On 04/25/2016 01:40 PM, Alex Bligh wrote:
> 
> On 25 Apr 2016, at 20:20, Eric Blake <ebl...@redhat.com> wrote:
> 
>>>>        } else if (fixedNewstyle) {
>>>
>>> So the above is for NewStyle (not fixedNewStyle)?
>>
>> The above is for fixedNewStyle when TLS is not negotiated yet; the 'else
>> if' is for fixedNewStyle after TLS has been negotiated.  Prior to TLS,
>> we have to special-case NBD_OPT_ABORT and actually quit.
> 
> OK. fixedNewStyle is defined as a prerequisite for TLS. I'm hoping
> nothing in Qemuland ever did non-fixed NewStyle, and assuming that's
> the case I would not even support it (certainly it won't play
> nicely with all the other stuff you've been doing).

Well, there were some last-minute patches that went into the 2.6
candidates that fixed qemu to actually be a fixedNewStyle server
(without commit 156f6a10, for example, qemu had the very bug of
disconnecting on unknown client options that fixedNewStyle was supposed
to prevent). Fortunately, qemu 2.5 is oldstyle only, and qemu 2.6 is the
first newstyle server, and I think I got the worst of the
interoperability bugs nailed in 2.6, whereas this series is focusing on
the feature enhancements for inclusion in 2.7.

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

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to