* XML serialization and deserialization of PCI VPD;
* PCI VPD capability flags added and used in relevant places;
* XML to XML tests for the added capability.
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Dmitrii Shcherbakov <dmitrii.shcherbakov@canonical.com>
Add helper functions to virpci to provide means of checking for a VPD
file presence and for VPD resource retrieval using the PCI VPD parser.
The added test assesses the basic functionality of VPD retrieval while
the full parser is tested by virpcivpdtest.
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Dmitrii Shcherbakov <dmitrii.shcherbakov@canonical.com>
Add support for deserializing the binary PCI/PCIe VPD format and storing
results in memory.
The VPD format is specified in "I.3. VPD Definitions" in PCI specs
(2.2+) and "6.28.1 VPD Format" PCIe 4.0. As section 6.28 in PCIe 4.0
notes, the PCI Local Bus and PCIe VPD formats are binary compatible
and PCIe 4.0 merely started incorporating what was already present in
PCI specs.
Linux kernel exposes a binary blob in the VPD format via sysfs since
v2.6.26 (commit 94e6108803469a37ee1e3c92dafdd1d59298602f) which requires
a parser to interpret.
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Dmitrii Shcherbakov <dmitrii.shcherbakov@canonical.com>
This will make it possible to limit changes to a single spot
later on, and is also just an overall nicer way to create and
destroy objects.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
There doesn't seem to be a reason for IOMMUs not to be handled
by this function.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This involves a bit of a hack, but is overall preferable to
forcing callers to pass non-const devdata as argument.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
These were generated using a QEMU binary built from commit
v6.1.0-1552-g362534a643
Notably, this causes the arguments of -device to be generated
in JSON format.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
This commit is related to 5de203f879 which I pushed a few days
ago. While that commit prioritized closing clients socket over
the rest of I/O process, this one goes one step further and
temporarily suspends processing new connection requests.
A brief recapitulation of the problem:
1) assume that libvirt is at the top of RLIMIT_NOFILE (that is no
new FDs can be opened).
2) we have a client trying to connect to a UNIX/TCP socket
Because of 2) our event loop sees POLLIN on the socket and thus
calls virNetServerServiceAccept(). But since no new FDs can be
opened (because of 1)) the request is not handled and we will get
the same event on next iteration. The poll() will exit
immediately because there is an event on the socket. Thus we end
up in an endless loop.
To break the loop and stop burning CPU cycles we can stop
listening for events on the socket and set up a timer tho enable
listening again after some time (I chose 5 seconds because of no
obvious reason).
There's another area where we play with temporarily suspending
accept() of new clients - when a client disconnects and we check
max_clients against number of current clients. Problem here is
that max_clients can be orders of magnitude larger than
RLIMIT_NOFILE but more importantly, what this code considers
client disconnect is not equal to closing client's FD.
A client disconnecting means that the corresponding client
structure is removed from the internal list of clients. Closing
of the client's FD is done from event loop - asynchronously.
To avoid this part stepping on the toes of my fix, let's make the
code NOP if socket timer (as described above) is active.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
The repo-lockdown service used to run as a bot outside GitHub, but has
now switched to using the GitHub Actions workflow framework. This
requires use of a new configuration file.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Some guest features that map to the -cpu arg are still added using
implicit syntax "feature" which is a deprecated shorthand for
"feature=on".
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
With the previous refactorings, there's no real benefit from the
qemuBuildCpuFeature helper method. Only one of the callers really
needs the CPU feature name re-writing logic, the others can just
use the right name directly.
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The -cpu arg gained support for feature=on|off syntax for the x86
emulator in 2.4.0
commit 38e5c119c2925812bd441450ab9e5e00fc79e662
Author: Eduardo Habkost <ehabkost@redhat.com>
Date: Mon Mar 23 17:29:32 2015 -0300
target-i386: Register QOM properties for feature flags
Most other targets gained this syntax even earlier in 1.4.1
commit 1590bbcb02921dfe8e3cf66e3a3aafd31193babf
Author: Andreas Färber <afaerber@suse.de>
Date: Mon Mar 3 23:33:51 2014 +0100
cpu: Implement CPUClass::parse_features() for the rest of CPUs
CPUs who do not provide their own implementation of feature parsing
will treat each option as a QOM property and set it to the supplied
value.
There appears no reason to keep supporting "+|-feature" syntax,
given the current minimum QEMU version.
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
QEMU switched from using underscores in x86 CPU features to hyphens
in the 2.8.0 series with two commits
commit fc7dfd205f3287893c436d932a167bffa30579c8 (HEAD, refs/bisect/bad)
Author: Eduardo Habkost <ehabkost@redhat.com>
Date: Fri Sep 30 15:49:40 2016 -0300
target-i386: Remove underscores from feat_names arrays
commit 54b8dc7c19cd781e96f1e9b001ca6001d804eb19
Author: Eduardo Habkost <ehabkost@redhat.com>
Date: Fri Sep 30 15:49:38 2016 -0300
target-i386: Register aliases for feature names with underscores
Libvirt names use underscores so we conditionally tranlate the
names when talking to new QEMU. Since the min QEMU was raised to
version 2.11.0, all QEMU versions we talk to expect hypens, so
the translation can be done unconditionally.
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
QEMU switched from using underscores in x86 CPU features to hyphens
in the 2.8.0 series with two commits
commit fc7dfd205f3287893c436d932a167bffa30579c8 (HEAD, refs/bisect/bad)
Author: Eduardo Habkost <ehabkost@redhat.com>
Date: Fri Sep 30 15:49:40 2016 -0300
target-i386: Remove underscores from feat_names arrays
commit 54b8dc7c19cd781e96f1e9b001ca6001d804eb19
Author: Eduardo Habkost <ehabkost@redhat.com>
Date: Fri Sep 30 15:49:38 2016 -0300
target-i386: Register aliases for feature names with underscores
Libvirt names use underscores so we conditionally tranlate the
names when talking to new QEMU. Since the min QEMU was raised to
version 2.11.0, all QEMU versions we talk to expect hypens, so
the translation can be done unconditionally.
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The non-released distros have reasonably frequent package installation
failures that can last for days at a time. This makes them unsuitable
for use as gating CI jobs.
This ensures all of the jobs in Debian Sid, Fedora Rawhide, openSUSE
Tumbleweed and FreeBSD Current are marked "allow-failure: true".
This means the jobs still run, but any failure will not be considered
fatal to the pipeline.
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The layering of the cross containers is fixed to move arch specific
ccache setup out of the common base layer.
A missing Cirrus CI variable substitution is added, though this is
irrelevant given libvirt's package list.
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Always fetch the stats for all backing chain members. Callers from
qemu_driver.c already always passed 'true' and the caller from the
migration code won't mind when we fetch all stats.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
All (proper) callers pass true so we can remove the argument.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Use automatic memory clearing and remove the cleanup label.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Use automatic memory clearing for the temporary variables and remove the
cleanup section.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Use automatic memory clearing for the temporary variable and remove the
cleanup section.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Use automatic memory clearing for the temporary variable and remove the
cleanup section.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Similarly to the fix to 'qemuDomainBlocksStatsGather' we should be
always fetching the full backing chain so that we can avoid any
automatic filter notes which would prevent us from fetching the stats
for the correct nodename.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
In certain cases such as when running a backup blockjob qemu installs a
filter node between the frontend and the top node of the backend of the
disk. The stats gathering code didn't instruct the monitor code to fetch
the stats for all the layers, so since the top layer now doesn't have
stats we were reporting wrong stats such as allocation.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2015281
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Using swtpm v0.7.0 we can run swtpm_setup to create default config files
for swtpm_setup and swtpm-localca in session mode. Now a user can start
a VM with an attached TPM without having to run this program on the
command line before. This program needs to run once.
This patch addresses the issue raised in
https://bugzilla.redhat.com/show_bug.cgi?id=2010649
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Libvirt will put the pid file of virtiofsd to per-domain directory.
However, the ownership of the per-domain directory is the user to run
the QEMU process and the user has the write permission of the directory.
If VM escape occurs, the attacker can
1. write arbitrary content to the pid file (if running QEMU using root),
then the attacker can kill any process by writing appropriate pid to
the pid file;
2. spoof the pid file (if running QEMU using a regular user), then the
virtiofsd process will never be cleared even if the VM is destroyed.
So, move the pid file of virtiofsd from per-domain directory to
stateDir.
Signed-off-by: Peng Liang <liangpeng10@huawei.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Libvirt will put the pid file of pr-helper to per-domain directory.
However, the ownership of the per-domain directory is the user to run
the QEMU process and the user has the write permission of the directory.
If VM escape occurs, the attacker can
1. write arbitrary content to the pid file (if running QEMU using root),
then the attacker can kill any process by writing appropriate pid to
the pid file;
2. spoof the pid file (if running QEMU using a regular user), then the
pr-helper process will never be cleared even if the VM is destroyed.
So, move the pid file of pr-helper from per-domain directory to
stateDir.
Signed-off-by: Peng Liang <liangpeng10@huawei.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
When parsing of the node device XML fails we'd still call the post-parse
and validation callbacks which makes no sense. Additionally the
callbacks were expecting a non-NULL pointer which leads to a crash.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2014139
Fixes: d5ae634ba2
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
Issuing simple QMP commands is pain as they need to be wrapped by the
JSON wrapper:
{ "execute": "COMMAND" }
and optionally also:
{ "execute": "COMMAND", "arguments":...}
For simple commands without arguments we can add syntax sugar to virsh
which allows simple usage of QMP and additionally prepares also for
passing through of the 'arguments' section:
virsh qemu-monitor-command $VM query-status
is equivalent to
virsh qemu-monitor-command $VM '{"execute":"query-status"}'
and
virsh qemu-monitor-command $VM query-named-block-nodes '{"flat":true}'
or
virsh qemu-monitor-command $VM query-named-block-nodes '"flat":true'
is equivalent to
virsh qemu-monitor-command $VM '{"execute":"query-named-block-nodes", "arguments":{"flat":true}}'
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Normally this would be considered an internal detail which we don't
document in the news, but in this case I'd like to make people aware of
the change so that they preferrably report them ASAP.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Now that we use JSON with -device we can validate it at least partially
(since the schema for 'device_add' is for now incomplete) against the
QMP schema.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Starting with QEMU-6.2 started accepting a JSON object as argument for
'-device' which will also become the only syntax considered stable by
qemu in the future.
