Loading
Commits on Source 81
-
Bastian Germann authored
On result = null, qemuAgentCommand will throw a NullPointerException. Prevent this by constructing a CString from it. Signed-off-by:Bastian Germann <bage@linutronix.de>
-
Daniel P. Berrangé authoredLibvirt changed from autotools to meson. All the containers need refreshing and the CI recipes updated. Signed-off-by:Daniel P. Berrangé <berrange@redhat.com>
-
Daniel P. Berrangé authored
This brings the repo into alignment with the main libvirt.git practice. Signed-off-by: -
Daniel P. Berrangé authored
This introduces Fedora 33 and removes some redundant packages. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
This makes the dockerfile name match the output container name Signed-off-by: -
Daniel P. Berrangé authored
This is primarily to fix the rename of the CentOS PowerTools repo. Signed-off-by: -
Erik Skultety authored
Changes contained in this refresh: - CentOS 7 is pulled from docker.io registry - PowerTools repo name regression fix for CentOS Stream - enhancement for the workaround fix for JDK installation fails on Debian 10 '-slim' containers Signed-off-by:Erik Skultety <eskultet@redhat.com>
-
Erik Skultety authored
Signed-off-by: -
Erik Skultety authored
Leap 15.1 reaches EOL on January 31st, 2021. Signed-off-by: -
SangHakLee authored
This make get suitable domain's xml for migration. Signed-off-by:SangHakLee <code.ryan.lee@gmail.com>
-
Daniel P. Berrangé authored
The current docker:dind container has broken default seccomp filter that results in clone3 being blocked, which in turn breaks Fedora 35 rawhide. This custom image has a workaround that causes the seccomp filter to return ENOSYS for clone3 instad of EPERM, thus triggering glibc to fallback to clone correctly. Signed-off-by: -
Daniel P. Berrangé authored
It is no longer possible to just install an extra package to run regular centos into centos stream. Signed-off-by: -
Daniel P. Berrangé authored
This platform is no longer supported by libvirt. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
This uses the command "lcitool manifest ci/manifest.yml" to re-generate all existing dockerfiles and gitlab CI config. Signed-off-by: -
Daniel P. Berrangé authored
We now use lcitool's manifest feature to generate files. Signed-off-by: -
Magnus Johansson authored
Signed-off-by:Magnus Johansson <magnus@glesys.se>
-
Signed-off-by: -
Magnus Johansson authored
Signed-off-by: -
Magnus Johansson authored
Signed-off-by: -
Magnus Johansson authored
Signed-off-by: -
Signed-off-by: -
Lars Dunemark authored
Signed-off-by:Lars Dunemark <lars.dunemark@glesys.se>
-
Signed-off-by: -
Signed-off-by: -
Lars Dunemark authored
Updating DomainEventID with new event id and adding registration of callback for Agent Lifecycle events and remove function for same callbacks. Signed-off-by: -
Daniel P. Berrangé authored
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: -
Peter Conrad authored
Signed-off-by:Peter Conrad <conrad@quisquis.de>
-
Signed-off-by: -
Marcus Sorensen authored
Signed-off-by:Marcus Sorensen <mls@apple.com>
-
-
Peter Conrad authored
Signed-off-by: -
Daniel P. Berrangé authored
These errors were causing javadoc to throw an exception when processing the docs. Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: -
Daniel P. Berrangé authored
Changed distros - centos-8 -> almalinux-8 - fedora-33 -> fedora-35 - fedora-34 -> fedora-36 - ubuntu-1804 -> ubuntu-2204 - opensuse-leap-152 -> opensuse-leap-153 Signed-off-by: -
Dmytro Klymenchuk authored
Signed-off-by:Dmytro Klymenchuk <d-kay23@ukr.net>
-
Dmytro Klymenchuk authored
Signed-off-by:Dmitry Klymenchuk <dmytro.klymenchuk@virtuozzo.com>
-
Daniel P. Berrangé authored
This refresh switches the CI for contributors to be triggered by merge requests. Pushing to a branch in a fork will no longer run CI pipelines, in order to avoid consuming CI minutes. To regain the original behaviour contributors can opt-in to a pipeline on push git push <remote> -o ci.variable=RUN_PIPELINE=1 This variable can also be set globally on the repository, though this is not recommended. Upstream repo pushes to branches will run CI. The use of containers has changed in this update, with only the upstream repo creating containers, in order to avoid consuming contributors' limited storage quotas. A fork with existing container images may delete them. Containers will be rebuilt upstream when pushing commits with CI changes to the default branch. Any other scenario with CI changes will simply install build pre-requisite packages in a throaway environment, using the ci/buildenv/ scripts. These scripts may also be used on a contributor's local machines. With pipelines triggered by merge requests, it is also now possible to workaround the inability of contributors to run pipelines if they have run out of CI quota. A project member can trigger a pipeline from the merge request, which will run in context of upstream, however, note this should only be done after reviewing the code for any malicious CI changes. Signed-off-by: -
Lars Dunemark authored
When deleting a DomainSnapshot the DomainSnapshotPointer was set to null preventing the free() method to release the underlaying resorces resulting in dangling connections to libvirtd. Signed-off-by: -
