qemu-kvm/kvm-multifd-Create-property-multifd-flush-after-each-sec.patch
Miroslav Rezanina 5e258fbe95 * Mon Jul 24 2023 Miroslav Rezanina <mrezanin@redhat.com> - 8.0.0-9
- kvm-scsi-fetch-unit-attention-when-creating-the-request.patch [bz#2176702]
- kvm-scsi-cleanup-scsi_clear_unit_attention.patch [bz#2176702]
- kvm-scsi-clear-unit-attention-only-for-REPORT-LUNS-comma.patch [bz#2176702]
- kvm-s390x-ap-Wire-up-the-device-request-notifier-interfa.patch [RHEL-794]
- kvm-multifd-Create-property-multifd-flush-after-each-sec.patch [bz#2196295]
- kvm-multifd-Protect-multifd_send_sync_main-calls.patch [bz#2196295]
- kvm-multifd-Only-flush-once-each-full-round-of-memory.patch [bz#2196295]
- kvm-net-socket-prepare-to-cleanup-net_init_socket.patch [RHEL-582]
- kvm-net-socket-move-fd-type-checking-to-its-own-function.patch [RHEL-582]
- kvm-net-socket-remove-net_init_socket.patch [RHEL-582]
- kvm-pcie-Add-hotplug-detect-state-register-to-cmask.patch [bz#2215819]
- kvm-spec-Build-DBUS-display.patch [bz#2207940]
- Resolves: bz#2176702
  ([RHEL9][virtio-scsi] scsi-hd cannot hot-plug successfully after hot-plug it repeatly)
- Resolves: RHEL-794
  (Backport s390x fixes from QEMU 8.1)
- Resolves: bz#2196295
  (Multifd flushes its channels 10 times per second)
- Resolves: RHEL-582
  ([passt][rhel 9.3] qemu core dump occurs when guest is shutdown after hotunplug/hotplug a passt interface)
- Resolves: bz#2215819
  (Migration test failed while guest with PCIe devices)
- Resolves: bz#2207940
  ([RFE] Enable qemu-ui-dbus subpackage)
2023-07-24 02:36:25 -04:00

128 lines
4.5 KiB
Diff

From 2a5ea92ca0a5dffad54e4d06a683f683996cea9a Mon Sep 17 00:00:00 2001
From: Juan Quintela <quintela@redhat.com>
Date: Tue, 21 Jun 2022 12:13:14 +0200
Subject: [PATCH 05/12] multifd: Create property
multifd-flush-after-each-section
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: quintela1 <quintela@redhat.com>
RH-MergeRequest: 186: Multifd flushes its channels 10 times per second
RH-Bugzilla: 2196295
RH-Acked-by: Peter Xu <peterx@redhat.com>
RH-Acked-by: Leonardo Brás <leobras@redhat.com>
RH-Commit: [1/3] 5bf5348e8be5b1d1629b859ce1ddb7aa0d72c0d6 (juan.quintela/c9s-qemu-kvm)
We used to flush all channels at the end of each RAM section
sent. That is not needed, so preparing to only flush after a full
iteration through all the RAM.
Default value of the property is false. But we return "true" in
migrate_multifd_flush_after_each_section() until we implement the code
in following patches.
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Acked-by: Peter Xu <peterx@redhat.com>
---
Rename each-iteration to after-each-section
Rename multifd-sync-after-each-section to
multifd-flush-after-each-section
Move to machine-8.0 (peter)
conflit hw_compat_8_0 and hw_compat_rhel_9_2
(cherry picked from commit 77c259a4cb1c9799754b48f570301ebf1de5ded8)
---
hw/core/machine.c | 2 ++
migration/migration.h | 12 ++++++++++++
migration/options.c | 13 +++++++++++++
migration/options.h | 1 +
4 files changed, 28 insertions(+)
diff --git a/hw/core/machine.c b/hw/core/machine.c
index 5abdc8c39b..5ea52317b9 100644
--- a/hw/core/machine.c
+++ b/hw/core/machine.c
@@ -54,6 +54,8 @@ const char *rhel_old_machine_deprecation =
"machine types for previous major releases are deprecated";
GlobalProperty hw_compat_rhel_9_2[] = {
+ /* hw_compat_rhel_9_2 from hw_compat_8_0 */
+ { "migration", "multifd-flush-after-each-section", "on"},
/* hw_compat_rhel_9_2 from hw_compat_7_2 */
{ "e1000e", "migrate-timadj", "off" },
/* hw_compat_rhel_9_2 from hw_compat_7_2 */
diff --git a/migration/migration.h b/migration/migration.h
index 7ccf460aa2..04c78c1fd6 100644
--- a/migration/migration.h
+++ b/migration/migration.h
@@ -411,6 +411,18 @@ struct MigrationState {
*/
bool preempt_pre_7_2;
+ /*
+ * flush every channel after each section sent.
+ *
+ * This assures that we can't mix pages from one iteration through
+ * ram pages with pages for the following iteration. We really
+ * only need to do this flush after we have go through all the
+ * dirty pages. For historical reasons, we do that after each
+ * section. This is suboptimal (we flush too many times).
+ * Default value is false. Setting this property has no effect
+ * until the patch that removes this comment. (since 8.1)
+ */
+ bool multifd_flush_after_each_section;
/*
* This decides the size of guest memory chunk that will be used
* to track dirty bitmap clearing. The size of memory chunk will
diff --git a/migration/options.c b/migration/options.c
index ccd7ef3907..5b0d080ecb 100644
--- a/migration/options.c
+++ b/migration/options.c
@@ -88,6 +88,8 @@ Property migration_properties[] = {
send_section_footer, true),
DEFINE_PROP_BOOL("decompress-error-check", MigrationState,
decompress_error_check, true),
+ DEFINE_PROP_BOOL("multifd-flush-after-each-section", MigrationState,
+ multifd_flush_after_each_section, true),
DEFINE_PROP_UINT8("x-clear-bitmap-shift", MigrationState,
clear_bitmap_shift, CLEAR_BITMAP_SHIFT_DEFAULT),
DEFINE_PROP_BOOL("x-preempt-pre-7-2", MigrationState,
@@ -344,6 +346,17 @@ bool migrate_zero_copy_send(void)
/* pseudo capabilities */
+bool migrate_multifd_flush_after_each_section(void)
+{
+ MigrationState *s = migrate_get_current();
+
+ /*
+ * Until the patch that remove this comment, we always return that
+ * the property is enabled.
+ */
+ return true || s->multifd_flush_after_each_section;
+}
+
bool migrate_postcopy(void)
{
return migrate_postcopy_ram() || migrate_dirty_bitmaps();
diff --git a/migration/options.h b/migration/options.h
index 0fc7be6869..271f49ae5f 100644
--- a/migration/options.h
+++ b/migration/options.h
@@ -60,6 +60,7 @@ bool migrate_zero_copy_send(void);
* check, but they are not a capability.
*/
+bool migrate_multifd_flush_after_each_section(void);
bool migrate_postcopy(void);
bool migrate_tls(void);
--
2.39.3