Jonathon Jongsma [Fri, 10 Jan 2020 23:32:13 +0000 (17:32 -0600)]
qemu: don't access vmdef within qemu_agent.c
In order to avoid holding an agent job and a normal job at the same
time, we want to avoid accessing the domain's definition while holding
the agent job. To achieve this, qemuAgentGetFSInfo() only returns the
raw information from the agent query to the caller. The caller can then
release the agent job and then proceed to look up the disk alias from
the vm definition. This necessitates moving a few helper functions to
qemu_driver.c and exposing the agent data structure (qemuAgentFSInfo) in
the header.
In addition, because the agent function no longer returns the looked-up
disk alias, we can't test the alias within qemuagenttest. Instead we
simply test that we parse and return the raw agent data correctly.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Jonathon Jongsma [Fri, 10 Jan 2020 23:32:12 +0000 (17:32 -0600)]
qemu: Don't store disk alias in qemuAgentDiskInfo
The qemuAgentDiskInfo structure is filled with information received from
the agent command response, except for the 'alias' field, which is
retrieved from the vm definition. Limit this structure only to data that
was received from the agent message.
This is another intermediate step in moving the responsibility for
searching the vmdef from qemu_agent.c to qemu_driver.c so that we can
avoid holding an agent job and a normal job at the same time.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Jonathon Jongsma [Fri, 10 Jan 2020 23:32:11 +0000 (17:32 -0600)]
qemu: store complete agent filesystem information
In an effort to avoid holding both an agent and normal job at the same
time, we shouldn't access the vm definition from within qemu_agent.c
(i.e. while the agent job is being held). In preparation, we need to
store the full filesystem disk information in qemuAgentDiskInfo. In a
following commit, we can pass this information back to the caller and
the caller can search the vm definition to match the filsystem disk to
an alias.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Jonathon Jongsma [Fri, 10 Jan 2020 23:32:10 +0000 (17:32 -0600)]
qemu: rename qemuAgentGetFSInfoInternalDisk()
The function name doesn't give a good idea of what the function does.
Rename to qemuAgentGetFSInfoFillDisks() to make it more obvious than it
is filling in the disk information in the fsinfo struct.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
tests: add host CPU data files for validating die_id
Only Cascadelake-AP CPUs appear to report "die_id" values != 0 on Linux
right now - AMD EPYC's don't report "die_id" (at least with Fedora 31
kernel). Lacking access to Cascadelake-AP CPUs, this test data was from
a Fedora 31 QEMU guest launched with
Ideally we'd replace this data with some from a real machine reporting
"die_id", to ensure we're not mislead by QEMU's impl.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Jiri Denemark <jdenemar@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
hostcpu: add support for reporting die_id in NUMA topology
Update the host CPU code to report the die_id in the NUMA topology
capabilities. On systems with multiple dies, this fixes the bug
where CPU cores can't be distinguished:
So core_id is now unique within a (socket_id, die_id) pair.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Jiri Denemark <jdenemar@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
qemu: add support for specifying CPU "dies" topology parameter
QEMU since 4.1.0 supports the "dies" parameter for -smp
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Jiri Denemark <jdenemar@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Jiri Denemark <jdenemar@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Jiri Denemark [Wed, 15 Jan 2020 14:24:55 +0000 (15:24 +0100)]
qemu: Don't emit SUSPENDED_POSTCOPY event on destination
When pause-before-switchover QEMU capability is enabled, we get STOP
event before MIGRATION event with postcopy-active state. To properly
handle post-copy migration and emit correct events commit v4.10.0-rc1-4-geca9d21e6c added a hack to
qemuProcessHandleMigrationStatus which translates the paused state
reason to VIR_DOMAIN_PAUSED_POSTCOPY and emits
VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY event when migration state changes
to post-copy.
However, the code was effective on both sides of migration resulting in
a confusing VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY event on the destination
host, where entering post-copy mode is already properly advertised by
VIR_DOMAIN_EVENT_RESUMED_POSTCOPY event.
virchrdev: Don't leak mutex if virChrdevAlloc() fails
This is only a theoretical leak, but in virChrdevAlloc() we
initialize a mutex and if creating a hash table fails,
then virChrdevFree() is called which because of incorrect check
doesn't deinit the mutex.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
virchrdev: Don't leak @dev member of virChrdevHashEntry struct
When opening a console to a domain, we put a tuple of {path,
virStreamPtr} into a hash table that's private to the domain.
This is to ensure only one client at most has the console stream
open. Later, when the console is closed, the tuple is removed
from the hash table and freed. Except, @path won't be freed.
