Hanna Czenczek [Fri, 2 Feb 2024 14:47:54 +0000 (15:47 +0100)]
block-backend: Allow concurrent context changes
Since AioContext locks have been removed, a BlockBackend's AioContext
may really change at any time (only exception is that it is often
confined to a drained section, as noted in this patch). Therefore,
blk_get_aio_context() cannot rely on its root node's context always
matching that of the BlockBackend.
In practice, whether they match does not matter anymore anyway: Requests
can be sent to BDSs from any context, so anyone who requests the BB's
context should have no reason to require the root node to have the same
context. Therefore, we can and should remove the assertion to that
effect.
In addition, because the context can be set and queried from different
threads concurrently, it has to be accessed with atomic operations.
Buglink: https://issues.redhat.com/browse/RHEL-19381 Suggested-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Hanna Czenczek <hreitz@redhat.com>
Message-ID: <20240202144755.671354-2-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>
The code is actually safe because the iohandler and qemu_aio_context
AioContext run under the Big QEMU Lock. Nevertheless, set a good example
and use aio_co_reschedule_self() so it's obvious that there is no race.
Suggested-by: Hanna Reitz <hreitz@redhat.com> Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org> Reviewed-by: Hanna Czenczek <hreitz@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20240206190610.107963-6-stefanha@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Acked-by: Markus Armbruster <armbru@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Stefan Hajnoczi [Tue, 6 Feb 2024 19:06:09 +0000 (14:06 -0500)]
virtio-blk: declare VirtIOBlock::rq with a type
The VirtIOBlock::rq field has had the type void * since its introduction
in commit 869a5c6df19a ("Stop VM on error in virtio-blk. (Gleb
Natapov)").
Perhaps this was done to avoid the forward declaration of
VirtIOBlockReq.
Hanna Czenczek <hreitz@redhat.com> pointed out the missing type. Specify
the actual type because there is no need to use void * here.
Suggested-by: Hanna Czenczek <hreitz@redhat.com> Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org> Reviewed-by: Hanna Czenczek <hreitz@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20240206190610.107963-5-stefanha@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The code is correct because both rq->vq and vq_rq[] depend on
num_queues, but this is indirect and not 100% obvious. Add an assertion.
Suggested-by: Hanna Czenczek <hreitz@redhat.com> Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org> Reviewed-by: Hanna Czenczek <hreitz@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20240206190610.107963-4-stefanha@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Stefan Hajnoczi [Tue, 6 Feb 2024 19:06:07 +0000 (14:06 -0500)]
virtio-blk: clarify that there is at least 1 virtqueue
It is not possible to instantiate a virtio-blk device with 0 virtqueues.
The following check is located in ->realize():
if (!conf->num_queues) {
error_setg(errp, "num-queues property must be larger than 0");
return;
}
Later on we access s->vq_aio_context[0] under the assumption that there
is as least one virtqueue. Hanna Czenczek <hreitz@redhat.com> noted that
it would help to show that the array index is already valid.
Add an assertion to document that s->vq_aio_context[0] is always
safe...and catch future code changes that break this assumption.
Suggested-by: Hanna Czenczek <hreitz@redhat.com> Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org> Reviewed-by: Hanna Czenczek <hreitz@redhat.com> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20240206190610.107963-3-stefanha@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Hanna Czenczek <hreitz@redhat.com> noticed that the safety of
`vq_aio_context[vq->value] = ctx;` with user-defined vq->value inputs is
not obvious.
The code is structured in validate() + apply() steps so input validation
is there, but it happens way earlier and there is nothing that
guarantees apply() can only be called with validated inputs.
This patch moves the validate() call inside the apply() function so
validation is guaranteed. I also added the bounds checking assertion
that Hanna suggested.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org> Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Message-ID: <20240206190610.107963-2-stefanha@redhat.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Peter Xu [Wed, 7 Feb 2024 00:54:02 +0000 (08:54 +0800)]
ci: Remove tag dependency for build-previous-qemu
The new build-previous-qemu job relies on QEMU release tag being present,
while that may not be always true for personal git repositories since by
default tag is not pushed. The job can fail on those CI kicks, as reported
by Peter Maydell.
Fix it by fetching the tags remotely from the official repository, as
suggested by Dan.
