On Fri, Mar 19, 2021 at 05:52:47PM +0300, Andrey Gruzdev wrote:

Added missing qemu_fflush() on buffer file holding precopy device state.
Increased initial QIOChannelBuffer allocation to 512KB to avoid reallocs.
Typical configurations often require >200KB for device state and VMDESC.

Signed-off-by: Andrey Gruzdev <andrey.gruzdev@virtuozzo.com>
---
 migration/migration.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/migration/migration.c b/migration/migration.c
index ca8b97baa5..32b48fe9f5 100644
--- a/migration/migration.c
+++ b/migration/migration.c
@@ -3812,7 +3812,7 @@ static void *bg_migration_thread(void *opaque)
      * with vCPUs running and, finally, write stashed non-RAM part of
      * the vmstate from the buffer to the migration stream.
      */
-    s->bioc = qio_channel_buffer_new(128 * 1024);
+    s->bioc = qio_channel_buffer_new(512 * 1024);
     qio_channel_set_name(QIO_CHANNEL(s->bioc), "vmstate-buffer");
     fb = qemu_fopen_channel_output(QIO_CHANNEL(s->bioc));
     object_unref(OBJECT(s->bioc));
@@ -3866,6 +3866,8 @@ static void *bg_migration_thread(void *opaque)
     if (qemu_savevm_state_complete_precopy_non_iterable(fb, false, false)) {
         goto fail;
     }
+    qemu_fflush(fb);

What will happen if the vmstates are bigger than 512KB?  Would the extra data
be dropped?

No, the buffer shall be reallocated and the original content shall be kept.

In that case, I'm wondering whether we'll need a qemu_file_get_error() after
the flush to detect it, and whether we need to retry with a bigger buffer size?

Thanks,

I think for QIOBufferChannel we don't need that, as I can see from the buffer channel
implementation.