qemu-devel
[Top][All Lists]
Advanced

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

[Qemu-devel] Re: [Bug 453617] Re: kvm hangs at 100% cpu when connecting


From: Jan Kiszka
Subject: [Qemu-devel] Re: [Bug 453617] Re: kvm hangs at 100% cpu when connecting to forwarded ports (when listed incorrectly on the command line)
Date: Fri, 21 May 2010 19:29:04 +0200
User-agent: Mozilla/5.0 (X11; U; Linux i686 (x86_64); de; rv:1.8.1.12) Gecko/20080226 SUSE/2.0.0.12-1.1 Thunderbird/2.0.0.12 Mnenhy/0.7.5.666

Michael Tokarev wrote:
> 21.05.2010 21:16, Jan Kiszka wrote:
>> Michael Tokarev wrote:
>>> 21.05.2010 19:42, Gleb Natapov wrote:
>>>>> ProblemType: Bug
>>>>> Architecture: amd64
>>>>> Date: Fri Oct 16 17:19:59 2009
>>>>> DistroRelease: Ubuntu 9.10
>>>> What is the point forwarding distro bugs here? Can we have upstream bug
>>>> tracker to report upstream issues?
>>> Um, this _is_ upstream issue, as far as I can see.
>>> It exists in qemu-kvm-0.12.4 at least, I just verified.
>> Well, this particular bug rather looks like a potential documentation
>> insufficiency:
>>
>> Instantiating two user space networking stacks on the _same_ vlan with
>> _identical_ parameters is just asking for troubles. If the reporter read
>> the manual and then did the mistake, we may have to clarify the docs
>> what "-net user" does.
> 
> Actually it's solved like that in ubuntu, by updating documentation.
> 
> It's a common trap - two pairs of -net nic -net tap without
> specifying vlan=NN and voila, you have a loop on your network.
> This has been tried/trapped on more than once, that's for sure :)

So, in the presence of -netdev, we may have to deemphasize on -net which
is now actually an expert knob. Doc updates welcome where required.

> 
>>> What's wrong with that forwarding?
>> IMHO so far nothing - as long as the ratio of pure downstream issues
>> remains low. I think we should give this channel a chance as it can be
>> advantageous to have more "real" end-user reports.
> 
> There's very few distro-specific bugreports usually.  At least on
> Debian.  Two observations: I forward or notify about bugs filed
> against Debian package (I maintain qemu-kvm there) anyway, and
> quite some bugs filed actually are either usage/operator errors
> or sort of misunderstanding or misuse, not real bugs.

If there are too much operator errors, that may actually point to some
usability and/or documentation issue.

> 
> But actually I were asking about this particular bugreport,
> not about general forwarding - which - at least - should be
> done with care :)

Right.

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux



reply via email to

[Prev in Thread] Current Thread [Next in Thread]