April 2026 - Devel - libvirt List Archives

[PATCH v2 0/6] util: log infor to help troubleshoot file/socket open failures
by Laine Stump 09 Apr '26

09 Apr '26

The first couple patches are just related cleanups I noticed while doing the others, then a couple patches to simplify adding more lines to the bannder printed at the top of the log, and then finally the last couple patches add some info to the banner at the top of the log that would be generally useful when trying to understand why some random file or socket failed to be created or opened. V2: I've fixed a few nits Peter found, and modified the format and contents of the banner based on suggestions from Peter and Daniel Laine Stump (6): util: log the name of the log directory that couldn't be created util: consistently use typedef virLogMetadata util: eliminate duplicate code in virLogVMessage util: make it easier to add lines to the log "init banner" util: add info to log banner about uid and user environment util: add info about g_get_user_*_dir directories to log banner src/util/virlog.c | 226 ++++++++++++++++++++++++++++++---------------- src/util/virlog.h | 4 +- tests/testutils.c | 2 +- 3 files changed, 150 insertions(+), 82 deletions(-) -- 2.53.0

2 11

[PATCH 0/6] util: log infor to help troubleshoot file/socket open failures
by Laine Stump 09 Apr '26

09 Apr '26

The first couple patches are just related cleanups I noticed while doing the others, then a couple patches to simplify adding more lines to the bannder printed at the top of the log, and then finally the last couple patches add some info to the banner at the top of the log that would be generally useful when trying to understand why some random file or socket failed to be created or opened. Ideas on what else to add (or what to remove, or how to better format it) are welcome! Laine Stump (6): util: log the name of the log directory that couldn't be created util: consistently use typedef virLogMetadata util: eliminate duplicate code in virLogVMessage util: make it easier to add lines to the log "init banner" util: add info to log banner about uid and user environment util: add info about g_get_user_*_dir directories to log banner src/util/virlog.c | 231 ++++++++++++++++++++++++++++++---------------- src/util/virlog.h | 4 +- tests/testutils.c | 2 +- 3 files changed, 155 insertions(+), 82 deletions(-) -- 2.53.0

4 20

[PATCH 0/2] util: Move udev settle wait to vNIC create functions
by Michal Privoznik 09 Apr '26

09 Apr '26

See 1/2 for in depth explanation. Michal Prívozník (2): util: Wait for udev to settle after creating vNIC virnetdevmacvlan: Drop udev busy loop from virNetDevMacVLanTapOpen() src/util/virnetdevmacvlan.c | 24 ++++++++++-------------- src/util/virnetdevtap.c | 6 ++++++ src/util/virnetdevveth.c | 4 ++++ 3 files changed, 20 insertions(+), 14 deletions(-) -- 2.52.0

4 7

[PATCH 00/16] qemu: reduce direct usage of cfg->user
by Cole Robinson 08 Apr '26

08 Apr '26

In the qemu driver, cfg->user and cfg->group are `user` and `group` from qemu.conf, the default qemu process user and owner for the security DAC driver. These are driver wide default configuration. They are used throughout driver code for storage access and launching sub processes. However code using cfg->user and cfg->group directly is usually subtly wrong, because these values can be overridden for individual VMs with XML like: <domain> <seclabel type='static' model='dac' relabel='yes'> <label>+1234:+4567</label> </seclabel> </domain> Most of the qemu storage code already gets this right (where these values can also be overridden at the <disk> level). This series fixes many other users by reusing the existing qemuDomainGetImageIds function used by the storage code. https://redhat.atlassian.net/browse/RHEL-119651 Cole Robinson (16): qemu: require non-NULL 'cfg' in qemuDomainGetImageIds() qemu: tweak qemuDomainGetImageIds() argument handling qemu: saveimage: use qemuDomainGetImageIds() qemu: domain taint: use qemuDomainGetImageIds() qemu: migrate to file: use qemuDomainGetImageIds() qemu: core dump: use qemuDomainGetImageIds() qemu: storage stat: use qemuDomainGetImageIds() qemu: tpm: use qemuDomainGetImageIds() qemu: block copy: use qemuDomainGetImageIds() qemu: cpu compare: use qemuDomainGetImageIds() qemu: nvram: use qemuDomainGetImageIds() qemu: simplify qemuProcessPrepareDomainStorage() args qemu: nbdkit: use qemuDomainGetImageIds() qemu: remove qemuDomainOpenFile() TODO comment qemu: don't check virParseOwnershipIds() error qemu: use qemuDomainGetImageIds() in qemuDomainOpenFile() src/qemu/qemu_backup.c | 12 +-- src/qemu/qemu_domain.c | 150 +++++++++++++++++++------------------- src/qemu/qemu_domain.h | 14 ++-- src/qemu/qemu_driver.c | 25 +++++-- src/qemu/qemu_hotplug.c | 6 +- src/qemu/qemu_migration.c | 5 +- src/qemu/qemu_process.c | 20 ++--- src/qemu/qemu_saveimage.c | 5 +- src/qemu/qemu_snapshot.c | 6 +- src/qemu/qemu_tpm.c | 6 +- 10 files changed, 130 insertions(+), 119 deletions(-) -- 2.53.0