Peter Krempa authoredReplace 'debian-10' with 'debian-11' and 'fedora-35' with 'fedora-37'. Signed-off-by:Peter Krempa <pkrempa@redhat.com>
-
Peter Krempa authored
Use 'download.libvirt.org' instead of 'www.libvirt.org' build.xml and also update '/data/www/maven2' to the new location '/data/ftp/libvirt/maven2'. Signed-off-by: -
Peter Krempa authored
Signed-off-by: -
Peter Krempa authored
Use the new URI 'download.libvirt.org' to refer to the source. Signed-off-by: -
https://libvirt.orgPeter Krempa authored
Drop the 'www' subdomain from links to libvirt.org. In all updated cases use https as protocol. Signed-off-by: -
Peter Krempa authored
Convert the rest of the links which were not using 'https'. Signed-off-by: -
Peter Krempa authored
The 'javadoc' which was originally hosted on libvirt.org is hopelessly outdated as nobody was updating it. Make sure it's up to date by generating it in gitlab. Signed-off-by: -
Andrea Bolognani authoredWhat's currently there makes for a barely functional Debian package, as it's missing a lot of the expected information such as copyright and changelog. Plus it contains hardcoded version information that are several years out of date. Even if this wasn't the case, there is a strong recommendation against attempting to maintain the Debian packaging for a project as part of the upstream repository. Signed-off-by:Andrea Bolognani <abologna@redhat.com>
-
Andrea Bolognani authored
libvirt-java has recently gained its own dedicated URL, but as of today we only use that to host the javadoc API reference under a subdirectory and navigating to the top-level leads to a GitLab 404 page. Prepare a very basic website for the project. The contents are the same as the current https://libvirt.org/java.html page, just reorganized a bit; the HTML, CSS and graphics are taken from the libvirt-php website. Signed-off-by:
-
Andrea Bolognani authored
Signed-off-by: -
Erik Skultety authored
Each respective project that lcitool knows about and currently maintains its list of package dependencies knows best what packages they actually depend on. If a new dependency is currently needed, first a change in lcitool is necessary before GitLab jobs and containers can be updated. Provided a mapping already exists in lcitool (which can quickly be added as an override via mappings.yml temporarily) we speed up the whole CI update process by one step. Note that starting this commit lcitool must be invoked as '$ lcitool -d/--data-dir ci/lcitool ...' to pick up the project dependency list correctly. Signed-off-by: -
-
Andrea Bolognani authored
The target list is updated to match what is currently supported by lcitool, notably bumping Fedora to 38 and 39. Signed-off-by: -
Andrea Bolognani authored
Recent versions of lcitool no longer expect this distinction. Signed-off-by: -
Andrea Bolognani authored
Currently the 'jobs' key is overwritten, so only the git job is produced. While at it, stop using 'template' for the default job, since the value we're using matches the default. Signed-off-by: -
Andrea Bolognani authored
Signed-off-by: -
Andrea Bolognani authored
Signed-off-by: -
Andrea Bolognani authored
We want the pages building job to run for MRs, to ensure that no breakages sneak in, but updated pages should only be published once the changes have been merged into master. Signed-off-by: -
Fabricio Duarte authored
Signed-off-by:Fabricio Duarte <fabricio.duarte.jr@gmail.com>
-
Daniel P. Berrangé authored
Notable changes * AlmaLinux 8 / CentOS Stream 8 replaced with 9 * Fedora 38 dropped, 40 added * Ubuntu 20.04 dropped, 24.04 added * Debian 11 dropped, 12 added Signed-off-by: -
Takashi Kajinami authored
... so that additional flags can be passed to Domain.migrateToURI. Fixes #8 Signed-off-by:
Takashi Kajinami <kajinamit@oss.nttdata.com>
-
Signed-off-by: -
This exposes virDomainGetCPUStats API so that cpu stats, along with memory stats, can be obtained. Signed-off-by: -
Signed-off-by: -
Mitsuru Kariya authored
free was missing for the pointer returned in domains by virConnectListAllDomains, so this has been fixed. cf. https://libvirt.org/html/libvirt-libvirt-domain.html#virConnectListAllDomains Signed-off-by:
Mitsuru Kariya <mitsuru.kariya@nttdata.com>
-
Francisco Manuel García Botella authoredFix a typo in Javadoc. When running autobuild.sh it raised a warning indicating missing final '>' in @See tag. Signed-off-by:
Francisco Manuel Garcia Botella <fmgb92@gmail.com>
-
Mitsuru Kariya authored
Although virConnCopyLastError and virCopyLastError are provided, using them requires freeing the string memory area with virResetError. Therefore, add virResetError. Also fix mistakes in the Javadoc "missig api in 0.1.0" list: - virConnCopyLastError is provided - virFreeError is not missing in 0.1.0 (it is listed as missing in 0.6.1) Signed-off-by: -
Takashi Kajinami authored
Expose virtDomainGetJobStats . This is useful especially when we want to query jobs actually running. Co-Authored-By:
-
Mitsuru Kariya authored
