[libvirt PATCH] NEWS: mention support for vhost-user-blk
by Pavel Hrdina
Signed-off-by: Pavel Hrdina <phrdina(a)redhat.com>
---
NEWS.rst | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/NEWS.rst b/NEWS.rst
index 440dbf2601..56a9afe749 100644
--- a/NEWS.rst
+++ b/NEWS.rst
@@ -47,6 +47,11 @@ v7.1.0 (unreleased)
to set the hostdev device's MAC address (which is a necessary
part of the alternate <interface type='hostdev'>).
+ * Introduce ``<disk type='vhostuser'>`` support
+
+ Introduces a support for QEMU vhost-user-blk device that can be used
+ together with qemu-storage-daemon to provide disks for VMs.
+
* **Improvements**
* **Bug fixes**
--
2.29.2
3 years, 6 months
[PATCH] NEWS: Mention snapshot quiesce rollback bugfix
by Peter Krempa
Signed-off-by: Peter Krempa <pkrempa(a)redhat.com>
---
NEWS.rst | 14 ++++++++++++++
1 file changed, 14 insertions(+)
diff --git a/NEWS.rst b/NEWS.rst
index 440dbf2601..8d639253ff 100644
--- a/NEWS.rst
+++ b/NEWS.rst
@@ -51,6 +51,20 @@ v7.1.0 (unreleased)
* **Bug fixes**
+ * qemu: Fix disk quiescing rollback when creating external snapshots
+
+ If the qemu guest agent call to freeze filesystems failed when creating
+ an external snapshot with ``VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE`` flag the
+ filesystems would be unconditionally thawed. This could cause problems when
+ the filesystems were frozen by an explicit call to ``virDomainFSFreeze``
+ since the guest agent then rejects any further freeze attempts once are
+ filesystems frozen, an explicit freeze followed by a quiesced snapshot
+ would fail and thaw filesystems.
+
+ Users are also encouraged to use ``virDomainFSFreeze/Thaw`` manually instead
+ of relying on ``VIR_DOMAIN_SNAPSHOT_CREATE_QUIESCE`` if they need finer
+ grained control.
+
v7.0.0 (2021-01-15)
===================
--
2.29.2
3 years, 6 months
[libvirt PATCH 0/2] followup fix for cpu quota limits
by Pavel Hrdina
Pavel Hrdina (2):
domain_validate: use defines for cpu period and quota limits
docs: use proper cpu quota value in our documentation
docs/formatdomain.rst | 8 ++++----
docs/manpages/virsh.rst | 2 +-
docs/schemas/domaincommon.rng | 2 +-
src/conf/domain_validate.c | 20 +++++++++++++-------
4 files changed, 19 insertions(+), 13 deletions(-)
--
2.29.2
3 years, 6 months
[PATCH 00/25] Clear pointers in virJSONValue(Object|Array)Append and other cleanups
by Peter Krempa
Peter Krempa (25):
virLockDaemonPreExecRestart: Refactor memory cleanup
virLogDaemonPreExecRestart: Refactor memory cleanup
virLogHandlerPreExecRestart: Refactor memory cleanup
virNetDaemonPreExecRestart: Refactor memory cleanup
virNetServerServicePreExecRestart: Refactor memory cleanup
virNetServerClientPreExecRestart: Refactor memory cleanup
virNetServerPreExecRestart: Drop error reporting from
virJSONValueObjectAppend* calls
virNetServerPreExecRestart: Refactor memory cleanup
virLockSpacePreExecRestart: Refactor memory cleanup
qemuAgentMakeCommand: Refactor memory cleanup
virJSONValueObjectInsert: Clear @value on successful insertion
virJSONValueCopy: Don't use virJSONValue(Object|Array)Append
virJSONValue(Array|Object)Append*: Simplify handling of appended
object
virJSONValueNewArrayFromBitmap: Refactor cleanup
virJSONValueObjectAddVArgs: Use autofree for the temporary bitmap
virJSONValueObjectAppend: Clear pointer when taking ownership of
passed value
qemuAgentMakeStringsArray: Refactor cleanup
virMACMapHashDumper: Refactor array addition
testQEMUSchemaValidateObjectMergeVariantMember: Fix theoretical leak
virJSONValueArrayAppend: Clear pointer when taking ownership of passed
value
qemuMonitorJSONTransactionAdd: Refactor cleanup
qemuAgentSetVCPUsCommand: Refactor cleanup
virJSONParserHandle*: Refactor memory cleanup and drop NULL checks
virJSONValueNewNumber: Take ownership of passed string
virJSONParserInsertValue: Take double pointer for @value
src/locking/lock_daemon.c | 77 ++++----
src/logging/log_daemon.c | 52 +++---
src/logging/log_handler.c | 40 ++--
src/network/leaseshelper.c | 3 +-
src/node_device/node_device_driver.c | 4 +-
src/qemu/qemu_agent.c | 89 ++++-----
src/qemu/qemu_block.c | 19 +-
src/qemu/qemu_command.c | 3 +-
src/qemu/qemu_firmware.c | 27 +--
src/qemu/qemu_migration_params.c | 4 +-
src/qemu/qemu_monitor_json.c | 58 +++---
src/rpc/virnetdaemon.c | 25 +--
src/rpc/virnetserver.c | 79 +++-----
src/rpc/virnetserverclient.c | 24 +--
src/rpc/virnetserverservice.c | 34 ++--
src/util/virjson.c | 261 +++++++++++----------------
src/util/virjson.h | 7 +-
src/util/virlease.c | 2 +-
src/util/virlockspace.c | 47 ++---
src/util/virmacmap.c | 13 +-
tests/testutilsqemuschema.c | 4 +-
21 files changed, 322 insertions(+), 550 deletions(-)
--
2.29.2
3 years, 6 months
[PATCH v2 00/15] qemu: migrate block bitmaps when migrating storage
by Peter Krempa
When we are copying storage we should also preserve the block dirty
bitmaps. This series implements their migration.
For standard migration with shared storage we let qemu flush the bitmaps
to disk and reload them on destination. There is possibility to migrate
them using the migration stream too but it's not implemented in this
series.
v2:
Patches from v1:
- patches 4-7 from v1 were pushed already
- patch 8 from v1 was dropped
This version:
- patch 1 was updated to capture the latest capabilities
- patch 4 is new, replaces patch 8 from v1
- all other patches apply review feedback from Jirka
- patch 14 has one additional change (thus I didn't carry Jirka's R-b):
The migration of bitmaps is enabled based on
QEMU_CAPS_MIGRATION_PARAM_BLOCK_BITMAP_MAPPING instead of
QEMU_CAPS_INCREMENTAL_BACKUP.
Peter Krempa (15):
qemucapabilitiesdata: Update test data for qemu-6.0 on x86_64
qemu: capabilities: Introduce
QEMU_CAPS_MIGRATION_PARAM_BLOCK_BITMAP_MAPPING
qemu: Probe whether an image is 'qcow2 v2' from
query-named-block-nodes
qemu: migration: Create qcow2 v3 images for
VIR_MIGRATE_NON_SHARED_DISK
qemu: monitor: Introduce qemuMonitorBitmapRemove
qemu: blockjob: Use qemuMonitorBitmapRemove for single bitmap removal
qemu: migration_params: Add infrastructure for 'dirty-bitmaps'
migration feature
qemu: migration_cookie: Add XML handling for setting up bitmap
migration
qemu: migration_cookie: Add helpers for transforming the cookie into
migration params
qemu: domain: Store list of temporary bitmaps for migration in status
XML
tests: qemustatusxml2xml: Add status XML from migration with bitmaps