2 37

[PATCH 0/6] qemu: sync initial dir perms with RPM spec
by Cole Robinson 08 Apr '26

08 Apr '26

The qemu driver creates a bunch of directories at startup, example /var/lib/libvirt/qemu/*. The initial mode mask on these directories is almost always specified as 0777, which leaves it up to umask to lock things down. As a result, if running from git, directories are usually created 0755 which seems overly permissive for a system daemon. This doesn't have much effect from RPM users at least, since the spec file pre-creates most of these directories with more limited permissions. This series syncs the code to match what we already specify in the spec file. Code is simplified first, and some missing dirs are added to the spec at the end. Cole Robinson (6): qemu: driver: split out qemuStateInitializeDirs qemu: driver: streamline dir creation qemu: driver: don't chown() dirname(cfg->channelTargetDir) qemu: driver: sync dir creation permissions with RPM spec qemu: driver: adjust mode mask for rdpStateDir qemu: driver: adjust mode mask for channelTargetDir libvirt.spec.in | 2 + src/qemu/qemu_driver.c | 249 ++++++++++++----------------------------- 2 files changed, 74 insertions(+), 177 deletions(-) -- 2.53.0

2 12

FYI: possible mailman service disruption 2026/04/08
by Daniel P. Berrangé 08 Apr '26

08 Apr '26

The hosting of the libvirt mailman instance will be moving to a new location tomorrow on 2026/04/08, at around 0900-1000 UTC. While we don't expect any significant loss of service, there could be minor short term, temporary disruption for sending emails while DNS records update. With regards, Daniel -- |: https://berrange.com ~~ https://hachyderm.io/@berrange :| |: https://libvirt.org ~~ https://entangle-photo.org :| |: https://pixelfed.art/berrange ~~ https://fstop138.berrange.com :|

1 1

[PATCH v2 0/2] remote: Remove dependency of virt-secret-init-encryption from libvirtd
by Arun Menon 08 Apr '26

08 Apr '26

The monolithic libvirtd.service currently has a dependency on virt-secret-init-encryption.service. This causes libvirtd to fail to start on systems where the secret driver is not installed or enabled, as systemd cannot satisfy the Requires= unit or the LoadCredentialEncrypted= path. See below, Requires=virt-secret-init-encryption.service LoadCredentialEncrypted=secrets-encryption-key:@localstatedir@/lib/libvirt/secrets/secrets-encryption-key This series accomplishes the following: The first commit decouples the secrets encryption logic from the main libvirtd service file. It is moved into a new systemd drop-in (10-secret.conf) which is only installed when libvirt is built with secret driver support. The override snippet is added to the daemon-driver-secret package. The second commit is a simple refactoring configuration_data of virt-secret-init-encryption. No functional change is intended. Arun Menon (2): remote: Move secrets encryption dependency to a systemd drop-in secret: Update virt_secret_init_encryption_conf initialization libvirt.spec.in | 2 ++ src/remote/libvirtd.service.in | 4 ---- src/remote/meson.build | 15 +++++++++++++++ src/remote/secret.conf.in | 7 +++++++ src/secret/meson.build | 6 +++--- 5 files changed, 27 insertions(+), 7 deletions(-) create mode 100644 src/remote/secret.conf.in -- 2.53.0

3 5

[PATCH] qemu: virtiofs: make fs->sock guard more consistent
by Cole Robinson 08 Apr '26