Reported-by: Peter Maydell <peter.maydell@linaro.org> Suggested-by: "Daniel P. Berrangé" <berrange@redhat.com> Reviewed-by: "Daniel P. Berrangé" <berrange@redhat.com> Link: https://lore.kernel.org/r/20240207005403.242235-3-peterx@redhat.com Signed-off-by: Peter Xu <peterx@redhat.com>
Peter Xu [Wed, 7 Feb 2024 00:54:01 +0000 (08:54 +0800)]
tests/migration-test: Stick with gicv3 in aarch64 test
Recently we introduced cross-binary migration test. It's always wanted
that migration-test uses stable guest ABI for both QEMU binaries in this
case, so that both QEMU binaries will be compatible on the migration
stream with the cmdline specified.
Switch to a static gic version "3" rather than using version "max", so that
GIC should be stable now across any future QEMU binaries for migration-test.
Here the version can actually be anything as long as the ABI is stable. We
choose "3" because it's the majority of what we already use in QEMU while
still new enough: "git grep gic-version=3" shows 6 hit, while version 4 has
no direct user yet besides "max".
Note that even with this change, aarch64 won't be able to work yet with
migration cross binary test, but then the only missing piece will be the
stable CPU model.
Fabiano Rosas [Tue, 6 Feb 2024 21:51:18 +0000 (18:51 -0300)]
migration/multifd: Add a synchronization point for channel creation
It is possible that one of the multifd channels fails to be created at
multifd_new_send_channel_async() while the rest of the channel
creation tasks are still in flight.
This could lead to multifd_save_cleanup() executing the
qemu_thread_join() loop too early and not waiting for the threads
which haven't been created yet, leading to the freeing of resources
that the newly created threads will try to access and crash.
Add a synchronization point after which there will be no attempts at
thread creation and therefore calling multifd_save_cleanup() past that
point will ensure it properly waits for the threads.
A note about performance: Prior to this patch, if a channel took too
long to be established, other channels could finish connecting first
and already start taking load. Now we're bounded by the
slowest-connecting channel.
Fabiano Rosas [Tue, 6 Feb 2024 21:51:17 +0000 (18:51 -0300)]
migration/multifd: Unify multifd and TLS connection paths
During multifd channel creation (multifd_send_new_channel_async) when
TLS is enabled, the multifd_channel_connect function is called twice,
once to create the TLS handshake thread and another time after the
asynchrounous TLS handshake has finished.
This creates a slightly confusing call stack where
multifd_channel_connect() is called more times than the number of
channels. It also splits error handling between the two callers of
multifd_channel_connect() causing some code duplication. Lastly, it
gets in the way of having a single point to determine whether all
channel creation tasks have been initiated.
Refactor the code to move the reentrancy one level up at the
multifd_new_send_channel_async() level, de-duplicating the error
handling and allowing for the next patch to introduce a
synchronization point common to all the multifd channel creation,
regardless of TLS.
Note that the previous code would never fail once p->c had been set.
This patch changes this assumption, which affects refcounting, so add
comments around object_unref to explain the situation.
Fabiano Rosas [Tue, 6 Feb 2024 21:51:16 +0000 (18:51 -0300)]
migration/multifd: Move multifd_send_setup into migration thread
We currently have an unfavorable situation around multifd channels
creation and the migration thread execution.
We create the multifd channels with qio_channel_socket_connect_async
-> qio_task_run_in_thread, but only connect them at the
multifd_new_send_channel_async callback, called from
qio_task_complete, which is registered as a glib event.
So at multifd_send_setup() we create the channels, but they will only
be actually usable after the whole multifd_send_setup() calling stack
returns back to the main loop. Which means that the migration thread
is already up and running without any possibility for the multifd
channels to be ready on time.
We currently rely on the channels-ready semaphore blocking
multifd_send_sync_main() until channels start to come up and release
it. However there have been bugs recently found when a channel's
creation fails and multifd_send_cleanup() is allowed to run while
other channels are still being created.
Let's start to organize this situation by moving the
multifd_send_setup() call into the migration thread. That way we
unblock the main-loop to dispatch the completion callbacks and
actually have a chance of getting the multifd channels ready for when
the migration thread needs them.
The next patches will deal with the synchronization aspects.
Note that this takes multifd_send_setup() out of the BQL.
