Bug 1008903

Just for completeness, can you attach the full domain xml and also paste in the qemu commandline and ensuing error messages from /var/log/libvirt/qemu/$guest.log?

Created attachment 799349 [details]
domain xml

Created attachment 799351 [details]
libvirt logfile

qemu is saying that the sata bus only supports a single unit. Is this true? (I couldn't find those error messages in my local copy of (upstream) qemu source)

I've also just found that when I add a 2nd sata controller and place both the primary hard disk and the cd drive on this controller, qemu *doesn't* complain.

Is this:

1) a "true-to-life" limitation of the integrated sata controller at 00:1f.2 of the q35 chipset?

2) a bug in qemu's emulation of that controller?

or perhaps 

3) an odd problem (bug?) in qemu caused by us not specifying a controller, instead relying on qemu to place the disks on the default controller?

If (1) then libvirt needs to be taught the limitation so that it only places a single disk on that controller. if (2) or (3) then this bug should be transferred to qemu.

(In reply to Laine Stump from comment #5)
> qemu is saying that the sata bus only supports a single unit. Is this true?
> (I couldn't find those error messages in my local copy of (upstream) qemu
> source)

Yes.  There are 6 busses though, one for each ahci port.  From 'info qtree':

      dev: ich9-ahci, id ""
        addr = 1f.2
        romfile = <null>
        rombar = 1
        multifunction = on
        command_serr_enable = on
        class SATA controller, addr 00:1f.2, pci id 8086:2922 (sub 1af4:1100)
        bar 4: i/o at 0xffffffffffffffff [0x1e]
        bar 5: mem at 0xffffffffffffffff [0xffe]
        bus: ide.5
          type IDE
        bus: ide.4
          type IDE
        bus: ide.3
          type IDE
        bus: ide.2
          type IDE
        bus: ide.1
          type IDE
        bus: ide.0
          type IDE

(In reply to Laine Stump from comment #6)
> I've also just found that when I add a 2nd sata controller and place both
> the primary hard disk and the cd drive on this controller, qemu *doesn't*
> complain.

Let me check ...

This happens because libvirt generates different command lines for the two cases:

built-in controller:

-device ide-hd,unit=0,drive=drive-sata0-0-0,id=sata0-0-0,bootindex=1
-device ide-cd,unit=2,drive=drive-sata0-0-2,id=sata0-0-2,bootindex=3

additional controller:

-device ide-hd,bus=ahci1.0,drive=drive-sata1-0-0,id=sata1-0-0,bootindex=1
-device ide-cd,bus=ahci1.2,drive=drive-sata1-0-2,id=sata1-0-2,bootindex=3

libvirt should use "bus=ide.$nr" instead of "unit=$nr" for the built-in controller.

So it could possibly be (3). I'll try a build that sets the alias for the first sata controller to ide.0 in this case and see if the behavior changes.

Yep, switching to using an explicit bus eliminates the error. I'll post a patch upstream tonight.

Fix posted upstream, waiting for review:

https://www.redhat.com/archives/libvir-list/2013-September/msg01177.html

Pushed upstream:

commit 30bb4c4b54ed3a23adc90c52540dd96b0cbcfbcc
Author: Laine Stump <laine>
Date:   Fri Sep 20 06:00:48 2013 -0400

    qemu: use "ide" as device name for implicit SATA controller on Q35

    The Q35 machinetype has an implicit SATA controller at 00:1F.2 which
    isn't given the "expected" id of ahci0 by qemu when it's created. The
    original suggested solution to this problem was to not specify any
    controller for the disks that use the default controller and just
    specify "unit=n" instead; qemu should then use the first IDE or SATA
    controller for the disk.
    
    Unfortunately, this "solution" is ignorant of the fact that in the
    case of SATA disks, the "unit" attribute in the disk XML is actually
    *not* being used for the unit, but is instead used to specify the
    "bus" number; each SATA controller has 6 buses, and each bus only
    allows a single unit. This makes it nonsensical to specify unit='n'
    where n is anything other than 0. It also means that the only way to
    connect more than a single device to the implicit SATA controller is
    to explicitly give the bus names, which happen to be "ide.$n", where
    $n can be replaced by the disk's "unit" number.

Reproduced with the packages:
libvirt-1.1.1-6.el7.x86_64

Verified with the packages:
libvirt-1.1.1-7.el7.x86_64

Test steps:
1. define a guest with xml:

    <disk type='file' device='disk'>
      <driver name='qemu' type='qcow2'/>
      <source file='/var/lib/libvirt/images/rh7-qcow2.img'/>
      <target dev='sda' bus='sata'/>
      <boot order='1'/>
      <alias name='sata0-0-0'/>
      <address type='drive' controller='0' bus='0' target='0' unit='0'/>
    </disk>
    <disk type='file' device='cdrom'>
      <driver name='qemu' type='raw'/>
      <target dev='sdc' bus='sata'/>
      <readonly/>
      <shareable/>
      <boot order='3'/>
      <alias name='sata0-0-2'/>
      <address type='drive' controller='0' bus='0' target='0' unit='2'/>
    </disk>
    <controller type='sata' index='0'>
      <alias name='sata0'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x1f' function='0x2'/>
    </controller>

2. virsh start the guest
# virsh start rh7.sata
Domain rh7.sata started
# virsh list --all
 Id    Name                           State
----------------------------------------------------
 2     rh7.sata                       running

# ps -ef|grep qemu| grep sata
......-drive file=/var/lib/libvirt/images/rh7-qcow2.img,if=none,id=drive-sata0-0-0,format=qcow2 -device ide-hd,bus=ide.0,drive=drive-sata0-0-0,id=sata0-0-0,bootindex=1 -drive if=none,media=cdrom,id=drive-sata0-0-2,readonly=on,format=raw -device ide-cd,bus=ide.2,drive=drive-sata0-0-2,id=sata0-0-2,bootindex=3
......

Test results:
Guest start with multiple sata disk/cdrom successfully, so change the status to VERIFIED.

This request was resolved in Red Hat Enterprise Linux 7.0.

Contact your manager or support representative in case you have further questions about the request.