Jay Gagnon wrote:
Just want to get this out before I start in on the resource add/mod/del stuff, to see if I'm on the right track. The idea here is that we are not supporting processor pinning, will support scheduling (weight/limit attributes), and are making Processor RASDs one-per-domain. Does this look sane?
This looks good to me. +1 Some of the Processor and EAFP tests fail due to these changes, but this is because the tests are expecting the InstanceID for the device to be <guest name>/<processor id> instead of <guest name>/proc. The commit log for the second patch is a bit misleading, since you're changing device_parsing.c which also impacts Virt_Device. But otherwise, this tests fine for me. -- Kaitlin Rupert IBM Linux Technology Center kaitlin@linux.vnet.ibm.com