February 2025 - Devel - libvirt List Archives

[PATCH v2 00/12] hw/arm/raspi: Allow creating any Raspberry Pi machine
by Philippe Mathieu-Daudé 22 Feb '25

22 Feb '25

Full rewrite of v1 [1], addressing Zoltan & Peter suggestion. Introduce a generic 'raspi' machine, which takes a 'model' and 'revision' properties, and any memory size. The 'board_rev' register is filled appropriately. Before, merge raspi4b.c within raspi.c (more is planned here with the MPCore refactor [2]). Regards, Phil. [1] https://lore.kernel.org/qemu-devel/20250201091528.1177-1-philmd@linaro.org/ [2] https://lore.kernel.org/qemu-devel/20231212162935.42910-1-philmd@linaro.org/ Philippe Mathieu-Daudé (12): hw/arm/raspi: Access SoC parent object using BCM283X_BASE() macro hw/arm/raspi: Merge model 4B with other models hw/arm/raspi: Unify RASPI_MACHINE types hw/arm/raspi: Pass board_rev as argument to raspi_base_machine_init() hw/arm/raspi: Consider processor id in types[] array hw/arm/raspi: Consider network interface for B models hw/arm/raspi: Check ramsize is within chipset aperture hw/arm/raspi: Introduce generic Raspberry Pi machine hw/arm/raspi: Have the generic machine take a 'revision' property hw/arm/raspi: List models creatable by the generic 'raspi' machine hw/arm/raspi: Deprecate old raspiX machine names hw/arm/raspi: Support more models docs/about/deprecated.rst | 13 + include/hw/arm/raspi_platform.h | 5 +- hw/arm/raspi.c | 383 ++++++++++++++++++++++-- hw/arm/raspi4b.c | 136 --------- tests/qtest/bcm2835-dma-test.c | 2 +- tests/qtest/bcm2835-i2c-test.c | 2 +- tests/qtest/boot-serial-test.c | 3 +- hw/arm/meson.build | 2 +- tests/functional/test_aarch64_raspi3.py | 5 +- tests/functional/test_aarch64_raspi4.py | 4 +- tests/functional/test_arm_raspi2.py | 4 +- 11 files changed, 385 insertions(+), 174 deletions(-) delete mode 100644 hw/arm/raspi4b.c -- 2.47.1

5 29

[PATCH] qemu: forbid a few unsupported things for vhost-user/passt interfaces
by Laine Stump 21 Feb '25

21 Feb '25

passt in vhost-user mode doesn't support using multiple queues. The path of the socket is auto-generated by libvirt for vhost-user/passt; it can't be set by the user The passt end of a vhost-user socket is always the server, and the qemu end is always a client. Signed-off-by: Laine Stump <laine(a)redhat.com> --- src/qemu/qemu_validate.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/src/qemu/qemu_validate.c b/src/qemu/qemu_validate.c index 3e3e368da3..53affcdcff 100644 --- a/src/qemu/qemu_validate.c +++ b/src/qemu/qemu_validate.c @@ -1824,6 +1824,22 @@ qemuValidateDomainDeviceDefNetwork(const virDomainNetDef *net, net->backend.type == VIR_DOMAIN_NET_BACKEND_PASST) { if (qemuValidateDomainDefVhostUserRequireSharedMemory(def, "interface type=\"vhostuser\" backend type=\"passt\"") < 0) return -1; + + if (net->driver.virtio.queues > 0) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("interface type=\"vhostuser\" backend type=\"passt\" does not support multiple queues")); + return -1; + } + if (net->data.vhostuser->data.nix.path) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("'path' attribute cannot be set for interface type=\"vhostuser\" backend type=\"passt\"")); + return -1; + } + if (net->data.vhostuser->data.nix.listen) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("mode=\"server\" is not supported for interface type=\"vhostuser\" backend type=\"passt\"")); + return -1; + } } if (net->type == VIR_DOMAIN_NET_TYPE_VDPA) { -- 2.48.1

3 4

[PATCH 00/13] virsh: Enhancements for hypervisor-cpu-* commands and their documentation
by Jiri Denemark 21 Feb '25

