- Devel - libvirt List Archives

[PATCH 0/2] ch: support NUMA and cpu topology configuration
by Thomas Prescher 31 Aug '26

31 Aug '26

This patch set adds NUMA cell configuration and cpu topology support to the cloud-hyervisor driver. The first patch adds NUMA awareness so users can map guest NUMA nodes to host NUMA nodes. The second patch allows the user to define a concrete vCPU topology that is presented to the guest. Stefan Kober (2): ch: support for numa awareness ch: support CPU topology configuration src/ch/ch_monitor.c | 163 +++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 161 insertions(+), 2 deletions(-) -- 2.53.0

2 4

[PATCH] NEWS: Document CVEs and a virsh crash fixed in this release
by Peter Krempa 31 Aug '26

31 Aug '26

From: Peter Krempa <pkrempa(a)redhat.com> Signed-off-by: Peter Krempa <pkrempa(a)redhat.com> --- NEWS.rst | 35 +++++++++++++++++++++++++++++++++++ 1 file changed, 35 insertions(+) diff --git a/NEWS.rst b/NEWS.rst index 829d6eca63..3a847c0107 100644 --- a/NEWS.rst +++ b/NEWS.rst @@ -13,6 +13,35 @@ v12.7.0 (unreleased) * **Security** + * CVE-2026-18917: Integer overflow in RPC handler for ``virNodeGetFreePages`` + + The RPC handler for ``virNodeGetFreePages`` didn't properly check for + overflows in multiplication of integers when calculating the amount of + returned data for validation and allocation of the return buffer. + + Specific values could thus pass validation and cause an undersized buffer + to be allocated. The hypervisor driver would then fill the undersized buffer + based on the values prior to multiplication and thus cause a write beyond + the end of the allocated buffer. + + This bug can be triggered via a read-only connection. + + * CVE-2026-77158: Double free of disks array in ``qemuAgentGetDisks`` + + The qemu hypervisor driver would double-free the array of parsed disks, on + error code paths leading to crash of the daemon. The error code path could + be reached when the qemu guest agent provided malformed data as response to + the ``guest-get-disks`` command. + + * CVE-2026-77159: ``chown()`` of swtpm log file follows symlinks + + When starting up a VM with a ``swtpm`` device configured, libvirt + ``chown()``-s the log file of swtpm to the (unprivileged) user/group + running the swtpm process. The problem is that the directory containing the + log files is also owned by the same user/group, thus users with access + to that directory could install a symlink to a privileged file that the + libvirt daemon would follow and chown the file pointed to. + * **Removed features** * **New features** @@ -77,6 +106,12 @@ v12.7.0 (unreleased) all datastores to find which one stores disks for the domain. But if a datastore wasn't mounted, then a crash would occur. This is now fixed. + * virsh: Don't crash when certain error messages are printed + + A bug in the error printing function could cause ``virsh`` to crash when + certain errors (e.g. disk not found in ``virsh detach-disk``) were being + reported. + v12.6.0 (2026-08-03) ==================== -- 2.55.0

2 1

[PULL 5/6] docs: deprecate virtio-crypto
by Alex Bennée 28 Aug '26

28 Aug '26

We have outstanding guest exploitable bugs in the code which is fairly complex and needs some serious love to clean it up. As off-CPU cryptography acceleration seems to be waning in popularity lets just the deprecate the sub-system with a view to removing it in a couple of cycles. We can also remove the backend code at that point as well. Link: https://gitlab.com/qemu-project/qemu/-/work_items/3625 Reviewed-by: Mohamed Mediouni <mohamed(a)unpredictable.fr> Reviewed-by: Thomas Huth <thuth(a)redhat.com> Message-ID: <20260824-fixes-for-11-2-v2-7-352c6b890402(a)linaro.org> Signed-off-by: Alex Bennée <alex.bennee(a)linaro.org> diff --git a/docs/about/deprecated.rst b/docs/about/deprecated.rst index 0c656a968fc..05e4ce8cf16 100644 --- a/docs/about/deprecated.rst +++ b/docs/about/deprecated.rst @@ -276,6 +276,24 @@ future release. Where no 'dh-params.pem' file is provided, the DH parameters will be automatically negotiated in accordance with RFC7919. +Devices +------- + +``virtio-crypto`` and cryptodev backends (since 11.2) +''''''''''''''''''''''''''''''''''''''''''''''''''''' + +The ``virtio-crypto`` device emulation is quite complex code with a +number of known flaws. It has never been migratable, so it is +unlikely to be used in any serious virtualization setting. Modern +ISAs provide on-CPU cryptography instructions (e.g. AES-NI/VAES, +armv8 crypto extensions), and the Linux kernel deprecated AF_ALG +and dropped its off-CPU accelerator support in Linux 7.2, removing +the primary userspace path for off-CPU crypto acceleration. The +time for this sort of off-load has passed. + +As this is the only device that uses the cryptodev backends these will +be removed at the same time the ``virtio-crypto`` device is. + Device options -------------- -- 2.47.3

