Re: [Xen-devel] Qemu upstream bugs with xen

[Fabio Fantoni <fabio.fantoni@xxxxxxxxxx>]

Il 18/01/2013 16:51, Stefano Stabellini ha scritto:
> On Fri, 18 Jan 2013, Fabio Fantoni wrote:
> > Il 18/01/2013 14:00, Stefano Stabellini ha scritto:
> > > On Fri, 18 Jan 2013, Fabio Fantoni wrote:
> > > > 1) Both linux and windows domU with qemu upstream have an additional
> > > > empty floppy and cdrom.
> > > > Is there a way to remove these additionals block devices?
> > > the best way would be to pass a device or a global command line option
> > > to QEMU to change the default for isa-fdc
> > >
> > >
> > > > 2) xl cd-eject and cd-insert are not working:
> > > >
> > > > xl -vvv cd-eject W7 hdb
> > > > libxl: debug: libxl.c:2389:libxl_cdrom_insert: ao 0x1b95990: create:
> > > > how=(nil) callback=(nil) poller=0x1b95930
> > > > libxl: debug: libxl_device.c:229:libxl__device_disk_set_backend: Disk
> > > > vdev=hdb spec.backend=phy
> > > > libxl: debug: libxl_device.c:175:disk_try_backend: Disk vdev=hdb,
> > > > backend phy unsuitable as phys path not a block device
> > > > libxl: error: libxl_device.c:269:libxl__device_disk_set_backend: no
> > > > suitable backend for disk hdb
> > > cd-eject should work by trying to insert an empty cdrom.
> > > In particular the disk format should be LIBXL_DISK_FORMAT_EMPTY, but I
> > > think this is not the case here.
> > > I guess it is a bug in parse_disk_config.
> > >
> > >
> > > > libxl: debug: libxl_event.c:1482:libxl__ao_abort: ao 0x1b95990: abort
> > > > libxl: debug: libxl_event.c:1472:libxl__ao__destroy: ao 0x1b95990: destroy
> > > > xc: debug: hypercall buffer: total allocations:4 total releases:4
> > > > xc: debug: hypercall buffer: current allocations:0 maximum allocations:2
> > > > xc: debug: hypercall buffer: cache current size:2
> > > > xc: debug: hypercall buffer: cache hits:1 misses:2 toobig:1
> > > >
> > > > xl -vvv cd-insert W7 hdb raw:/mnt/vm/iso/QUANTAL.iso
> > > > libxl: debug: libxl.c:2389:libxl_cdrom_insert: ao 0x1d07990: create:
> > > > how=(nil) callback=(nil) poller=0x1d079f0
> > > > libxl: debug: libxl_device.c:229:libxl__device_disk_set_backend: Disk
> > > > vdev=hdb spec.backend=phy
> > > > libxl: error: libxl_device.c:243:libxl__device_disk_set_backend: Disk
> > > > vdev=hdb failed to stat: raw:/mnt/vm/iso/QUANTAL.iso: No such file or
> > > > directory
> > > It looks like it is trying to stat "raw:/mnt/vm/iso/QUANTAL.iso" instead
> > > of "/mnt/vm/iso/QUANTAL.iso".  The format parsing must be wrong.
> > >
> > >
> > > > libxl: debug: libxl_event.c:1482:libxl__ao_abort: ao 0x1d07990: abort
> > > > libxl: debug: libxl_event.c:1472:libxl__ao__destroy: ao 0x1d07990: destroy
> > > > xc: debug: hypercall buffer: total allocations:4 total releases:4
> > > > xc: debug: hypercall buffer: current allocations:0 maximum allocations:2
> > > > xc: debug: hypercall buffer: cache current size:2
> > > > xc: debug: hypercall buffer: cache hits:1 misses:2 toobig:1
> > > >
> > > > Seem there is parsing error about backend and device/iso path, for now I
> > > > not found solution for fix it on code.
> > I've added -nodefaults option to qemu upstream to remove empty floppy
> > and cdrom with success. Please could this be added by default on libxl
> > on qemu upstream starting?
> The problem with -nodefaults is that it changes a lot of other things
> aside from the floppy drive:
>
> default_serial = 0;
> default_parallel = 0;
> default_virtcon = 0;
> default_monitor = 0;
> default_vga = 0;
> default_net = 0;
> default_floppy = 0;
> default_cdrom = 0;
> default_sdcard = 0;
>
> libxl should always set the right command line options for vga and
> network, and most of the other settings don't matter for us.
> However I am concerned about the monitor and the cdrom drive.
>
> I thought that we always wanted to have a cdrom drive on HVM guests,
> even if empty.
>
> It would be better to find a way to change only the floppy default
> setting.
It seems there is no other way to disable default floppy only, what can 
we do about it?
http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg00338.html

> > About cd-eject cd-insert parsing bug can you solve please?
> Yes, however I am a bit busy at the moment so it might take some time.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel