Some drivers might want to make use of auxiliary helper threads during VM
state loading, for example to make sure that their blocking (sync) I/O
operations don't block the rest of the migration process.
Add a migration core managed thread pool to facilitate this use case.
The migration core will wait for these threads to finish before
(re)starting the VM at destination.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Link: https://lore.kernel.org/qemu-devel/b09fd70369b6159c75847e69f235cb908b02570c.1741124640.git.maciej.szmigiero@oracle.com
Signed-off-by: Cédric Le Goater <clg@redhat.com>
This QEMU_VM_COMMAND sub-command and its switchover_start SaveVMHandler is
used to mark the switchover point in main migration stream.
It can be used to inform the destination that all pre-switchover main
migration stream data has been sent/received so it can start to process
post-switchover data that it might have received via other migration
channels like the multifd ones.
Add also the relevant MigrationState bit stream compatibility property and
its hw_compat entry.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Zhang Chen <zhangckid@gmail.com> # for the COLO part
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Link: https://lore.kernel.org/qemu-devel/311be6da85fc7e49a7598684d80aa631778dcbce.1741124640.git.maciej.szmigiero@oracle.com
Signed-off-by: Cédric Le Goater <clg@redhat.com>
Remove all instances of _fd_ from the migration generic code. These
functions have grown over time and the _fd_ part is now just
confusing.
migration_fd_error() -> migration_error() makes it a little
vague. Since it's only used for migration_connect() failures, change
it to migration_connect_set_error().
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-ID: <20250213175927.19642-4-farosas@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
There's no point passing the error into migration cancel only for it
to call migrate_set_error().
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-ID: <20250213175927.19642-2-farosas@suse.de>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
We're currently changing the way the source multifd migration handles
the shutdown of the multifd channels when TLS is in use to perform a
clean termination by calling gnutls_bye().
Older src QEMUs will always close the channel without terminating the
TLS session. New dst QEMUs treat an unclean termination as an error.
Add multifd_clean_tls_termination (default true) that can be switched
on the destination whenever a src QEMU <= 9.2 is in use.
(Note that the compat property is only strictly necessary for src
QEMUs older than 9.1. Due to synchronization coincidences, src QEMUs
9.1 and 9.2 can put the destination in a condition where it doesn't
see the unclean termination. Still, make the property more inclusive
to facilitate potential backports.)
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
The general expectation is that header files should follow the same
file/path naming scheme as the corresponding source file. There are
various historical exceptions to this practice in QEMU, with one of
the most notable being the include/qapi/qmp/ directory. Most of the
headers there correspond to source files in qobject/.
This patch corrects most of that inconsistency by creating
include/qobject/ and moving the headers for qobject/ there.
This also fixes MAINTAINERS for include/qapi/qmp/dispatch.h:
scripts/get_maintainer.pl now reports "QAPI" instead of "No
maintainers found".
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Acked-by: Halil Pasic <pasic@linux.ibm.com> #s390x
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20241118151235.2665921-2-armbru@redhat.com>
[Rebased]
Block devices have an individual active state, a single global flag
can't cover this correctly. This becomes more important as we allow
users to manually manage which nodes are active or inactive.
Now that it's allowed to call bdrv_inactivate_all() even when some
nodes are already inactive, we can remove the flag and just
unconditionally call bdrv_inactivate_all() and, more importantly,
bdrv_activate_all() before we make use of the nodes.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20250204211407.381505-5-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
QEMU machine has a property "suppress-vmdesc". When it is enabled, QEMU
will stop attaching JSON VM description at the end of the precopy migration
stream (postcopy is never affected because postcopy never attach that).
However even if it's suppressed by the user, the source QEMU will still
construct the JSON descriptions, which is a complete waste of CPU and
memory resources.
To avoid it, only create the JSON writer object if suppress-vmdesc is not
specified.
Luckily, vmstate_save() already supports vmdesc==NULL, so only a few spots
that are left to be prepared that vmdesc can be NULL now.
When at it, move the init / destroy of the JSON writer object to start /
end of the migration - the JSON writer object is a sub-struct of migration
state, and that looks like the only object that was dynamically allocated /
destroyed within migration process. Make it the same as the rest objects
that migration uses.
Signed-off-by: Peter Xu <peterx@redhat.com>
Tested-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Juraj Marcin <jmarcin@redhat.com>
Link: https://lore.kernel.org/r/20250114230746.3268797-3-peterx@redhat.com
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Add the cpr-transfer migration mode, which allows the user to transfer
a guest to a new QEMU instance on the same host with minimal guest pause
time, by preserving guest RAM in place, albeit with new virtual addresses
in new QEMU, and by preserving device file descriptors. Pages that were
locked in memory for DMA in old QEMU remain locked in new QEMU, because the
descriptor of the device that locked them remains open.
cpr-transfer preserves memory and devices descriptors by sending them to
new QEMU over a unix domain socket using SCM_RIGHTS. Such CPR state cannot
be sent over the normal migration channel, because devices and backends
are created prior to reading the channel, so this mode sends CPR state
over a second "cpr" migration channel. New QEMU reads the cpr channel
prior to creating devices or backends. The user specifies the cpr channel
in the channel arguments on the outgoing side, and in a second -incoming
command-line parameter on the incoming side.
The user must start old QEMU with the the '-machine aux-ram-share=on' option,
which allows anonymous memory to be transferred in place to the new process
by transferring a memory descriptor for each ram block. Memory-backend
objects must have the share=on attribute, but memory-backend-epc is not
supported.
The user starts new QEMU on the same host as old QEMU, with command-line
arguments to create the same machine, plus the -incoming option for the
main migration channel, like normal live migration. In addition, the user
adds a second -incoming option with channel type "cpr". This CPR channel
must support file descriptor transfer with SCM_RIGHTS, i.e. it must be a
UNIX domain socket.
To initiate CPR, the user issues a migrate command to old QEMU, adding
a second migration channel of type "cpr" in the channels argument.
Old QEMU stops the VM, saves state to the migration channels, and enters
the postmigrate state. New QEMU mmap's memory descriptors, and execution
resumes.
The implementation splits qmp_migrate into start and finish functions.
Start sends CPR state to new QEMU, which responds by closing the CPR
channel. Old QEMU detects the HUP then calls finish, which connects the
main migration channel.
In summary, the usage is:
qemu-system-$arch -machine aux-ram-share=on ...
start new QEMU with "-incoming <main-uri> -incoming <cpr-channel>"
Issue commands to old QEMU:
migrate_set_parameter mode cpr-transfer
{"execute": "migrate", ...
{"channel-type": "main"...}, {"channel-type": "cpr"...} ... }
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Link: https://lore.kernel.org/r/1736967650-129648-17-git-send-email-steven.sistare@oracle.com
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Export migrate_uri_parse for use outside migration internals, and define
a method migrate_is_uri that indicates when migrate_uri_parse should
be used.
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1736967650-129648-12-git-send-email-steven.sistare@oracle.com
Signed-off-by: Fabiano Rosas <farosas@suse.de>
This patch proposes a flag to maintain disk activation status globally. It
mostly rewrites disk activation mgmt for QEMU, including COLO and QMP
command xen_save_devices_state.
Backgrounds
===========
We have two problems on disk activations, one resolved, one not.
Problem 1: disk activation recover (for switchover interruptions)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
When migration is either cancelled or failed during switchover, especially
when after the disks are inactivated, QEMU needs to remember re-activate
the disks again before vm starts.
It used to be done separately in two paths: one in qmp_migrate_cancel(),
the other one in the failure path of migration_completion().
It used to be fixed in different commits, all over the places in QEMU. So
these are the relevant changes I saw, I'm not sure if it's complete list:
- In 2016, commit fe904ea824 ("migration: regain control of images when
migration fails to complete")
- In 2017, commit 1d2acc3162 ("migration: re-active images while migration
been canceled after inactive them")
- In 2023, commit 6dab4c93ec ("migration: Attempt disk reactivation in
more failure scenarios")
Now since we have a slightly better picture maybe we can unify the
reactivation in a single path.
One side benefit of doing so is, we can move the disk operation outside QMP
command "migrate_cancel". It's possible that in the future we may want to
make "migrate_cancel" be OOB-compatible, while that requires the command
doesn't need BQL in the first place. This will already do that and make
migrate_cancel command lightweight.
Problem 2: disk invalidation on top of invalidated disks
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
This is an unresolved bug for current QEMU. Link in "Resolves:" at the
end. It turns out besides the src switchover phase (problem 1 above), QEMU
also needs to remember block activation on destination.
