Jim Fehlig [Thu, 28 Oct 2021 21:58:16 +0000 (15:58 -0600)]
libxl: Free data returned from libxl_userdata_retrieve
Found via valgrind
==15016== 3,701 bytes in 2 blocks are definitely lost in loss record 975 of 1,009
==15016== at 0x4C2A2AF: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==15016== by 0x1FCD30CB: libxl_read_file_contents (in /usr/lib64/libxenlight.so.4.12.0)
==15016== by 0x1FCCA58A: ??? (in /usr/lib64/libxenlight.so.4.12.0)
==15016== by 0x1FCCA6C2: libxl_userdata_retrieve (in /usr/lib64/libxenlight.so.4.12.0)
==15016== by 0x1FA42A5A: libxlReconnectDomain (libxl_driver.c:394)
==15016== by 0x53BAC99: virDomainObjListHelper (virdomainobjlist.c:802)
==15016== by 0x530842F: virHashForEach (virhash.c:575)
==15016== by 0x53BC0E0: virDomainObjListForEach (virdomainobjlist.c:817)
==15016== by 0x1FA423C4: libxlReconnectDomains (libxl_driver.c:468)
==15016== by 0x1FA423C4: libxlStateInitialize (libxl_driver.c:778)
==15016== by 0x54E8E9E: virStateInitialize (libvirt.c:657)
==15016== by 0x12DBFA: daemonRunStateInit (remote_daemon.c:797)
==15016== by 0x535BF79: virThreadHelper (virthread.c:206)
Signed-off-by: Jim Fehlig <jfehlig@suse.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Bihong Yu [Mon, 25 Oct 2021 09:04:55 +0000 (17:04 +0800)]
qemu_process: continue to process fakereboot after restarting libvirtd
During the vm rebooting, the vm could be paused if the libvirtd is
restarted for some reason, which is not expected. We need continue
fakereboot process if fakereboot flags is true and the vm is in
paused-user status.
Signed-off-by: Bihong Yu <yubihong@huawei.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Bihong Yu [Mon, 25 Oct 2021 09:04:54 +0000 (17:04 +0800)]
qemu_process: set fakereboot flags false after processing fakereboot over
During the vm rebooting, the vm could be shut down if the libvirtd is
restarted for some reason, which is not expected. We move set
fakereboot flags false after processing fakereboot over, so we can
ensure that fakereboot process have been executed.
Signed-off-by: Bihong Yu <yubihong@huawei.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
This patch improves readability of the function and makes the
code look cleaner by removing the 'else' branches after return
and reordering of the 'if' branches.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Pavel Hrdina [Tue, 9 Nov 2021 14:00:53 +0000 (15:00 +0100)]
qemu_command: do not use host-nodes for system memory
Commit 88957116c9d3cb4705380c3702c9d4315fb500bb switched to use
memory-backend-* for regular VM memory as well. That change indirectly
started using 'host-nodes' for system memory which results in QEMU
calling mbind() to bind the system memory to specific NUMA node if the
VM XML contains the configuration similar to this:
virnetsocket: pass HOME and XDG_RUNTIME_DIR to ssh
openssh supports environment variable expansion in its ssh_config
file[1]. These two environment variables can be used to
expand paths for ssh sockets and other files.
Some forward declarations in bridge_driver.c are not needed
really. They only create a noise when trying to jump onto the
correct tag. Drop them.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com> Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
Tim Wiederhake [Mon, 2 Aug 2021 14:44:22 +0000 (16:44 +0200)]
virQEMUCaps: Add host cpuid information
Many things can affect the availability of cpu flags (e.g. software
upgrades, kernel versions, kernel command line, etc.) and invalidate the
cached capabilities without notice. Add CPUID information to the
capabilities cache.
Signed-off-by: Tim Wiederhake <twiederh@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Tim Wiederhake [Wed, 13 Oct 2021 10:53:43 +0000 (12:53 +0200)]
cpu: Change virCPUArchDataParse to take xmlNodePtr
The function does not need a full xmlXPathContextPtr any longer and a
later patch will require a call to this function with only a xmlNodePtr
available.
Signed-off-by: Tim Wiederhake <twiederh@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Fri, 5 Nov 2021 09:13:07 +0000 (10:13 +0100)]
syntax-check: Fix regex for sc_require_attribute_cleanup_initialization
When I was cleaning up the regex after we removed most of our custom
autofree helpers I've forgot to delete one closing brace, thus the regex
was not matching anything.
Fixes: 65f702020e8 Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
remote_daemon: Validate tcp_min_ssf value only if found in config
If there is no tcp_min_ssf value set in daemon config we still
compare it against the default (56 which corresponds to DES) and
if the value is below our expected minimum (112 which corresponds
to 3DES) an error is reported and the daemon refuses to start.
This is not what we want. What we want is to check the value iff
the value was specified in the config file.
