Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Bug#1043544: qemu kvm rtl8139 interface breaks with linux-image-5.10.0-24-amd64
13 views
Skip to first unread message
Jamin Hoggard
Aug 12, 2023, 1:50:04 PM8/12/23
to
Package: linux-image-amd64
Version: linux-image-5.10.0-24-amd64
Upon updating a server hosting qemu-system-x86_64 / kvm virtual
machines (VMs) from linux-image-5.10.0-23-amd64 to
linux-image-5.10.0-24-amd64 and rebooting, all VM OSes saw the rtl8139
network interface as unplugged (and the network link down). Trying to
change the interface state via the QEMU monitor command "set_link
<netid> off" then "set_link <netid> on" runs, but the VM OS still sees
the interface as unplugged. Rebooting the host OS or VMs had no
effect.
The rtl8139 interfaces for the VMs work fine booting using earlier
kernel versions.
Server info:
CPU: AMD EPYC 7282 16-Core Processor
microcode: microcode updated early to new patch_level=0x0830107a
The same problem was not observed on an Intel server using virtio
network interfaces with the same kernel update.
I suspect the problem may be caused by the patches for INCEPTION
because the EPYC 7282 processor is affected by that (and Zenbleed),
but that may be incorrect.
Version: linux-image-5.10.0-24-amd64
Upon updating a server hosting qemu-system-x86_64 / kvm virtual
machines (VMs) from linux-image-5.10.0-23-amd64 to
linux-image-5.10.0-24-amd64 and rebooting, all VM OSes saw the rtl8139
network interface as unplugged (and the network link down). Trying to
change the interface state via the QEMU monitor command "set_link
<netid> off" then "set_link <netid> on" runs, but the VM OS still sees
the interface as unplugged. Rebooting the host OS or VMs had no
effect.
The rtl8139 interfaces for the VMs work fine booting using earlier
kernel versions.
Server info:
CPU: AMD EPYC 7282 16-Core Processor
microcode: microcode updated early to new patch_level=0x0830107a
The same problem was not observed on an Intel server using virtio
network interfaces with the same kernel update.
I suspect the problem may be caused by the patches for INCEPTION
because the EPYC 7282 processor is affected by that (and Zenbleed),
but that may be incorrect.
0 new messages