Fabiano Rosas [Tue, 6 Feb 2024 21:51:14 +0000 (18:51 -0300)]
migration/multifd: Remove p->running
We currently only need p->running to avoid calling qemu_thread_join()
on a non existent thread if the thread has never been created.
However, there are at least two bugs in this logic:
1) On the sending side, p->running is set too early and
qemu_thread_create() can be skipped due to an error during TLS
handshake, leaving the flag set and leading to a crash when
multifd_send_cleanup() calls qemu_thread_join().
2) During exit, the multifd thread clears the flag while holding the
channel lock. The counterpart at multifd_send_cleanup() reads the flag
outside of the lock and might free the mutex while the multifd thread
still has it locked.
Fix the first issue by setting the flag right before creating the
thread. Rename it from p->running to p->thread_created to clarify its
usage.
Fix the second issue by not clearing the flag at the multifd thread
exit. We don't have any use for that.
Note that these bugs are straight-forward logic issues and not race
conditions. There is still a gap for races to affect this code due to
multifd_send_cleanup() being allowed to run concurrently with the
thread creation loop. This issue is solved in the next patches.
Cc: qemu-stable <qemu-stable@nongnu.org> Fixes: 29647140157a ("migration/tls: add support for multifd tls-handshake") Reported-by: Avihai Horon <avihaih@nvidia.com> Reported-by: chenyuhui5@huawei.com Reviewed-by: Peter Xu <peterx@redhat.com> Signed-off-by: Fabiano Rosas <farosas@suse.de> Link: https://lore.kernel.org/r/20240206215118.6171-3-farosas@suse.de Signed-off-by: Peter Xu <peterx@redhat.com>
Avihai Horon [Thu, 25 Jan 2024 16:25:12 +0000 (18:25 +0200)]
migration: Fix logic of channels and transport compatibility check
The commit in the fixes line mistakenly modified the channels and
transport compatibility check logic so it now checks multi-channel
support only for socket transport type.
Thus, running multifd migration using a transport other than socket that
is incompatible with multi-channels (such as "exec") would lead to a
segmentation fault instead of an error message.
For example:
(qemu) migrate_set_capability multifd on
(qemu) migrate -d "exec:cat > /tmp/vm_state"
Segmentation fault (core dumped)
Fix it by checking multi-channel compatibility for all transport types.
Cc: qemu-stable <qemu-stable@nongnu.org> Fixes: d95533e1cdcc ("migration: modify migration_channels_and_uri_compatible() for new QAPI syntax") Signed-off-by: Avihai Horon <avihaih@nvidia.com> Reviewed-by: Peter Xu <peterx@redhat.com> Link: https://lore.kernel.org/r/20240125162528.7552-2-avihaih@nvidia.com Signed-off-by: Peter Xu <peterx@redhat.com>
After console_sshd_config(), the SSH server needs to be nudged to pick
up the new configs. The scripts for the other BSD flavors already do
this with a reboot, but a simple reload is sufficient.
Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-ID: <20240206002344.12372-3-iii@linux.ibm.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
There is a 2s delay during connection, and ConnectTimeout is set to 1.
Raising it makes the issue go away, but we can do better. The server
truss shows:
target/s390x: Prefer fast cpu_env() over slower CPU QOM cast macro
Mechanical patch produced running the command documented
in scripts/coccinelle/cpu_env.cocci_template header.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240129164514.73104-25-philmd@linaro.org> Signed-off-by: Thomas Huth <thuth@redhat.com>
Convert to Binary - counterparts of the already implemented Convert
to Decimal (CVD*) instructions.
Example from the Principles of Operation: 25594C becomes 63FA.
Co-developed-by: Pavel Zbitskiy <pavel.zbitskiy@gmail.com> Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240205205830.6425-3-iii@linux.ibm.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
Mark Kanda [Wed, 31 Jan 2024 16:53:27 +0000 (10:53 -0600)]
oslib-posix: initialize backend memory objects in parallel
QEMU initializes preallocated backend memory as the objects are parsed from
the command line. This is not optimal in some cases (e.g. memory spanning
multiple NUMA nodes) because the memory objects are initialized in series.
Allow the initialization to occur in parallel (asynchronously). In order to
ensure optimal thread placement, asynchronous initialization requires prealloc
context threads to be in use.
