[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [libvirt] [PATCH v3 libvirt 3/3] qemu: add support for MTP filesystem



On 08/20/2014 05:46 AM, Michal Privoznik wrote:
> On 19.08.2014 22:11, Giuseppe Scrivano wrote:
>> Generate the qemu command line option:
>>
>> -device 'usb-mtp,root=$SRC,desc=$TARGET'
>>
>> from the definition XML:
>>
>>      <filesystem type='mount'>
>>        <source dir='$SRC'/>
>>        <target dir='$TARGET'/>
>>        <model type='mtp'/>
>>      </filesystem>
>>
>> Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1121781
>>
>> Signed-off-by: Giuseppe Scrivano <gscrivan redhat com>
>> ---

>> +            virBufferAddLit(&opt, "usb-mtp");
>> +            virBufferAsprintf(&opt, ",root=%s,desc=%s", fs->src,
>> fs->dst);
> 
> Reading qemu sources the root property is called 'x-root'. And indeed
> domain fails to start:

Ouch.  Qemu has explicitly documented that anything starting with x- is
unstable, and may change.  We probably should NOT be targetting this in
libvirt API just yet, in case qemu changes their mind.  It's nice that
you've done the patch as a proof of concept, but I don't think we want
it in the tree yet :(


> 
>     usb: mtp: tag root property as experimental
> 
>     Reason: we don't want commit to that interface yet.  Possibly
>     the implementation will be switched over to use fsdev.
> 
>     Suggested-by: Paolo Bonzini <pbonzini redhat com>
>     Signed-off-by: Gerd Hoffmann <kraxel redhat com>
> 
> diff --git a/hw/usb/dev-mtp.c b/hw/usb/dev-mtp.c
> index 1b51a90..384d4a5 100644
> --- a/hw/usb/dev-mtp.c
> +++ b/hw/usb/dev-mtp.c
> @@ -1090,7 +1090,7 @@ static const VMStateDescription vmstate_usb_mtp = {
>  };
> 
>  static Property mtp_properties[] = {
> -    DEFINE_PROP_STRING("root", MTPState, root),
> +    DEFINE_PROP_STRING("x-root", MTPState, root),
>      DEFINE_PROP_STRING("desc", MTPState, desc),
>      DEFINE_PROP_END_OF_LIST(),
>  };
> 
> 
> Question that pops up immediately: do we want to commit to something
> that even qemu developers don't believe yet? What will happen when qemu
> decides to switch to 'root' attribute again? Libvirt will have to adapt
> which won't work with older qemus supporting 'x-root' only.

Yep - anything in qemu with x- is not a candidate for libvirt support
(for the longest time, we refused to support RDMA migration because it
still had an x- prefix).

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org

Attachment: signature.asc
Description: OpenPGP digital signature


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]