tests: qemumigrationcookie: Add testing for block dirty bitmap
migration
qemu: migration: Clean up temporary bitmaps when cancelling a
migration
qemu: migration: Migrate block dirty bitmaps corresponding to
checkpoints
qemu: capabilities: Enable QEMU_CAPS_INCREMENTAL_BACKUP
src/qemu/qemu_blockjob.c | 24 +-
src/qemu/qemu_capabilities.c | 9 +-
src/qemu/qemu_capabilities.h | 1 +
src/qemu/qemu_domain.c | 87 +-
src/qemu/qemu_domain.h | 15 +
src/qemu/qemu_migration.c | 369 ++++++++-
src/qemu/qemu_migration_cookie.c | 261 ++++++
src/qemu/qemu_migration_cookie.h | 42 +
src/qemu/qemu_migration_params.c | 29 +
src/qemu/qemu_migration_params.h | 5 +
src/qemu/qemu_monitor.c | 13 +
src/qemu/qemu_monitor.h | 8 +
src/qemu/qemu_monitor_json.c | 35 +
src/qemu/qemu_monitor_json.h | 6 +
tests/meson.build | 2 +-
tests/qemublocktest.c | 2 +
tests/qemublocktestdata/bitmap/synthetic.json | 2 +-
tests/qemublocktestdata/bitmap/synthetic.out | 1 +
.../caps_6.0.0.x86_64.replies | 741 ++++++++++--------
.../caps_6.0.0.x86_64.xml | 23 +-
.../nbd-bitmaps-xml2xml-in.xml | 52 ++
.../nbd-bitmaps-xml2xml-migparams.json | 25 +
.../nbd-bitmaps-xml2xml-out.xml | 51 ++
tests/qemumigrationcookiexmltest.c | 166 +++-
tests/qemumonitorjsontest.c | 2 +
.../migration-out-nbd-bitmaps-in.xml | 574 ++++++++++++++
.../migration-out-nbd-bitmaps-out.xml | 1 +
tests/qemustatusxml2xmltest.c | 1 +
28 files changed, 2172 insertions(+), 375 deletions(-)
create mode 100644 tests/qemumigrationcookiexmldata/nbd-bitmaps-xml2xml-in.xml
create mode 100644 tests/qemumigrationcookiexmldata/nbd-bitmaps-xml2xml-migparams.json
create mode 100644 tests/qemumigrationcookiexmldata/nbd-bitmaps-xml2xml-out.xml
create mode 100644 tests/qemustatusxml2xmldata/migration-out-nbd-bitmaps-in.xml
create mode 120000 tests/qemustatusxml2xmldata/migration-out-nbd-bitmaps-out.xml
--
2.29.2
3 years, 6 months
[PATCH 0/4] qemuxml2*test: Significantly improve performance by caching qemu capabilities and qapi schema
by Peter Krempa
This results in 1/3 of the original execution time.
Peter Krempa (4):
qemuxml2argvtest: Cache QAPI schema between tests
testCompareXMLToArgvValidateSchema: Improve and fix helper for testing
everything
testQemuInfoSetArgs: Use curly braces in else section
qemu*xml2*test: Cache capabilities between tests
tests/qemustatusxml2xmltest.c | 3 ++-
tests/qemuxml2argvtest.c | 26 +++++++++++++++++++-------
tests/qemuxml2xmltest.c | 3 ++-
tests/testutilsqemu.c | 24 +++++++++++++++++++-----
tests/testutilsqemu.h | 2 ++
5 files changed, 44 insertions(+), 14 deletions(-)
--
2.29.2
3 years, 6 months
[PATCH v2] qemu: Fix libvirt hang due to early TPM device stop
by Stefan Berger
This patch partially reverts commit 5cde9dee where the qemuExtDevicesStop()
was moved to a location before the QEMU process is stopped. It may be
alright to tear down some devices before QEMU is stopped, but it doesn't work
for the external TPM (swtpm) which assumes that QEMU sends it a signal to stop
it before libvirt may try to clean it up. So this patch moves the
virFileDeleteTree() calls after the call to qemuExtDevicesStop() so that the
pid file of virtiofsd is not deleted before that call.
Afftected libvirt versions are 6.10 and 7.0.
