Bug 1961326

Hi Eric， Could you please help evaluate if this feature is still needed? If yes, we can extend the stale date to +6M. Or it will be closed as won't fix automatically. Thank you!

extend the stale date as the priority for arm increased

Yes thank you, the feature is still needed.

Hi, Andrea, is there any progress or plan for this new feature?

(In reply to Hu Shuai (Fujitsu) from comment #11)
> Hi, Andrea, is there any progress or plan for this new feature?

Reassigning to Kristina, who will look into it in the coming weeks.

Hi Kristina, the qemu enablement was tracked by https://bugzilla.redhat.com/show_bug.cgi?id=1747467. There, you will be able to find the test setup we used. Compared to x86 integration the specificity is the pvpanic device is a pcie device and not an ISA one. So it needs to be integrated like other PCIe end-points (maybe the virtio-iommu-pci device laterly introduced can be used as an example? But better to ask Andrea)

v2 of patches posted upstream:

https://listman.redhat.com/archives/libvir-list/2023-February/238009.html

merged upstream as:

4bae0c8ae4 NEWS: Document new pvpanic-pci device
c98e49f82e docs: document panic device 'pvpanic-pci'
78744415fe tests: add case for pvpanic-pci without address
9f52df3a70 qemu: assign PCI address to device pvpanic-pci
f4367059d2 tests: add test cases for device pvpanic-pci
46ef87e10e conf: add panic model 'pvpanic'
741624a1a6 qemu: introduce QEMU_CAPS_DEVICE_PANIC_PCI

v9.0.0-247-g4bae0c8ae4

Test with upstream libvirt.
```
# git describe
v9.1.0-rc1-1-g6af048339f
```
Env:
  kernel-5.14.0-268.el9.aarch64
  libvirt-9.1.0-1.el9.aarch64

1. The guest can be defined and started with device panic.
  1.1 Add panic device to the guest xml
```
# cat avocado-vt-vm1.xml | grep panic
    <panic model='pvpanic'/>
```
  1.2 Define && start the guest
```
# virsh define avocado-vt-vm1.xml
Domain 'avocado-vt-vm1' defined from avocado-vt-vm1.xml
# virsh start avocado-vt-vm1
Domain 'avocado-vt-vm1' started
```
  1.3 Check the panic device of running guest
```
# virsh dumpxml avocado-vt-vm1 | grep /panic -B2
    <panic model='pvpanic'>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x03' function='0x0'/>
    </panic>
```


2. The guest can be shut off when triggering the guest crash through 'echo c > /proc/sysrq-trigger'.
On guest:
```
# systemctl stop kdump
# echo 1 > /proc/sys/kernel/sysrq
# echo c > /proc/sysrq-trigger
```
On host:
```
# virsh domstate avocado-vt-vm1 --reason
shut off (crashed)
```
Log from /var/log/libvirt/qemu/avocado-vt-vm1.log
```
2023-02-23 07:47:37.216+0000: shutting down, reason=crashed
2023-02-23T07:47:37.220955Z qemu-kvm: terminating on signal 15 from pid 4590 (<unknown process>)
```

Change status to verified based on above comment

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (Moderate: libvirt security, bug fix, and enhancement update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHSA-2023:6409