Commit Graph

44025 Commits

Author SHA1 Message Date
Michal Privoznik
6bcd263011 virDomainObjListAdd: Transfer definition ownership
Upon successful return from virDomainObjListAdd() the
virDomainObj is the owner of secret definition. To make this
ownership transfer even more visible, lets pass the definition as
a double pointer and use g_steal_pointer().

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
2021-11-24 13:12:20 +01:00
Michal Privoznik
900fb1a315 virStoragePoolObjListAdd: Transfer definition ownership
Upon successful return from virStoragePoolObjListAdd() the
virStoragePoolObj is the owner of secret definition. To make this
ownership transfer even more visible, lets pass the definition as
a double pointer and use g_steal_pointer().

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
2021-11-24 13:12:20 +01:00
Michal Privoznik
8196a213b4 virSecretObjListAdd: Transfer definition ownership
Upon successful return from virSecretObjListAdd() the
virSecretObj is the owner of secret definition. To make this
ownership transfer even more visible, lets pass the definition as
a double pointer and use g_steal_pointer().

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
2021-11-24 13:12:20 +01:00
Michal Privoznik
10c68f5dd4 virInterfaceObjListAssignDef: Transfer definition ownership
Upon successful return from virInterfaceObjListAssignDef() the
virInterfaceObj is the owner of secret definition. To make this
ownership transfer even more visible, lets pass the definition as
a double pointer and use g_steal_pointer().

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
2021-11-24 13:12:20 +01:00
Daniel P. Berrangé
e18fff6c85 util: fix cache invalidation of swtpm capabilities
The check for whether the swtpm binary was modified is checking pointers
to the mtime field in two distinct structs, so will always compare
different. This resulted in re-probing swtpm capabilities every time,
as many as 20 times for a single VM launch.

Fixes:

  commit 01cf7a1bb9
  Author: Stefan Berger <stefanb@us.ibm.com>
  Date:   Thu Jul 25 14:22:04 2019 -0400

    tpm: Check whether previously found executables were updated

Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-11-24 11:31:16 +00:00
Daniel P. Berrangé
4f08dfa5f4 ci: display installed packages at start of build
When a build fails it is helpful to know what packages were installed,
because by the time we look at the build job output, the original
container image might have changed.

Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-11-24 11:17:10 +00:00
Xu Chao
6fac961b08 util: virExec may blocked by reading pipe if grandchild prematurely exit
When VIR_EXEC_DAEMON is set, if virPidFileAcquirePath/virSetInherit failed,
then pipesync[0] can not be closed when granchild process exit, because
pipesync[1] still opened in child process. and then saferead in child
process may blocked forever, and left grandchild process in defunct state.

Signed-off-by: Xu Chao <xu.chao6@zte.com.cn>
Signed-off-by: Yi Wang <wang.yi59@zte.com.cn>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-11-24 11:59:50 +01:00
Peter Krempa
c1a85daf99 util: xml: Remove virXMLPropStringLimit and virXPathStringLimit
The functions have very difficult semantics where callers are not able
to tell whether the property is missing or failed the length check. Only
the latter produces errors.

Since usage of the functions was phased out, remove them completely to
avoid further broken code.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:39 +01:00
Peter Krempa
01ab6513bd virSecurityLabelDefParseXML: Don't use virXMLPropStringLimit
The function produces an error which is ignored in this code path.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:39 +01:00
Peter Krempa
f3a8f26339 virSecurityDeviceLabelDefParseXML: Don't use 'virXPathStringLimit'
virXPathStringLimit doesn't give callers a way to differentiate between
the queried XPath being empty and the length limit being exceeded.

This means that the callers is completely ignoring the error.

Move the length check into the caller.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:39 +01:00
Peter Krempa
33f2cc0712 virSecurityDeviceLabelDefParseXML: Use automatic memory clearing for temp strings
Apart from code simplification the refactor of 'model' fixes an unlikely
memory leak of the string if a duplicate model is found.