1 0

[libvirt PATCH] storage: logical: don't write past buffer bounds
by Ján Tomko 27 Aug '26

27 Aug '26

From: Ján Tomko <jtomko(a)redhat.com> If parsed number of extents is 0, the calculated buffer size won't hold the first regex_unit. Closes: https://gitlab.com/libvirt/libvirt/-/work_items/913 Signed-off-by: Ján Tomko <jtomko(a)redhat.com> --- src/storage/storage_backend_logical.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/storage/storage_backend_logical.c b/src/storage/storage_backend_logical.c index 51e9337820..81f9d62d8b 100644 --- a/src/storage/storage_backend_logical.c +++ b/src/storage/storage_backend_logical.c @@ -161,12 +161,12 @@ virStorageBackendLogicalParseVolExtents(virStorageVolDef *vol, /* Allocate space for 'nextents' regex_unit strings plus a comma for each */ regex = g_new0(char, nextents * (strlen(regex_unit) + 1) + 1); - strcat(regex, regex_unit); - for (i = 1; i < nextents; i++) { + for (i = 0; i < nextents; i++) { /* "," is the separator of "devices" field */ - strcat(regex, ","); strcat(regex, regex_unit); + strcat(regex, ","); } + regex[strlen(regex) - 1] = '\0'; re = g_regex_new(regex, 0, 0, &err); if (!re) { -- 2.55.0

3 3

[PATCH 0/7] vmx: Add more controllers and parse PCI addresses
by Michal Privoznik 27 Aug '26

27 Aug '26

This is a prerequisite of the following issue: https://redhat.atlassian.net/browse/RHEL-147630 It's not fixing the issue per-se, so you'll not find any reference to the issue in commit messages. But my gut feeling says let's expose more controllers and their PCI addresses in the domain XML. Michal Prívozník (7): vmx: Add SCSI controllers vmx: Parse PCI address for SCSI controllers vmx: Add SATA controllers vmx: Parse PCI address for SATA controllers vmx: Add NVMe controllers vmx: Parse PCI address for NVMe controllers vmx: Parse PCI address for ethernet devices src/vmx/vmx.c | 173 ++++++++++++++------- tests/vmx2xmldata/esx-in-the-wild-10.xml | 5 +- tests/vmx2xmldata/esx-in-the-wild-11.xml | 5 +- tests/vmx2xmldata/esx-in-the-wild-12.xml | 9 +- tests/vmx2xmldata/esx-in-the-wild-13.xml | 5 +- tests/vmx2xmldata/esx-in-the-wild-14.xml | 5 +- tests/vmx2xmldata/esx-in-the-wild-15.xml | 5 +- tests/vmx2xmldata/esx-in-the-wild-16.xml | 11 +- tests/vmx2xmldata/esx-in-the-wild-17.xml | 9 +- tests/vmx2xmldata/esx-in-the-wild-5.xml | 5 +- tests/vmx2xmldata/esx-in-the-wild-6.xml | 5 +- tests/vmx2xmldata/esx-in-the-wild-7.xml | 5 +- tests/vmx2xmldata/esx-in-the-wild-8.xml | 18 ++- tests/vmx2xmldata/esx-in-the-wild-9.xml | 8 +- tests/vmx2xmldata/fusion-in-the-wild-1.xml | 6 +- tests/vmx2xmldata/gsx-in-the-wild-1.xml | 1 + tests/vmx2xmldata/gsx-in-the-wild-2.xml | 1 + tests/vmx2xmldata/gsx-in-the-wild-3.xml | 1 + tests/vmx2xmldata/gsx-in-the-wild-4.xml | 1 + 19 files changed, 206 insertions(+), 72 deletions(-) -- 2.54.0

2 9

[PATCH 0/2] Fix double free of stream error reply aborting libvirtd
by Ross Golder 27 Aug '26

