We used to do off-by-one fixup for pss->page when finished one host huge page
transfer. That seems to be unnecesary at all. Drop it.
Cc: Keqian Zhu <zhukeqian1@huawei.com>
Cc: Kunkun Jiang <jiangkunkun@huawei.com>
Cc: Andrey Gruzdev <andrey.gruzdev@virtuozzo.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Provide information on the number of bytes copied in the pre-copy,
downtime and post-copy phases of migration.
Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Replace direct manipulation of ram_counters.transferred with a
function.
Signed-off-by: David Edmondson <david.edmondson@oracle.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
postcopy_send_discard_bm_ram() always return zero. Since it can't
fail, simplify and do not return anything.
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: David Edmondson <david.edmondson@oracle.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
It will just never fail. Drop those return values where they're constantly
zeros.
A tiny touch-up on the tracepoint so trace_ram_postcopy_send_discard_bitmap()
is called after the logic itself (which sounds more reasonable).
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Right now we loop ramblocks for twice, the 1st time chunk the dirty bits with
huge page information; the 2nd time we send the discard ranges. That's not
necessary - we can do them in a single loop.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This function calls three functions:
- postcopy_discard_send_init(ms, block->idstr);
- postcopy_chunk_hostpages_pass(ms, block);
- postcopy_discard_send_finish(ms);
However only the 2nd function call is meaningful. It's major role is to make
sure dirty bits are applied in host-page-size granule, so there will be no
partial dirty bits set for a whole host page if huge pages are used.
The 1st/3rd call are for latter when we want to send the disgard ranges.
They're mostly no-op here besides some tracepoints (which are misleading!).
Drop them, then we can directly drop postcopy_chunk_hostpages() as a whole
because we can call postcopy_chunk_hostpages_pass() directly.
There're still some nice comments above postcopy_chunk_hostpages() that explain
what it does. Copy it over to the caller's site.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
It always return zero, because it just can't go wrong so far. Simplify the
code with no functional change.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
I planned to add "#ifdef DEBUG_POSTCOPY" around the function too because
otherwise it'll be compiled into qemu binary even if it'll never be used. Then
I found that maybe it's easier to just drop it for good..
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Just a removal of an unused comment.
a0a8aa147a did many fixes and removed the parameter named "ms", but forget to remove the corresponding comment in function named "ram_save_host_page".
Signed-off-by: Xu Zheng <xuzheng@cmss.chinamobile.com>
Signed-off-by: Mao Zhongyi <maozhongyi@cmss.chinamobile.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
We always need to call it when we find a zero page, so put it in a
single place.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Remove the mask in the call to ram_release_pages(). Nothing else does
it, and if the offset has that bits set, we have a lot of trouble.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Remove the pages argument. And s/pages/page/
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
---
- Use 1LL instead of casts (philmd)
- Change the whole 1ULL for TARGET_PAGE_SIZE
We only need last_stage in two places and we are passing it all
around. Just add a field to RAMState that passes it.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
---
Repeat subject (philmd suggestion)
It just calls buffer_is_zero(). Just change the callers.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
The code to acquire bitmap_mutex is added in the commit of
"63268c4970a5f126cc9af75f3ccb8057abef5ec0". There is no
need to acquire bitmap_mutex in colo_flush_ram_cache(). This
is because the colo_flush_ram_cache only be called on the COLO
secondary VM, which is the destination side.
On the COLO secondary VM, only the COLO thread will touch
the bitmap of ram cache.
Signed-off-by: Lei Rao <lei.rao@intel.com>
Reviewed-by: Zhang Chen <chen.zhang@intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
if we don't reset the auto-converge counter,
it will continue to run with COLO running,
and eventually the system will hang due to the
CPU throttle reaching DEFAULT_MIGRATE_MAX_CPU_THROTTLE.
Signed-off-by: Lei Rao <lei.rao@intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>
Tested-by: Lukas Straub <lukasstraub2@web.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
When flushing memory from ram cache to ram during every checkpoint
on secondary VM, we can copy continuous chunks of memory instead of
4096 bytes per time to reduce the time of VM stop during checkpoint.
Signed-off-by: Lei Rao <lei.rao@intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Tested-by: Lukas Straub <lukasstraub2@web.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
One might set dump-guest-core=off to make coredumps smaller and
still allow to debug many qemu bugs. Extend this option to the colo
cache.
Signed-off-by: Lukas Straub <lukasstraub2@web.de>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This avoids to call migrate_get_current() in the caller function
whereas migration_cancel() already needs the pointer to the current
migration state.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
We already don't ever migrate memory that corresponds to discarded ranges
as managed by a RamDiscardManager responsible for the mapped memory region
of the RAMBlock.
virtio-mem uses this mechanism to logically unplug parts of a RAMBlock.
Right now, we still populate zeropages for the whole usable part of the
RAMBlock, which is undesired because:
1. Even populating the shared zeropage will result in memory getting
consumed for page tables.
2. Memory backends without a shared zeropage (like hugetlbfs and shmem)
will populate an actual, fresh page, resulting in an unintended
memory consumption.
Discarded ("logically unplugged") parts have to remain discarded. As
these pages are never part of the migration stream, there is no need to
track modifications via userfaultfd WP reliably for these parts.
Further, any writes to these ranges by the VM are invalid and the
behavior is undefined.
Note that Linux only supports userfaultfd WP on private anonymous memory
for now, which usually results in the shared zeropage getting populated.
The issue will become more relevant once userfaultfd WP supports shmem
and hugetlb.
Acked-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Let's factor out prefaulting/populating to make further changes easier to
review and add a comment what we are actually expecting to happen. While at
it, use the actual page size of the ramblock, which defaults to
qemu_real_host_page_size for anonymous memory. Further, rename
ram_block_populate_pages() to ram_block_populate_read() as well, to make
it clearer what we are doing.
In the future, we might want to use MADV_POPULATE_READ to speed up
population.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Let's use QEMU_ALIGN_DOWN() and friends to make the code a bit easier to
read.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Currently, when someone (i.e., the VM) accesses discarded parts inside a
RAMBlock with a RamDiscardManager managing the corresponding mapped memory
region, postcopy will request migration of the corresponding page from the
source. The source, however, will never answer, because it refuses to
migrate such pages with undefined content ("logically unplugged"): the
pages are never dirty, and get_queued_page() will consequently skip
processing these postcopy requests.
Especially reading discarded ("logically unplugged") ranges is supposed to
work in some setups (for example with current virtio-mem), although it
barely ever happens: still, not placing a page would currently stall the
VM, as it cannot make forward progress.
Let's check the state via the RamDiscardManager (the state e.g.,
of virtio-mem is migrated during precopy) and avoid sending a request
that will never get answered. Place a fresh zero page instead to keep
the VM working. This is the same behavior that would happen
automatically without userfaultfd being active, when accessing virtual
memory regions without populated pages -- "populate on demand".
For now, there are valid cases (as documented in the virtio-mem spec) where
a VM might read discarded memory; in the future, we will disallow that.
Then, we might want to handle that case differently, e.g., warning the
user that the VM seems to be mis-behaving.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
We don't want to migrate memory that corresponds to discarded ranges as
managed by a RamDiscardManager responsible for the mapped memory region of
the RAMBlock. The content of these pages is essentially stale and
without any guarantees for the VM ("logically unplugged").
Depending on the underlying memory type, even reading memory might populate
memory on the source, resulting in an undesired memory consumption. Of
course, on the destination, even writing a zeropage consumes memory,
which we also want to avoid (similar to free page hinting).
Currently, virtio-mem tries achieving that goal (not migrating "unplugged"
memory that was discarded) by going via qemu_guest_free_page_hint() - but
it's hackish and incomplete.
For example, background snapshots still end up reading all memory, as
they don't do bitmap syncs. Postcopy recovery code will re-add
previously cleared bits to the dirty bitmap and migrate them.
Let's consult the RamDiscardManager after setting up our dirty bitmap
initially and when postcopy recovery code reinitializes it: clear
corresponding bits in the dirty bitmaps (e.g., of the RAMBlock and inside
KVM). It's important to fixup the dirty bitmap *after* our initial bitmap
sync, such that the corresponding dirty bits in KVM are actually cleared.
As colo is incompatible with discarding of RAM and inhibits it, we don't
have to bother.
Note: if a misbehaving guest would use discarded ranges after migration
started we would still migrate that memory: however, then we already
populated that memory on the migration source.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
since dirty ring has been introduced, there are two methods
to track dirty pages of vm. it seems that "logging" has
a hint on the method, so rename the global_dirty_log to
global_dirty_tracking would make description more accurate.
dirty rate measurement may start or stop dirty tracking during
calculation. this conflict with migration because stop dirty
tracking make migration leave dirty pages out then that'll be
a problem.
make global_dirty_tracking a bitmask can let both migration and
dirty rate measurement work fine. introduce GLOBAL_DIRTY_MIGRATION
and GLOBAL_DIRTY_DIRTY_RATE to distinguish what current dirty
tracking aims for, migration or dirty rate.
Signed-off-by: Hyman Huang(黄勇) <huangy81@chinatelecom.cn>
Message-Id: <9c9388657cfa0301bd2c1cfa36e7cf6da4aeca19.1624040308.git.huangy81@chinatelecom.cn>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
The parameter is unused, let's drop it.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
When skipping free pages to send, their corresponding dirty bits in the
memory region dirty bitmap need to be cleared. Otherwise the skipped
pages will be sent in the next round after the migration thread syncs
dirty bits from the memory region dirty bitmap.