==234102== 60 bytes in 5 blocks are definitely lost in loss record 436 of 651
==234102== at 0x4836753: malloc (vg_replace_malloc.c:307)
==234102== by 0x5549110: g_malloc (in /usr/lib64/libglib-2.0.so.0.6000.6)
==234102== by 0x5562D1E: g_strdup (in /usr/lib64/libglib-2.0.so.0.6000.6)
==234102== by 0x4A5A917: virChrdevOpen (virchrdev.c:412)
==234102== by 0x17B64645: qemuDomainOpenConsole (qemu_driver.c:17309)
==234102== by 0x4BC8031: virDomainOpenConsole (libvirt-domain.c:9662)
==234102== by 0x13F854: remoteDispatchDomainOpenConsole (remote_daemon_dispatch_stubs.h:9211)
==234102== by 0x13F72F: remoteDispatchDomainOpenConsoleHelper (remote_daemon_dispatch_stubs.h:9178)
==234102== by 0x4AB0685: virNetServerProgramDispatchCall (virnetserverprogram.c:430)
==234102== by 0x4AB01F0: virNetServerProgramDispatch (virnetserverprogram.c:302)
==234102== by 0x4AB700B: virNetServerProcessMsg (virnetserver.c:136)
==234102== by 0x4AB70CB: virNetServerHandleJob (virnetserver.c:153)
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
Michal Privoznik [Mon, 13 Jan 2020 10:07:53 +0000 (11:07 +0100)]
qemu: Stop domain on failed restore
When resuming a domain from a save file, we read the domain XML
from the file, add it onto our internal list of domains, start
the qemu process, let it load the incoming migration stream and
resume its vCPUs afterwards. If anything goes wrong, the domain
object is removed from the list of domains and error is returned
to the caller. However, the qemu process might be left behind -
if resuming vCPUs fails (e.g. because qemu is unable to acquire
write lock on a disk) then due to a bug the qemu process is not
killed but the domain object is removed from the list.
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1718707 Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Julio Faracco [Fri, 27 Dec 2019 20:36:25 +0000 (17:36 -0300)]
virsh: Expose virDomainGetHostnameFlags
Our virsh already has 'domhostname' command. Add '--source'
argument to it so that users can chose between 'lease' and
'agent' sources. Also, implement completer for the argument.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
Julio Faracco [Thu, 9 Jan 2020 10:15:05 +0000 (11:15 +0100)]
lxc: Implement virDomainGetHostnameFlags
Since there is no guest agent in LXC world (yet), we can
implement _LEASE flag only.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
Julio Faracco [Thu, 9 Jan 2020 12:19:21 +0000 (13:19 +0100)]
qemu: Implement virDomainGetHostnameFlags
We have to keep the default - querying the agent if no flag is
set.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
Julio Faracco [Thu, 9 Jan 2020 10:12:37 +0000 (11:12 +0100)]
Introduce source flags to virDomainGetHostname()
There is a lots of possibilities to retrieve hostname information
from domain. Libvirt could use lease information from dnsmasq to
get current hostname too. QEMU supports QEMU-agent but it can use
lease source.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
tests: Avoid GCC over-optimization caused by _GL_ARG_NONNULL attributes.
Reported by Jim Meyering in
<https://lists.gnu.org/archive/html/bug-gnulib/2020-01/msg00040.html>.
* lib/stdlib.in.h (GNULIB_defined_canonicalize_file_name): New macro.
(GNULIB_defined_ptsname_r): New macro.
* tests/test-canonicalize.c (_GL_ARG_NONNULL): Define to empty.
(main): Disable the NULL argument test if canonicalize_file_name does
not come from gnulib.
* tests/test-canonicalize-lgpl.c (_GL_ARG_NONNULL): Define to empty.
(main): Disable the NULL argument test if canonicalize_file_name does
not come from gnulib.
* tests/test-ptsname_r.c (_GL_ARG_NONNULL): Define to empty.
(test_errors): Disable the NULL argument test if ptsname_r does not come
from gnulib.
since they fix a build failure caused by the gnulib tests failing
on ppc64le, as reported in
Ján Tomko [Tue, 14 Jan 2020 07:06:34 +0000 (08:06 +0100)]
virbuftest: use field names when initalizing test info
Allow adding new fields without changing all the macros.
