On an Oracle VM Server for SPARC system, the igb SR-IOV virtual functions cause the I/O domain to panic when the root domain is rebooted.
The following error message is displayed:
root@racn1:~# May 22 16:20:58 racn1 in.mpathd: The link has gone down on net1 May 22 16:20:58 racn1 in.mpathd: IP interface failure detected on net1 of group ipmp0 May 22 16:21:01 racn1 in.mpathd: The link has come up on net1. panic[cpu10]/thread=2a100cc5c40: BAD TRAP: type=30 rp=2a100cc54a0 addr=0 mmu_fsr=4. sched: data access exception: ... ... ... syncing file systems... done dumping to /dev/zvol/dsk/rpool/dump, offset 65536, content: kernel sections
The behavior of I/O domains with PCIe endpoint devices is unpredictable when the primary domain is rebooted while the I/O domains are running.
Workaround: Choose one of the following workarounds:
Shut down the domains on the system that have PCIe endpoint devices assigned to them, before you shut down the root domain. Ensure that the domains are properly shut down before you halt, reboot, or shut down the root domain.
Configure a domain dependency relationship between the root domain and the domains that have PCIe endpoint devices assigned to them. This dependency relationship ensures that domains with PCIe endpoint devices are automatically restarted when the root domain reboots for any reason. The dependency relationship forcibly resets the domains.
For more information, see Oracle VM Server for SPARC 3.2 Administration Guide.