21 Feb '25

This series originally started as a documentation update for hypervisor-cpu-compare and hypervisor-cpu-baseline commands with an idea to print a warning if the commands are used suboptimally. The rest is a refactor and a lot of fixes of (mostly) error messages printed by virsh. Jiri Denemark (13): docs: Clarify documentation of virsh hypervisor-cpu-compare docs: Clarify documentation of virsh hypervisor-cpu-baseline virsh: Do not format messages twice virsh: Make messages printed by vshError properly translatable virsh: Refactor vshError virsh: Introduce vshWarn virsh: Warn when hypervisor-cpu-* is used with host CPU virsh: Do not require \n in vshDebug messages virsh: Properly mark all error messages for translation virsh: Avoid using translated messages without format virsh: Drop extra newlines at the end of error messages virsh: Let prohibit_newline_at_end_of_diagnostic check pass build: Enable syntax checks for vshError and vshWarn build-aux/syntax-check.mk | 3 +- docs/manpages/virsh.rst | 47 ++++++++----- tools/virsh-domain-monitor.c | 16 ++--- tools/virsh-domain.c | 131 ++++++++++++++++++----------------- tools/virsh-host.c | 22 ++++-- tools/virsh-interface.c | 6 +- tools/virsh-network.c | 10 +-- tools/virsh-nodedev.c | 6 +- tools/virsh-nwfilter.c | 10 +-- tools/virsh-pool.c | 10 +-- tools/virsh-secret.c | 8 +-- tools/virsh-snapshot.c | 12 ++-- tools/virsh-util.c | 16 ++--- tools/virsh-volume.c | 24 +++---- tools/virsh.c | 10 +-- tools/virt-admin.c | 16 ++--- tools/vsh.c | 113 +++++++++++++++++------------- tools/vsh.h | 6 +- 18 files changed, 254 insertions(+), 212 deletions(-) -- 2.48.1

2 14

[PULL 01/41] docs/about: Change notes on x86 machine type deprecation into a general one
by Michael S. Tsirkin 21 Feb '25

21 Feb '25

From: Thomas Huth <thuth(a)redhat.com> We now have a general note about versioned machine types getting deprecated and removed at the beginning of the deprecated.rst file, so we should also have a general note about this in removed-features.rst (which will also apply to versioned non-x86 machine types) instead of listing individual old machine types in the document. Signed-off-by: Thomas Huth <thuth(a)redhat.com> Message-Id: <20250116064644.65670-1-thuth(a)redhat.com> Reviewed-by: Zhao Liu <zhao1.liu(a)intel.com> Reviewed-by: Michael S. Tsirkin <mst(a)redhat.com> Signed-off-by: Michael S. Tsirkin <mst(a)redhat.com> --- docs/about/deprecated.rst | 7 ------- docs/about/removed-features.rst | 11 +++++------ 2 files changed, 5 insertions(+), 13 deletions(-) diff --git a/docs/about/deprecated.rst b/docs/about/deprecated.rst index 4a3c302962..7b42d6eecc 100644 --- a/docs/about/deprecated.rst +++ b/docs/about/deprecated.rst @@ -236,13 +236,6 @@ deprecated; use the new name ``dtb-randomness`` instead. The new name better reflects the way this property affects all random data within the device tree blob, not just the ``kaslr-seed`` node. -``pc-i440fx-2.4`` up to ``pc-i440fx-2.12`` (since 9.1) -'''''''''''''''''''''''''''''''''''''''''''''''''''''' - -These old machine types are quite neglected nowadays and thus might have -various pitfalls with regards to live migration. Use a newer machine type -instead. - PPC 405 ``ref405ep`` machine (since 9.1) '''''''''''''''''''''''''''''''''''''''' diff --git a/docs/about/removed-features.rst b/docs/about/removed-features.rst index c6616ce05e..156c0c253c 100644 --- a/docs/about/removed-features.rst +++ b/docs/about/removed-features.rst @@ -972,6 +972,11 @@ from Linux in 2021, and is not supported anymore by QEMU either. System emulator machines ------------------------ +Note: Versioned machine types that have been introduced in a QEMU version +that has initially been released more than 6 years before are considered +obsolete and will be removed without further notice in this document. +Please use newer machine types instead. + ``s390-virtio`` (removed in 2.6) '''''''''''''''''''''''''''''''' @@ -1006,12 +1011,6 @@ mips ``fulong2e`` machine alias (removed in 6.0) This machine has been renamed ``fuloong2e``. -``pc-0.10`` up to ``pc-i440fx-2.3`` (removed in 4.0 up to 9.0) -'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''' - -These machine types were very old and likely could not be used for live -migration from old QEMU versions anymore. Use a newer machine type instead. - Raspberry Pi ``raspi2`` and ``raspi3`` machines (removed in 6.2) '''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''' -- MST

1 0

[libvirt PATCHv1 0/4] do not format empty metadata element
by Ján Tomko 21 Feb '25