Fixes: 58a48cff840 Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Fri, 5 Nov 2021 08:48:52 +0000 (09:48 +0100)]
qemuTPMEmulatorReconfigure: Fix two build issues
1) 'activePcrBanksStr' is not initialized:
../../../libvirt/src/qemu/qemu_tpm.c: In function ‘qemuExtTPMStart’:
/usr/include/glib-2.0/glib/glib-autocleanups.h:28:3: error: ‘activePcrBanksStr’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
28 | g_free (*pp);
| ^~~~~~~~~~~~
../../../libvirt/src/qemu/qemu_tpm.c:613:22: note: ‘activePcrBanksStr’ was declared here
613 | g_autofree char *activePcrBanksStr;
| ^~~~~~~~~~~~~~~~~
Stefan Berger [Wed, 3 Nov 2021 17:04:23 +0000 (13:04 -0400)]
qemu: tpm: Extend TPM domain XML with PCR banks to activate
Extend the TPM backend XML with a node 'active_pcr_banks' that allows a
user to specify the PCR banks to activate before starting a VM. Valid
choices for PCR banks are sha1, sha256, sha384 and sha512. When the XML
node is provided, the set of active PCR banks is 'enforced' by running
swtpm_setup before every start of the VM. The activation requires that
swtpm_setup v0.7 or later is installed and may not have any effect
otherwise.
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=2016599 Signed-off-by: Stefan Berger <stefanb@linux.ibm.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
At this point, we're no longer using the availability of the
ZFS programs at build time to decide whether to enable ZFS
support, so the only purpose of these find_program() calls is
to record their absolute paths.
However, the virCommand facilities that we're ultimately using
to run them are already capable of performing this lookup at
runtime, and in fact that's exactly what we already do in the
case of, for example, vstorage.
Drop the build time lookups and always perform them at runtime.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com>
Commit 73a2ff36163e already improved the situation a lot. This
pushes things even further.
If the user or, more likely, the distro packager explicitly
asked for ZFS support to be enabled, then we should comply with
that request regardless of whether the necessary programs are
available at build time.
This is particularly important in the context of Debian, where
ZFS cannot be a build dependency of libvirt due to licensing
issues but it can still be an optional runtime dependency.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com>
Ján Tomko [Tue, 2 Nov 2021 15:17:36 +0000 (16:17 +0100)]
qemu: always assume QEMU_CAPS_SPICE_UNIX
The presence of this capability depends on QEMU being compiled
with spice that has the SPICE_ADDR_FLAG_UNIX_ONLY constant.
It was added by spice commit 5365caeaae released in spice v0.12.6,
which is older than the spice version on our supported architectures.
Signed-off-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com> Reviewed-by: Peter Krempa <pkrempa@redhat.com>
This commit aims to address the bug reported in [1] and [2].
If the profile is corrupted (0-size) the VM cannot be launched.
To overcome this, check if the profile exists and if it has 0 size
remove it.
Peter Krempa [Fri, 29 Oct 2021 14:04:45 +0000 (16:04 +0200)]
qemuDomainGetStatsBlockExportDisk: Report stats also for helper images
Add stat entries also for the mirror destination and the backup job
scratch/target file. This is possible with '-blockdev' as we use unique
index for the entries.
The stats are reported when the VIR_CONNECT_GET_ALL_DOMAINS_STATS_BACKING
is used.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2017928 Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Mon, 1 Nov 2021 10:35:41 +0000 (11:35 +0100)]
qemuMonitorJSONQueryBlockstats: query stats for helper images
Use the 'query-nodes' flag to return all stats. The flag was introduced
prior to qemu-2.11 so we can always use it, but we invoke it only when
querying stats. The other invocation is used for detecting the nodenames
which is fragile code.
The images without a frontend don't have the device field so the
extraction code checks need to be relaxed.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Mon, 1 Nov 2021 11:42:39 +0000 (12:42 +0100)]
virDomainBackupDefFormat: Propagate private data callbacks
The formatter for the backup job data didn't pass the virDomainXMLOption
struct to the disk formatter which meant that the private data of the
disk source were not formatted.
This didn't pose a problem for now as the blockjob list remembered the
nodenames for the jobs, but the backup source lost them.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
While being great semantic patching tool, coccinelle fails to
understand some of macros we use (including those provided by
glib). What they have in common is use of __attribute__ under the
hood. We store a list of such macros in a file. But in there,
g_auto() macro is not defined properly. Indeed, g_auto(type)
declares a local variable of given type, for instance from
cocci's POV:
lib: Use G_N_ELEMENTS instead of sizeof()/sizeof()
For statically declared arrays one can use G_N_ELEMENTS() instead
of explicit sizeof(array) / sizeof(item). I've noticed couple of
places where the latter was used.
I am not fixing every occurrence because we have some places
which do not use glib (examples and NSS module).