Otherwise the compiler complains that not all have been initialized:
../../tests/virbuftest.c:419:5: error: missing field 'arg' initializer [-Werror,-Wmissing-field-initializers]
DO_TEST_ESCAPE("<td></td><td></td>",
^
../../tests/virbuftest.c:414:56: note: expanded from macro 'DO_TEST_ESCAPE'
struct testBufAddStrData info = { data, expect }; \
Signed-off-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
Thomas Huth [Mon, 13 Jan 2020 08:04:57 +0000 (09:04 +0100)]
qemu_capabilities: Do not report USB as subsystem type if it is not available
libvirt currently always reports that USB is available as a bus subsystem
type when running "virsh domcapabilities". However, this is not always
true, for example the qemu-system-s390x binary normally never has support
for USB. Thus we should only report that USB is available if there is
also a USB host controller available where we can attach USB devices.
Reported-by: Sebastian Mitterle <smitterl@redhat.com> Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=1759849 Signed-off-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Thomas Huth [Mon, 13 Jan 2020 08:30:14 +0000 (09:30 +0100)]
domain_conf: Do not use USB by default for <input> devices on s390x
When trying to specify an input device on s390x without bus like this:
<input type='keyboard'/>
... then libvirt currently complains:
error: unsupported configuration: USB is disabled for this domain,
but USB devices are present in the domain XML
This is somewhat confusing since the user did not specify an USB
device here. Since USB is not available on s390x, we should default
to the "virtio" bus here instead.
Peter Krempa [Fri, 10 Jan 2020 16:25:16 +0000 (17:25 +0100)]
conf: Always format storage source auth and encryption under <source> for backing files
Historically there are two places where we format authentication and
encryption for a disk. The logich which formats it for backing files was
flawed though and didn't format it at all. This worked if the image
became a backing file through the means of a snapshot but not directly.
Force formatting of the source and encryption for any non-disk case to
fix the issue.
This caused problems in many places as we use the formatter to copy the
definition. Effectively any copy lost the secret definition.
Commit v5.10.0-269-g62065a6cb5 moved NUMA validation code to domain
definition time and appropriately adjusted affected test cases except
for hugepages-default-system-size. And since we don't mock
virGetSystemPageSizeKB in our tests, hugepages-default-system-size test
would fail on architectures (ppc64le) with default page size other than
4KiB.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
In v5.0.0-rc1~94 we switched from one huge switch() to an array
for translating error numbers into error messages. However, the
array is declared to have VIR_ERR_NUMBER_LAST items which makes
it impossible to spot this place by compile checking when adding
new error number.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
When debugging tests under GDB/valgrind there is a significant
delay each time an execve is done as they scan shared libraries
once again. For tests which use many mock libraries, we have
been invoking execve many times which makes the debug experience
horrible. This changes our framework to activate the full
set of mock libraries in one single execve.
Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Michael Weiser [Fri, 3 Jan 2020 18:43:25 +0000 (19:43 +0100)]
docs: Add snapshot-revert qemu managedsave force
Add documentation for additional reason why snapshot-revert might need
to be forced. This explains why restoring an inactive snapshot while
there is managed saved state is refused by default.
Signed-off-by: Michael Weiser <michael.weiser@gmx.de> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Michael Weiser [Fri, 3 Jan 2020 18:43:24 +0000 (19:43 +0100)]
docs: Reformat snapshot-revert force reasons
Reformat explanations of the snapshot-revert force reasons in
preparation for more to be added. This is a simple reformat without any
wording changes.
Signed-off-by: Michael Weiser <michael.weiser@gmx.de> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Michael Weiser [Fri, 3 Jan 2020 18:43:23 +0000 (19:43 +0100)]
qemu: Warn of restore with managed save being risky
Internal snapshots of a non-running domain do not carry any memory state
and restoring such a snapshot will not replace existing saved memory
state. This allows a scenario, where a user first suspends a domain into
managedsave, restores a non-running snapshot and then resumes the domain
from managedsave. After that, the guest system will run with its
previous memory state atop a different disk state. The most obvious
possible fallout from this is extensive file system corruption. Swap
content and RAID bitmaps might also be off.
This has been discussed[1] and fixed[2] from the end-user perspective for
virt-manager.
This patch marks the restore operation as risky at the libvirt level,
requiring the user to remove the saved memory state first or force the
operation.
Signed-off-by: Michael Weiser <michael.weiser@gmx.de> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Michael Weiser [Fri, 3 Jan 2020 18:39:56 +0000 (19:39 +0100)]
docs: Harmonize hypervisor names for QEMU and LXC
Trivially replace usages of qemu and lxc in the virsh manpage with their
more heavily used and (according to Wikipedia) correct upper-case
spellings QEMU and LXC.
