[v2,3/3] multifd: Send using asynchronous write on nocomp to send RAM pages.

Commit Message

Leonardo Bras Sept. 22, 2021, 5:03 a.m. UTC

Change multifd nocomp version to use asynchronous write for RAM pages, and
benefit of MSG_ZEROCOPY when it's available.

The asynchronous flush happens on cleanup only, before destroying the QIOChannel.

This will work fine on RAM migration because the RAM pages are not usually freed,
and there is no problem on changing the pages content between async_send() and
the actual sending of the buffer, because this change will dirty the page and
cause it to be re-sent on a next iteration anyway.

Signed-off-by: Leonardo Bras <[e-mail]>
---
 migration/multifd.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Patch

diff --git a/migration/multifd.c b/migration/multifd.c
index 377da78f5b..d247207a0a 100644
--- a/migration/multifd.c
+++ b/migration/multifd.c
@@ -105,7 +105,7 @@  static int nocomp_send_prepare(MultiFDSendParams *p, uint32_t used,
  */
 static int nocomp_send_write(MultiFDSendParams *p, uint32_t used, Error **errp)
 {
-    return qio_channel_writev_all(p->c, p->pages->iov, used, errp);
+    return qio_channel_async_writev_all(p->c, p->pages->iov, used, errp);
 }
 
 /**
@@ -546,6 +546,7 @@  void multifd_save_cleanup(void)
         MultiFDSendParams *p = &multifd_send_state->params[i];
         Error *local_err = NULL;
 
+        qio_channel_async_flush(p->c, NULL);
         socket_send_channel_destroy(p->c);
         p->c = NULL;
         qemu_mutex_destroy(&p->mutex);