Cc: David Hildenbrand <david@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Reported-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Wei Wang <wei.w.wang@intel.com>
Message-Id: <20210722083055.23352-1-wei.w.wang@intel.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
migration uses QIOChannel typed qemufiles. In follow up patches, we'll need
the capability to identify this fact, so that we can get the backing QIOChannel
from a QEMUFile.
We can also define types for QEMUFile but so far since we only need to be able
to identify QIOChannel, introduce a boolean which is simpler.
Introduce another helper qemu_file_get_ioc() to return the ioc backend of a
qemufile if has_ioc is set.
No functional change.
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20210722175841.938739-5-peterx@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Accessing from_dst_file is potentially racy in current code base like below:
if (s->from_dst_file)
do_something(s->from_dst_file);
Because from_dst_file can be reset right after the check in another
thread (rp_thread). One example is migrate_fd_cancel().
Use the same qemu_file_lock to protect it too, just like to_dst_file.
When it's safe to access without lock, comment it.
There's one special reference in migration_thread() that can be replaced by
the newly introduced rp_thread_created flag.
Reported-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Lukas Straub <lukasstraub2@web.de>
Message-Id: <20210722175841.938739-3-peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
with Peter's fixup
Taking the mutex every time for each dirty bit to clear is too slow, especially
we'll take/release even if the dirty bit is cleared. So far it's only used to
sync with special cases with qemu_guest_free_page_hint() against migration
thread, nothing really that serious yet. Let's move the lock to be upper.
There're two callers of migration_bitmap_clear_dirty().
For migration, move it into ram_save_iterate(). With the help of MAX_WAIT
logic, we'll only run ram_save_iterate() for no more than 50ms-ish time, so
taking the lock once there at the entry. It also means any call sites to
qemu_guest_free_page_hint() can be delayed; but it should be very rare, only
during migration, and I don't see a problem with it.
For COLO, move it up to colo_flush_ram_cache(). I think COLO forgot to take
that lock even when calling ramblock_sync_dirty_bitmap(), where another example
is migration_bitmap_sync() who took it right. So let the mutex cover both the
ramblock_sync_dirty_bitmap() and migration_bitmap_clear_dirty() calls.
It's even possible to drop the lock so we use atomic operations upon rb->bmap
and the variable migration_dirty_pages. I didn't do it just to still be safe,
also not predictable whether the frequent atomic ops could bring overhead too
e.g. on huge vms when it happens very often. When that really comes, we can
keep a local counter and periodically call atomic ops. Keep it simple for now.
Cc: Wei Wang <wei.w.wang@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hailiang Zhang <zhang.zhanghailiang@huawei.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Juan Quintela <quintela@redhat.com>
Cc: Leonardo Bras Soares Passos <lsoaresp@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20210630200805.280905-1-peterx@redhat.com>
Reviewed-by: Wei Wang <wei.w.wang@intel.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Fixes commit 3d0684b2ad ("ram: Update
all functions comments")
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210708162159.18045-1-olaf@aepfle.de>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Let's introduce RAM_NORESERVE, allowing mmap'ing with MAP_NORESERVE. The
new flag has the following semantics:
"
RAM is mmap-ed with MAP_NORESERVE. When set, reserving swap space (or huge
pages if applicable) is skipped: will bail out if not supported. When not
set, the OS will do the reservation, if supported for the memory type.
"
Allow passing it into:
- memory_region_init_ram_nomigrate()
- memory_region_init_resizeable_ram()
- memory_region_init_ram_from_file()
... and teach qemu_ram_mmap() and qemu_anon_ram_alloc() about the flag.
Bail out if the flag is not supported, which is the case right now for
both, POSIX and win32. We will add Linux support next and allow specifying
RAM_NORESERVE via memory backends.
The target use case is virtio-mem, which dynamically exposes memory
inside a large, sparse memory area to the VM.
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Acked-by: Eduardo Habkost <ehabkost@redhat.com> for memory backend and machine core
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210510114328.21835-9-david@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
We never read or write beyond the used_length of memory blocks when
migrating. Make this clearer by using offset_in_ramblock() consistently.
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210429112708.12291-11-david@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Resizing while migrating is dangerous and does not work as expected.
The whole migration code works with the usable_length of a ram block and
does not expect this value to change at random points in time.
In the case of postcopy, relying on used_length is racy as soon as the
guest is running. Also, when used_length changes we might leave the
uffd handler registered for some memory regions, reject valid pages
when migrating and fail when sending the recv bitmap to the source.
Resizing can be trigger *after* (but not during) a reset in
ACPI code by the guest
- hw/arm/virt-acpi-build.c:acpi_ram_update()
- hw/i386/acpi-build.c:acpi_ram_update()
Let's remember the original used_length in a separate variable and
use it in relevant postcopy code. Make sure to update it when we resize
during precopy, when synchronizing the RAM block sizes with the source.
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210429112708.12291-9-david@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Add two new helper functions. This will come in come handy once we want to
handle ram block resizes while postcopy is active.
Note that ram_block_from_stream() will already print proper errors.
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210429112708.12291-8-david@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
dgilbert: Added brackets in host_page_from_ram_block_offset
to cause uintptr_t to cast the sum, to fix armhf-cross build
In case we grow our RAM after ram_postcopy_incoming_init() (e.g., when
synchronizing the RAM block state with the migration source), the resized
part would not get discarded. Let's perform that when being notified
about a resize while postcopy has been advised, but is not listening
yet. With precopy, the process is as following:
1. VM created
- RAM blocks are created
2. Incomming migration started
- Postcopy is advised
- All pages in RAM blocks are discarded
3. Precopy starts
- RAM blocks are resized to match the size on the migration source.
- RAM pages from precopy stream are loaded
- Uffd handler is registered, postcopy starts listening
4. Guest started, postcopy running
- Pagefaults get resolved, pages get placed
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210429112708.12291-7-david@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Resizing while migrating is dangerous and does not work as expected.
The whole migration code works on the usable_length of ram blocks and does
not expect this to change at random points in time.
In the case of precopy, the ram block size must not change on the source,
after syncing the RAM block list in ram_save_setup(), so as long as the
guest is still running on the source.
Resizing can be trigger *after* (but not during) a reset in
ACPI code by the guest
- hw/arm/virt-acpi-build.c:acpi_ram_update()
- hw/i386/acpi-build.c:acpi_ram_update()
Use the ram block notifier to get notified about resizes. Let's simply
cancel migration and indicate the reason. We'll continue running on the
source. No harm done.
Update the documentation. Postcopy will be handled separately.
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210429112708.12291-5-david@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Manual merge
Starting from pss->page, ram_save_host_page() will check every page
and send the dirty pages up to the end of the current host page or
the boundary of used_length of the block. If the host page size is
a huge page, the step "check" will take a lot of time.
It will improve performance to use migration_bitmap_find_dirty().
Tested on Kunpeng 920; VM parameters: 1U 4G (page size 1G)
The time of ram_save_host_page() in the last round of ram saving:
before optimize: 9250us after optimize: 34us
Signed-off-by: Keqian Zhu <zhukeqian1@huawei.com>
Signed-off-by: Kunkun Jiang <jiangkunkun@huawei.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <20210316125716.1243-3-jiangkunkun@huawei.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
When the host page is a huge page and something is sent in the
current iteration, migration_rate_limit() should be executed.
If not, it can be omitted.
Signed-off-by: Keqian Zhu <zhukeqian1@huawei.com>
Signed-off-by: Kunkun Jiang <jiangkunkun@huawei.com>
Reviewed-by: David Edmondson <david.edmondson@oracle.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20210316125716.1243-2-jiangkunkun@huawei.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
The bulk stage is kind of weird: migration_bitmap_find_dirty() will
indicate a dirty page, however, ram_save_host_page() will never save it, as
migration_bitmap_clear_dirty() detects that it is not dirty.
We already fill the bitmap in ram_list_init_bitmaps() with ones, marking
everything dirty - it didn't used to be that way, which is why we needed
an explicit first bulk stage.
Let's simplify: make the bitmap the single source of thuth. Explicitly
handle the "xbzrle_enabled after first round" case.
Regarding XBZRLE (implicitly handled via "ram_bulk_stage = false" right
now), there is now a slight change in behavior:
- Colo: When starting, it will be disabled (was implicitly enabled)
until the first round actually finishes.
- Free page hinting: When starting, XBZRLE will be disabled (was implicitly
enabled) until the first round actually finished.
- Snapshots: When starting, XBZRLE will be disabled. We essentially only
do a single run, so I guess it will never actually get disabled.
Postcopy seems to indirectly disable it in ram_save_page(), so there
shouldn't be really any change.
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Juan Quintela <quintela@redhat.com>
Cc: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Cc: Andrey Gruzdev <andrey.gruzdev@virtuozzo.com>
Cc: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210216105039.40680-1-david@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Stop including cpu.h in files that don't need it.
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210416171314.2074665-4-thuth@redhat.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Stop including sysemu/sysemu.h in files that don't need it.
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20210416171314.2074665-2-thuth@redhat.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Rename 'bs' to commonly used 'block' in migration/ram.c background
snapshot code.