Fixes: 5cde9dee8c70b17c458d031ab6cf71dce476eea2
Cc: Masayoshi Mizuma <m.mizuma(a)jp.fujitsu.com>
Signed-off-by: Stefan Berger <stefanb(a)linux.ibm.com>
---
src/qemu/qemu_process.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c
index 7feb35e609..d930ff9a74 100644
--- a/src/qemu/qemu_process.c
+++ b/src/qemu/qemu_process.c
@@ -7659,11 +7659,6 @@ void qemuProcessStop(virQEMUDriverPtr driver,
/* Do this before we delete the tree and remove pidfile. */
qemuProcessKillManagedPRDaemon(vm);
- qemuExtDevicesStop(driver, vm);
-
- virFileDeleteTree(priv->libDir);
- virFileDeleteTree(priv->channelTargetDir);
-
ignore_value(virDomainChrDefForeach(vm->def,
false,
qemuProcessCleanupChardevDevice,
@@ -7677,10 +7672,15 @@ void qemuProcessStop(virQEMUDriverPtr driver,
qemuDomainCleanupRun(driver, vm);
+ qemuExtDevicesStop(driver, vm);
+
qemuDBusStop(driver, vm);
vm->def->id = -1;
+ virFileDeleteTree(priv->libDir);
+ virFileDeleteTree(priv->channelTargetDir);
+
/* Stop autodestroy in case guest is restarted */
qemuProcessAutoDestroyRemove(driver, vm);
--
2.26.2
3 years, 6 months
[PATCH] qemu: Fix libvirt hang due to early TPM device stop
by Stefan Berger
This patch partially reverts commit 5cde9dee where the qemuExtDevicesStop()
was moved to a location before the QEMU process is stopped. It may be
alright to tear down some devices before QEMU is stopped, but it doesn't work
for the external TPM (swtpm) which assumes that QEMU sends it a signal to stop
it before libvirt may try to clean it up. So this patch introduces a function
qemuExtDevicesStopEarly() to accommodate those devices that can/must
be torn down early, which seems to be the case with virtio and its virtiofsd,
and recreates the location of the call to qemuExtDevicesStop() from before the
regression for all other devices.
Fixes: 5cde9dee8c70b17c458d031ab6cf71dce476eea2
Cc: Masayoshi Mizuma <m.mizuma(a)jp.fujitsu.com>
Signed-off-by: Stefan Berger <stefanb(a)linux.ibm.com>
---
src/qemu/qemu_extdevice.c | 27 ++++++++++++++++++++-------
src/qemu/qemu_extdevice.h | 4 ++++
src/qemu/qemu_process.c | 5 ++++-
3 files changed, 28 insertions(+), 8 deletions(-)
diff --git a/src/qemu/qemu_extdevice.c b/src/qemu/qemu_extdevice.c
index 8fe7ceaa10..96fdc9a589 100644
--- a/src/qemu/qemu_extdevice.c
+++ b/src/qemu/qemu_extdevice.c
@@ -205,7 +205,27 @@ qemuExtDevicesStart(virQEMUDriverPtr driver,
return 0;
}
+/* qemuExtDevicesStopEarly stops devices that may be stopped
+ * before QEMU terminates
+ */
+void
+qemuExtDevicesStopEarly(virQEMUDriverPtr driver,
+ virDomainObjPtr vm)
+{
+ virDomainDefPtr def = vm->def;
+ size_t i;
+ for (i = 0; i < def->nfss; i++) {
+ virDomainFSDefPtr fs = def->fss[i];
+
+ if (fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS)
+ qemuVirtioFSStop(driver, vm, fs);
+ }
+}
+
+/* qemuExtDevicesStop stops devices that may only be stopped
+ * after QEMU terminated
+ */
void
qemuExtDevicesStop(virQEMUDriverPtr driver,
virDomainObjPtr vm)
@@ -236,13 +256,6 @@ qemuExtDevicesStop(virQEMUDriverPtr driver,
if (actualType == VIR_DOMAIN_NET_TYPE_ETHERNET && net->downscript)
virNetDevRunEthernetScript(net->ifname, net->downscript);
}
-
- for (i = 0; i < def->nfss; i++) {
- virDomainFSDefPtr fs = def->fss[i];
-
- if (fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS)