27 Aug '26

libvirtd aborts with "free(): invalid pointer" whenever a stream write fails, and then crash-loops until systemd's start limit is reached, leaving the host with no management plane. Running guests are unaffected but become unmanageable. The cause is an ambiguous return contract. daemonStreamHandleWriteData() reports a failed stream write through virNetServerProgramSendReplyError(), which takes ownership of the message and queues it on client->tx, and then returns that function's result - 0 on success. Its caller reads 0 as "no reply was sent" and, for VIR_NET_CONTINUE, clears the message and queues it a second time. Since the message is by then the sole element of client->tx, virNetMessageQueuePush() walks to the tail - itself - and links it to itself. virNetMessageQueueServe() then hands out the same pointer twice and virNetServerClientDispatchWrite() frees it twice. Patch 1 is the fix: give the handlers a distinct return value meaning "already queued, do not touch msg again". daemonStreamHandleHole() had the same defect and is fixed alongside. Note the requeue test has to change from "ret > 0" to "ret == 1", otherwise the new value would requeue a message which is already on client->tx. Patch 2 is independent hardening: make virNetMessageQueuePush() refuse a push that would corrupt the list, so a caller bug of this shape surfaces as a log message rather than as heap corruption. It includes a regression test that reproduces the self-cycle deterministically. Note that virNetMessageClear() memsets the whole message including ->next, so a queued message can appear unlinked. That is why patch 2 walks the queue rather than testing msg->next, and why this class of bug is easy to miss by inspection. There is no ABI change; src/libvirt_remote.syms is untouched. Evidence -------- - valgrind memcheck: "Invalid free()" with the previous free at the *same* call site, plus 13 invalid reads and 6 invalid writes as the "while (client->tx)" loop re-reads the freed block. Exactly one invalid free per occurrence, as a self-cycle predicts. - A production core dump whose crash IP is the return address of the same call to virNetMessageFree(), resolving to virnetserverclient.c:1374. - Instrumented builds logging both push sites: virNetServerProgramSendError:168 followed by daemonStreamHandleWrite:797. Two cautions for anyone reproducing this: Under valgrind the daemon does *not* abort - memcheck replaces the allocator, so glibc's malloc_printerr never runs. It logs and continues, and the fault is easy to mistake for "works fine". Freed virNetMessage blocks are promptly reused by malloc, so the same address legitimately reappears as a new message moments after being freed. In RPC debug logs that closely resembles a use-after-free and is not one. Testing ------- Reproduced and fixed on three production hypervisors running 12.5.0 with these patches backported. One host had logged 5197 aborts beforehand; across all three there have been zero aborts and zero hardening warnings since. The hardening warning firing before the patch 1 fix and never firing after it is the direct evidence that the root cause, and not just the symptom, is addressed. Full test suite passes on master (306 ok, 1 expected fail, 0 failures). CI on a personal fork is green: 22 jobs passed, 0 failed, covering Fedora 43/44/rawhide, CentOS Stream 9/10, Ubuntu 24.04/26.04 (including clang), openSUSE Tumbleweed/Leap 16, Debian 13, armv7l and mingw32/64, plus check-dco and codestyle. Reported as https://gitlab.com/libvirt/libvirt/-/issues/902 Ross Golder (2): remote: don't queue the stream error reply twice rpc: refuse to queue a message that is already queued src/remote/remote_daemon_stream.c | 51 +++++++++++++----- src/rpc/virnetmessage.c | 43 +++++++++++++++ tests/virnetmessagetest.c | 87 +++++++++++++++++++++++++++++++ 3 files changed, 169 insertions(+), 12 deletions(-) -- 2.53.0

2 4

[PATCH] NEWS: Document features/improvements/bug fixes I've participated in
by Michal Privoznik 27 Aug '26

27 Aug '26

From: Michal Privoznik <mprivozn(a)redhat.com> There are some features/improvements/bug fixes I've either contributed or reviewed/merged. Document them for upcoming release. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- NEWS.rst | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) diff --git a/NEWS.rst b/NEWS.rst index 89e470e29d..acd4c78dbe 100644 --- a/NEWS.rst +++ b/NEWS.rst @@ -47,12 +47,31 @@ v12.7.0 (unreleased) <wwn>01000000efbeadde</wwn> </disk> + * Add guest device info to ``virDomainGetGuestInfo()`` + + QEMU guest agent has 'guest-get-devices` command, which returns information + on guest devices (driver name, version, release date, and so on). This is + now exposed via ``virDomainGetGuestInfo()`` API (accessible via ``guestinfo + --devices`` virsh command). + + * **Improvements** * **Bug fixes** * bhyve: fix domain API calls hanging after domain shutdown + * remote: Fix stale virStream status + + In a few cases it might happen that daemon held incorrect state of a + virStream leading to an infinite loop. This is now fixed. + + * esx: Don't crash when parsing unmounted datastore + + When libvirt constructs domain XML for an ESX domain it traverses through + all datastores to find which one stores disks for the domain. But if a + datastore wasn't mounted, then a crash would occur. This is now fixed. + v12.6.0 (2026-08-03) ==================== -- 2.54.0

2 1

[PATCH] NEWS: document added qemu poll-weigth support
by Boris Fiuczynski 27 Aug '26

27 Aug '26

Signed-off-by: Boris Fiuczynski <fiuczy(a)linux.ibm.com> --- NEWS.rst | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/NEWS.rst b/NEWS.rst index 89e470e29d..bce158af0a 100644 --- a/NEWS.rst +++ b/NEWS.rst @@ -47,6 +47,12 @@ v12.7.0 (unreleased) <wwn>01000000efbeadde</wwn> </disk> + * qemu: Add poll-weight support for iothreads + + The qemu driver now supports the iothread poll-weight property, + which controls the weight of the most recent event interval in the + adaptive polling duration calculation. + * **Improvements** * **Bug fixes** -- 2.54.0

2 1

[PATCH] ci: Add job to check if all mentioned CVEs have corresponding git tags
by Peter Krempa 27 Aug '26