Consider two continuous migration in a row, where the VM was always paused.
In that scenario, the disks are not activated even until migration
completed in the 1st round. When the 2nd round starts, if QEMU doesn't
know the status of the disks, it needs to try inactivate the disk again.
Here the issue is the block layer API bdrv_inactivate_all() will crash a
QEMU if invoked on already inactive disks for the 2nd migration. For
detail, see the bug link at the end.
Implementation
==============
This patch proposes to maintain disk activation with a global flag, so we
know:
- If we used to inactivate disks for migration, but migration got
cancelled, or failed, QEMU will know it should reactivate the disks.
- On incoming side, if the disks are never activated but then another
migration is triggered, QEMU should be able to tell that inactivate is
not needed for the 2nd migration.
We used to have disk_inactive, but it only solves the 1st issue, not the
2nd. Also, it's done in completely separate paths so it's extremely hard
to follow either how the flag changes, or the duration that the flag is
valid, and when we will reactivate the disks.
Convert the existing disk_inactive flag into that global flag (also invert
its naming), and maintain the disk activation status for the whole
lifecycle of qemu. That includes the incoming QEMU.
Put both of the error cases of source migration (failure, cancelled)
together into migration_iteration_finish(), which will be invoked for
either of the scenario. So from that part QEMU should behave the same as
before. However with such global maintenance on disk activation status, we
not only cleanup quite a few temporary paths that we try to maintain the
disk activation status (e.g. in postcopy code), meanwhile it fixes the
crash for problem 2 in one shot.
For freshly started QEMU, the flag is initialized to TRUE showing that the
QEMU owns the disks by default.
For incoming migrated QEMU, the flag will be initialized to FALSE once and
for all showing that the dest QEMU doesn't own the disks until switchover.
That is guaranteed by the "once" variable.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2395
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20241206230838.1111496-7-peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Headers in include/sysemu/ are not only related to system
*emulation*, they are also used by virtualization. Rename
as system/ which is clearer.
Files renamed manually then mechanical change using sed tool.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Lei Yang <leiyang@redhat.com>
Message-Id: <20241203172445.28576-1-philmd@linaro.org>
Multifd receive threads run on the destination side.
Correct the thread name marco to indicate the same.
Fixes: e620b1e477 ("migration: Put thread names together with macros")
Signed-off-by: Prasad Pandit <pjp@fedoraproject.org>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20241127111528.167330-1-ppandit@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
It's only used within migration/, so it shouldn't be exported.
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20241024213056.1395400-3-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
When VM is configured with huge memory, the current throttle logic
doesn't look like to scale, because migration_trigger_throttle()
is only called for each iteration, so it won't be invoked for a long
time if one iteration can take a long time.
The periodic dirty sync aims to fix the above issue by synchronizing
the ramblock from remote dirty bitmap and, when necessary, triggering
the CPU throttle multiple times during a long iteration.
This is a trade-off between synchronization overhead and CPU throttle
impact.
Signed-off-by: Hyman Huang <yong.huang@smartx.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/f61f1b3653f2acf026901103e1c73d157d38b08f.1729146786.git.yong.huang@smartx.com
[peterx: make prev_cnt global, and reset for each migration]
Signed-off-by: Peter Xu <peterx@redhat.com>
Keep migration thread names together, so it's easier to see a list of all
possible migration threads.
Still two functional changes below besides the macro defintions:
- There's one dirty rate thread that we overlooked before, now we add
that too and name it as "mig/dirtyrate" following the old rules.
- The old name "mig/src/rp-thr" has "-thr" but it may not be useful if
it's a thread name anyway, while "rp" can be slightly hard to read.
Taking this chance to rename it to "mig/src/return", hopefully a better
name.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Acked-by: Hyman Huang <yong.huang@smartx.com>
Reviewed-by: Zhang Chen <chen.zhang@intel.com>
Link: https://lore.kernel.org/r/20241011153652.517440-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
QEMU uses "int" in most cases even if it stores MigrationStatus. I don't
know why, so let's try to do that right and see what blows up..
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
The 'compress' migration capability enables the old compression code
which has shown issues over the years and is thought to be less stable
and tested than the more recent multifd-based compression. The old
compression code has been deprecated in 8.2 and now is time to remove
it.
Deprecation commit 864128df46 ("migration: Deprecate old compression
method").