08 Apr '26

When fs->sock is set, virtiofs is externally managed, and most of qemu_virtiofs.c should do nothing. Some qemu_virtiofs.c functions handle this case, but some require the caller to guard against it. Standardize on making this the responsibility of qemu_virtiofs.c Signed-off-by: Cole Robinson <crobinso(a)redhat.com> --- src/qemu/qemu_extdevice.c | 8 +++----- src/qemu/qemu_hotplug.c | 18 ++++++++---------- src/qemu/qemu_virtiofs.c | 9 +++++++++ 3 files changed, 20 insertions(+), 15 deletions(-) diff --git a/src/qemu/qemu_extdevice.c b/src/qemu/qemu_extdevice.c index 28cea52980..6e16c6be29 100644 --- a/src/qemu/qemu_extdevice.c +++ b/src/qemu/qemu_extdevice.c @@ -225,7 +225,7 @@ qemuExtDevicesStart(virQEMUDriver *driver, for (i = 0; i < def->nfss; i++) { virDomainFSDef *fs = def->fss[i]; - if (fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS && !fs->sock) { + if (fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS) { if (qemuVirtioFSStart(driver, vm, fs) < 0) return -1; } @@ -315,8 +315,7 @@ qemuExtDevicesStop(virQEMUDriver *driver, for (i = 0; i < def->nfss; i++) { virDomainFSDef *fs = def->fss[i]; - if (!fs->sock && - fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS) + if (fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS) qemuVirtioFSStop(driver, vm, fs); } @@ -473,8 +472,7 @@ qemuExtDevicesSetupCgroup(virQEMUDriver *driver, for (i = 0; i < def->nfss; i++) { virDomainFSDef *fs = def->fss[i]; - if (!fs->sock && - fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS && + if (fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS && qemuVirtioFSSetupCgroup(vm, fs, cgroup) < 0) return -1; } diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index 9439948089..deb02c20be 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -3361,17 +3361,15 @@ qemuDomainAttachFSDevice(virQEMUDriver *driver, if (!(devprops = qemuBuildVHostUserFsDevProps(fs, vm->def, charAlias, priv))) goto cleanup; - if (!fs->sock) { - if (qemuVirtioFSPrepareDomain(driver, fs) < 0) - goto cleanup; + if (qemuVirtioFSPrepareDomain(driver, fs) < 0) + goto cleanup; - if (qemuVirtioFSStart(driver, vm, fs) < 0) - goto cleanup; - started = true; + if (qemuVirtioFSStart(driver, vm, fs) < 0) + goto cleanup; + started = true; - if (qemuVirtioFSSetupCgroup(vm, fs, priv->cgroup) < 0) - goto cleanup; - } + if (qemuVirtioFSSetupCgroup(vm, fs, priv->cgroup) < 0) + goto cleanup; qemuDomainObjEnterMonitor(vm); @@ -5506,7 +5504,7 @@ qemuDomainRemoveFSDevice(virQEMUDriver *driver, if (rc < 0) return -1; - if (!fs->sock && fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS) + if (fs->fsdriver == VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS) qemuVirtioFSStop(driver, vm, fs); if ((idx = virDomainFSDefFind(vm->def, fs)) >= 0) diff --git a/src/qemu/qemu_virtiofs.c b/src/qemu/qemu_virtiofs.c index 91d58703f1..cbc3bc0bc2 100644 --- a/src/qemu/qemu_virtiofs.c +++ b/src/qemu/qemu_virtiofs.c @@ -228,6 +228,9 @@ qemuVirtioFSStart(virQEMUDriver *driver, g_autoptr(domainLogContext) logContext = NULL; int rc; + if (fs->sock) + return 0; + if (!virFileIsExecutable(fs->binary)) { virReportError(VIR_ERR_INTERNAL_ERROR, _("virtiofsd binary '%1$s' is not executable"), @@ -327,6 +330,9 @@ qemuVirtioFSStop(virQEMUDriver *driver G_GNUC_UNUSED, g_autofree char *pidfile = NULL; virErrorPtr orig_err; + if (fs->sock) + return; + virErrorPreserveLast(&orig_err); if (!(pidfile = qemuVirtioFSCreatePidFilename(vm, fs->info.alias))) @@ -361,6 +367,9 @@ qemuVirtioFSSetupCgroup(virDomainObj *vm, pid_t pid = -1; int rc; + if (fs->sock) + return 0; + if (!cgroup) return 0; -- 2.53.0

2 2

[PATCH 0/4] Fix memory leaks
by Pavel Hrdina 07 Apr '26

07 Apr '26

Pavel Hrdina (4): qemuBlockStorageSourceGetNFSProps: Fix memory handeling nodeDeviceEventSubmit: Fix memory leak virStorageVolPoolRefreshDataFree: Fix memory leak qemuDomainSetThrottleGroup: Fix memory leak src/node_device/node_device_udev.c | 3 ++- src/qemu/qemu_block.c | 4 ++-- src/qemu/qemu_driver.c | 2 ++ src/storage/storage_driver.c | 1 + 4 files changed, 7 insertions(+), 3 deletions(-) -- 2.53.0

2 8

[PATCH 0/2] qemu: allow changing link state on vhostuser interfaces
by Laine Stump 07 Apr '26

07 Apr '26

(and a 2nd incidental patch to log an error and fail if someone tries to change the <backend type>) Laine Stump (2): qemu: log error on attempts to change backend type of live network interface qemu: allow update-device of vhostuser network devices src/qemu/qemu_hotplug.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) -- 2.53.0

2 5