Signed-off-by: Mark Kanda <mark.kanda@oracle.com>
Message-ID: <20240131165327.3154970-2-mark.kanda@oracle.com> Tested-by: Mario Casquero <mcasquer@redhat.com> Signed-off-by: David Hildenbrand <david@redhat.com>
memory-device: reintroduce memory region size check
We used to check that the memory region size is multiples of the overall
requested address alignment for the device memory address.
We removed that check, because there are cases (i.e., hv-balloon) where
devices unconditionally request an address alignment that has a very large
alignment (i.e., 32 GiB), but the actual memory device size might not be
multiples of that alignment.
However, this change:
(a) allows for some practically impossible DIMM sizes, like "1GB+1 byte".
(b) allows for DIMMs that partially cover hugetlb pages, previously
reported in [1].
Both scenarios don't make any sense: we might even waste memory.
So let's reintroduce that check, but only check that the
memory region size is multiples of the memory region alignment (i.e.,
page size, huge page size), but not any additional memory device
requirements communicated using md->get_min_alignment().
The following examples now fail again as expected:
(a) 1M with 2M THP
qemu-system-x86_64 -m 4g,maxmem=16g,slots=1 -S -nodefaults -nographic \
-object memory-backend-ram,id=mem1,size=1M \
-device pc-dimm,id=dimm1,memdev=mem1
-> backend memory size must be multiple of 0x200000
(b) 1G+1byte
qemu-system-x86_64 -m 4g,maxmem=16g,slots=1 -S -nodefaults -nographic \
-object memory-backend-ram,id=mem1,size=1073741825B \
-device pc-dimm,id=dimm1,memdev=mem1
-> backend memory size must be multiple of 0x200000
(c) Unliagned hugetlb size (2M)
qemu-system-x86_64 -m 4g,maxmem=16g,slots=1 -S -nodefaults -nographic \
-object memory-backend-file,id=mem1,mem-path=/dev/hugepages/tmp,size=511M \
-device pc-dimm,id=dimm1,memdev=mem1
backend memory size must be multiple of 0x200000
(d) Unliagned hugetlb size (1G)
qemu-system-x86_64 -m 4g,maxmem=16g,slots=1 -S -nodefaults -nographic \
-object memory-backend-file,id=mem1,mem-path=/dev/hugepages1G/tmp,size=2047M \
-device pc-dimm,id=dimm1,memdev=mem1
-> backend memory size must be multiple of 0x40000000
Note that this fix depends on a hv-balloon change to communicate its
additional alignment requirements using get_min_alignment() instead of
through the memory region.
Peter Xu [Fri, 2 Feb 2024 10:28:57 +0000 (18:28 +0800)]
migration/multifd: Optimize sender side to be lockless
When reviewing my attempt to refactor send_prepare(), Fabiano suggested we
try out with dropping the mutex in multifd code [1].
I thought about that before but I never tried to change the code. Now
maybe it's time to give it a stab. This only optimizes the sender side.
The trick here is multifd has a clear provider/consumer model, that the
migration main thread publishes requests (either pending_job/pending_sync),
while the multifd sender threads are consumers. Here we don't have a lot
of complicated data sharing, and the jobs can logically be submitted
lockless.
Arm the code with atomic weapons. Two things worth mentioning:
- For multifd_send_pages(): we can use qatomic_load_acquire() when trying
to find a free channel, but that's expensive if we attach one ACQUIRE per
channel. Instead, keep the qatomic_read() on reading the pending_job
flag as we do already, meanwhile use one smp_mb_acquire() after the loop
to guarantee the memory ordering.
- For pending_sync: it doesn't have any extra data required since now
p->flags are never touched, it should be safe to not use memory barrier.
That's different from pending_job.
Provide rich comments for all the lockless operations to state how they are
paired. With that, we can remove the mutex.
Thomas Huth [Wed, 17 Jan 2024 14:10:54 +0000 (15:10 +0100)]
docs/about: Deprecate the old "power5+" and "power7+" CPU names
For consistency we should drop the names with a "+" in it in the
long run.
