On Fri, Sep 20, 2013 at 04:35:12PM +0200, Gerd Hoffmann wrote: > Hi, > > > 1) Every SATA controller has 6 buses, and each bus allows a single > > target (or "unit"). > > That is how it is modeled in qemu, largely due to reusing the ide code. > > On real hardware you have a sata controller with a number of ports > (happens to be 6 for the one emulated by qemu, but that may be different > on other controllers). You can plug a single device into each port. > There is not really a concept of a sata bus. Ok, so in libvirt terminology then, using 'unit' to map to the 'port' is what we want to be doing, and ignore the 'bus' attribute for sata drives. > > (or are we just stuck with the current state of unit being interpreted > > as bus for sata addresses? :-/) > > I'd say continuing using 'unit' is fine. When changing the way the > address is specified in xml I'd rather go for something like this: > > <address type=sata controller=0 port=2 /> We already use type=drive for addressing sata drives (and indeed any other drives attached to disk controllers), so we won't be inventing a new address type. Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :| -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list