Signed-off-by: Andrey Gruzdev <andrey.gruzdev@virtuozzo.com>
Reported-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210401092226.102804-5-andrey.gruzdev@virtuozzo.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
This commit solves the issue with userfault_fd WP feature that
background snapshot is based on. For any never poluated or discarded
memory page, the UFFDIO_WRITEPROTECT ioctl() would skip updating
PTE for that page, thereby loosing WP setting for it.
So we need to pre-fault pages for each RAM block to be protected
before making a userfault_fd wr-protect ioctl().
Fixes: 278e2f551a (migration: support
UFFD write fault processing in ram_save_iterate())
Signed-off-by: Andrey Gruzdev <andrey.gruzdev@virtuozzo.com>
Reported-by: David Hildenbrand <david@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210401092226.102804-4-andrey.gruzdev@virtuozzo.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
dgilbert:
Bodged ifdef __linux__ on ram_write_tracking_prepare, should really
go in a stub
The generic 'migrate_set_parameters' command handle all types of param.
Only the QMP commands were documented in the deprecations page, but the
rationale for deprecating applies equally to HMP, and the replacements
exist. Furthermore the HMP commands are just shims to the QMP commands,
so removing the latter breaks the former unless they get re-implemented.
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Replaced various qemu_mutex_lock calls and their respective
qemu_mutex_unlock calls with QEMU_LOCK_GUARD macro. This simplifies
the code by eliminating the respective qemu_mutex_unlock calls.
Signed-off-by: Mahmoud Mandour <ma.mandourr@gmail.com>
Message-Id: <20210311031538.5325-7-ma.mandourr@gmail.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Postcopy may also be advised for dirty-bitmap migration only, in which
case the remote page size will not be available and we'll instead read
bogus data, blocking migration with a mismatch error if the VM uses
hugepages.
Fixes: 58110f0acb ("migration: split common postcopy out of ram postcopy")
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Message-Id: <20210204163522.13291-1-s.reiter@proxmox.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
73af8dd8d7 "migration: Make xbzrle_cache_size a migration
parameter" (v2.11.0) made the new parameter unsigned (QAPI type
'size', uint64_t in C). It neglected to update existing code, which
continues to use int64_t.
migrate_xbzrle_cache_size() returns the new parameter. Adjust its
return type.
QMP query-migrate-cache-size returns migrate_xbzrle_cache_size().
Adjust its return type.
migrate-set-parameters passes the new parameter to
xbzrle_cache_resize(). Adjust its parameter type.
xbzrle_cache_resize() passes it on to cache_init(). Adjust its
parameter type.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20210202141734.2488076-3-armbru@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
In this particular implementation the same single migration
thread is responsible for both normal linear dirty page
migration and procesing UFFD page fault events.
Processing write faults includes reading UFFD file descriptor,
finding respective RAM block and saving faulting page to
the migration stream. After page has been saved, write protection
can be removed. Since asynchronous version of qemu_put_buffer()
is expected to be used to save pages, we also have to flush
migraion stream prior to un-protecting saved memory range.
Write protection is being removed for any previously protected
memory chunk that has hit the migration stream. That's valid
for pages from linear page scan along with write fault pages.
Signed-off-by: Andrey Gruzdev <andrey.gruzdev@virtuozzo.com>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20210129101407.103458-4-andrey.gruzdev@virtuozzo.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
fixup pagefault.address cast for 32bit
Add new capability to 'qapi/migration.json' schema.
Update migrate_caps_check() to validate enabled capability set
against introduced one. Perform checks for required kernel features
and compatibility with guest memory backends.
Signed-off-by: Andrey Gruzdev <andrey.gruzdev@virtuozzo.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20210129101407.103458-2-andrey.gruzdev@virtuozzo.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
The '%u' conversion specifier is for decimal notation.
When prefixing a format with '0x', we want the hexadecimal
specifier ('%x').
Inspired-by: Dov Murik <dovmurik@linux.vnet.ibm.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20201103112558.2554390-5-philmd@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
we should set ram_bulk_stage to false after ram_state_init,
otherwise the bitmap will be unused in migration_bitmap_find_dirty.
all pages in ram cache will be flushed to the ram of secondary guest
for each checkpoint.
Signed-off-by: Lei Rao <lei.rao@intel.com>
Signed-off-by: Derek Su <dereksu@qnap.com>
Signed-off-by: Zhang Chen <chen.zhang@intel.com>
Reviewed-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Reviewed-by: Zhang Chen <chen.zhang@intel.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Bihong Yu <yubihong@huawei.com>
Reviewed-by: Chuan Zheng <zhengchuan@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <1603163448-27122-7-git-send-email-yubihong@huawei.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Bihong Yu <yubihong@huawei.com>
Reviewed-by: Chuan Zheng <zhengchuan@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <1603163448-27122-6-git-send-email-yubihong@huawei.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Bihong Yu <yubihong@huawei.com>
Reviewed-by: Chuan Zheng <zhengchuan@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <1603163448-27122-4-git-send-email-yubihong@huawei.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Bihong Yu <yubihong@huawei.com>
Reviewed-by: Chuan Zheng <zhengchuan@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <1603163448-27122-3-git-send-email-yubihong@huawei.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
RAMBLOCK_FOREACH_MIGRATABLE is need in dirtyrate measure,
move the existing definition up into migration/ram.h
Signed-off-by: Chuan Zheng <zhengchuan@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: David Edmondson <david.edmondson@oracle.com>
Reviewed-by: Li Qiang <liq3ea@gmail.com>
Message-Id: <1600237327-33618-6-git-send-email-zhengchuan@huawei.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
I found that there are many spelling errors in the comments of qemu,
so I used the spellcheck tool to check the spelling errors
and finally found some spelling errors in the migration folder.
Signed-off-by: zhaolichang <zhaolichang@huawei.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20200917075029.313-3-zhaolichang@huawei.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
move the vcpu throttling functionality into its own module.
This functionality is not specific to any accelerator,
and it is used currently by migration to slow down guests to try to
have migrations converge, and by the cocoa MacOS UI to throttle speed.
cpu-throttle contains the controls to adjust and inspect throttle
settings, start (set) and stop vcpu throttling, and the throttling
function itself that is run periodically on vcpus to make them take a nap.
Execution of the throttling function on all vcpus is triggered by a timer,
registered at module initialization.
No functionality change.
Signed-off-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Message-Id: <20200629093504.3228-3-cfontana@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
real_dirty_pages becomes equal to total ram size after dirty log sync
in ram_init_bitmaps, the reason is that the bitmap of ramblock is
initialized to be all set, so old path counts them as "real dirty" at
beginning.
This causes wrong dirty rate and false positive throttling.
Signed-off-by: Keqian Zhu <zhukeqian1@huawei.com>
Message-Id: <20200622032037.31112-1-zhukeqian1@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
It's reported an error of implicit conversion from "unsigned long" to
"double" when compiling with Clang 10. Simply make the encoding rate 0
when the encoded_size is 0.
Fixes: e460a4b1a4
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reported-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Wei Wang <wei.w.wang@intel.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20200617201309.1640952-3-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
If we suceed in receiving ram state, but fail receiving the device
state, there will be a mismatch between the two.
Fix this by flushing the ram cache only after the vmstate has been
received.
Signed-off-by: Lukas Straub <lukasstraub2@web.de>
Message-Id: <3289d007d494cb0e2f05b1cf4ae6a78d300fede3.1589193382.git.lukasstraub2@web.de>
Reviewed-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Users may need to check the xbzrle encoding rate to know if the guest
memory is xbzrle encoding-friendly, and dynamically turn off the
encoding if the encoding rate is low.
Signed-off-by: Yi Sun <yi.y.sun@intel.com>
Signed-off-by: Wei Wang <wei.w.wang@intel.com>
Message-Id: <1588208375-19556-1-git-send-email-wei.w.wang@intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Let's consolidate resetting the variables.
Cc: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Cc: Juan Quintela <quintela@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20200421085300.7734-10-david@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Fixup for context conflicts with 91ba442
At the tail stage of throttling, the Guest is very sensitive to
CPU percentage while the @cpu-throttle-increment is excessive
usually at tail stage.
If this parameter is true, we will compute the ideal CPU percentage
used by the Guest, which may exactly make the dirty rate match the
dirty rate threshold. Then we will choose a smaller throttle increment
between the one specified by @cpu-throttle-increment and the one
generated by ideal CPU percentage.
Therefore, it is compatible to traditional throttling, meanwhile
the throttle increment won't be excessive at tail stage. This may
make migration time longer, and is disabled by default.