Message-ID: <20240117141054.73841-3-thuth@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Cédric Le Goater <clg@kaod.org> Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
Thomas Huth [Wed, 17 Jan 2024 14:10:53 +0000 (15:10 +0100)]
target/ppc/cpu-models: Rename power5+ and power7+ for new QOM naming rules
The character "+" is now forbidden in QOM device names (see commit b447378e1217 - "Limit type names to alphanumerical and some few special
characters"). For the "power5+" and "power7+" CPU names, there is
currently a hack in type_name_is_valid() to still allow them for
compatibility reasons. However, there is a much nicer solution for this:
Simply use aliases! This way we can still support the old names without
the need for the ugly hack in type_name_is_valid().
Message-ID: <20240117141054.73841-2-thuth@redhat.com> Reviewed-by: Cédric Le Goater <clg@kaod.org> Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
Sven Schnelle [Sun, 28 Jan 2024 20:22:14 +0000 (21:22 +0100)]
hw/scsi/lsi53c895a: add missing decrement of reentrancy counter
When the maximum count of SCRIPTS instructions is reached, the code
stops execution and returns, but fails to decrement the reentrancy
counter. This effectively renders the SCSI controller unusable
because on next entry the reentrancy counter is still above the limit.
This bug was seen on HP-UX 10.20 which seems to trigger SCRIPTS
loops.
Fixes: b987718bbb ("hw/scsi/lsi53c895a: Fix reentrancy issues in the LSI controller (CVE-2023-0330)") Signed-off-by: Sven Schnelle <svens@stackframe.org>
Message-ID: <20240128202214.2644768-1-svens@stackframe.org> Reviewed-by: Thomas Huth <thuth@redhat.com> Tested-by: Helge Deller <deller@gmx.de> Signed-off-by: Thomas Huth <thuth@redhat.com>
As reported correctly by Fabiano [1] (while per Fabiano, it sourced back to
Elena's initial report in Oct 2023), MultiFDSendParams.packet_num is buggy
to be assigned and stored. Consider two consequent operations of: (1)
queue a job into multifd send thread X, then (2) queue another sync request
to the same send thread X. Then the MultiFDSendParams.packet_num will be
assigned twice, and the first assignment can get lost already.
To avoid that, we move the packet_num assignment from p->packet_num into
where the thread will fill in the packet. Use atomic operations to protect
the field, making sure there's no race.
Note that atomic fetch_add() may not be good for scaling purposes, however
multifd should be fine as number of threads should normally not go beyond
16 threads. Let's leave that concern for later but fix the issue first.
There's also a trick on how to make it always work even on 32 bit hosts for
uint64_t packet number. Switching to uintptr_t as of now to simply the
case. It will cause packet number to overflow easier on 32 bit, but that
shouldn't be a major concern for now as 32 bit systems is not the major
audience for any performance concerns like what multifd wants to address.
We also need to move multifd_send_state definition upper, so that
multifd_send_fill_packet() can reference it.
Peter Xu [Fri, 2 Feb 2024 10:28:53 +0000 (18:28 +0800)]
migration/multifd: Cleanup multifd_save_cleanup()
Shrink the function by moving relevant works into helpers: move the thread
join()s into multifd_send_terminate_threads(), then create two more helpers
to cover channel/state cleanups.
Add a TODO entry for the thread terminate process because p->running is
still buggy. We need to fix it at some point but not yet covered.
Peter Xu [Fri, 2 Feb 2024 10:28:52 +0000 (18:28 +0800)]
migration/multifd: Rewrite multifd_queue_page()
The current multifd_queue_page() is not easy to read and follow. It is not
good with a few reasons:
- No helper at all to show what exactly does a condition mean; in short,
readability is low.
- Rely on pages->ramblock being cleared to detect an empty queue. It's
slightly an overload of the ramblock pointer, per Fabiano [1], which I
also agree.
- Contains a self recursion, even if not necessary..
Rewrite this function. We add some comments to make it even clearer on
what it does.
Peter Xu [Fri, 2 Feb 2024 10:28:48 +0000 (18:28 +0800)]
migration/multifd: Forbid spurious wakeups
Now multifd's logic is designed to have no spurious wakeup. I still
remember a talk to Juan and he seems to agree we should drop it now, and if
my memory was right it was there because multifd used to hit that when
still debugging.
Let's drop it and see what can explode; as long as it's not reaching
soft-freeze.
Peter Xu [Fri, 2 Feb 2024 10:28:47 +0000 (18:28 +0800)]
migration/multifd: Move header prepare/fill into send_prepare()
This patch redefines the interfacing of ->send_prepare(). It further
simplifies multifd_send_thread() especially on zero copy.
Now with the new interface, we require the hook to do all the work for
preparing the IOVs to send. After it's completed, the IOVs should be ready
to be dumped into the specific multifd QIOChannel later.
Peter Xu [Fri, 2 Feb 2024 10:28:43 +0000 (18:28 +0800)]
migration/multifd: Rename p->num_packets and clean it up
This field, no matter whether on src or dest, is only used for debugging
purpose.
They can even be removed already, unless it still more or less provide some
accounting on "how many packets are sent/recved for this thread". The
other more important one is called packet_num, which is embeded in the
multifd packet headers (MultiFDPacket_t).
So let's keep them for now, but make them much easier to understand, by
doing below:
- Rename both of them to packets_sent / packets_recved, the old
name (num_packets) are waaay too confusing when we already have
MultiFDPacket_t.packets_num.
- Avoid worrying on the "initial packet": we know we will send it, that's
good enough. The accounting won't matter a great deal to start with 0 or
with 1.
- Move them to where we send/recv the packets. They're:
- multifd_send_fill_packet() for senders.
- multifd_recv_unfill_packet() for receivers.
Peter Xu [Fri, 2 Feb 2024 10:28:41 +0000 (18:28 +0800)]
migration/multifd: Simplify locking in sender thread
The sender thread will yield the p->mutex before IO starts, trying to not
block the requester thread. This may be unnecessary lock optimizations,
because the requester can already read pending_job safely even without the
lock, because the requester is currently the only one who can assign a
task.
Drop that lock complication on both sides:
(1) in the sender thread, always take the mutex until job done
(2) in the requester thread, check pending_job clear lockless
Peter Xu [Fri, 2 Feb 2024 10:28:40 +0000 (18:28 +0800)]
migration/multifd: Separate SYNC request with normal jobs
Multifd provide a threaded model for processing jobs. On sender side,
there can be two kinds of job: (1) a list of pages to send, or (2) a sync
request.
The sync request is a very special kind of job. It never contains a page
array, but only a multifd packet telling the dest side to synchronize with
sent pages.
Before this patch, both requests use the pending_job field, no matter what
the request is, it will boost pending_job, while multifd sender thread will
decrement it after it finishes one job.
However this should be racy, because SYNC is special in that it needs to
set p->flags with MULTIFD_FLAG_SYNC, showing that this is a sync request.
Consider a sequence of operations where:
- migration thread enqueue a job to send some pages, pending_job++ (0->1)
- [...before the selected multifd sender thread wakes up...]
- migration thread enqueue another job to sync, pending_job++ (1->2),
setup p->flags=MULTIFD_FLAG_SYNC
- multifd sender thread wakes up, found pending_job==2
- send the 1st packet with MULTIFD_FLAG_SYNC and list of pages
- send the 2nd packet with flags==0 and no pages
This is not expected, because MULTIFD_FLAG_SYNC should hopefully be done
after all the pages are received. Meanwhile, the 2nd packet will be
completely useless, which contains zero information.
I didn't verify above, but I think this issue is still benign in that at
least on the recv side we always receive pages before handling
MULTIFD_FLAG_SYNC. However that's not always guaranteed and just tricky.
One other reason I want to separate it is using p->flags to communicate
between the two threads is also not clearly defined, it's very hard to read
and understand why accessing p->flags is always safe; see the current impl
of multifd_send_thread() where we tried to cache only p->flags. It doesn't
need to be that complicated.
This patch introduces pending_sync, a separate flag just to show that the
requester needs a sync. Alongside, we remove the tricky caching of
p->flags now because after this patch p->flags should only be used by
multifd sender thread now, which will be crystal clear. So it is always
thread safe to access p->flags.
With that, we can also safely convert the pending_job into a boolean,
because we don't support >1 pending jobs anyway.
Always use atomic ops to access both flags to make sure no cache effect.
When at it, drop the initial setting of "pending_job = 0" because it's
always allocated using g_new0().
Peter Xu [Fri, 2 Feb 2024 10:28:39 +0000 (18:28 +0800)]
migration/multifd: Drop MultiFDSendParams.normal[] array
This array is redundant when p->pages exists. Now we extended the life of
p->pages to the whole period where pending_job is set, it should be safe to
always use p->pages->offset[] rather than p->normal[]. Drop the array.
