A new kvm guest on a CentOS6.5 x84_64 host is set up to make use of the serial port (virt-install ... --serial dev,path=/dev/ttyS0)
<serial type='dev'> <source path='/dev/ttyS0'/> <target port='0'/> </serial> <console type='dev'> <source path='/dev/ttyS0'/> <target type='serial' port='0'/> </console>
When I launch the vm, setroubleshootd goes into overdrive and creates massive logs, and /var/log/messages starts growing significantly also. The reason is that as soon as the vm starts, ownership and context of /dev/ttyS0 on the host are being reset. Why is that, and how can I prevent it from happening?
Should be:
crw-rw----. root dialout system_u:object_r:tty_device_t:s0 /dev/ttyS0
Changes to:
crw-rw----. qemu qemu unconfined_u:object_r:svirt_image_t:s0:c9,c796 /dev/ttyS0
When I launch the vm, setroubleshootd goes into overdrive and creates massive logs, and /var/log/messages starts growing significantly also. The reason is that as soon as the vm starts, ownership and context of /dev/ttyS0 on the host are being reset. Why is that, and how can I prevent it from happening?
Should be:
crw-rw----. root dialout system_u:object_r:tty_device_t:s0 /dev/ttyS0
Changes to:
crw-rw----. qemu qemu unconfined_u:object_r:svirt_image_t:s0:c9,c796 /dev/ttyS0
One way of doing this: https://bugzilla.redhat.com/show_bug.cgi?id=596553#c4 (no. 1). This avoids resetting of group ownership to root instead of dialout.
In addition, I had to add qemu to the disk group to access the /dev/dm* device that serves as vm disk.
Also found qemu_use_comm and set it to on, for what it's worth ...
-
lhecking＠users.sourceforge.net