Acked-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
The block migration is considered obsolete and has been deprecated in
8.2. Remove the migrate command option that enables it. This only
affects the QMP and HMP commands, the feature can still be accessed by
setting the migration 'block' capability. The whole feature will be
removed in a future patch.
Deprecation commit 8846b5bfca ("migration: migrate 'blk' command
option is deprecated.").
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Now we do set MIGRATION_FAILED state, but don't give a chance to
orchestrator to query migration state and get the error.
Let's provide a possibility for QMP-based orchestrators to get an error
like with outgoing migration.
For hmp_migrate_incoming(), let's enable the new behavior: HMP is not
and ABI, it's mostly intended to use by developer and it makes sense
not to stop the process.
For x-exit-preconfig, let's keep the old behavior:
- it's called from init(), so here we want to keep current behavior by
default
- it does exit on error by itself as well
So, if we want to change the behavior of x-exit-preconfig, it should be
another patch.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Acked-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Migration code needs no private fields of the coroutine backend.
Include the "regular" coroutine.h header.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Delete the MigrationState parameter from migration_is_setup_or_active
and move it to the public API in misc.h.
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Link: https://lore.kernel.org/r/1710179338-294359-3-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>
When migration for cpr is initiated, stop the vm and set state
RUN_STATE_FINISH_MIGRATE before ram is saved. This eliminates the
possibility of ram and device state being out of sync, and guarantees
that a guest in the suspended state remains suspended, because qmp_cont
rejects a cont command in the RUN_STATE_FINISH_MIGRATE state.
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1708622920-68779-11-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Now that the migration state reference counting is correct, further
wrap the bottom half dispatch process to avoid future issues.
Move BH creation and scheduling together and wrap the dispatch with an
intermediary function that will ensure we always keep the ref/unref
balanced.
Also move the responsibility of deleting the BH into the wrapper and
remove the now unnecessary pointers.
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240119233922.32588-6-farosas@suse.de
Signed-off-by: Peter Xu <peterx@redhat.com>
It's always used to compare against another uint64_t. Make it always clear
that it's never a negative.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240117075848.139045-2-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
migrate_max_downtime() has been removed long ago, but its declaration
was mistakenly left. Remove it.
Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20231231093016.14204-2-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Integrate MigrateChannelList with all transport backends
(socket, exec and rdma) for both src and dest migration
endpoints for hmp migration.
Suggested-by: Aravind Retnakaran <aravind.retnakaran@nutanix.com>
Signed-off-by: Het Gala <het.gala@nutanix.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Message-ID: <20231023182053.8711-14-farosas@suse.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Normally the postcopy recover phase should only exist for a super short
period, that's the duration when QEMU is trying to recover from an
interrupted postcopy migration, during which handshake will be carried out
for continuing the procedure with state changes from PAUSED -> RECOVER ->
POSTCOPY_ACTIVE again.
Here RECOVER phase should be super small, that happens right after the
admin specified a new but working network link for QEMU to reconnect to
dest QEMU.
However there can still be case where the channel is broken in this small
RECOVER window.
If it happens, with current code there's no way the src QEMU can got kicked
out of RECOVER stage. No way either to retry the recover in another channel
when established.
This patch allows the RECOVER phase to fail itself too - we're mostly
ready, just some small things missing, e.g. properly kick the main
migration thread out when sleeping on rp_sem when we found that we're at
RECOVER stage. When this happens, it fails the RECOVER itself, and
rollback to PAUSED stage. Then the user can retry another round of
recovery.
To make it even stronger, teach QMP command migrate-pause to explicitly
kick src/dst QEMU out when needed, so even if for some reason the migration
thread didn't got kicked out already by a failing rethrn-path thread, the
admin can also kick it out.
This will be an super, super corner case, but still try to cover that.
One can try to test this with two proxy channels for migration:
(a) socat unix-listen:/tmp/src.sock,reuseaddr,fork tcp:localhost:10000
(b) socat tcp-listen:10000,reuseaddr,fork unix:/tmp/dst.sock
So the migration channel will be:
(a) (b)
src -> /tmp/src.sock -> tcp:10000 -> /tmp/dst.sock -> dst
Then to make QEMU hang at RECOVER stage, one can do below:
(1) stop the postcopy using QMP command postcopy-pause
(2) kill the 2nd proxy (b)
(3) try to recover the postcopy using /tmp/src.sock on src
(4) src QEMU will go into RECOVER stage but won't be able to continue
from there, because the channel is actually broken at (b)
Before this patch, step (4) will make src QEMU stuck in RECOVER stage,
without a way to kick the QEMU out or continue the postcopy again. After
this patch, (4) will quickly fail qemu and bounce back to PAUSED stage.