Fix the following issues in Domain.getJobStats: - Fix an exception that occurred when the virTypedParameter[] had zero elements. virDomainGetJobStats returns successfully with *params == NULL and *nparams == 0 when *type == VIR_DOMAIN_JOB_NONE. However, calling Structure.toArray(n) with n == 0 causes an ArrayIndexOutOfBoundsException. This is avoided by directly returning an empty array when *nparams == 0. - Fix the issue where the first element of virTypedParameter[] was not being read. The Structure constructor does not automatically read fields. This is fixed by explicitly calling the virTypedParameter.read() method for the first element. Note: Structure.toArray(n) automatically reads fields (i.e., calls the virTypedParameter.read() method) for all elements except the first. - Fix the issue where the value was not being read when virTypedParameter was a String. Union does not perform pointer-to-value reads (such as char* -> String) unless it is the active element. This is fixed by explicitly reading the String when virTypedParameter.type == VIR_TYPED_STRING. - Fix a potential crash in virTypedParamsFree when handling String parameters. When passing a Structure received from a native function as an argument to another native function: - String fields directly embedded in the Structure are not updated in native memory if their values haven't changed. - However, Unions embedded in the Structure always update native memory for the active element, even if the value hasn't changed. Therefore, if the active element of the Union is a String, JNA allocates new memory for the string and overwrites the received char* pointer. This not only causes the original char* pointer to be leaked, but also leads to double-free issues: once by virTypedParamsFree and again by GC on the JNA-allocated memory. This is avoided by changing the argument type of virTypedParamsFree from virTypedParameter[] to Pointer. Additionally, extract the conversion from a pointer to a virTypedParameter array to a TypedParameter array into a static method of TypedParameter, as there are other functions that also receive virTypedParameter arrays. Signed-off-by: -
The 5th argument `flags` of `virDomainMigrate3` is `unsigned int`, not `unsigned long`. Therefore, the Java binding should use `int` instead of `NativeLong`. To maintain backward compatibility, the version taking `NativeLong` as an argument has been deprecated, and a new version taking `int` has been added. Although not fully verified, it is likely that no practical issue occurs on supported platforms because: - On 32-bit platforms, `long` and `int` are the same size. - On 64-bit platforms, the 5th argument is passed via a register. Signed-off-by: -
Created the bindings to manage the domain checkpoints of a domain. Added tests to check the functionality of these bindings. Signed-off-by: -
Added tests to these bindings. Signed-off-by: -
Pavel Hrdina authoredRemoved old OSes: - Fedora 39, Fedora 40 Added new OSes: - AlmaLinux 10 - CentOS Stream 10 - Debian 13 - Fedora 42, Fedora 43 Signed-off-by:Pavel Hrdina <phrdina@redhat.com>
-
Takashi Kajinami authored
This parameter is useful to save disk space during volume data copy in live migration with disk data copy. Signed-off-by: -
... so that all of the flags currently available in libvirt can be defined more easily. Signed-off-by: -
The unit of memory size is not kilobytes (blocks of 1000) but kibibytes (blocks of 1024). Update the documentation to clarify this. Signed-off-by: -
Mitsuru Kariya authored
virDomainGetMetadata returns a dynamically allocated char* that the caller must free. The JNA binding incorrectly mapped it as String, which reads the value but never frees the native memory. Changed to CString, which automatically frees the native memory on toString(), consistent with other similar bindings (e.g. virDomainGetXMLDesc, virDomainGetOSType). Signed-off-by: -
The C struct defines state as unsigned char (1 byte), but the JNA mapping used int (4 bytes). This causes two problems: - On big-endian systems, the 1-byte state value is stored at the beginning of the field, but reading it as a 4-byte int interprets it as the most significant byte, producing a completely wrong value (e.g. state=1 would be read as 0x01000000). - On little-endian systems, the state byte is correctly placed in the least significant position, but the remaining 3 padding bytes are also read as part of the int. These padding bytes may contain garbage, corrupting the value. In practice this was masked by libvirt zero-initializing the struct, but it is not guaranteed by the C specification. Changed to byte to match the C type, and added a bitmask (& 0xFF) when converting to the DomainState enum index to handle the unsigned-to-signed conversion correctly. Signed-off-by: -
Both blockPeek and memoryPeek split large requests into 64K chunks to work around remote driver limitations, but neither method incremented the offset between iterations. This caused every chunk to read from the same starting position, resulting in corrupted data for any request larger than 64KB. Signed-off-by: -
C unsigned long is platform-dependent (4 bytes on 32-bit, 8 bytes on 64-bit), but Java long is always 8 bytes. This mismatch causes incorrect struct layout and calling convention issues on 32-bit systems. - virDomainBlockJobInfo.bandwidth: long -> NativeLong (fixes struct layout so cur/end fields are read at correct offsets) - virDomainBlockCommit bandwidth parameter: long -> NativeLong (fixes ABI mismatch in function call) Signed-off-by: -
Daniel P. Berrangé authored
Signed-off-by: