- * home thread that runs the bs AioContext. Calling them from
- * another thread in another AioContext would cause deadlocks.
+ * More specifically, these functions use BDRV_POLL_WHILE(bs), which requires
+ * the caller to be either in the main thread or directly in the home thread
+ * that runs the bs AioContext. Calling them from another thread in another
+ * AioContext would cause deadlocks.
I am not sure whether deadlocks are still possible. Maybe they have just
moved to the fine-grained locks that have replaced the AioContext. Since
I am not sure if the deadlocks are gone, I have kept the substance
unchanged and just removed mention of the AioContext.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-ID: <20231205182011.1976568-15-stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Delete these functions because nothing calls these functions anymore.
I introduced these APIs in commit 98563fc3ec ("aio: add
aio_context_acquire() and aio_context_release()") in 2014. It's with a
sigh of relief that I delete these APIs almost 10 years later.
Thanks to Paolo Bonzini's vision for multi-queue QEMU, we got an
understanding of where the code needed to go in order to remove the
limitations that the original dataplane and the IOThread/AioContext
approach that followed it.
Emanuele Giuseppe Esposito had the splendid determination to convert
large parts of the codebase so that they no longer needed the AioContext
lock. This was a painstaking process, both in the actual code changes
required and the iterations of code review that Emanuele eked out of
Kevin and me over many months.
Kevin Wolf tackled multitudes of graph locking conversions to protect
in-flight I/O from run-time changes to the block graph as well as the
clang Thread Safety Analysis annotations that allow the compiler to
check whether the graph lock is being used correctly.
And me, well, I'm just here to add some pizzazz to the QEMU multi-queue
block layer :). Thank you to everyone who helped with this effort,
including Eric Blake, code reviewer extraordinaire, and others who I've
forgotten to mention.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-ID: <20231205182011.1976568-11-stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This is the big patch that removes
aio_context_acquire()/aio_context_release() from the block layer and
affected block layer users.
There isn't a clean way to split this patch and the reviewers are likely
the same group of people, so I decided to do it in one patch.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Paul Durrant <paul@xen.org>
Message-ID: <20231205182011.1976568-7-stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Stop acquiring/releasing the AioContext lock in
bdrv_graph_wrlock()/bdrv_graph_unlock() since the lock no longer has any
effect.
The distinction between bdrv_graph_wrunlock() and
bdrv_graph_wrunlock_ctx() becomes meaningless and they can be collapsed
into one function.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20231205182011.1976568-6-stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_graph_wrunlock() calls aio_poll(), which may run callbacks that
have a nested event loop. Nested event loops can depend on other
iothreads making progress, so in order to allow them to make progress it
must not hold the AioContext lock of another thread while calling
aio_poll().
This introduces a @bs parameter to bdrv_graph_wrunlock() whose
AioContext is temporarily dropped (which matches bdrv_graph_wrlock()),
and a bdrv_graph_wrunlock_ctx() that can be used if the BlockDriverState
doesn't necessarily exist any more when unlocking.
This also requires a change to bdrv_schedule_unref(), which was relying
on the incorrectly taken lock. It needs to take the lock itself now.
While this is a separate bug, it can't be fixed a separate patch because
otherwise the intermediate state would either deadlock or try to release
a lock that we don't even hold.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20231115172012.112727-3-kwolf@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
[kwolf: Fixed up bdrv_schedule_unref()]
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Almost all functions that access bs->file already take the graph
lock now. Add locking to the remaining users and finally annotate the
struct field itself as protected by the graph lock.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20231027155333.420094-25-kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
This adds GRAPH_RDLOCK to some driver callbacks that are already called
with the graph lock held, and which will need the annotation because
they access bs->file, but don't have it yet.
This also covers a few callbacks that were not marked GRAPH_RDLOCK
before, but where updating BlockDriver is trivially possible.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20231027155333.420094-21-kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
bdrv_change_backing_file() is called both inside and outside coroutine
context. This makes it difficult for it to take the graph lock
internally. It also means that driver implementations need to be able to
run outside of coroutines, too. Switch it to the usual model with a
coroutine based implementation and a co_wrapper instead. The new
function is marked GRAPH_RDLOCK.
As the co_wrapper now runs the function in the AioContext of the node
(as it should always have done), this is not GLOBAL_STATE_CODE() any
more.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20231027155333.420094-20-kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Almost all functions that access bs->backing already take the graph
lock now. Add locking to the remaining users and finally annotate the
struct field itself as protected by the graph lock.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20231027155333.420094-18-kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Instead of taking the writer lock internally, require callers to already
hold it when calling bdrv_replace_node(). Its callers may already want
to hold the graph lock and so wouldn't be able to call functions that
take it internally.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20231027155333.420094-17-kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Instead of taking the writer lock internally, require callers to already
hold it when calling bdrv_set_backing_hd_drained(). Basically everthing
in the function needs the lock and its callers may already want to hold
the graph lock and so wouldn't be able to call functions that take it
internally.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20231027155333.420094-14-kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