Admin can also kick QEMU from (4) into PAUSED when needed using
migrate-pause when needed.
After bouncing back to PAUSED stage, one can recover again.
Reported-by: Xiaohui Li <xiaohli@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2111332
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231017202633.296756-3-peterx@redhat.com>
rp_state.error was a boolean used to show error happened in return path
thread. That's not only duplicating error reporting (migrate_set_error),
but also not good enough in that we only do error_report() and set it to
true, we never can keep a history of the exact error and show it in
query-migrate.
To make this better, a few things done:
- Use error_setg() rather than error_report() across the whole lifecycle
of return path thread, keeping the error in an Error*.
- With above, no need to have mark_source_rp_bad(), remove it, alongside
with rp_state.error itself.
- Use migrate_set_error() to apply that captured error to the global
migration object when error occured in this thread.
- Do the same when detected qemufile error in source return path
We need to re-export qemu_file_get_error_obj() to do the last one.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231017202633.296756-2-peterx@redhat.com>
Provide a helper for non-COLO use case of migration to stop a VM. This
prepares for adding some downtime relevant tracepoints to migration, where
they may or may not apply to COLO.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231030163346.765724-5-peterx@redhat.com>
Helper to say if we are doing a migration over rdma.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Li Zhijian <lizhijian@fujitsu.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231011203527.9061-2-quintela@redhat.com>
Migration bandwidth is a very important value to live migration. It's
because it's one of the major factors that we'll make decision on when to
switchover to destination in a precopy process.
This value is currently estimated by QEMU during the whole live migration
process by monitoring how fast we were sending the data. This can be the
most accurate bandwidth if in the ideal world, where we're always feeding
unlimited data to the migration channel, and then it'll be limited to the
bandwidth that is available.
However in reality it may be very different, e.g., over a 10Gbps network we
can see query-migrate showing migration bandwidth of only a few tens of
MB/s just because there are plenty of other things the migration thread
might be doing. For example, the migration thread can be busy scanning
zero pages, or it can be fetching dirty bitmap from other external dirty
sources (like vhost or KVM). It means we may not be pushing data as much
as possible to migration channel, so the bandwidth estimated from "how many
data we sent in the channel" can be dramatically inaccurate sometimes.
With that, the decision to switchover will be affected, by assuming that we
may not be able to switchover at all with such a low bandwidth, but in
reality we can.
The migration may not even converge at all with the downtime specified,
with that wrong estimation of bandwidth, keeping iterations forever with a
low estimation of bandwidth.
The issue is QEMU itself may not be able to avoid those uncertainties on
measuing the real "available migration bandwidth". At least not something
I can think of so far.
One way to fix this is when the user is fully aware of the available
bandwidth, then we can allow the user to help providing an accurate value.
For example, if the user has a dedicated channel of 10Gbps for migration
for this specific VM, the user can specify this bandwidth so QEMU can
always do the calculation based on this fact, trusting the user as long as
specified. It may not be the exact bandwidth when switching over (in which
case qemu will push migration data as fast as possible), but much better
than QEMU trying to wildly guess, especially when very wrong.
A new parameter "avail-switchover-bandwidth" is introduced just for this.
So when the user specified this parameter, instead of trusting the
estimated value from QEMU itself (based on the QEMUFile send speed), it
trusts the user more by using this value to decide when to switchover,
assuming that we'll have such bandwidth available then.
Note that specifying this value will not throttle the bandwidth for
switchover yet, so QEMU will always use the full bandwidth possible for
sending switchover data, assuming that should always be the most important
way to use the network at that time.
This can resolve issues like "unconvergence migration" which is caused by
hilarious low "migration bandwidth" detected for whatever reason.
Reported-by: Zhiyi Guo <zhguo@redhat.com>
Reviewed-by: Joao Martins <joao.m.martins@oracle.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231010221922.40638-1-peterx@redhat.com>
It's just a simple wrapper for rp_sem on either wait() or kick(), make it
even clearer on how it is used. Prepared to be used even for other things.
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-ID: <20231004220240.167175-8-peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Introduce a helper to detect whether MigrationState.error is set for
whatever reason.
This is preparation work for any thread (e.g. source return path thread) to
setup errors in an unified way to MigrationState, rather than relying on
its own way to set errors (mark_source_rp_bad()).
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231004220240.167175-3-peterx@redhat.com>
Replace the return path retry logic with finishing and restarting the
thread. This fixes a race when resuming the migration that leads to a
segfault.
Currently when doing postcopy we consider that an IO error on the
return path file could be due to a network intermittency. We then keep
the thread alive but have it do cleanup of the 'from_dst_file' and
wait on the 'postcopy_pause_rp' semaphore. When the user issues a
migrate resume, a new return path is opened and the thread is allowed
to continue.
There's a race condition in the above mechanism. It is possible for
the new return path file to be setup *before* the cleanup code in the
return path thread has had a chance to run, leading to the *new* file
being closed and the pointer set to NULL. When the thread is released
after the resume, it tries to dereference 'from_dst_file' and crashes:
Thread 7 "return path" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffd1dbf700 (LWP 9611)]
0x00005555560e4893 in qemu_file_get_error_obj (f=0x0, errp=0x0) at ../migration/qemu-file.c:154
154 return f->last_error;
(gdb) bt
#0 0x00005555560e4893 in qemu_file_get_error_obj (f=0x0, errp=0x0) at ../migration/qemu-file.c:154
#1 0x00005555560e4983 in qemu_file_get_error (f=0x0) at ../migration/qemu-file.c:206
#2 0x0000555555b9a1df in source_return_path_thread (opaque=0x555556e06000) at ../migration/migration.c:1876
#3 0x000055555602e14f in qemu_thread_start (args=0x55555782e780) at ../util/qemu-thread-posix.c:541
#4 0x00007ffff38d76ea in start_thread (arg=0x7fffd1dbf700) at pthread_create.c:477
#5 0x00007ffff35efa6f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
Here's the race (important bit is open_return_path happening before
migration_release_dst_files):
migration | qmp | return path
--------------------------+-----------------------------+---------------------------------
qmp_migrate_pause()
shutdown(ms->to_dst_file)
f->last_error = -EIO
migrate_detect_error()
postcopy_pause()
set_state(PAUSED)
wait(postcopy_pause_sem)
qmp_migrate(resume)
migrate_fd_connect()
resume = state == PAUSED
open_return_path <-- TOO SOON!
set_state(RECOVER)
post(postcopy_pause_sem)
(incoming closes to_src_file)
res = qemu_file_get_error(rp)
migration_release_dst_files()
ms->rp_state.from_dst_file = NULL
post(postcopy_pause_rp_sem)
postcopy_pause_return_path_thread()
wait(postcopy_pause_rp_sem)
rp = ms->rp_state.from_dst_file
goto retry
qemu_file_get_error(rp)
SIGSEGV
-------------------------------------------------------------------------------------------
We can keep the retry logic without having the thread alive and
waiting. The only piece of data used by it is the 'from_dst_file' and
it is only allowed to proceed after a migrate resume is issued and the
semaphore released at migrate_fd_connect().
Move the retry logic to outside the thread by waiting for the thread
to finish before pausing the migration.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-8-farosas@suse.de>
We hit intermit CI issue on failing at migration-test over the unit test
preempt/plain:
qemu-system-x86_64: Unable to read from socket: Connection reset by peer
Memory content inconsistency at 5b43000 first_byte = bd last_byte = bc current = 4f hit_edge = 1
**
ERROR:../tests/qtest/migration-test.c:300:check_guests_ram: assertion failed: (bad == 0)
(test program exited with status code -6)
Fabiano debugged into it and found that the preempt thread can quit even
without receiving all the pages, which can cause guest not receiving all
the pages and corrupt the guest memory.
To make sure preempt thread finished receiving all the pages, we can rely
on the page_requested_count being zero because preempt channel will only
receive requested page faults. Note, not all the faulted pages are required
to be sent via the preempt channel/thread; imagine the case when a
requested page is just queued into the background main channel for
migration, the src qemu will just still send it via the background channel.