- qemuVirtioFSStop(driver, vm, fs);
- }
}
diff --git a/src/qemu/qemu_extdevice.h b/src/qemu/qemu_extdevice.h
index 49373a15a1..df0cbd6993 100644
--- a/src/qemu/qemu_extdevice.h
+++ b/src/qemu/qemu_extdevice.h
@@ -51,6 +51,10 @@ int qemuExtDevicesStart(virQEMUDriverPtr driver,
ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2)
G_GNUC_WARN_UNUSED_RESULT;
+void qemuExtDevicesStopEarly(virQEMUDriverPtr driver,
+ virDomainObjPtr vm)
+ ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2);
+
void qemuExtDevicesStop(virQEMUDriverPtr driver,
virDomainObjPtr vm)
ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2);
diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c
index 7feb35e609..b5b2ddef4c 100644
--- a/src/qemu/qemu_process.c
+++ b/src/qemu/qemu_process.c
@@ -7659,7 +7659,7 @@ void qemuProcessStop(virQEMUDriverPtr driver,
/* Do this before we delete the tree and remove pidfile. */
qemuProcessKillManagedPRDaemon(vm);
- qemuExtDevicesStop(driver, vm);
+ qemuExtDevicesStopEarly(driver, vm);
virFileDeleteTree(priv->libDir);
virFileDeleteTree(priv->channelTargetDir);
@@ -7677,6 +7677,9 @@ void qemuProcessStop(virQEMUDriverPtr driver,
qemuDomainCleanupRun(driver, vm);
+ /* tear down external devices after QEMU is gone */
+ qemuExtDevicesStop(driver, vm);
+
qemuDBusStop(driver, vm);
vm->def->id = -1;
--
2.26.2
3 years, 6 months
[PATCH] event-test: Properly terminate strings printed from callbacks
by Kristina Hanicova
Stdio was buffering strings in functions:
myDomainEventBlockJobCallback,
myDomainEventBlockThresholdCallback,
myDomainEventMemoryFailureCallback. It caused flushing the
printed strings from callbacks at the end of a run, not
gradually. The solution is to add \n at the end of each string.
Signed-off-by: Kristina Hanicova <khanicov(a)redhat.com>
---
examples/c/misc/event-test.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/examples/c/misc/event-test.c b/examples/c/misc/event-test.c
index f164e825e1..76d4f3f6e8 100644
--- a/examples/c/misc/event-test.c
+++ b/examples/c/misc/event-test.c
@@ -938,7 +938,7 @@ myDomainEventBlockJobCallback(virConnectPtr conn G_GNUC_UNUSED,
const char *eventName = opaque;
printf("%s EVENT: Domain %s(%d) block job callback '%s' disk '%s', "
- "type '%s' status '%s'",
+ "type '%s' status '%s'\n",
__func__, virDomainGetName(dom), virDomainGetID(dom), eventName,
disk, blockJobTypeToStr(type), blockJobStatusToStr(status));
return 0;
@@ -956,7 +956,7 @@ myDomainEventBlockThresholdCallback(virConnectPtr conn G_GNUC_UNUSED,
{
/* Casts to uint64_t to work around mingw not knowing %lld */
printf("%s EVENT: Domain %s(%d) block threshold callback dev '%s'(%s), "
- "threshold: '%" PRIu64 "', excess: '%" PRIu64 "'",
+ "threshold: '%" PRIu64 "', excess: '%" PRIu64 "'\n",
__func__, virDomainGetName(dom), virDomainGetID(dom),
dev, NULLSTR(path), (uint64_t)threshold, (uint64_t)excess);
return 0;
@@ -972,7 +972,7 @@ myDomainEventMemoryFailureCallback(virConnectPtr conn G_GNUC_UNUSED,
void *opaque G_GNUC_UNUSED)
{
printf("%s EVENT: Domain %s(%d) memory failure: recipient '%d', "
- "aciont '%d', flags '%d'", __func__, virDomainGetName(dom),
+ "aciont '%d', flags '%d'\n", __func__, virDomainGetName(dom),
virDomainGetID(dom), recipient, action, flags);
return 0;
}
--
2.29.2
3 years, 6 months