Signed-off-by: Keqian Zhu <zhukeqian1@huawei.com>
Message-Id: <20200413101508.54793-1-zhukeqian1@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Silent static analyzer warning
Remove dead assignments
Support -chardev serial on macOS
Update MAINTAINERS
Some cosmetic changes
-----BEGIN PGP SIGNATURE-----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=ibgH
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/vivier2/tags/trivial-branch-for-5.1-pull-request' into staging
trivial patches (20200504)
Silent static analyzer warning
Remove dead assignments
Support -chardev serial on macOS
Update MAINTAINERS
Some cosmetic changes
# gpg: Signature made Mon 04 May 2020 16:45:18 BST
# gpg: using RSA key CD2F75DDC8E3A4DC2E4F5173F30C38BD3F2FBE3C
# gpg: issuer "laurent@vivier.eu"
# gpg: Good signature from "Laurent Vivier <lvivier@redhat.com>" [full]
# gpg: aka "Laurent Vivier <laurent@vivier.eu>" [full]
# gpg: aka "Laurent Vivier (Red Hat) <lvivier@redhat.com>" [full]
# Primary key fingerprint: CD2F 75DD C8E3 A4DC 2E4F 5173 F30C 38BD 3F2F BE3C
* remotes/vivier2/tags/trivial-branch-for-5.1-pull-request:
hw/timer/pxa2xx_timer: Add assertion to silent static analyzer warning
hw/timer/stm32f2xx_timer: Remove dead assignment
hw/gpio/aspeed_gpio: Remove dead assignment
hw/isa/i82378: Remove dead assignment
hw/ide/sii3112: Remove dead assignment
hw/input/adb-kbd: Remove dead assignment
hw/i2c/pm_smbus: Remove dead assignment
blockdev: Remove dead assignment
block: Avoid dead assignment
Compress lines for immediate return
chardev: Add macOS to list of OSes that support -chardev serial
MAINTAINERS: Update Keith Busch's email address
elf_ops: Don't try to g_mapped_file_unref(NULL)
hw/mem/pc-dimm: Fix line over 80 characters warning
hw/mem/pc-dimm: Print slot number on error at pc_dimm_pre_plug()
MAINTAINERS: Mark the LatticeMico32 target as orphan
timer/exynos4210_mct: Remove redundant statement in exynos4210_mct_write()
display/blizzard: use extract16() for fix clang analyzer warning in blizzard_draw_line16_32()
scsi/esp-pci: add g_assert() for fix clang analyzer warning in esp_pci_io_write()
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
- ran regexp "qemu_mutex_lock\(.*\).*\n.*if" to find targets
- replaced result with QEMU_LOCK_GUARD if all unlocks at function end
- replaced result with WITH_QEMU_LOCK_GUARD if unlock not at end
Signed-off-by: Daniel Brodsky <dnbrdsky@gmail.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-id: 20200404042108.389635-3-dnbrdsky@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Compress two lines into a single line if immediate return statement is found.
It also remove variables progress, val, data, ret and sock
as they are no longer needed.
Remove space between function "mixer_load" and '(' to fix the
checkpatch.pl error:-
ERROR: space prohibited between function name and open parenthesis '('
Done using following coccinelle script:
@@
local idexpression ret;
expression e;
@@
-ret =
+return
e;
-return ret;
Signed-off-by: Simran Singhal <singhalsimran0@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20200401165314.GA3213@simran-Inspiron-5558>
[lv: in handle_aiocb_write_zeroes_unmap() move "int ret" inside the #ifdef]
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
local_err is used again in migration_bitmap_sync_precopy() after
precopy_notify(), so we must zero it. Otherwise try to set
non-NULL local_err will crash.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20200324153630.11882-6-vsementsov@virtuozzo.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
It is only need to record bitmap of dirty pages while goes
into COLO stage.
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Message-Id: <20200224065414.36524-6-zhang.zhanghailiang@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
This patch will reduce the downtime of VM for the initial process,
Previously, we copied all these memory in preparing stage of COLO
while we need to stop VM, which is a time-consuming process.
Here we optimize it by a trick, back-up every page while in migration
process while COLO is enabled, though it affects the speed of the
migration, but it obviously reduce the downtime of back-up all SVM'S
memory in COLO preparing stage.
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Message-Id: <20200224065414.36524-5-zhang.zhanghailiang@huawei.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
minor typo fixes
Currently, if the bytes_dirty_period is more than the 50% of
bytes_xfer_period, we start or increase throttling.
If we make this percentage higher, then we can tolerate higher
dirty rate during migration, which means less impact on guest.
The side effect of higher percentage is longer migration time.
We can make this parameter configurable to switch between mig-
ration time first or guest performance first.
The default value is 50 and valid range is 1 to 100.
Signed-off-by: Keqian Zhu <zhukeqian1@huawei.com>
Message-Id: <20200224023142.39360-1-zhukeqian1@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
It will be used later.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
---
No comp value needs to be zero.
We need to change the full chain to pass the Error parameter.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
If the multifd_send_threads is not created when migration is failed,
multifd_save_cleanup would be called twice. In this senario, the
multifd_send_state is accessed after it has been released, the result
is that the source VM is crashing down.
Here is the coredump stack:
Program received signal SIGSEGV, Segmentation fault.
0x00005629333a78ef in multifd_send_terminate_threads (err=err@entry=0x0) at migration/ram.c:1012
1012 MultiFDSendParams *p = &multifd_send_state->params[i];
#0 0x00005629333a78ef in multifd_send_terminate_threads (err=err@entry=0x0) at migration/ram.c:1012
#1 0x00005629333ab8a9 in multifd_save_cleanup () at migration/ram.c:1028
#2 0x00005629333abaea in multifd_new_send_channel_async (task=0x562935450e70, opaque=<optimized out>) at migration/ram.c:1202
#3 0x000056293373a562 in qio_task_complete (task=task@entry=0x562935450e70) at io/task.c:196
#4 0x000056293373a6e0 in qio_task_thread_result (opaque=0x562935450e70) at io/task.c:111
#5 0x00007f475d4d75a7 in g_idle_dispatch () from /usr/lib64/libglib-2.0.so.0
#6 0x00007f475d4da9a9 in g_main_context_dispatch () from /usr/lib64/libglib-2.0.so.0
#7 0x0000562933785b33 in glib_pollfds_poll () at util/main-loop.c:219
#8 os_host_main_loop_wait (timeout=<optimized out>) at util/main-loop.c:242
#9 main_loop_wait (nonblocking=nonblocking@entry=0) at util/main-loop.c:518
#10 0x00005629334c5acf in main_loop () at vl.c:1810
#11 0x000056293334d7bb in main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:4471
If the multifd_send_threads is not created when migration is failed.
In this senario, we don't call multifd_save_cleanup in multifd_new_send_channel_async.
Signed-off-by: Zhimin Feng <fengzhimin1@huawei.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
If we do a cancel, we got out without one error, but we can't do the
rest of the output as in a normal situation.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
We transmit ram_addr_t always as uint64_t. Be consistent in its
use (on 64bit system, it is always uint64_t problem is 32bits).
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Added type conversions to ram_addr_t before all left shifts of page
indexes to TARGET_PAGE_BITS, to correct overflows when the page
address was 4Gb and more.
Signed-off-by: Alexey Romko <nevilad@yahoo.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
One multifd will lock all the other multifds' IOChannel mutex to inform them
to quit by setting p->quit or shutting down p->c. In this senario, if some
multifds had already been terminated and multifd_load_cleanup/multifd_save_cleanup
had destroyed their mutex, it could cause destroyed mutex access when trying
lock their mutex.
Here is the coredump stack:
#0 0x00007f81a2794437 in raise () from /usr/lib64/libc.so.6
#1 0x00007f81a2795b28 in abort () from /usr/lib64/libc.so.6
#2 0x00007f81a278d1b6 in __assert_fail_base () from /usr/lib64/libc.so.6
#3 0x00007f81a278d262 in __assert_fail () from /usr/lib64/libc.so.6
#4 0x000055eb1bfadbd3 in qemu_mutex_lock_impl (mutex=0x55eb1e2d1988, file=<optimized out>, line=<optimized out>) at util/qemu-thread-posix.c:64
#5 0x000055eb1bb4564a in multifd_send_terminate_threads (err=<optimized out>) at migration/ram.c:1015
#6 0x000055eb1bb4bb7f in multifd_send_thread (opaque=0x55eb1e2d19f8) at migration/ram.c:1171
#7 0x000055eb1bfad628 in qemu_thread_start (args=0x55eb1e170450) at util/qemu-thread-posix.c:502
#8 0x00007f81a2b36df5 in start_thread () from /usr/lib64/libpthread.so.0
#9 0x00007f81a286048d in clone () from /usr/lib64/libc.so.6
To fix it up, let's destroy the mutex after all the other multifd threads had
been terminated.
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Signed-off-by: Ying Fang <fangying1@huawei.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
One multifd channel will shutdown all the other multifd's IOChannel when it
fails to receive an IOChannel. In this senario, if some multifds had not
received its IOChannel yet, it would try to shutdown its IOChannel which could
cause nullptr access at qio_channel_shutdown.
Here is the coredump stack:
#0 object_get_class (obj=obj@entry=0x0) at qom/object.c:908
#1 0x00005563fdbb8f4a in qio_channel_shutdown (ioc=0x0, how=QIO_CHANNEL_SHUTDOWN_BOTH, errp=0x0) at io/channel.c:355
#2 0x00005563fd7b4c5f in multifd_recv_terminate_threads (err=<optimized out>) at migration/ram.c:1280
#3 0x00005563fd7bc019 in multifd_recv_new_channel (ioc=ioc@entry=0x556400255610, errp=errp@entry=0x7ffec07dce00) at migration/ram.c:1478
#4 0x00005563fda82177 in migration_ioc_process_incoming (ioc=ioc@entry=0x556400255610, errp=errp@entry=0x7ffec07dce30) at migration/migration.c:605
#5 0x00005563fda8567d in migration_channel_process_incoming (ioc=0x556400255610) at migration/channel.c:44
#6 0x00005563fda83ee0 in socket_accept_incoming_migration (listener=0x5563fff6b920, cioc=0x556400255610, opaque=<optimized out>) at migration/socket.c:166
#7 0x00005563fdbc25cd in qio_net_listener_channel_func (ioc=<optimized out>, condition=<optimized out>, opaque=<optimized out>) at io/net-listener.c:54
#8 0x00007f895b6fe9a9 in g_main_context_dispatch () from /usr/lib64/libglib-2.0.so.0
#9 0x00005563fdc18136 in glib_pollfds_poll () at util/main-loop.c:218
#10 0x00005563fdc181b5 in os_host_main_loop_wait (timeout=1000000000) at util/main-loop.c:241
#11 0x00005563fdc183a2 in main_loop_wait (nonblocking=nonblocking@entry=0) at util/main-loop.c:517
#12 0x00005563fd8edb37 in main_loop () at vl.c:1791
#13 0x00005563fd74fd45 in main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:4473
To fix it up, let's check p->c before calling qio_channel_shutdown.
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Signed-off-by: Ying Fang <fangying1@huawei.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
We don't support multifd during postcopy, but user still could enable
both multifd and postcopy. This leads to migration failure.
Skip multifd during postcopy.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This is a preparation for the next patch:
not use multifd during postcopy.
Without enabling postcopy, everything looks good. While after enabling
postcopy, migration may fail even not use multifd during postcopy. The
reason is the pages is not properly cleared and *old* target page will
continue to be transferred.
After clean pages, migration succeeds.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
postcopy requires to place a whole host page, while migration thread
migrate memory in target page size. This makes postcopy need to collect
all target pages in one host page before placing via userfaultfd.
To enable compress during postcopy, there are two problems to solve:
1. Random order for target page arrival
2. Target pages in one host page arrives without interrupt by target
page from other host page
The first one is handled by previous cleanup patch.
This patch handles the second one by:
1. Flush compress thread for each host page
2. Wait for decompress thread for before placing host page
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
After using number of target page received to track one host page, we
could have the capability to handle random order target page arrival in
one host page.
This is a preparation for enabling compress during postcopy.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
For the first target page, all_zero is set to true for this round check.
After target_pages introduced, we could leverage this variable instead
of checking the address offset.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
In postcopy, it requires to place whole host page instead of target
page.
Currently, it relies on the page offset to decide whether this is the
last target page. We also can count the target page number during the
iteration. When the number of target page equals
(host page size / target page size), this means it is the last target
page in the host page.
This is a preparation for non-ordered target page transmission.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Compress is not supported with postcopy, it is safe to wait for
decompress thread just in precopy.
This is a preparation for later patch.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
In this case, page_buffer content would not be used.
Skip this to save some time.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Usually, incoming migration coroutine yields to the main loop
while its IO-channel is waiting for data to receive. But there is a case
when RAM migration and data receive have the same speed: VM with huge
zeroed RAM. In this case, IO-channel won't read and thus the main loop
is stuck and for instance, it doesn't respond to QMP commands.
For this case, yield periodically, but not too often, so as not to
affect the speed of migration.
Signed-off-by: Yury Kotov <yury-kotov@yandex-team.ru>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
When using hugepages, rate limiting is necessary within each huge
page, since a 1G huge page can take a significant time to send, so
you end up with bursty behaviour.
Fixes: 4c011c37ec ("postcopy: Send whole huge pages")
Reported-by: Lin Ma <LMa@suse.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
ram_save_queue_pages() has an 'err' label that can be replaced by
'return -1' instead.
Same thing with ram_discard_range(), and in this case we can also
get rid of the 'ret' variable and return either '-1' on error
or the result of ram_block_discard_range().
CC: Juan Quintela <quintela@redhat.com>
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
If we are exiting due to an error/finish/.... Just don't try to even
touch the channel with one IO operation.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Fill everything with zero, so the padding fields are also initialized.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Use WITH_RCU_READ_LOCK_GUARD to avoid exiting colo_init_ram_cache
without releasing RCU.
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
../migration/ram.c: In function ‘multifd_recv_thread’:
/home/elmarco/src/qq/include/qapi/error.h:165:5: error: ‘block’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
165 | error_setg_internal((errp), __FILE__, __LINE__, __func__, \
| ^~~~~~~~~~~~~~~~~~~
../migration/ram.c:818:15: note: ‘block’ was declared here
818 | RAMBlock *block;
| ^~~~~
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Switch to ram block writeback for pmem migration.
Signed-off-by: Beata Michalska <beata.michalska@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Acked-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-id: 20191121000843.24844-4-beata.michalska@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
When we found an available channel in multifd_send_pages(), its
pages->used is cleared and then attached to multifd_send_state.
It is not necessary to do this twice.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20191011085050.17622-5-richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
MultiFDPacket_t's magic and version field never changes during
migration, so move these two fields in setup stage.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20191011085050.17622-4-richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
multifd_send_fill_packet() prepares meta data for following pages to
transfer. It would be more proper to fill pages->allocated instead of
static max value, especially we want to support flexible packet size.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20191011085050.17622-3-richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20191011085050.17622-2-richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
During migration, a tmp page is allocated so that we could place a whole
host page during postcopy.
Currently the page is allocated during load stage, this is a little bit
late. And more important, if we failed to allocate it, the error is not
checked properly. Even it is NULL, we would still use it.
This patch moves the allocation to setup stage and if failed error
message would be printed and caller would notice it.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Use the automatic read unlocker in migration/ram.c
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20191007143642.301445-4-dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Use the automatic rcu_read unlocker to fix a missing unlock.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20191007143642.301445-3-dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
This is a cleanup for previous removal of unsentmap.
The sent parameter is not necessary now.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190819061843.28642-4-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Commit f3f491fcd6 ('Postcopy: Maintain unsentmap') introduced
unsentmap to track not yet sent pages.
This is not necessary since:
* unsentmap is a sub-set of bmap before postcopy start
* unsentmap is the summation of bmap and unsentmap after canonicalizing
This patch just removes it.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190819061843.28642-3-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
All pages, either partially sent or partially dirty, will be discarded in
postcopy_send_discard_bm_ram(), since we update the unsentmap to be
unsentmap = unsentmap | dirty in ram_postcopy_send_discard_bitmap().
This is not necessary to do discard when canonicalizing bitmap. And by
doing so, we separate the page discard into two individual steps:
* canonicalize bitmap
* discard page
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190819061843.28642-2-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Commit 78dd48df3 removed the last caller of register_savevm_live for an
instantiable device (rather than a single system wide device);
so trim out the parameter.
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20190822115433.12070-1-dgilbert@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
When encounter error, multifd_send_thread should always notify who pay
attention to it before exit. Otherwise it may block migration_thread
at multifd_send_sync_main forever.
Error as follow:
-------------------------------------------------------------------------------
(gdb) bt
#0 0x00007f4d669dfa0b in do_futex_wait.constprop.1 () from /lib64/libpthread.so.0
#1 0x00007f4d669dfa9f in __new_sem_wait_slow.constprop.0 () from /lib64/libpthread.so.0
#2 0x00007f4d669dfb3b in sem_wait@@GLIBC_2.2.5 () from /lib64/libpthread.so.0
#3 0x0000562ccf0a5614 in qemu_sem_wait (sem=sem@entry=0x562cd1b698e8) at util/qemu-thread-posix.c:319
#4 0x0000562ccecb4752 in multifd_send_sync_main (rs=<optimized out>) at /qemu/migration/ram.c:1099
#5 0x0000562ccecb95f4 in ram_save_iterate (f=0x562cd0ecc000, opaque=<optimized out>) at /qemu/migration/ram.c:3550
#6 0x0000562ccef43c23 in qemu_savevm_state_iterate (f=0x562cd0ecc000, postcopy=false) at migration/savevm.c:1189
#7 0x0000562ccef3dcf3 in migration_iteration_run (s=0x562cd09fabf0) at migration/migration.c:3131
#8 migration_thread (opaque=opaque@entry=0x562cd09fabf0) at migration/migration.c:3258
#9 0x0000562ccf0a4c26 in qemu_thread_start (args=<optimized out>) at util/qemu-thread-posix.c:502
#10 0x00007f4d669d9e25 in start_thread () from /lib64/libpthread.so.0
#11 0x00007f4d6670635d in clone () from /lib64/libc.so.6
(gdb) f 4
#4 0x0000562ccecb4752 in multifd_send_sync_main (rs=<optimized out>) at /qemu/migration/ram.c:1099
1099 qemu_sem_wait(&p->sem_sync);
(gdb) list
1094 }
1095 for (i = 0; i < migrate_multifd_channels(); i++) {
1096 MultiFDSendParams *p = &multifd_send_state->params[i];
1097
1098 trace_multifd_send_sync_main_wait(p->id);
1099 qemu_sem_wait(&p->sem_sync);
1100 }
1101 trace_multifd_send_sync_main(multifd_send_state->packet_num);
1102 }
1103
(gdb) p i
$1 = 0
(gdb) p multifd_send_state->params[0].pending_job
$2 = 2 //It means the job before MULTIFD_FLAG_SYNC has already fail
(gdb) p multifd_send_state->params[0].quit
$3 = true
Signed-off-by: Ivan Ren <ivanren@tencent.com>
Message-Id: <1567044996-2362-1-git-send-email-ivanren@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
There is a race between TCG and accesses to the dirty log:
vCPU thread reader thread
----------------------- -----------------------
TLB check -> slow path
notdirty_mem_write
write to RAM
set dirty flag
clear dirty flag
TLB check -> fast path
read memory
write to RAM
Fortunately, in order to fix it, no change is required to the
vCPU thread. However, the reader thread must delay the read after
the vCPU thread has finished the write. This can be approximated
conservatively by run_on_cpu, which waits for the end of the current
translation block.
A similar technique is used by KVM, which has to do a synchronous TLB
flush after doing a test-and-clear of the dirty-page flags.
Reported-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20190814020218.1868-6-quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
This makes easy to debug things because when you want for all threads
to arrive at that semaphore, you know which one your are waiting for.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20190814020218.1868-3-quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20190814020218.1868-2-quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Rename for better understanding of the code.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190808033155.30162-1-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Multifd sync will send MULTIFD_FLAG_SYNC flag info to destination, add
these bytes to ram_counters record.
Signed-off-by: Ivan Ren <ivanren@tencent.com>
Suggested-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <1564464816-21804-4-git-send-email-ivanren@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Limit the speed of multifd migration through common speed limitation
qemu file.
Signed-off-by: Ivan Ren <ivanren@tencent.com>
Message-Id: <1564464816-21804-3-git-send-email-ivanren@tencent.com>
Reviewed-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Use QEMU_IS_ALIGNED for the check, it would be more consistent with
other align calculations.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190806004648.8659-3-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
The purpose of the calculation is to find a HostPage which is partially
dirty.
* fixup_start_addr points to the start of the HostPage to discard
* run_start points to the next HostPage to check
While in the middle stage, there would two cases for run_start:
* aligned with HostPage means this is not partially dirty
* not aligned means this is partially dirty
When it is aligned, no work and calculation is necessary. run_start
already points to the start of next HostPage and is ready to continue.
When it is not aligned, the calculation could be simplified with:
* fixup_start_addr = QEMU_ALIGN_DOWN(run_start, host_ratio)
* run_start = QEMU_ALIGN_UP(run_start, host_ratio)
By doing so, run_start always points to the next HostPage to check.
fixup_start_addr always points to the HostPage to discard.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190806004648.8659-2-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
In postcopy-ram.c, we provide three functions to discard certain
RAMBlock range:
* postcopy_discard_send_init()
* postcopy_discard_send_range()
* postcopy_discard_send_finish()
Currently, we allocate/deallocate PostcopyDiscardState for each RAMBlock
on sending discard information to destination. This is not necessary and
the same data area could be reused for each RAMBlock.
This patch defines PostcopyDiscardState a static variable. By doing so:
1) avoid memory allocation and deallocation to the system
2) avoid potential failure of memory allocation
3) hide some details for their users
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190724010721.2146-1-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
After cleanup, it would be clear to audience there are two cases
ram_load:
* precopy
* postcopy
And it is not necessary to check postcopy_running on each iteration for
precopy.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20190725002023.2335-3-richardw.yang@linux.intel.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
It is not reasonable to continue when version_id mismatch.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190722075339.25121-2-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
RAMBlock->used_length is always passed to migration_bitmap_sync_range(),
which could be retrieved from RAMBlock.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190718012547.16373-1-richardw.yang@linux.intel.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
This means it is not necessary to spare an extra variable to hold this
condition. Use host_offset directly is fine.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190710050814.31344-3-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Use the same way for run_end to calculate run_start, which saves one
operation.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190710050814.31344-2-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Since we break the loop when there is no more page to discard, we are
sure the following process would find some page to discard.
It is not necessary to check it again.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190627020822.15485-4-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
When one is equal or bigger then end, it means there is no page to
discard. Just break the loop in this case instead of processing it.
No functional change, just refactor it a little.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190627020822.15485-3-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
If one equals end, it means we have gone through the whole bitmap.
Use a more restrict check to skip a unnecessary condition.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190627020822.15485-2-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
When migrate_cancel a multifd migration, if run sequence like this:
[source] [destination]
multifd_send_sync_main[finish]
multifd_recv_thread wait &p->sem_sync
shutdown to_dst_file
detect error from_src_file
send RAM_SAVE_FLAG_EOS[fail] [no chance to run multifd_recv_sync_main]
multifd_load_cleanup
join multifd receive thread forever
will lead destination qemu hung at following stack:
pthread_join
qemu_thread_join
multifd_load_cleanup
process_incoming_migration_co
coroutine_trampoline
Signed-off-by: Ivan Ren <ivanren@tencent.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <1561468699-9819-4-git-send-email-ivanren@tencent.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
When we 'migrate_cancel' a multifd migration, live_migration thread may
hung forever at some points, because of multifd_send_thread has already
exit for socket error:
1. multifd_send_pages may hung at qemu_sem_wait(&multifd_send_state->
channels_ready)
2. multifd_send_sync_main my hung at qemu_sem_wait(&multifd_send_state->
sem_sync)
Signed-off-by: Ivan Ren <ivanren@tencent.com>
Message-Id: <1561468699-9819-3-git-send-email-ivanren@tencent.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
---
Remove spurious not needed bits
When we 'migrate_cancel' a multifd migration, live_migration thread may
go into endless loop in multifd_send_pages functions.
Reproduce steps:
(qemu) migrate_set_capability multifd on
(qemu) migrate -d url
(qemu) [wait a while]
(qemu) migrate_cancel
Then may get live_migration 100% cpu usage in following stack:
pthread_mutex_lock
qemu_mutex_lock_impl
multifd_send_pages
multifd_queue_page
ram_save_multifd_page
ram_save_target_page
ram_save_host_page
ram_find_and_save_block
ram_find_and_save_block
ram_save_iterate
qemu_savevm_state_iterate
migration_iteration_run
migration_thread
qemu_thread_start
start_thread
clone
Signed-off-by: Ivan Ren <ivanren@tencent.com>
Message-Id: <1561468699-9819-2-git-send-email-ivanren@tencent.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reproduce the problem:
migrate
migrate_cancel
migrate
Error happen for memory migration
The reason as follows:
1. qemu start, ram_list.dirty_memory[DIRTY_MEMORY_MIGRATION] all set to
1 by a series of cpu_physical_memory_set_dirty_range
2. migration start:ram_init_bitmaps
- memory_global_dirty_log_start: begin log diry
- memory_global_dirty_log_sync: sync dirty bitmap to
ram_list.dirty_memory[DIRTY_MEMORY_MIGRATION]
- migration_bitmap_sync_range: sync ram_list.
dirty_memory[DIRTY_MEMORY_MIGRATION] to RAMBlock.bmap
and ram_list.dirty_memory[DIRTY_MEMORY_MIGRATION] is set to zero
3. migration data...
4. migrate_cancel, will stop log dirty
5. migration start:ram_init_bitmaps
- memory_global_dirty_log_start: begin log diry
- memory_global_dirty_log_sync: sync dirty bitmap to
ram_list.dirty_memory[DIRTY_MEMORY_MIGRATION]
- migration_bitmap_sync_range: sync ram_list.
dirty_memory[DIRTY_MEMORY_MIGRATION] to RAMBlock.bmap
and ram_list.dirty_memory[DIRTY_MEMORY_MIGRATION] is set to zero
Here RAMBlock.bmap only have new logged dirty pages, don't contain
the whole guest pages.
Signed-off-by: Ivan Ren <ivanren@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <1563115879-2715-1-git-send-email-ivanren@tencent.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Commit 6b6712efcc ('ram: Split dirty bitmap by RAMBlock') changes the
parameter of postcopy_send_discard_bm_ram(), while left the document
part untouched.
This patch correct the document and fix two typo by hand.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20190715020549.15018-1-richardw.yang@linux.intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
By removing the share ram check, qemu is able to migrate
to private destination ram when x-ignore-shared capability
is on. Then we can create multiple destination VMs based
on the same source VM.
This changes the x-ignore-shared migration capability to
work similar to Lai's original bypass-shared-memory
work(https://lists.gnu.org/archive/html/qemu-devel/2018-04/msg00003.html)
which enables kata containers (https://katacontainers.io)
to implement the VM templating feature.
An example usage in kata containers(https://katacontainers.io):
1. Start the source VM:
qemu-system-x86 -m 2G \
-object memory-backend-file,id=mem0,size=2G,share=on,mem-path=/tmpfs/template-memory \
-numa node,memdev=mem0
2. Stop the template VM, set migration x-ignore-shared capability,
migrate "exec:cat>/tmpfs/state", quit it
3. Start target VM:
qemu-system-x86 -m 2G \
-object memory-backend-file,id=mem0,size=2G,share=off,mem-path=/tmpfs/template-memory \
-numa node,memdev=mem0 \
-incoming defer
4. connect to target VM qmp, set migration x-ignore-shared capability,
migrate_incoming "exec:cat /tmpfs/state"
5. create more target VMs repeating 3 and 4
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Yury Kotov <yury-kotov@yandex-team.ru>
Cc: Jiangshan Lai <laijs@hyper.sh>
Cc: Xu Wang <xu@hyper.sh>
Signed-off-by: Peng Tao <tao.peng@linux.alibaba.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <1560494113-1141-1-git-send-email-tao.peng@linux.alibaba.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Currently we are doing log_clear() right after log_sync() which mostly
keeps the old behavior when log_clear() was still part of log_sync().
This patch tries to further optimize the migration log_clear() code
path to split huge log_clear()s into smaller chunks.
We do this by spliting the whole guest memory region into memory
chunks, whose size is decided by MigrationState.clear_bitmap_shift (an
example will be given below). With that, we don't do the dirty bitmap
clear operation on the remote node (e.g., KVM) when we fetch the dirty
bitmap, instead we explicitly clear the dirty bitmap for the memory
chunk for each of the first time we send a page in that chunk.
Here comes an example.
Assuming the guest has 64G memory, then before this patch the KVM
ioctl KVM_CLEAR_DIRTY_LOG will be a single one covering 64G memory.
If after the patch, let's assume when the clear bitmap shift is 18,
then the memory chunk size on x86_64 will be 1UL<<18 * 4K = 1GB. Then
instead of sending a big 64G ioctl, we'll send 64 small ioctls, each
of the ioctl will cover 1G of the guest memory. For each of the 64
small ioctls, we'll only send if any of the page in that small chunk
was going to be sent right away.
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20190603065056.25211-12-peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
cpu_physical_memory_sync_dirty_bitmap() has one RAMBlock* as
parameter, which means that it must be with RCU read lock held
already. Taking it again inside seems redundant. Removing it.
Instead comment on the functions about the RCU read lock.
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20190603065056.25211-2-peterx@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
In case we gets a queued page, the order of block is interrupted. We may
not rely on the complete_round flag to say we have already searched the
whole blocks on the list.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20190605010828.6969-1-richardw.yang@linux.intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Notification from recv thread is not ordered, which means we may be
notified by one MultiFDRecvParams but adjust packet_num for another.
Move the adjustment after we are sure each recv thread are sync-ed.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <20190604023540.26532-1-richardw.yang@linux.intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
When we are not in the last_stage, we need to update the cache if page
is not the same.
Currently this procedure is scattered in two places and mixed with
encoding status check.
This patch extract this general step out to make the code a little bit
easy to read.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20190610004159.20966-1-richardw.yang@linux.intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
On receiving RAM_SAVE_FLAG_EOS, multifd_recv_sync_main() is called to
synchronize receive threads. Current synchronization mechanism is to wait
for each channel's sem_sync semaphore. This semaphore is triggered by a
packet with MULTIFD_FLAG_SYNC flag. While in current implementation, we
don't do multifd_send_sync_main() to send such packet when
blk_mig_bulk_active() is true.
This will leads to the receive threads won't notify
multifd_recv_sync_main() by sem_sync. And multifd_recv_sync_main() will
always wait there.
[Note]: normal migration test works, while didn't test the
blk_mig_bulk_active() case. Since not sure how to produce this
situation.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20190612014337.11255-1-richardw.yang@linux.intel.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
'postocpy' should be 'postcopy'.
CC: qemu-trivial@nongnu.org
Signed-off-by: Li Qiang <liq3ea@163.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20190525062832.18009-1-liq3ea@163.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
During migration, we would sync bitmap from ram_list.dirty_memory to
RAMBlock.bmap in cpu_physical_memory_sync_dirty_bitmap().
Since we set RAMBlock.bmap and ram_list.dirty_memory both to all 1, this
means at the first round this sync is meaningless and is a duplicated
work.
Leaving RAMBlock->bmap blank on allocating would have a side effect on
migration_dirty_pages, since it is calculated from the result of
cpu_physical_memory_sync_dirty_bitmap(). To keep it right, we need to
set migration_dirty_pages to 0 in ram_state_init().
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Besides init and destroy, MultiFDSendParams.sem_sync is not really used.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190510233729.15554-1-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Since the ram bitmap and the unsent bitmap are split by RAMBlock
in commit 6b6712e, it's better to update the comments about them.
Signed-off-by: Yi Wang <wang.yi59@zte.com.cn>
Message-Id: <1555311089-18610-1-git-send-email-wang.yi59@zte.com.cn>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
We can eliminate to pass 0.
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20190430034412.12935-2-richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Coverity points out (CID 1400442) that in this code:
if (packet->pages_alloc > p->pages->allocated) {
multifd_pages_clear(p->pages);
multifd_pages_init(packet->pages_alloc);
}
we free p->pages in multifd_pages_clear() but continue to
use it in the following code. We also leak memory, because
multifd_pages_init() returns the pointer to a new MultiFDPages_t
struct but we are ignoring its return value.
Fix both of these bugs by adding the missing assignment of
the newly created struct to p->pages.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-id: 20190409151830.6024-1-peter.maydell@linaro.org
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
I found upstream codes conflict with COLO and lead to crash,
and I located to this patch:
commit 386a907b37
Author: Wei Wang <wei.w.wang@intel.com>
Date: Tue Dec 11 16:24:49 2018 +0800
migration: use bitmap_mutex in migration_bitmap_clear_dirty
My colleague Wei's patch add bitmap_mutex in migration_bitmap_clear_dirty,
but COLO didn't initialize the bitmap_mutex. So we always get an error
when COLO start up. like that:
qemu-system-x86_64: util/qemu-thread-posix.c:64: qemu_mutex_lock_impl: Assertion `mutex->initialized' failed.
This patch add the bitmap_mutex initialize and destroy in COLO
lifecycle.
Signed-off-by: Zhang Chen <chen.zhang@intel.com>
Message-Id: <20190329222951.28945-1-chen.zhang@intel.com>
Reviewed-by: Wei Wang <wei.w.wang@intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Add some padding.
MultifdInit_t is padded to 64 bytes.
MultiFDPacket_t is padded to 320bytes (64 * 5).
Signed-off-by: Juan Quintela <quintela@redhat.com>
We moved from 64KB to 512KB, as it makes less locking contention
without any downside in testing.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This way we can change the packet size in the future and everything
will work. We choose an arbitrary big number (100 times configured
size) as a limit about how big we will reallocate.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Libvirt don't want to expose (and explain it). From now on we measure
the number of packages in bytes instead of pages, so it is the same
independently of architecture. We choose the page size of x86.
Notice that in the following patch we make this variable.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
We need to send this field when we add compression support. As we are
still on x- stage, we can do this kind of changes.
Signed-off-by: Juan Quintela <quintela@redhat.com>
It really indicates what is the number of allocated pages for one
packet. Once there rename "used" to "pages_used".
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
We send packages without pages sometimes for sysnchronizanion. The
iov functions do the right thing, but we will be changing this code in
future patches.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This patch adds the free page optimization enable flag, and a function
to set this flag. When the free page optimization is enabled, not all
the pages are needed to be sent in the bulk stage.
Why using a new flag, instead of directly disabling ram_bulk_stage when
the optimization is running?
Thanks for Peter Xu's reminder that disabling ram_bulk_stage will affect
the use of compression. Please see save_page_use_compression. When
xbzrle and compression are used, if free page optimizaion causes the
ram_bulk_stage to be disabled, save_page_use_compression will return
false, which disables the use of compression. That is, if free page
optimization avoids the sending of half of the guest pages, the other
half of pages loses the benefits of compression in the meantime. Using a
new flag to let migration_bitmap_find_dirty skip the free pages in the
bulk stage will avoid the above issue.
Signed-off-by: Wei Wang <wei.w.wang@intel.com>
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
CC: Juan Quintela <quintela@redhat.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Peter Xu <peterx@redhat.com>
Message-Id: <1544516693-5395-7-git-send-email-wei.w.wang@intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
This patch adds a notifier chain for the memory precopy. This enables various
precopy optimizations to be invoked at specific places.
Signed-off-by: Wei Wang <wei.w.wang@intel.com>
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
CC: Juan Quintela <quintela@redhat.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Peter Xu <peterx@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <1544516693-5395-6-git-send-email-wei.w.wang@intel.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
This patch adds an API to clear bits corresponding to guest free pages
from the dirty bitmap. Spilt the free page block if it crosses the QEMU
RAMBlock boundary.
Signed-off-by: Wei Wang <wei.w.wang@intel.com>
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
CC: Juan Quintela <quintela@redhat.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Peter Xu <peterx@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <1544516693-5395-5-git-send-email-wei.w.wang@intel.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
The bitmap mutex is used to synchronize threads to update the dirty
bitmap and the migration_dirty_pages counter. For example, the free
page optimization clears bits of free pages from the bitmap in an
iothread context. This patch makes migration_bitmap_clear_dirty update
the bitmap and counter under the mutex.
Signed-off-by: Wei Wang <wei.w.wang@intel.com>
CC: Dr. David Alan Gilbert <dgilbert@redhat.com>
CC: Juan Quintela <quintela@redhat.com>
CC: Michael S. Tsirkin <mst@redhat.com>
CC: Peter Xu <peterx@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <1544516693-5395-4-git-send-email-wei.w.wang@intel.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
If ignore-shared capability is set then skip shared RAMBlocks during the
RAM migration.
Also, move qemu_ram_foreach_migratable_block (and rename) to the
migration code, because it requires access to the migration capabilities.
Signed-off-by: Yury Kotov <yury-kotov@yandex-team.ru>
Message-Id: <20190215174548.2630-4-yury-kotov@yandex-team.ru>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
It introduces a new statistic, pages-per-second, as bandwidth or mbps is
not enough to measure the performance of posting pages out as we have
compression, xbzrle, which can significantly reduce the amount of the
data size, instead, pages-per-second is the one we want
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Message-Id: <20190111063732.10484-2-xiaoguangrong@tencent.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
With typo's Eric spotted fixed
multifd_save_cleanup() takes an Error ** argument and returns an
error code even though it can't actually fail. Its callers
dutifully check for failure. Remove the useless argument and return
value, and simplify the callers.