21 Feb '25

https://issues.redhat.com/browse/RHEL-27172 Ján Tomko (4): conf: networkobj: fix indentation conf: obj: remove extra empty line conf: metadata: ignore empty metadata element conf: metadata: remove metadata node if all metadata is removed src/conf/domain_conf.c | 8 +++++-- src/conf/network_conf.c | 3 ++- src/conf/virnetworkobj.c | 48 +++++++++++++++++++++------------------- 3 files changed, 33 insertions(+), 26 deletions(-) -- 2.48.1

2 6

[PATCH 0/2] docs: two fixes
by Peter Krempa 20 Feb '25

20 Feb '25

Please see individual patches. Peter Krempa (2): docs: formatdomain: Document few NVRAM config limitations docs: formatdomain: Mention that vhostuser interface with mode='server' waits for connection docs/formatdomain.rst | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) -- 2.48.1

2 5

[PATCH v2] NEWS: Improve mention of vTPM transient VM crash fix in v11.0.0
by Peter Krempa 20 Feb '25

20 Feb '25

The original NEWS entry for the vTPM transient VM crash was rather vague and non-actionable. As the bug is still actively experienced by users [1] of distros that didn't yet ship an update to v11.0.0 and is hit by relatively common usage improve the entry to mention situations when it happens and link to upstream bug reports containing workarounds. [1]: https://gitlab.com/libvirt/libvirt/-/issues/746 Signed-off-by: Peter Krempa <pkrempa(a)redhat.com> --- v2: Instead of listing the workarounds link to the upstream issues. NEWS.rst | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/NEWS.rst b/NEWS.rst index 7984f358f3..96e6ee9ada 100644 --- a/NEWS.rst +++ b/NEWS.rst @@ -116,10 +116,17 @@ v11.0.0 (2025-01-15) * **Bug fixes** - * qemu: tpm: do not update profile name for transient domains + * qemu: tpm: Fix crash on startup of transient domains with vTPM - Fix a possible crash when starting a transient domain which was - introduced in the previous release. + A bug was introduced in `v10.10.0 (2024-12-02)`_ when a transient VM with + a vTPM device is started either via ``virsh create domain.xml`` or when + creating VMs via ``virt-manager`` which uses transient definition for the + initial installation the libvirt daemon (``virtqemud`` or ``libvirtd``) can + crash. + + More information and workarounds were documented in upstream issues + `#746 <https://gitlab.com/libvirt/libvirt/-/issues/746>`__ and + `#715 <https://gitlab.com/libvirt/libvirt/-/issues/715>`__. * qemu: Fix snapshot to not delete disk image with internal snapshot -- 2.48.1

1 0

[PATCH] NEWS: Improve mention of vTPM transient VM crash fix in v11.0.0
by Peter Krempa 20 Feb '25

20 Feb '25

The original NEWS entry for the vTPM transient VM crash was rather vague and non-actionable. As the bug is still actively experienced by users [1] of distros that didn't yet ship an update to v11.0.0 and is hit by relatively common usage improve the entry to mention situations when it happens a bit more close and provide workarounds for users who are not able to update. [1]: https://gitlab.com/libvirt/libvirt/-/issues/746 Signed-off-by: Peter Krempa <pkrempa(a)redhat.com> --- NEWS.rst | 29 ++++++++++++++++++++++++++--- 1 file changed, 26 insertions(+), 3 deletions(-) diff --git a/NEWS.rst b/NEWS.rst index 7dc6a3fa37..117287044a 100644 --- a/NEWS.rst +++ b/NEWS.rst @@ -97,10 +97,33 @@ v11.0.0 (2025-01-15) * **Bug fixes** - * qemu: tpm: do not update profile name for transient domains + * qemu: tpm: Fix crash on startup of transient domains with vTPM - Fix a possible crash when starting a transient domain which was - introduced in the previous release. + A bug was introduced in `v10.10.0 (2024-12-02)`_ when a transient VM with + a vTPM device is started either via ``virsh create domain.xml`` or when + creating VMs via ``virt-manager`` which uses transient definition for the + initial installation the libvirt daemon (``virtqemud`` or ``libvirtd``) can + crash. + + Note that vTPM is auto-added in many cases by ``virt-install`` based on your + guest OS that is about to be installed. + + The bug is fixed in this release. Following workarounds are possible if + upgrade is currently not available: + + - make the VM persistent instead of starting it as transient:: + + virsh define domain.xml + virsh start.xml + + - disable vTPM if practical in your deployment, either by dropping ``<tpm`` + element or:: + + virt-install --tpm none ... + + To obtain the XML ``virt-install`` would use for the above steps you can use:: + + virt-install --print-xml ... * qemu: Fix snapshot to not delete disk image with internal snapshot -- 2.48.1