Alongside, the normal_num is also redundant, which is the same to
p->pages->num.
This doesn't apply to recv side, because there's no extra buffering on recv
side, so p->normal[] array is still needed.
Peter Xu [Fri, 2 Feb 2024 10:28:38 +0000 (18:28 +0800)]
migration/multifd: Postpone reset of MultiFDPages_t
Now we reset MultiFDPages_t object in the multifd sender thread in the
middle of the sending job. That's not necessary, because the "*pages"
struct will not be reused anyway until pending_job is cleared.
Move that to the end after the job is completed, provide a helper to reset
a "*pages" object. Use that same helper when free the object too.
This prepares us to keep using p->pages in the follow up patches, where we
may drop p->normal[].
Merge them into the global one. The replacement is done by changing all
p->quit checks into the global var check. The global check doesn't need
any lock.
A few more things done on top of this altogether:
- multifd_send_terminate_threads()
Moving the xchg() of &multifd_send_state->exiting upper, so as to cover
the tracepoint, migrate_set_error() and migrate_set_state().
- multifd_send_sync_main()
In the 2nd loop, add one more check over the global var to make sure we
don't keep the looping if QEMU already decided to quit.
- multifd_tls_outgoing_handshake()
Use multifd_send_terminate_threads() to set the error state. That has
a benefit of updating MigrationState.error to that error too, so we can
persist that 1st error we hit in that specific channel.
- multifd_new_send_channel_async()
Take similar approach like above, drop the migrate_set_error() because
multifd_send_terminate_threads() already covers that. Unwrap the helper
multifd_new_send_channel_cleanup() along the way; not really needed.
William Roche [Tue, 30 Jan 2024 19:06:40 +0000 (19:06 +0000)]
migration: prevent migration when VM has poisoned memory
A memory page poisoned from the hypervisor level is no longer readable.
The migration of a VM will crash Qemu when it tries to read the
memory address space and stumbles on the poisoned page with a similar
stack trace:
Program terminated with signal SIGBUS, Bus error.
#0 _mm256_loadu_si256
#1 buffer_zero_avx2
#2 select_accel_fn
#3 buffer_is_zero
#4 save_zero_page
#5 ram_save_target_page_legacy
#6 ram_save_host_page
#7 ram_find_and_save_block
#8 ram_save_iterate
#9 qemu_savevm_state_iterate
#10 migration_iteration_run
#11 migration_thread
#12 qemu_thread_start
To avoid this VM crash during the migration, prevent the migration
when a known hardware poison exists on the VM.
hv-balloon: use get_min_alignment() to express 32 GiB alignment
Let's implement the get_min_alignment() callback for memory devices, and
copy for the device memory region the alignment of the host memory
region. This mimics what virtio-mem does, and allows for re-introducing
proper alignment checks for the memory region size (where we don't care
about additional device requirements) in memory device core.
Message-ID: <20240117135554.787344-2-david@redhat.com> Reviewed-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com> Signed-off-by: David Hildenbrand <david@redhat.com>
Better constraint for tcg_out_cmp, based on the comparison.
We can't yet remove the fallback to load constants into a
scratch because of tcg_out_cmp2, but that path should not
be as frequent.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Paolo Bonzini [Tue, 9 Jan 2024 22:30:32 +0000 (09:30 +1100)]
tcg/i386: Use TEST r,r to test 8/16/32 bits
Just like when testing against the sign bits, TEST r,r can be used when the
immediate is 0xff, 0xff00, 0xffff, 0xffffffff.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231028194522.245170-12-richard.henderson@linaro.org>
[PMD: Split from bigger patch, part 2/2] Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231108145244.72421-2-philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231028194522.245170-12-richard.henderson@linaro.org>
[PMD: Split from bigger patch, part 1/2] Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231108145244.72421-1-philmd@linaro.org>
tcg/aarch64: Generate CBNZ for TSTNE of UINT32_MAX
... and the inverse, CBZ for TSTEQ.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Test the sign bit for LT/GE vs 0, and TSTNE/EQ vs a power of 2.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240119224737.48943-2-philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
In order to ease next commit review, modify tcg_out_brcond()
to switch over TCGCond. No logical change intended.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240119224737.48943-1-philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
target/alpha: Use TCG_COND_TST{EQ,NE} for BLB{C,S}
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231028194522.245170-33-richard.henderson@linaro.org>
[PMD: Split from bigger patch, part 2/2] Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231108205247.83234-2-philmd@linaro.org>
target/alpha: Pass immediate value to gen_bcond_internal()
Simplify gen_bcond() by passing an immediate value.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231028194522.245170-33-richard.henderson@linaro.org>
[PMD: Split from bigger patch, part 1/2] Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231108205247.83234-1-philmd@linaro.org>
Fold constant comparisons.