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Fei Li <fli@suse.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20190113140849.38339-4-lifei1214@126.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
In our current code, when multifd is used during migration, if there
is an error before the destination receives all new channels, the
source keeps running, however the destination does not exit but keeps
waiting until the source is killed deliberately.
Fix this by dumping the specific error and let users decide whether
to quit from the destination side when failing to receive packet via
some channel. And update the comment for multifd_recv_new_channel().
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Fei Li <fli@suse.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Message-Id: <20190113140849.38339-3-lifei1214@126.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Most list head structs need not be given a name. In most cases the
name is given just in case one is going to use QTAILQ_LAST, QTAILQ_PREV
or reverse iteration, but this does not apply to lists of other kinds,
and even for QTAILQ in practice this is only rarely needed. In addition,
we will soon reimplement those macros completely so that they do not
need a name for the head struct. So clean up everything, not giving a
name except in the rare case where it is necessary.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Don't need to flush all VM's ram from cache, only
flush the dirty pages since last checkpoint
Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
Signed-off-by: Zhang Chen <chen.zhang@intel.com>
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
During the time of VM's running, PVM may dirty some pages, we will transfer
PVM's dirty pages to SVM and store them into SVM's RAM cache at next checkpoint
time. So, the content of SVM's RAM cache will always be same with PVM's memory
after checkpoint.
Instead of flushing all content of PVM's RAM cache into SVM's MEMORY,
we do this in a more efficient way:
Only flush any page that dirtied by PVM since last checkpoint.
In this way, we can ensure SVM's memory same with PVM's.
Besides, we must ensure flush RAM cache before load device state.
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
We record the address of the dirty pages that received,
it will help flushing pages that cached into SVM.
Here, it is a trick, we record dirty pages by re-using migration
dirty bitmap. In the later patch, we will start the dirty log
for SVM, just like migration, in this way, we can record both
the dirty pages caused by PVM and SVM, we only flush those dirty
pages from RAM cache while do checkpoint.
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
Signed-off-by: Zhang Chen <chen.zhang@intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
We should not load PVM's state directly into SVM, because there maybe some
errors happen when SVM is receving data, which will break SVM.
We need to ensure receving all data before load the state into SVM. We use
an extra memory to cache these data (PVM's ram). The ram cache in secondary side
is initially the same as SVM/PVM's memory. And in the process of checkpoint,
we cache the dirty pages of PVM into this ram cache firstly, so this ram cache
always the same as PVM's memory at every checkpoint, then we flush this cached ram
to SVM after we receive all PVM's state.
Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Signed-off-by: Zhang Chen <zhangckid@gmail.com>
Signed-off-by: Zhang Chen <chen.zhang@intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Taking the address of a field in a packed struct is a bad idea, because
it might not be actually aligned enough for that pointer type (and
thus cause a crash on dereference on some host architectures). Newer
versions of clang warn about this:
migration/ram.c:651:19: warning: taking address of packed member 'magic' of class or structure 'MultiFDInit_t' may result in an unaligned pointer value [-Waddress-of-packed-member]
migration/ram.c:652:19: warning: taking address of packed member 'version' of class or structure 'MultiFDInit_t' may result in an unaligned pointer value [-Waddress-of-packed-member]
migration/ram.c:737:19: warning: taking address of packed member 'magic' of class or structure 'MultiFDPacket_t' may result in an unaligned pointer value [-Waddress-of-packed-member]
migration/ram.c:745:19: warning: taking address of packed member 'version' of class or structure 'MultiFDPacket_t' may result in an unaligned pointer value [-Waddress-of-packed-member]
migration/ram.c:755:19: warning: taking address of packed member 'size' of class or structure 'MultiFDPacket_t' may result in an unaligned pointer value [-Waddress-of-packed-member]
Avoid the bug by not using the "modify in place" byteswapping
functions.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20180925161924.7832-1-peter.maydell@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Add judgement in compress_threads_save_cleanup() to check whether the
static CompressParam *comp_param has been allocated. If not, just
return; or else segmentation fault will occur when using the NULL
comp_param's parameters. One test case can reproduce this is: set
the compression on and migrate to a wrong nonexistent host IP address.
Our current code does not judge before handling comp_param[idx]'s quit
and cond that whether they have been initialized. If not initialized,
"qemu_mutex_lock_impl: Assertion `mutex->initialized' failed." will
occur. Fix this by squashing the terminate_compression_threads() into
compress_threads_save_cleanup() and employing the existing judgement
condition. One test case can reproduce this error is: set the
compression on and fail to fully setup the default eight compression
thread in compress_threads_save_setup().
Signed-off-by: Fei Li <fli@suse.com>
Message-Id: <20180925091440.18910-1-fli@suse.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
It avoids to touch compression locks if xbzrle and compression
are both enabled
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20180906070101.27280-4-xiaoguangrong@tencent.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Currently, it includes:
pages: amount of pages compressed and transferred to the target VM
busy: amount of count that no free thread to compress data
busy-rate: rate of thread busy
compressed-size: amount of bytes after compression
compression-rate: rate of compressed size
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Message-Id: <20180906070101.27280-3-xiaoguangrong@tencent.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
flush_compressed_data() needs to wait all compression threads to
finish their work, after that all threads are free until the
migration feeds new request to them, reducing its call can improve
the throughput and use CPU resource more effectively
We do not need to flush all threads at the end of iteration, the
data can be kept locally until the memory block is changed or
memory migration starts over in that case we will meet a dirtied
page which may still exists in compression threads's ring
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20180906070101.27280-2-xiaoguangrong@tencent.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
ram_find_and_save_block() can return negative if any error hanppens,
however, it is completely ignored in current code
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20180903092644.25812-5-xiaoguangrong@tencent.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
As Peter pointed out:
| - xbzrle_counters.cache_miss is done in save_xbzrle_page(), so it's
| per-guest-page granularity
|
| - RAMState.iterations is done for each ram_find_and_save_block(), so
| it's per-host-page granularity
|
| An example is that when we migrate a 2M huge page in the guest, we
| will only increase the RAMState.iterations by 1 (since
| ram_find_and_save_block() will be called once), but we might increase
| xbzrle_counters.cache_miss for 2M/4K=512 times (we'll call
| save_xbzrle_page() that many times) if all the pages got cache miss.
| Then IMHO the cache miss rate will be 512/1=51200% (while it should
| actually be just 100% cache miss).
And he also suggested as xbzrle_counters.cache_miss_rate is the only
user of rs->iterations we can adapt it to count target guest page
numbers
After that, rename 'iterations' to 'target_page_count' to better reflect
its meaning
Suggested-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Message-Id: <20180903092644.25812-3-xiaoguangrong@tencent.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
The generated qapi_event_send_FOO() take an Error ** argument. They
can't actually fail, because all they do with the argument is passing it
to functions that can't fail: the QObject output visitor, and the
@qmp_emit callback, which is either monitor_qapi_event_queue() or
event_test_emit().
Drop the argument, and pass &error_abort to the QObject output visitor
and @qmp_emit instead.
Suggested-by: Eric Blake <eblake@redhat.com>
Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20180815133747.25032-4-peterx@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Commit message rewritten, update to qapi-code-gen.txt corrected]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Try to hold src_page_req_mutex only if the queue is not
empty
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Detecting zero page is not a light work, moving it to the thread to
speed the main thread up, btw, handling ram_release_pages() for the
zero page is moved to the thread as well
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
It is not used and cleans the code up a little
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
It will be used by the compression threads
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
The compressed page is not normal page
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Instead of putting the main thread to sleep state to wait for
free compression thread, we can directly post it out as normal
page that reduces the latency and uses CPUs more efficiently
A parameter, compress-wait-thread, is introduced, it can be
enabled if the user really wants the old behavior
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This patch implements bi-directional RDMA QIOChannel. Because different
threads may access RDMAQIOChannel currently, this patch use RCU to protect it.
Signed-off-by: Lidong Chen <lidongchen@tencent.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Currently, the default maximum CPU throttle for migration is
99(CPU_THROTTLE_PCT_MAX). This is too big and can make a remarkable
performance effect for the guest. We see a lot of packets latency
exceed 500ms when the CPU_THROTTLE_PCT_MAX reached. This patch set
adds a new max-cpu-throttle parameter to limit the CPU throttle.
Signed-off-by: Li Qiang <liq3ea@gmail.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Because we need to make sure the pmem kind memory data is synced
after migration, we choose to call pmem_persist() when the migration
finish. This will make sure the data of pmem is safe and will not
lose if power is off.
Signed-off-by: Junyan He <junyan.he@intel.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The nvdimm kind memory does not support post copy now.
We disable post copy if we have nvdimm memory and print some
log hint to user.
Signed-off-by: Junyan He <junyan.he@intel.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
We shouldn't update the received bitmap if we're the source VM. This
fixes a breakage when release-ram is enabled on postcopy.
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20180723123305.24792-2-peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
I would guess it won't happen normally, but this should ease Coverity.
>>> CID 1394385: Integer handling issues (OVERFLOW_BEFORE_WIDEN)
>>> Potentially overflowing expression "pages->used * 8192U" with type "unsigned int" (32 bits, unsigned) is evaluated using 32-bit arithmetic, and then used in a context that expects an expression of type "uint64_t" (64 bits, unsigned).
854 transferred = pages->used * TARGET_PAGE_SIZE + p->packet_len;
Fixes: CID 1394385
CC: Juan Quintela <quintela@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20180720034713.11711-1-peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