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com>
virpcivpdtest: Declare variables at multiple lines
In testPCIVPDResourceCustomCompareIndex() there are two variables
declared at one line. They are both g_autoptr() decorated which
makes it worse, because coccinelle fails to parse that.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com>
I've noticed one function inside virpcivpd.c, namely
virPCIVPDParseVPDLargeResourceFields() that declares some
variables at the top level even though they are used only inside
a loop in which they have to be freed explicitly.
Bringing variable declarations into the loop allows us to make
the code nicer.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com>
that have strange attitude towards g_auto* variables. The first
problem is that variables are declared at the top level despite
being used inside a loop. The second problem is use of g_free()
in combination with g_steal_pointer() even though we have
VIR_FREE() which does exactly that.
Bringing variable declarations into their respective loops allows
us to make the code nicer.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com>
The first part of the version string contains the name that the
rst2html5 command was invoked as, which can differ based on the
operating system: on FreeBSD, for example, it's rst2html5.py
instead of just rst2html5.
Fix our detection logic so that it works regardless of the
specific name used for the docutils-provided rst2html5 command.
Fixes: cf0c9e186565e886a0016b2b269088b3eed3d26d Signed-off-by: Andrea Bolognani <abologna@redhat.com>
While invalid values need to be ignored when presenting VPD data to the
user, it would be good to attempt to parse a valid portion of the VPD
instead of marking it invalid as a whole.
Based on a mailing list discussion, the set of accepted characters is
extended to the set of printable ASCII characters.
The particular example encountered on real hardware was multi-faceted:
* "N/A" strings present in read-only fields. This would not be a useful
valid value for a field (especially if a unique serial number is
expected), however, it was decided to delegate handling of those kinds
of values to higher-level software;
* "4W/1W PCIeG2x4" - looks like some vendors use even more printable
characters in the ASCII range than we currently allow. Since the
PCI/PCIe VPD specs mention alphanumeric characters without specifying
the full character set, it looks like this is ambiguous for vendors
and they tend to use printable ASCII characters;
* 0xFF bytes present in VPD-W field values. Those bytes do not map to
printable ASCII code points and were probably used by the vendor as
placeholders. Ignoring the whole VPD because of that would be too
strict.
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Signed-off-by: Dmitrii Shcherbakov <dmitrii.shcherbakov@canonical.com>
* RV and RW fields must be at the last position in their respective
section (per the conditions in the spec). Therefore, the parser now
stops iterating over fields as soon as it encounters one of those
fields and checks whether the end of the resource has been reached;
* The lack of the RW field is not treated as a parsing error since we
can still extract valid data even though this is a PCI/PCIe VPD spec
violation;
* Individual fields must have a valid length - the parser needs to check
for invalid length values that violate boundary conditions of the
resource.
* A zero-length field may be the last one in the resource, however, the
boundary check is currently too strict to allow that.
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Signed-off-by: Dmitrii Shcherbakov <dmitrii.shcherbakov@canonical.com>
Michal Privoznik [Thu, 23 Sep 2021 12:32:24 +0000 (14:32 +0200)]
lib: Introduce and use g_autoptr() for virInterfaceDef
There are a lot of places where we call virInterfaceDefFree()
explicitly. We can define autoptr cleanup macro and annotate
declarations with g_autoptr() and remove plenty of those explicit
free calls.
This also fixes a memory leak in udevInterfaceGetXMLDesc() which
called virInterfaceDefFree() only in successful path.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Tim Wiederhake <twiederh@redhat.com>
Stefan Berger [Mon, 1 Nov 2021 17:23:39 +0000 (13:23 -0400)]
qemu: Move code to add encryption options for swtpm_setup into function
Move the code that adds encryption options for the swtpm_setup command
line into its own function.
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Fri, 17 Sep 2021 14:21:43 +0000 (16:21 +0200)]
testQEMUSchemaValidateEnum: Refactor logic to simplify switching to new QMP schema format
QEMU-6.2 is reporting enum values in the new 'members' array which we'll
be switching to. Rewrite the logic so that adding the new checker is
more straightforward.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Han Han [Wed, 27 Oct 2021 08:52:33 +0000 (16:52 +0800)]
virsh: Fix ambiguous output in metadata-change event
When you set metadata with type element like the following:
dom.setMetadata(libvirt.VIR_DOMAIN_METADATA_ELEMENT, "<test/>", 'abc', "HAHAH", 0)
Then for `virsh event --all`, then it will output this message:
event 'metadata-change' for domain 'rhel9': element HAHAH
The message is ambiguous since it looks like the params for
metadata-change event is the element HAHAH. Actually that means the type is
element while the url is HAHAH. Let's make it more clear.
Signed-off-by: Han Han <hhan@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Report an error if the new hotplug is not supported and remove the
alternate code paths.
The modern cpu-hotplug code was introduced in qemu-2.7. We keep the
capability so that proper errors are reported in case a platform doesn't
support hotplug.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
projects / thirdparty / libvirt.git / log