Signed-off-by: Michael Weiser <michael.weiser@gmx.de> Suggested-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Jiri Denemark [Thu, 9 Jan 2020 14:40:14 +0000 (15:40 +0100)]
qemu: Don't use NULL path from qemuDomainGetHostdevPath
Commit v5.10.0-290-g3a4787a301 refactored qemuDomainGetHostdevPath to
return a single path rather than an array of paths. When the function is
called on a missing device, it will now return NULL in @path rather than
a NULL array with zero items and the callers need to be adapted
properly.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Jiri Denemark [Tue, 17 Dec 2019 13:31:10 +0000 (14:31 +0100)]
virsh migrate: Require --tls for --tls-destination
--tls-destination would be just ignored unless --tls is not specified,
which is correct, but let's provide a bit of a guidance is a user
forgets to add --tls.
This is just a virsh-only check targeted to end users as we don't
currently have such checks at the API level for migration parameters
that depend on flags.
bhyveargv2xmlmock calls virBhyveCapsBuild which in turn
calls virCPUProbeHost, probing the real host CPU. This
causes a test failure if the host CPU happens to contain
the 'arch-capabilities' feature as it triggers a call
to virHostCPUGetMSR() which fails on FreeBSD.
Fortunately we already have convenient code for mocking
the host CPU probing.
Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
schema: Allow iSCSI source to have interleaved children
There is no need to require users to produce iSCSI disk source
following our ordering of children elements. In fact, we don't
even accept our own order in the schema :(.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Laine Stump <laine@redhat.com>
bhyve_parse_command.c: Don't jump onto non-existent label in CONSUME_ARG()
In v5.10.0-508-gfbf3f3d86a, the 'error' label was removed from
bhyveParseBhyveCommandLine(), however the CONSUME_ARG() macro
still uses it. Fix the macro to return an error instead.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Ján Tomko [Wed, 8 Jan 2020 09:20:39 +0000 (10:20 +0100)]
qemu: fix implicit fallthrough warning
src/qemu/qemu_domain_address.c:680:13: error: this statement may fall through [-Werror=implicit-fallthrough=]
switch ((virDomainFSModel) dev->data.fs->model) {
Signed-off-by: Ján Tomko <jtomko@redhat.com> Fixes: f363af7e351c6cac0ff442390d0dab283cbd9fae
vz: Don't try to jump on a non-existent label in prlsdkRemoveBootDevices()
Commit v5.10.0-522-g9000b2f298 was too aggressive and removed the
'error' label from prlsdkRemoveBootDevices() even though it's
used. Luckily, it's used only from one place and we have an
alternative for it that doesn't require the label.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
qemu_firmware: Accept int in qemuFirmwareOSInterfaceTypeFromOsDefFirmware()
The point of this function is to translate virDomainOsDefFirmware
enum to qemuFirmwareOSInterface enum. However, with my commit v5.10.0-507-g8e1804f9f6 we are passing a variable type of
virDomainLoader enum. Make the function accept both enums and
make the enum members correspond to each other.
This fixes clang build.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Michal Privoznik [Tue, 17 Dec 2019 13:08:42 +0000 (14:08 +0100)]
remote_daemon: Initialize host boot time global variable
This is not strictly needed, but it makes sure we initialize the
@bootTime global variable. Thing is, in order to validate XATTRs
and prune those set in some previous runs of the host, a
timestamp is recorded in XATTRs. The host boot time was unique
enough so it was chosen as the timestamp value. And to avoid
querying and parsing /proc/uptime every time, the query function
does that only once and stores the boot time in a global
variable. However, the only time the query function is called is
in a child process that does lock files and changes seclabels. So
effectively, we are doing exactly what we wanted to prevent from
happening.
The fix is simple, call the virHostBootTimeInit() function which
sets the global variable.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Cole Robinson <crobinso@redhat.com>
gather_scsi_host_cap() in node_device_hal.c can be greatly
simplified, given that the 'out' label is always getting
hit regardless of 'retval', which can also be eliminated.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
Michal Privoznik [Tue, 17 Dec 2019 16:45:50 +0000 (17:45 +0100)]
qemu_firmware: Try to autofill for old style UEFI specification
While we discourage people to use the old style of specifying
UEFI for their domains (the old style is putting path to the FW
image under /domain/os/loader/ whilst the new one is using
/domain/os/@firmware), some applications might have not adapted
yet. They still rely on libvirt autofilling NVRAM path and
figuring out NVRAM template when using the old way (notably
virt-install does this). We must preserve backcompat for this
previously supported config approach. However, since we really
want distro maintainers to leave --with-loader-nvram configure
option and rely on JSON descriptors, we need to implement
autofilling of NVRAM template for the old way too.
projects / thirdparty / libvirt.git / log