Here instead of spinning over reading the count, we add a condvar so the
main thread can wait on it if that unusual case happened, without burning
the cpu for no good reason, even if the duration is short; so even if we
spin in this rare case is probably fine. It's just better to not do so.
The condvar is only used when that special case is triggered. Some memory
ordering trick is needed to guarantee it from happening (against the
preempt thread status field), so the main thread will always get a kick
when that triggers correctly.
Closes: https://gitlab.com/qemu-project/qemu/-/issues/1886
Debugged-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-2-farosas@suse.de>
Add a new .save_prepare() handler to struct SaveVMHandlers. This handler
is called early, even before migration starts, and can be used by
devices to perform early checks.
Refactor migrate_init() to be able to return errors and call
.save_prepare() from there.
Suggested-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
The functions in target.c are not static, yet they don't have a proper
migration prefix. Add such prefix.
Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
The only migrate_fd_error() call sites are in "migration/migration.c",
which is also where we define migrate_fd_error(). Make the function
static, and remove its declaration from "migration/migration.h".
Cc: Juan Quintela <quintela@redhat.com> (maintainer:Migration)
Cc: Leonardo Bras <leobras@redhat.com> (reviewer:Migration)
Cc: Peter Xu <peterx@redhat.com> (reviewer:Migration)
Cc: qemu-trivial@nongnu.org
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2018404
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Currently, VFIO bytes_transferred is not reset properly:
1. bytes_transferred is not reset after a VM snapshot (so a migration
following a snapshot will report incorrect value).
2. bytes_transferred is a single counter for all VFIO devices, however
upon migration failure it is reset multiple times, by each VFIO
device.
Fix it by introducing a new function vfio_reset_bytes_transferred() and
calling it during migration and snapshot start.
Remove existing bytes_transferred reset in VFIO migration state
notifier, which is not needed anymore.
Fixes: 3710586caa ("qapi: Add VFIO devices migration stats in Migration stats")
Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
Implement switchover ack logic. This prevents the source from stopping
the VM and completing the migration until an ACK is received from the
destination that it's OK to do so.
To achieve this, a new SaveVMHandlers handler switchover_ack_needed()
and a new return path message MIG_RP_MSG_SWITCHOVER_ACK are added.
The switchover_ack_needed() handler is called during migration setup in
the destination to check if switchover ack is used by the migrated
device.
When switchover is approved by all migrated devices in the destination
that support this capability, the MIG_RP_MSG_SWITCHOVER_ACK return path
message is sent to the source to notify it that it's OK to do
switchover.
Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Tested-by: YangHang Liu <yanghliu@redhat.com>
Acked-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
No logic change here, only refactoring. That's a preparation for next
commit where we finally restore the stopped vm state on migration
failure or cancellation.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20230517123752.21615-5-vsementsov@yandex-team.ru>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Originally, migration_incoming_co was introduced by
25d0c16f62
"migration: Switch to COLO process after finishing loadvm"
to be able to enter from COLO code to one specific yield point, added
by 25d0c16f62.
Later in 923709896b
"migration: poll the cm event for destination qemu"
we reused this variable to wake the migration incoming coroutine from
RDMA code.
That was doubtful idea. Entering coroutines is a very fragile thing:
you should be absolutely sure which yield point you are going to enter.
I don't know how much is it safe to enter during qemu_loadvm_state()
which I think what RDMA want to do. But for sure RDMA shouldn't enter
the special COLO-related yield-point. As well, COLO code doesn't want
to enter during qemu_loadvm_state(), it want to enter it's own specific
yield-point.
As well, when in 8e48ac9586
"COLO: Add block replication into colo process" we added
bdrv_invalidate_cache_all() call (now it's called activate_all())
it became possible to enter the migration incoming coroutine during
that call which is wrong too.
So, let't make these things separate and disjoint: loadvm_co for RDMA,
non-NULL during qemu_loadvm_state(), and colo_incoming_co for COLO,
non-NULL only around specific yield.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20230515130640.46035-3-vsementsov@yandex-team.ru>
Signed-off-by: Juan Quintela <quintela@redhat.com>
have_colo_incoming_thread variable is unused. colo_incoming_thread can
be local.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Zhang Chen <chen.zhang@intel.com>
Message-Id: <20230428194928.1426370-6-vsementsov@yandex-team.ru>
Signed-off-by: Juan Quintela <quintela@redhat.com>