27 Aug '26

From: Peter Krempa <pkrempa(a)redhat.com> Introduce a job which will run with the scheduled libvirt pipeline runs which will check if all CVEs mentioned in the git commit messages, NEWS entries and for all libvirt security notices published at https://security.libvirt.org have corresponding git tags. Since git commit messages do sometimes mention CVE not related to libvirt, the '.cve_exceptions' file lists allows to add overrides. Since the upstream repository has signed tags, the tagging is done by Jirka, so it doesn't make sense to run this check inside the checkout or as a regular CI pipeline run, so it's implemented directly in the CI job rather than as a standalone tool. The job also creates an artifact containing all tagged CVEs in the repository. This can later be used e.g. to see which CVEs are missing a libvirt security notice. Signed-off-by: Peter Krempa <pkrempa(a)redhat.com> --- .cve_exceptions | 46 +++++++++++++++++++++++++++++++++++++++++++++ .gitlab-ci.yml | 50 +++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 96 insertions(+) create mode 100644 .cve_exceptions In order for this to pass the following CVEs need to be tagged (the order of the commit IDs is as 'git log' shows): commitlog CVE-2026-15268 f74495c7a157618e3df6fa61079678084c4cc685 CVE-2026-18917 5a62cbf2907d4590283597b46da9c0f41e7b4d4f CVE-2026-61477 3cfc77963b512d809348fca07f97fe924fac9a05 cb8974b923e3c40cde96f0c7bceaa638f7f9c72b 289ffa796d737a79a4c05d07232ebd75def9a12a d44836a1dc6771ac22f69755fc69bf730f0eec87 CVE-2026-61478 68da70aae766c6271b8d3b466374d3cc7d1a8afb CVE-2026-63622 801160fd414ca2cc402bc01ead09b7ed4c3b8f5b CVE-2026-63623 69335a484768d550854da1133d5490074695e825 News CVE-2021-4147 5c5df5310f72be4878a71ace47074c54e0d1a27d security.libvirt.org CVE-2009-0036 be33b189a5e579509b5025d72b7f283401ef9dc1 CVE-2010-2237 68719c4bddb85fbcc931a5b7d99ac7c8a0af09b0 CVE-2010-2239 187da82fea598aa017bdfc82a1622f9c8d75ee74 CVE-2016-10746 506e9d6c2d4baaf580d489fff0690c0ff2ff588f CVE-2019-20485 d61f95cf6a6fbd564e104c168d325581acd9cd8d 4cc90c2e62df653e909ad31fd810224bf8bcf913 d9605abed68af4b65c4e04386346a2fd372d7ab9 95f5ac9ae52455e9da47afc95fa31c9456ac27ae 1cb8bc52c1035573a0c1a87f724a6c7dfee82f12 0a9893121187c0c3f9807e9164366e1f6977619c a663a860819287e041c3de672aad1d8543098ecc e005c95f56fee9ed780be7f8db103d690bd34cbd CVE-2019-3840 7cfd1fbb1332ae5df678b9f41a62156cb2e88c73 CVE-2020-10703 dfff16a7c261f8d28e3abe60a47165f845fa952f CVE-2020-14301 a5b064bf4b17a9884d7d361733737fb614ad8979 CPU CVE (covered by LSNs) CVE-2017-5715 6d4a3cd42781babed7d29b061e220ebff24dd43e 6b7e7d1cc24a28a9f5ece8626f807189647d14b4 2e3b220a874e558e54678afd7cf49466fe605e09 30b381cfdd5e92e5afa6de09f0fe533353e71d07 203c92e9cc2db854199b39ef3ffcc10406d3c59e 7dd85ff62d7080b52d4d175f53ad5eb11cdcfb9c 7f83eefa9e6940c83579d31941efd07fab1b90c8 49bffcb3cc1850d332b9648c686a7be18de9e708 7bb4ce9761dfbd1620ddffb26fbd6f0ff1fedf3f b2042020c32b74069fa5365b5e966537aaba8cf6 24d504396c3c05eff87d29173a224e2faaeb2637 CVE-2018-3639 9267342206ce17f6933d57a3128cdc504d5945c9 1dbca2eccad58d91a5fd33962854f1a653638182 CVE-2019-11135 f411b7ef68221e82dec0129aaf2f2a26a8987504 07aaced4e6ea6db8b27f44636f51cafa6f1847a8 diff --git a/.cve_exceptions b/.cve_exceptions new file mode 100644 index 0000000000..335085e1bd --- /dev/null +++ b/.cve_exceptions @@ -0,0 +1,46 @@ +This file contains CVE IDs mentioned in either the commit log or the NEWS which +are not actually CVEs in libvirt and is used for the sanity check ensuring that +we have git tags for all CVE patches. + +This file can contain any other text explaining given CVE as it's filtered for +just CVE ids for matching. + +Commit messages mention the following non-libvirt CVEs: + + - CVE-2011-4127: Kernel privilege escalation via SG_IO + + Mentioned as justification to introduce new disk access mode 'lun' + + - CVE-2012-3386: GNU Automake granting world-writable permissions to extraction directory + + Mentioned as justification for fixing a syntax check failure. + + - CVE-2013-0242: Buffer overflow in glibc's regexec code + + gnulib update + + - CVE-2013-4288: race condition in polkit allowing bypass of restrictions + + The libvirt patch references the required pokit fix + + - CVE-2014-6458: unspecified vulnerability in Oracle Java + + mentioned by mistake, same CVE id from 2013 was intended + + - CVE-2022-24765: git for windows safe directory handling + + CI needed to adapt to this fix + +Libvirt adjecent CVEs: + - CVE-2020-15708: libvirtd.sock being configured as world-writable + + This was a packaging/deployment bug in Ubuntu 20.10 allowing anyone access + the system connection as they shipped default socket permissions. Libvirt + later changed the default to harden against this + +The following CVEs mentioned in our NEWS or commitlog are referencing CPU bugs +that libvirt covered but didn't publish a LSN for: + - CVE-2018-12126: Microarchitectural Store Buffer Data Sampling (MSBDS) + - CVE-2018-12127: Microarchitectural Fill Buffer Data Sampling (MFBDS) + - CVE-2018-12130: Microarchitectural Load Port Data Sampling (MLPDS) + - CVE-2019-11091: Microarchitectural Data Sampling Uncacheable Memory (MDSUM) diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml index 04bfef8ed7..7121248f7b 100644 --- a/.gitlab-ci.yml +++ b/.gitlab-ci.yml @@ -173,3 +173,53 @@ coverity: - if: '$CI_PROJECT_NAMESPACE == $RUN_UPSTREAM_NAMESPACE && $CI_PIPELINE_SOURCE == "schedule" && $CI_COMMIT_REF_NAME == $CI_DEFAULT_BRANCH' when: on_success - when: never + +# Check that we have CVE git tags for CVEs mentioned in commit messages and NEWS +# and expose the list of CVE tags +cve_tags: + image: $CI_REGISTRY/$RUN_UPSTREAM_NAMESPACE/libvirt/ci-$NAME:latest + variables: + NAME: almalinux-9 + GIT_STRATEGY: clone + GIT_DEPTH: "" + stage: sanity_checks + needs: + - job: x86_64-almalinux-9-container + optional: true + artifacts: false + rules: + - if: $CI_PIPELINE_SOURCE == "schedule" + when: always + - when: never + script: + # fetch list of git tags and exceptions + - git tag | grep -o 'CVE-[0-9]\+-[0-9]\+' > cve-tags + - grep -o 'CVE-[0-9]\+-[0-9]\+' .cve_exceptions > cve-allowed-exceptions + - sort -u cve-tags cve-allowed-exceptions > cve-allowed + # lookup CVE ids in 'git-log' + - git log | grep -o '[Cc][Vv][Ee]-[0-9]\+-[0-9]\+' | sort -u > cves-found-log + - comm -2 -3 cves-found-log cve-allowed > cves-missing-tags-log + # lookup CVE ids in 'NEWS.rst' + - grep -o '[Cc][Vv][Ee]-[0-9]\+-[0-9]\+' NEWS.rst | sort -u > cves-found-news + - comm -2 -3 cves-found-news cve-allowed > cves-missing-tags-news + # lookup CVE ids in index of security notices + - curl -fsSL https://security.libvirt.org/index.xml | grep -o 'CVE-[0-9]\+-[0-9]\+' | sort -u > cves-found-notices + - comm -2 -3 cves-found-notices cve-allowed > cves-missing-tags-notices + - if [ -s cves-missing-tags-log ] || [ -s cves-missing-tags-news ] || [ -s cves-missing-tags-notices ]; + then + echo "Following CVE's are either missing a git tag or an exception in '.cve_exceptions:"; + echo "Source 'git log'"; + cat cves-missing-tags-log; + echo "Source 'NEWS.rst'"; + cat cves-missing-tags-news; + echo "Source 'security.libvirt.org/index.xml'"; + cat cves-missing-tags-notices; + false; + fi + artifacts: + expose_as: 'CVE tags' + name: 'cve_tags' + when: always + expire_in: 30 days + paths: + - cve-tags -- 2.55.0

2 2

[PATCH] qemu: Fix crash in qemuProcessStop due to NULL eventThread
by Sahoo, Pritam Srichandan 26 Aug '26

26 Aug '26

Hi, The attached patch fixes a NULL pointer dereference crash in the QEMU driver that was observed . Issue: libvirtd crashes with SIGSEGV in virEventThreadStop(evt=0x0) during VM shutdown. The crash occurs in qemuProcessStop() when it is invoked from the QEMU monitor EOF event worker thread. Root Cause: There is a race condition when two threads enter qemuProcessStop() concurrently for the same domain: - Thread A (shutdown path) frees priv->eventThread via g_steal_pointer(). - Thread B (EOF event worker) passes the "if (priv->eventThread)" NULL check while the pointer is still valid, but the VM object is then unlocked before virEventThreadStop() is called. In that window, Thread A clears priv->eventThread, so by the time Thread B calls virEventThreadStop(priv->eventThread), the pointer is already NULL, causing the SIGSEGV. The existing "if (priv->eventThread)" guard is insufficient because the VM object lock is released between the check and the actual call to virEventThreadStop(). Crash backtrace: #0 virEventThreadStop (evt=0x0) #1 qemuProcessStop (reason=VIR_DOMAIN_SHUTOFF_SHUTDOWN, asyncJob=VIR_ASYNC_JOB_NONE) #2 processMonitorEOFEvent (driver=..., vm=...) #3 qemuProcessEventHandler (data=..., opaque=...) #4 virThreadPoolWorker (opaque=...) Fix: Take a GObject reference on priv->eventThread before unlocking the VM object, use the local reference for virEventThreadStop(), then release it with g_object_unref() afterwards. This ensures the pointer remains valid for virEventThreadStop() even if another thread clears priv->eventThread concurrently. The affected code is present in current master (introduced in commit 0888784f387, 2024-07-25): https://gitlab.com/libvirt/libvirt/-/commit/0888784f387 (https://gitlab.com/libvirt/libvirt/-/commit/0888784f387) The crash is not reproducible on demand due to the narrow timing window of the race, but it is deterministic once the window is hit — virEventThreadStop(evt=0x0) will always SIGSEGV. The fix has been build-tested. Thanks, Pritam Srichandan Sahoo

1 0