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
From 43e203f8fe19702cc9813d109499e4a7165f6c6f Mon Sep 17 00:00:00 2001
From: Pritam Srichandan Sahoo <[email protected]>
Date: Mon, 17 Aug 2026 08:22:22 +0000
Subject: [PATCH 1/1] qemu: Fix crash in qemuProcessStop due to NULL
 eventThread

When a VM shuts down, the monitor EOF event is queued to the worker
thread pool. If another thread enters qemuProcessStop() first and
frees priv->eventThread via g_steal_pointer(), the second call from
the EOF event worker thread hits a NULL dereference when calling
virEventThreadStop(priv->eventThread).

The existing 'if (priv->eventThread)' guard is insufficient because
the VM object is unlocked between the check and the call to
virEventThreadStop(), allowing a concurrent thread to clear the
pointer in the meantime.

Fix this by taking a GObject reference to eventThread before
unlocking the VM. This ensures the pointer remains valid for
virEventThreadStop() even if another thread clears
priv->eventThread concurrently.

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=...)

Signed-off-by: Pritam Srichandan Sahoo <[email protected]>
---
 src/qemu/qemu_process.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c
index 0d9b8bcb93..642e18eeb1 100644
--- a/src/qemu/qemu_process.c
+++ b/src/qemu/qemu_process.c
@@ -8850,13 +8850,15 @@ void qemuProcessStop(virQEMUDriver *driver,
      * the global domain object list code depends on it (and it can't actually
      * check 'priv->beingDestroyed as that's private). */
     if (priv->eventThread) {
+        virEventThread *eventThread = g_object_ref(priv->eventThread);
         /* Explicitly set priv->beingDestroyed. While it's done in
          * qemuProcessBeginStopJob(), qemuProcessStop() is called from places
          * where stop job is not acquired. */
         priv->beingDestroyed = true;
         virObjectUnlock(vm);
-        virEventThreadStop(priv->eventThread);
+        virEventThreadStop(eventThread);
         virObjectLock(vm);
+        g_object_unref(eventThread);
     }
 
     if (priv->agent) {
-- 
2.53.0

Reply via email to