While the coversion of 'label' variable may seem unnecessary it will
come in handy in the next patch.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:39 +01:00
Peter Krempa
a0e84f21b2 virSecurityLabelDefParseXML: Don't use 'virXPathStringLimit'
virXPathStringLimit doesn't give callers a way to differentiate between
the queried XPath being empty and the length limit being exceeded.

This means that callers are either overwriting the error message or
ignoring it altogether.

Move the length checks into the caller.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:39 +01:00
Peter Krempa
8f9bc6e5f6 virNodeDeviceCapVPDParseCustomFields: Don't use 'virXPathStringLimit'
virXPathStringLimit doesn't give callers a way to differentiate between
the queried XPath being empty and the length limit being exceeded.

This means that callers are overwriting the error message.

Move the length checks into the caller.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:39 +01:00
Peter Krempa
712a04bca1 virSecurityLabelDefParseXML: Remove pointless 'error' label
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:39 +01:00
Peter Krempa
1e67130b63 virSecurityLabelDefParseXML: Use automatic freeing for 'seclabel'
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:39 +01:00
Peter Krempa
0cb3e162a6 virSecurityLabelDefParseXML: Don't reuse temporary string 'p'
Use separate variables for 'model' and 'relabel' properties.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:38 +01:00
Peter Krempa
b63c70810c virSecurityLabelDefParseXML: Directly assign strings into appropriate variables
'seclabel->label', 'seclabel->imagelabel' and 'seclabel->baselabel' are
populated by stealing the pointer from the 'p' temporary string. Remove
the extra step.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:38 +01:00
Peter Krempa
f7ff8556ad virSecurityLabelDef: Declare 'type' as 'virDomainSeclabelType'
Use the appropriate enum type instead of an int and fix the XML parser
and one missing fully populated switch.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:38 +01:00
Peter Krempa
396ce0b568 util: seclabel: Define autoptr cleanup func for virSecurityLabelDef and virSecurityDeviceLabelDef
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-24 09:20:38 +01:00
Roman Bogorodskiy
537c8e5e24 meson: fix cpuset_getaffinity() detection
The cpuset_getaffinity() function is checked in sys/cpuset.h to see if
BSD CPU affinity APIs are available. This check requires including
sys/param.h to work properly, otherwise the test program fails with
unrelated errors like:

/usr/include/sys/cpuset.h:155:1: error: unknown type name
'__BEGIN_DECLS'
__BEGIN_DECLS
^
/usr/include/sys/cpuset.h:156:12: error: unknown type name 'cpusetid_t';
did you mean 'cpuset_t'?
int     cpuset(cpusetid_t *);

and so forth.

Signed-off-by: Roman Bogorodskiy <bogorodskiy@gmail.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-11-24 10:43:28 +04:00
Kristina Hanicova
fa7023f4eb qemu: Remove unnecessary variables and labels
This patch removes variables such as 'ret', 'rc' and others which
are easily replaced. Therefore, making the code look cleaner and
easier to understand.

Signed-off-by: Kristina Hanicova <khanicov@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-11-23 18:14:01 +01:00
Daniel P. Berrangé
c7edcb320b ci: run a mingw64 job on stable Fedora
Both of the current mingw jobs are marked as 'allow_failure' because
they are running against Fedora rawhide which is an unstable distro.

We need at least one mingw job to be gating to more reliably detect
problems.

This introduces dockerfiles for both mingw variants on Fedora 35
and sets the mingw64 build to run on Fedora 34, and mingw32 on
Fedora rawhide.

Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-11-23 16:44:18 +00:00
Daniel P. Berrangé
3ca8611455 ci: refresh variables/dockerfiles with latest content
- The Cirrus CI variables are now sorted
  - The dockerfiles update commands changed for some distros
  - Meson in CentOS is now new enough to use

Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-11-23 16:44:15 +00:00
Daniel P. Berrangé
4e2dcfcb82 ci: replace Fedora 33 with Fedora 35
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-11-23 16:43:57 +00:00
Ján Tomko
05cd957836 tests: pcivpdtest: check return value of virCreateAnonymousFile
Fixes: 59c1bc3a0e
Fixes: 43820e4b80
Fixes: 600f580d62
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Kristína Hanicová <khanicov@redhat.com>
2021-11-23 16:59:36 +01:00
Ján Tomko
096412f1ba ch: fix logic in virCHMonitorBuildPtyJson
There is a leftover 'ptys' variable, which we only assign
to and one assignment to 'content', where we add an empty
'pty' object.

