Many of these should be cleaned up with proper qdev-/QOM-ification.
Right now there are many catch-all headers in include/hw/ARCH depending
on cpu.h, and this makes it necessary to compile these files per-target.
However, fixing this does not belong in these patches.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
As one consequence, strtok() -- which modifies its argument -- is replaced
with g_strsplit().
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Anthony Liguori <aliguori@us.ibm.com>
Message-id: 1363821803-3380-6-git-send-email-lersek@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
This will remove an unneeded copy of guest memory pages.
For the page header and device state we still copy the data to the
static buffer the other option is to allocate the memory on demand
which is more expensive.
Signed-off-by: Orit Wasserman <owasserm@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
at the beginning of migration all pages are marked dirty and
in the first round a bulk migration of all pages is performed.
currently all these pages are copied to the page cache regardless
of whether they are frequently updated or not. this doesn't make sense
since most of these pages are never transferred again.
this patch changes the XBZRLE transfer to only be used after
the bulk stage has been completed. that means a page is added
to the page cache the second time it is transferred and XBZRLE
can benefit from the third time of transfer.
since the page cache is likely smaller than the number of pages
it's also likely that in the second round the page is missing in the
cache due to collisions in the bulk phase.
on the other hand a lot of unnecessary mallocs, memdups and frees
are saved.
the following results have been taken earlier while executing
the test program from docs/xbzrle.txt. (+) with the patch and (-)
without. (thanks to Eric Blake for reformatting and comments)
+ total time: 22185 milliseconds
- total time: 22410 milliseconds
Shaved 0.3 seconds, better than 1%!
+ downtime: 29 milliseconds
- downtime: 21 milliseconds
Not sure why downtime seemed worse, but probably not the end of the world.
+ transferred ram: 706034 kbytes
- transferred ram: 721318 kbytes
Fewer bytes sent - good.
+ remaining ram: 0 kbytes
- remaining ram: 0 kbytes
+ total ram: 1057216 kbytes
- total ram: 1057216 kbytes
+ duplicate: 108556 pages
- duplicate: 105553 pages
+ normal: 175146 pages
- normal: 179589 pages
+ normal bytes: 700584 kbytes
- normal bytes: 718356 kbytes
Fewer normal bytes...
+ cache size: 67108864 bytes
- cache size: 67108864 bytes
+ xbzrle transferred: 3127 kbytes
- xbzrle transferred: 630 kbytes
...and more compressed pages sent - good.
+ xbzrle pages: 117811 pages
- xbzrle pages: 21527 pages
+ xbzrle cache miss: 18750
- xbzrle cache miss: 179589
And very good improvement on the cache miss rate.
+ xbzrle overflow : 0
- xbzrle overflow : 0
Signed-off-by: Peter Lieven <pl@kamp.de>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
avoid searching for dirty pages just increment the
page offset. all pages are dirty anyway.
Signed-off-by: Peter Lieven <pl@kamp.de>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@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 first round of ram transfer is special since all pages
are dirty and thus all memory pages are transferred to
the target. this patch adds a boolean variable to track
this stage.
Signed-off-by: Peter Lieven <pl@kamp.de>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
virtually all dup pages are zero pages. remove
the special is_dup_page() function and use the
optimized buffer_find_nonzero_offset() function
instead.
here buffer_find_nonzero_offset() is used directly
to avoid the unnecssary additional checks in
buffer_is_zero().
raw performace gain checking 1 GByte zeroed memory
over is_dup_page() is approx. 10-12% with SSE2
and 8-10% with unsigned long arithmedtic.
Signed-off-by: Peter Lieven <pl@kamp.de>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
vector optimizations will now be used at various places
not just in is_dup_page() in arch_init.c
Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Juan Quintela <quintela@redhat.com>
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>
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>
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>
Before this fix we couldn't load a guest from
XBZRLE compressed file.
For example:
The user activated the XBZRLE capability
The user run migrate -d "exec:gzip -c > vm.gz"
The user won't be able to load vm.gz and get an error.
Signed-off-by: Orit Wasserman <owasserm@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Juan Quintela <quintela@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>
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>
It could only return 0 if we only found dirty xbzrle pages that hadn't
changed (i.e. they were written with the same content). We don't care
about that case, it is the same than nothing dirty.
So now the return of the function is how much have it written, nothing
else. Adjust callers.
And we also made ram_save_iterate() return the number of transferred
bytes, not the number of transferred pages.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Instead of testing each page individually, we search what is the next
dirty page with a bitmap operation. We have to reorganize the code to
move from a "for" loop, to a while(dirty) loop.
Signed-off-by: Juan Quintela <quintela@redhat.com>
This avoids having to do two walks over the dirty bitmap, once reading
the dirty bits, and anthoer cleaning them.
Signed-off-by: Juan Quintela <quintela@redhat.com>
This is the last block from where we have sent data.
Signed-off-by: Orit Wasserman <owasserm@redhat.com>
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>
Add the new mutex that protects shared state between ram_save_live
and the iothread. If the iothread mutex has to be taken together
with the ramlist mutex, the iothread shall always be _outside_.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Umesh Deshpande <udeshpan@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
This will be used to detect if last_block might have become invalid
across different calls to ram_save_live.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Umesh Deshpande <udeshpan@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
Most of the time, only 2 items will be active (from/to for a string operation,
or code/data). But TCG guests likely won't have gigabytes of memory, so
this actually goes down to 1 item.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Various header files rely on qemu-char.h including qemu-config.h or
main-loop.h, but they really do not need qemu-char.h at all (particularly
interesting is the case of the block layer!). Clean this up, and also
add missing inclusions of qemu-char.h itself.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
madvise(DONTNEED) will throw away the contents of the whole page at the
given address, even if the given length is less than the page size. One
can argue about whether that's the correct behaviour, but that's what it's
done for a long time in Linux at least.
That means that the madvise() in ram_load(), on a setup where
TARGET_PAGE_SIZE is smaller than the host page size, can throw away data
in guest pages adjacent to the one it's actually processing right now,
leading to guest memory corruption on an incoming migration.
This patch therefore, disables the madvise() if the host page size is
larger than TARGET_PAGE_SIZE. This means we don't get the benefits of that
madvise() in this case, but a more complete fix is more difficult to
accomplish. This at least fixes the guest memory corruption.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reported-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
The code for migrating (or savevm-ing) memory pages starts off by creating
a dirty bitmap and filling it with 1s. Except, actually, because bit
addresses are 0-based it fills every bit except bit 0 with 1s and puts an
extra 1 beyond the end of the bitmap, potentially corrupting unrelated
memory. Oops. This patch fixes it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
This patch creates a migration bitmap, which is periodically kept in
sync with the qemu bitmap. A separate copy of the dirty bitmap for the
migration limits the amount of concurrent access to the qemu bitmap
from iothread and migration thread (which requires taking the big
lock).
We use the qemu bitmap type. We have to "undo" the dirty_pages
counting optimization on the general dirty bitmap and do the counting
optimization with the migration local bitmap.
Signed-off-by: Umesh Deshpande <udeshpan@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
Helper that we use each time that we need to syncronize the migration
bitmap with the other dirty bitmaps.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
It just test if the dirty bit is set, and clears it.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
It just marks a region of memory as dirty.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