Canonicalize "tst x,x" to equality vs zero.
Canonicalize "tst x,sign" to sign test vs zero.
Fold double-word comparisons with zero parts.
Fold setcond of "tst x,pow2" to a bit extract.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
* tag 'pull-qapi-2024-02-03' of https://repo.or.cz/qemu/armbru:
qga/qapi-schema: Move command description right after command name
qga: Move type description right after type name
qapi: Elide "Potential additional modes" from generated docs
qapi: Drop redundant documentation of conditional
qapi: Drop redundant documentation of inherited members
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
* tag 'pull-tcg-20240202-2' of https://gitlab.com/rth7680/qemu: (58 commits)
linux-user/aarch64: Add padding before __kernel_rt_sigreturn
target/sparc: Remove FSR_FTT_NMASK, FSR_FTT_CEXC_NMASK
target/sparc: Split fcc out of env->fsr
target/sparc: Remove cpu_fsr
target/sparc: Split cexc and ftt from env->fsr
target/sparc: Merge check_ieee_exceptions with FPop helpers
target/sparc: Clear cexc and ftt in do_check_ieee_exceptions
target/sparc: Split ver from env->fsr
target/sparc: Introduce cpu_get_fsr, cpu_put_fsr
target/sparc: Remove qt0, qt1 temporaries
target/sparc: Use i128 for Fdmulq
target/sparc: Use i128 for FdTOq, FxTOq
target/sparc: Use i128 for FsTOq, FiTOq
target/sparc: Use i128 for FCMPq, FCMPEq
target/sparc: Use i128 for FqTOd, FqTOx
target/sparc: Use i128 for FqTOs, FqTOi
target/sparc: Use i128 for FADDq, FSUBq, FMULq, FDIVq
target/sparc: Use i128 for FSQRTq
target/sparc: Inline FNEG, FABS
target/sparc: Introduce gen_{load,store}_fpr_Q
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
qga/qapi-schema: Move command description right after command name
Documentation of commands guest-ssh-get-authorized-keys,
guest-ssh-add-authorized-keys, and guest-ssh-remove-authorized-keys
describes the command's purpose after its arguments. Everywhere else,
we do it the other way round. Move it for consistency.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240129115008.674248-6-armbru@redhat.com> Reviewed-by: Konstantin Kostiuk <kkostiuk@redhat.com>
Documentation of type BlockdevOptionsIscsi describes the type's
purpose after its members. Everywhere else, we do it the other way
round. Move it for consistency.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240129115008.674248-5-armbru@redhat.com> Reviewed-by: Konstantin Kostiuk <kkostiuk@redhat.com>
qapi: Elide "Potential additional modes" from generated docs
Documentation of BlockExportRemoveMode has
Potential additional modes to be added in the future:
hide: Just hide export from new clients, leave existing connections
as is. Remove export after all clients are disconnected.
soft: Hide export from new clients, answer with ESHUTDOWN for all
further requests from existing clients.
I think this is useful only for developers. Elide it from generated
documentation by turning it into a TODO section.
This effectively reverts my own commit b71fd73cc45 (Revert "qapi:
BlockExportRemoveMode: move comments to TODO"). At the time, I was
about to elide TODO sections from the generated manual, I wasn't sure
about this one, and decided to avoid change. And now I've made up my
mind.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240129115008.674248-4-armbru@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
This command is only supported on s390 architecture.
and
If
~~
"TARGET_S390X"
The former became redundant in commit 901a34a400a (qapi: add 'If:'
section to generated documentation) added the latter. Drop the
former.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240129115008.674248-3-armbru@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
projects / qemu.git / log