Since libvirt was recently converted to generate the properties via JSON
to begin wit we can start using it on the commandline as well, by simply
enabling the QEMU_CAPS_DEVICE_JSON capability, which we do by probing
for the 'json-cli' feature flag of 'device_add'.
Normally a change which changes a commandline output should be happening
only after the impacted real-caps test files are forked in the version
preceding the change, but in this case it's not necessary as the logic
for generating the device properties stays identical and we just change
the output format (avoid conversion). Additionally we still have a lot
of tests validating the conversion to the old commandline options.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Use the 'allowIncomplete' argument of testQEMUSchemaValidateCommand to
validate at least properties which are already described by the schema.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The QMP schema for 'device_add' is not complete yet. Allow validation of
incomplete schema so that we can enable at least some validation. Once
there's more schema in the future all present members are still
validated.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Update to v6.1.0-1510-gc148a05721 which most notably adds the 'json-cli'
feature for 'device_add' QMP command meaning that -device accepts JSON.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
There are two variables (@vm and @domflags) in qemuConnectGetAllDomainStats()
that are used only within the for() loop but declared for entire function.
Bring them into the loop to make it obvious they are not used outside of it.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
Our general pattern is to initialize @ret to -1 and set it to 0
only at the end of a function. Some functions in
objecteventtest.c do not follow this pattern.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
The sole purpose of the lifecycleEventCounter_reset() function is
to zero out given lifecycleEventCounter struct. Well, we can let
the compiler zero it out when declaring a variable and just
remove the function.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
I think that virsh domstats problem on qemu < 5.2.0 is what users want
to find which version fixes.
Signed-off-by: Hiroki Narukawa <hnarukaw@yahoo-corp.jp>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
query-dirty-rate command is used for virsh domstats by default, but this
is available only on qemu >=5.2.0.
By this commit, qemu domain stats will check capabilities requirements before issuing actual query.
Signed-off-by: Hiroki Narukawa <hnarukaw@yahoo-corp.jp>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
One of qemuDomainGetStatsWorkers requires capabilities to run.
This commit adds capability information to qemuDomainGetStatsWorkers.
Signed-off-by: Hiroki Narukawa <hnarukaw@yahoo-corp.jp>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
query-dirty-rate command is used for virsh domstats by default, but this
is available only on qemu >=5.2.0.
In this commit, add capability flag for query-dirty-rate first.
Signed-off-by: Hiroki Narukawa <hnarukaw@yahoo-corp.jp>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
There have been countless reports from users concerned about the following
error reported by libvirtd when qemu domains are shutdown
internal error: End of file from qemu monitor
While the error is harmless, users often mistaken it for real problem with
their deployments. EOF from the monitor can't be entirely ignored since
other threads may be using the monitor and must be able to detect the EOF
condition.
One potential fix is to delay reporting EOF until the monitor is used
after EOF is detected. This patch adds a 'goteof' member to the
qemuMonitor structure, which is set when EOF is detected on the monitor
socket. If another thread later tries to send data on the monitor, the
EOF error is reported.
Signed-off-by: Jim Fehlig <jfehlig@suse.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
There are few functions in virnetsocket.c where an object/memory
is freed by explicit call. Use g_autoptr()/g_autofree/VIR_AUTOCLOSE
to do that automatically.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The aim of virNetSocketNewConnectCommand() is to execute passed
command and attach socket pair/pipe to it so that client socket
can be opened (this is used for connections with alternative
transports, e.g. ssh). The virCommand is created in a caller and
then passed to virNetSocketNewConnectCommand() where it is freed
using virCommandFree(). This approach is wrong on two levels:
1) The deallocation happens on a different level than allocation,
2) There's a WIN32 stub that just reports an error and doesn't
free the command.
However, with g_autoptr() trickery the command can be freed in
caller.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Commit ad209e7d adds QEMU_CAPS_VIRTIO_BLK_QUEUE_SIZE capability, but
the following commit 2d6d67e1 missed to use it and uses
QEMU_CAPS_VIRTIO_BLK_NUM_QUEUES instead.
This commit fixes the mistake.
Signed-off-by: Hiroki Narukawa <hnarukaw@yahoo-corp.jp>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The capability reflects whether QEMU is capable of -device
virtio-*,ats=. Since the property was introduced in QEMU commit
v2.9.0-rc0~162^2~32 we can safely assume the property is always
present as the minimal version required is 2.11.0.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>