qemu-kvm/kvm-migration-ram-Implement-save_postcopy_prepare.patch
Miroslav Rezanina ebd0a23ab5 * Mon Jul 28 2025 Miroslav Rezanina <mrezanin@redhat.com> - 10.0.0-8
- kvm-migration-multifd-move-macros-to-multifd-header.patch [RHEL-59697]
- kvm-migration-refactor-channel-discovery-mechanism.patch [RHEL-59697]
- kvm-migration-Add-save_postcopy_prepare-savevm-handler.patch [RHEL-59697]
- kvm-migration-ram-Implement-save_postcopy_prepare.patch [RHEL-59697]
- kvm-tests-qtest-migration-consolidate-set-capabilities.patch [RHEL-59697]
- kvm-migration-write-zero-pages-when-postcopy-enabled.patch [RHEL-59697]
- kvm-migration-enable-multifd-and-postcopy-together.patch [RHEL-59697]
- kvm-migration-Add-qtest-for-migration-over-RDMA.patch [RHEL-59697]
- kvm-qtest-migration-rdma-Enforce-RLIMIT_MEMLOCK-128MB-re.patch [RHEL-59697]
- kvm-qtest-migration-rdma-Add-test-for-rdma-migration-wit.patch [RHEL-59697]
- kvm-tests-qtest-migration-add-postcopy-tests-with-multif.patch [RHEL-59697]
- kvm-file-posix-Fix-aio-threads-performance-regression-af.patch [RHEL-96854]
- kvm-block-remove-outdated-comments-about-AioContext-lock.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-read-locked-bdrv_reopen_.patch [RHEL-88561]
- kvm-block-snapshot-move-drain-outside-of-read-locked-bdr.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-read-locked-bdrv_inactiv.patch [RHEL-88561]
- kvm-block-mark-bdrv_parent_change_aio_context-GRAPH_RDLO.patch [RHEL-88561]
- kvm-block-mark-change_aio_ctx-callback-and-instances-as-.patch [RHEL-88561]
- kvm-block-mark-bdrv_child_change_aio_context-GRAPH_RDLOC.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-bdrv_change_aio_context-.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-bdrv_try_change_aio_cont.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-bdrv_attach_child_common.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-bdrv_set_backing_hd_drai.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-bdrv_root_attach_child.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-bdrv_attach_child.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-quorum_add_child.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-bdrv_root_unref_child.patch [RHEL-88561]
- kvm-block-move-drain-outside-of-quorum_del_child.patch [RHEL-88561]
- kvm-blockdev-drain-while-unlocked-in-internal_snapshot_a.patch [RHEL-88561]
- kvm-blockdev-drain-while-unlocked-in-external_snapshot_a.patch [RHEL-88561]
- kvm-block-mark-bdrv_drained_begin-and-friends-as-GRAPH_U.patch [RHEL-88561]
- kvm-iotests-graph-changes-while-io-remove-image-file-aft.patch [RHEL-88561]
- kvm-iotests-graph-changes-while-io-add-test-case-with-re.patch [RHEL-88561]
- Resolves: RHEL-59697
  (Allow multifd+postcopy features being enabled together, but only use multifd during precopy )
- Resolves: RHEL-96854
  (Performance Degradation(aio=threads) between Upstream Commit b75c5f9 and 984a32f)
- Resolves: RHEL-88561
  (qemu graph deadlock during job-dismiss)
2025-07-28 02:29:30 -04:00

86 lines
3.3 KiB
Diff

From d5b76b77dc891f0bec211a6d00b099a2979223ee Mon Sep 17 00:00:00 2001
From: Peter Xu <peterx@redhat.com>
Date: Fri, 11 Apr 2025 17:15:31 +0530
Subject: [PATCH 04/33] migration/ram: Implement save_postcopy_prepare()
RH-Author: Prasad Pandit <None>
RH-MergeRequest: 390: migration: allow to enable multifd+postcopy features together, but use multifd during precopy only
RH-Jira: RHEL-59697
RH-Acked-by: Juraj Marcin <None>
RH-Acked-by: Miroslav Rezanina <mrezanin@redhat.com>
RH-Commit: [4/11] 4df55b4e65458ffccf48df5de3afad0b38cded51 (pjp/cs-qemu-kvm)
Implement save_postcopy_prepare(), preparing for the enablement
of both multifd and postcopy.
Jira: https://issues.redhat.com/browse/RHEL-59697
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Prasad Pandit <pjp@fedoraproject.org>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Message-ID: <20250411114534.3370816-5-ppandit@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
(cherry picked from commit ad8d82ffbb8b8034f58a570911e6e9c6328c9384)
Signed-off-by: Prasad Pandit <ppandit@redhat.com>
---
migration/ram.c | 37 +++++++++++++++++++++++++++++++++++++
1 file changed, 37 insertions(+)
diff --git a/migration/ram.c b/migration/ram.c
index 21d2f87ff1..856769a77c 100644
--- a/migration/ram.c
+++ b/migration/ram.c
@@ -4515,6 +4515,42 @@ static int ram_resume_prepare(MigrationState *s, void *opaque)
return 0;
}
+static bool ram_save_postcopy_prepare(QEMUFile *f, void *opaque, Error **errp)
+{
+ int ret;
+
+ if (migrate_multifd()) {
+ /*
+ * When multifd is enabled, source QEMU needs to make sure all the
+ * pages queued before postcopy starts have been flushed.
+ *
+ * The load of these pages must happen before switching to postcopy.
+ * It's because loading of guest pages (so far) in multifd recv
+ * threads is still non-atomic, so the load cannot happen with vCPUs
+ * running on the destination side.
+ *
+ * This flush and sync will guarantee that those pages are loaded
+ * _before_ postcopy starts on the destination. The rationale is,
+ * this happens before VM stops (and before source QEMU sends all
+ * the rest of the postcopy messages). So when the destination QEMU
+ * receives the postcopy messages, it must have received the sync
+ * message on the main channel (either RAM_SAVE_FLAG_MULTIFD_FLUSH,
+ * or RAM_SAVE_FLAG_EOS), and such message would guarantee that
+ * all previous guest pages queued in the multifd channels are
+ * completely loaded.
+ */
+ ret = multifd_ram_flush_and_sync(f);
+ if (ret < 0) {
+ error_setg(errp, "%s: multifd flush and sync failed", __func__);
+ return false;
+ }
+ }
+
+ qemu_put_be64(f, RAM_SAVE_FLAG_EOS);
+
+ return true;
+}
+
void postcopy_preempt_shutdown_file(MigrationState *s)
{
qemu_put_be64(s->postcopy_qemufile_src, RAM_SAVE_FLAG_EOS);
@@ -4534,6 +4570,7 @@ static SaveVMHandlers savevm_ram_handlers = {
.load_setup = ram_load_setup,
.load_cleanup = ram_load_cleanup,
.resume_prepare = ram_resume_prepare,
+ .save_postcopy_prepare = ram_save_postcopy_prepare,
};
static void ram_mig_ram_block_resized(RAMBlockNotifier *n, void *host,
--
2.39.3