This capability allows you to disable dynamic chunk registration
for better throughput on high-performance links.
For example, using an 8GB RAM virtual machine with all 8GB of memory in
active use and the VM itself is completely idle using a 40 gbps infiniband link:
1. x-rdma-pin-all disabled total time: approximately 7.5 seconds @ 9.5 Gbps
2. x-rdma-pin-all enabled total time: approximately 4 seconds @ 26 Gbps
These numbers would of course scale up to whatever size virtual machine
you have to migrate using RDMA.
Enabling this feature does *not* have any measurable affect on
migration *downtime*. This is because, without this feature, all of the
memory will have already been registered already in advance during
the bulk round and does not need to be re-registered during the successive
iteration rounds.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Tested-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
These are the prototypes and implementation of new hooks that
RDMA takes advantage of to perform dynamic page registration.
An optional hook is also introduced for a custom function
to be able to override the default save_page function.
Also included are the prototypes and accessor methods used by
arch_init.c which invoke funtions inside savevm.c to call out
to the hooks that may or may not have been overridden
inside of QEMUFileOps.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
RDMA uses this to flush the control channel before sending its
own message to handle page registrations.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
QEMUFileRDMA also has read and write modes. This function is now
shared to reduce code duplication.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This exposes throughput (in megabits/sec) through QMP.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
RDMA writes happen asynchronously, and thus the performance accounting
also needs to be able to occur asynchronously. This allows anybody
to call into savevm.c to update both f->pos as well as into arch_init.c
to update the acct_info structure with up-to-date values when
the RDMA transfer actually completes.
Reviewed-by: Juan Quintela <quintela@redhat.com>
Tested-by: Chegu Vinod <chegu_vinod@hp.com>
Tested-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Michael R. Hines <mrhines@us.ibm.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Instead of breaking up RAM state into many small chunks, pass the iovec
to the block layer for better performance.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Add support for migrating two dimensional arrays, by defining
a set of new macros VMSTATE_*_2DARRAY paralleling the existing
VMSTATE_*_ARRAY macros. 2D arrays are handled the same for actual
state serialization; the only difference is that the type check
has to change for a 2D array.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Message-id: 1363975375-3166-2-git-send-email-peter.maydell@linaro.org
Macro could be used to migrate a dynamically allocated buffer of known size.
Signed-off-by: Igor Mitsyanko <i.mitsyanko@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 1362923278-4080-2-git-send-email-i.mitsyanko@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This allows us to add a buffer to the iovec to send without copying it
into the static buffer, the buffer will be sent later when qemu_fflush is called.
Signed-off-by: Orit Wasserman <owasserm@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This will allow us to write an iovec
Signed-off-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
during bulk stage of ram migration if a page is a
zero page do not send it at all.
the memory at the destination reads as zero anyway.
even if there is an madvise with QEMU_MADV_DONTNEED
at the target upon receipt of a zero page I have observed
that the target starts swapping if the memory is overcommitted.
it seems that the pages are dropped asynchronously.
this patch also updates QMP to return the number of
skipped pages in MigrationStats.
Signed-off-by: Peter Lieven <pl@kamp.de>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
The VMSTATE_BUFFER_MULTIPLY macro is misnamed - it actually specifies
a variably sized buffer with VMS_VBUFFER, so should be named
VMSTATE_VBUFFER_MULTIPLY. This patch fixes this (the macro had no current
users under either name).
In addition, unlike the other VMSTATE_VBUFFER variants, this macro did not
specify VMS_POINTER. This patch fixes this bug as well.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Currently the savevm code contains a VMSTATE_STRUCT_VARRAY_POINTER_INT32
helper (a variably sized array with the number of elements in an int32_t),
but not VMSTATE_STRUCT_VARRAY_POINTER_UINT32 (... with the number of
elements in a uint32_t). This patch (trivially) fixes the deficiency.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Juan Quintela <quintela@redhat.com>
The current savevm code includes VMSTATE helpers for a number of commonly
used data types, but not for the float64 type used by the internal floating
point emulation code. This patch fixes the deficiency.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This adds an _EQUAL VMSTATE helper for target_ulongs, defined in terms of
VMSTATE_UINT32_EQUAL or VMSTATE_UINT64_EQUAL as appropriate.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Juan Quintela <quintela@redhat.com>
The savevm code already includes a number of *_EQUAL helpers which act as
sanity checks verifying that the configuration of the saved state matches
that of the machine we're loading into to work. Variants already exist
for 8 bit 16 bit and 32 bit integers, but not 64 bit integers. This patch
fills that hole, adding a UINT64 version.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This avoids adding a duplicate stub for CONFIG_USER_ONLY.
Suggested-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
The page cache frees all data on finish, on resize and
if there is collision on insert. So it should be the caches
responsibility to dup the data that is stored in the cache.
Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Juan Quintela <quintela@redhat.com>
The indirection is useless now. Backends can open s->file directly.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Rate limiting is now simply a byte counter; client call
qemu_file_rate_limit() manually to determine if they have to exit.