3 3

[PATCH RFC 00/13] qemu: Add support for iothread to virtqueue mapping for 'virtio-scsi'
by Peter Krempa 20 Feb '25

20 Feb '25

The first part of the series refactors the existing code for reuse and then uses the new helpers to implement the feature. Note that this series is in RFC state as the qemu patches are still being discussed. Thus also the capability bump is not final. Also note that we should discuss the libvirt interface perhaps as it turns out that 'virtio-scsi' has two internal queues that need to be mapped as well. For now I've solved this administratively by instructing users to also add mapping for queue '0' and '1' which are the special ones in case of virtio-scsi. qemu-patches: https://mail.gnu.org/archive/html/qemu-devel/2025-02/msg02810.html Peter Krempa (13): conf: Rename 'virDomainDiskIothreadDef' to 'virDomainIothreadMappingDef' conf: domain: Extract code for parsing and formatting iotrhead mapping definition hypervisor: domain: Extract code for checking iothread usage qemu: command: Rename 'qemuBuildDiskDeviceIothreadMappingProps' to 'qemuBuildIothreadMappingProps' qemu: validate: Extract iothread mapping validation code qemuValidateCheckSCSIControllerIOThreads: Return '0' and '-1' instead of bools conf: schemas: Rename 'diskDriverIothreads' to 'iothreadMapping' conf: Validate that iohtreads are used only with 'virtio-scsi' controllers qemucapabilitiestest: Update 'caps_10.0.0_x86_64' to XXXXXX qemu: capabilities: Introduce QEMU_CAPS_VIRTIO_SCSI_IOTHREAD_MAPPING conf: Add support for iothread to queue mapping config for 'virtio-scsi' qemu: Implement support for iothread <-> virtqueue mapping for 'virtio-scsi' controllers qemuxmlconftest: Add 'iothreads-virtio-scsi-mapping' case docs/formatdomain.rst | 33 +++ src/conf/domain_conf.c | 157 +++++++----- src/conf/domain_conf.h | 11 +- src/conf/domain_validate.c | 19 ++ src/conf/schemas/domaincommon.rng | 7 +- src/hypervisor/domain_driver.c | 34 +-- src/qemu/qemu_capabilities.c | 2 + src/qemu/qemu_capabilities.h | 1 + src/qemu/qemu_command.c | 12 +- src/qemu/qemu_domain.c | 4 +- src/qemu/qemu_validate.c | 234 ++++++++++-------- .../caps_10.0.0_x86_64.replies | 12 +- .../caps_10.0.0_x86_64.xml | 3 +- ...r-virtio-serial-iothread.x86_64-latest.err | 1 + .../controller-virtio-serial-iothread.xml | 27 ++ ...ads-virtio-scsi-mapping.x86_64-latest.args | 39 +++ ...eads-virtio-scsi-mapping.x86_64-latest.xml | 54 ++++ .../iothreads-virtio-scsi-mapping.xml | 46 ++++ tests/qemuxmlconftest.c | 3 + 19 files changed, 506 insertions(+), 193 deletions(-) create mode 100644 tests/qemuxmlconfdata/controller-virtio-serial-iothread.x86_64-latest.err create mode 100644 tests/qemuxmlconfdata/controller-virtio-serial-iothread.xml create mode 100644 tests/qemuxmlconfdata/iothreads-virtio-scsi-mapping.x86_64-latest.args create mode 100644 tests/qemuxmlconfdata/iothreads-virtio-scsi-mapping.x86_64-latest.xml create mode 100644 tests/qemuxmlconfdata/iothreads-virtio-scsi-mapping.xml -- 2.48.1

3 20

virsh console hangs when ssh connection is lost
by Olaf Hering 20 Feb '25

20 Feb '25

Hello, the command 'virsh -c qemu+ssh://root@remotehost/system console vm' from libvirt 10.0.0 just hangs when the remotehost is rebooted. It prints error: Disconnected from qemu+ssh://root@remotehost/system due to end of file and waits for the user to press return. Then it prints error: internal error: client socket is closed and the command terminates as expected. I tried to add -k and -K, but that does not help. If for some reason the vm is shutdown manually via 'virsh shutdown --domain vm' on remotehost or via "poweroff", then 'virsh console' also terminates properly. Is there a way to avoid such a hang state if the ssh connection drops? It seems the simple reproducer is to send SIGTERM to the sshd child process. Olaf

2 1