Remove 'ptys'.

Fixes: 93accefd9e
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Kristína Hanicová <khanicov@redhat.com>
2021-11-23 16:59:36 +01:00
Ján Tomko
f5dd918978 vbox: fix vboxCapsInit
There is a stray mis-indented 'return NULL' left after a recent
refactor.

Fixes: c18d9e23fa
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Kristína Hanicová <khanicov@redhat.com>
2021-11-23 16:59:36 +01:00
Ján Tomko
097d3a4f56 tools: virt-host-validate: fix memory leak
virHostValidateGetCPUFlags returns an allocated virBitmap and
it needs to be freed.

Fixes: a0ec7165e3
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Kristína Hanicová <khanicov@redhat.com>
2021-11-23 16:59:36 +01:00
Martin Kletzander
edd1fd8ca9 Use virProcessGetStat
This eliminates one incorrect parsing implementation which relied on the
command field not having a closing bracket.  This possibility is already
tested against in the virProcessGetStat() tests.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 16:43:08 +01:00
Martin Kletzander
e370d4056b util: Add virProcessGetStat
This reads and separates all fields from /proc/<pid>/stat or
/proc/<pid>/task/<tid>/stat as there are easy mistakes to be done in the
implementation.  Some tests are added to show it works correctly.  No number
parsing is done as it would be unused for most of the fields most, if not all,
of the time.  No struct is used for the result as the length can vary (new
fields can be added in the future).

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 16:43:08 +01:00
Pavel Hrdina
2fe7622d19 virsh: man: update snapshot-revert description
We've changed the behavior of this API that from now on it will always
restart the VM process and we are no longer able to revert to snapshots
created by libvirt older then 0.9.5.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 16:03:19 +01:00
Pavel Hrdina
4b3c0d1aba qemu_monitor: remove unused load snapshot code
Recent cleanup of snapshot revert code made these function unused.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 16:03:07 +01:00
Koichi Murase
097296c30b bash-completion: fix variable leaks of "IFS" and "word"
Signed-off-by: Koichi Murase <myoga.murase@gmail.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-11-23 13:35:36 +01:00
Martin Kletzander
00c0ba5de3 util: Check for pkttyagent availability properly
It does not need a tty to work, it opens its controlling terminal for user
interaction and with this patch even crazy things like this work:

  echo 'list --name' | virsh -q >/dev/null

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 12:51:09 +01:00
Martin Kletzander
32eae6fd31 util: Report errors in all code paths in virPolkitAgentCreate
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 12:51:09 +01:00
Martin Kletzander
0ac13b189f virsh: Do not try connecting first time without polkit agent
Trying to connect once without a polkit agent will generate an error on the
server side which seems too rough given it only serves the purpose of the client
(virsh in this case) to figure out that an agent is needed.  Thankfully we can
just try running the agent.  It does not break anything as we are running it
with `--fallback`, which makes sure it does not replace an existing agent in
case there is one already registered.

The second piece of code trying to start the polkit text agent is kept in order
to _really_ try out starting the agent (and error out when failing to do so)
just in case the agent was not available the first time it was ran.  Even though
it should not happen it avoids a very rare race condition and really does not
add much complexity.

Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1945501

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 12:51:09 +01:00
Martin Kletzander
32d100ca5c util: Add virPolkitAgentAvailable
With this function we can decide whether to try running the polkit text agent
only if it is available, removing a potential needless error saying that the
agent binary does not exist, which is useful especially when running the agent
before knowing whether it is going to be needed.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 12:51:09 +01:00
Martin Kletzander
e4ebe391d7 util: Tiny reword fix in comment
Automatic "Ptr " -> " *" also wreaked havoc in comments.  Fix it and while at it
reword the sentence so it is clear that the object is newly allocated.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 12:51:09 +01:00
Martin Kletzander
e20ad73780 virsh: Remove needless variable
It only redundantly reflects whether pkagent != NULL.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-11-23 12:51:09 +01:00
Roman Bogorodskiy
c07cf0a686 meson: improve CPU affinity routines check
Recently, FreeBSD has got sched_get/setaffinity(3) implementations and
the sched.h header as well [1]. To make these routines visible,
users have to define _WITH_CPU_SET_T.

This breaks current detection. Specifically, meson sees the
sched_getaffinity() symbol and defines WITH_SCHED_GETAFFINITY. This
define unlocks Linux implementation of virProcessSetAffinity() and other
functions, which fails to build on FreeBSD because cpu_set_t is not
visible as _WITH_CPU_SET_T is not defined.

For now, change detection to the following:

 - Instead of checking sched_getaffinity(), check if 'cpu_set_t' is
   available through sched.h
 - Explicitly check the sched.h header instead of assuming its presence
   if WITH_SCHED_SETSCHEDULER is defined

1:
https://cgit.freebsd.org/src/commit/?id=43736b71dd051212d5c55be9fa21c45993017fbb
https://cgit.freebsd.org/src/commit/?id=160b4b922b6021848b6b48afc894d16b879b7af2
https://cgit.freebsd.org/src/commit/?id=90fa9705d5cd29cf11c5dc7319299788dec2546a

Signed-off-by: Roman Bogorodskiy <bogorodskiy@gmail.com>
Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
2021-11-23 15:11:00 +04:00
Pavel Hrdina
9b0a6d959b domain_snapshot: update virDomainRevertToSnapshot description
We've changed the behavior of this API that from now on it will always
restart the VM process and we are no longer able to revert to snapshots
created by libvirt older then 0.9.5.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:42:07 +01:00
Pavel Hrdina
5956eddf9f test: snapshot revert: drop error the VM must be restarted
This will always happen so there is no need to error out and require
usage of FORCE flag.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:42:05 +01:00
Pavel Hrdina
a4942cfa3a test: snapshot revert: fix emitting events
Now that we always emulate restarting the VM process events are emitted
differently so we need to update the code and the comment as well.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:42:03 +01:00
Pavel Hrdina
74ea062be6 test: snapshot revert: drop unused code
Now that we always emulate VM process stop we can drop the unused code
and simply the logic.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:42:00 +01:00
Pavel Hrdina
5bed5db990 test: snapshot revert: always emulate VM process stop
Reflect the same change in test driver as in QEMU driver because the
compatibility check code isn't perfect.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:41:57 +01:00
Pavel Hrdina
2ef6b1cecf test: snapshot revert: properly emulate starting CPUs
When active snapshot is reverted we stop CPUs in order to load the
snapshot but we never start the CPUs again.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:41:55 +01:00
Pavel Hrdina
ca04290be3 test: snapshot revert: always error out if VM XML is missing
We should have this check even if FORCE flag is used because later we
unconditionally copy the `snap->def->dom` and error out if there is no
copy created. The test driver will always save the VM XML when creating
new snapshot.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:41:52 +01:00
Pavel Hrdina
6c86b92f14 qemu_snapshot: revert: drop error that QEMU process must be restarted
This will always happen so there is no need to error out and require
usage of FORCE flag.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:41:50 +01:00
Pavel Hrdina
d8a4acab96 qemu_snapshot: revert: fix emitting events
Now that we always restart the QEMU process events are emitted
differently so we need to update the code and the comment as well.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:41:48 +01:00
Pavel Hrdina
f33ce12e9c qemu_snapshot: revert: drop unused loadvm code
Now that we always restart QEMU process the loadvm code is unused and
can be dropped.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-11-23 11:41:45 +01:00