So it is possible and simple to move the functionality to QEMUFile.
This makes the remaining functionality of s->file redundant;
in the next patch we can remove it and write directly to s->migration_file.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Second, drop the file descriptor indirection, and write directly to the
QEMUFile.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
As a start, use QEMUFile to store the destination and close it.
qemu_get_fd gets a file descriptor that will be used by the write
callbacks.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
There is no reason for outgoing exec migration to do popen manually
anymore (the reason used to be that we needed the FILE* to make it
non-blocking). Use qemu_popen_cmd.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Buffering was needed because blocking writes could take a long time
and starve other threads seeking to grab the big QEMU mutex.
Now that all writes (except within _complete callbacks) are done
outside the big QEMU mutex, we do not need buffering at all.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Only the migration_bitmap_sync() call needs the iothread lock.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This makes it possible to do blocking writes directly to the socket,
with no buffer in the middle. For RAM, only the migration_bitmap_sync()
call needs the iothread lock. For block migration, it is needed by
the block layer (including bdrv_drain_all and dirty bitmap access),
but because some code is shared between iterate and complete, all of
mig_save_device_dirty is run with the lock taken.
In the savevm case, the iterate callback runs within the big lock.
This is annoying because it complicates the rules. Luckily we do not
need to do anything about it: the RAM iterate callback does not need
the iothread lock, and block migration never runs during savevm.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This groups together the callbacks that later will have similar
locking rules.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Perform final cleanup in a bottom half, and add joining the thread to
the series of cleanup actions.
migrate_fd_error remains for connection error, but it doesn't need
to cleanup anything anymore.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Always use qemu_file_get_error to detect errors, since that is how
QEMUFile itself drops I/O after an error occurs. There is no need
to propagate and check return values all the time.
Also remove the "complete" member, since we know that it is set (via
migrate_fd_cleanup) only when the state changes.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Right now, migration cannot entirely rely on QEMUFile's automatic
drop of I/O after an error, because it does its "real" I/O outside
the put_buffer callback. To fix this until buffering is gone, expose
qemu_file_set_error which we will use in buffered_flush.
Similarly, buffered_flush is not a complete flush because some data may
still reside in the QEMUFile's own buffer. This somewhat complicates the
process of closing the migration thread. Again, when buffering is gone
buffered_flush will disappear and calling qemu_fflush will not be needed;
in the meanwhile, we expose the function for use in migration.c.
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
We removed the calculation in commit e4ed1541ac
Now we add it back. We need to create dirty_bytes_rate because we
can't include cpu-all.h from migration.c, and there is no other way to
include TARGET_PAGE_SIZE.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Migration .save_live_iterate() functions return the number of bytes
transferred. The easiest way of doing this is by calling qemu_ftell(f)
at the beginning and end of the function to calculate the difference.
Make qemu_ftell() public so that block-migration will be able to use it.
Also adjust the ftell calculation for writable files where buf_offset
does not include buf_size.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-id: 1360661835-28663-2-git-send-email-stefanha@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Reviewed-by: Eric Blake <eblake@redhat.com>
Avoid splitting the state of outgoing migration, more or less arbitrarily,
between two data structures. QEMUFileBuffered anyway is used only during
migration.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
This only moves the code (also from buffered_file.h to migration.h).
Fix whitespace until checkpatch is happy.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Code just now does (simplified for clarity)
if (qemu_savevm_state_iterate(s->file) == 1) {
vm_stop_force_state(RUN_STATE_FINISH_MIGRATE);
qemu_savevm_state_complete(s->file);
}
Problem here is that qemu_savevm_state_iterate() returns 1 when it
knows that remaining memory to sent takes less than max downtime.
But this means that we could end spending 2x max_downtime, one
downtime in qemu_savevm_iterate, and the other in
qemu_savevm_state_complete.
Changed code to:
pending_size = qemu_savevm_state_pending(s->file, max_size);
DPRINTF("pending size %lu max %lu\n", pending_size, max_size);
if (pending_size >= max_size) {
ret = qemu_savevm_state_iterate(s->file);
} else {
vm_stop_force_state(RUN_STATE_FINISH_MIGRATE);
qemu_savevm_state_complete(s->file);
}
So what we do is: at current network speed, we calculate the maximum
number of bytes we can sent: max_size.
Then we ask every save_live section how much they have pending. If
they are less than max_size, we move to complete phase, otherwise we
do an iterate one.
This makes things much simpler, because now individual sections don't
have to caluclate the bandwidth (it was implossible to do right from
there).
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Now that we have a thread, and blocking writes, we don't need it.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Move all the writes to the migration_thread, and make writings
blocking. Notice that are still using the iothread for everything
that we do.
Signed-off-by: Juan Quintela <quintela@redhat.com>
We want the file assignment to happen before the thread is created to
avoid locking, so we just do it before creating the thread.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Peter Maydell