Re: Ask for suggestions for CVE-2019-12928

[Markus Armbruster <armbru@redhat.com>]

"Dr. David Alan Gilbert" <dgilbert@redhat.com> writes:

> * 江芳杰 (18401698361@126.com) wrote:
>> Hi:
>>     Sorry to bother you~
>>     I have read the discussions about CVE--2019-12928 ( https://lists.gnu.org/archive/html/qemu-devel/2019-07/msg01153.html).
>> But, for the scenario of PC users, which is no requirement of network access to QMP, there are some mitigating proposes.
>> 1. Modify the compilation options to disable QMP.
>> 2. Modify command line parsing function to discard the QMP parameters with network configurations.
>> 3. PC manager or other manage software make sure only the trusted user can use QMP.
>> 4. Other ideas?
>
> QMP is a useful part of QEMU - so we don't want to do 1 - we need it to
> let things control QEMU; including configuring complex setups.

Compiling out QMP gains you exactly nothing unless you also compile out
HMP.  And then you're left without a way to monitor a running QEMU.
Similarly useful (but not nearly as secure) as not running QEMU at all
;)

> The important part is (3) - anything that runs a qemu must make sure it
> wires the QMP up securely; e.g. using unix sockets with appropriate
> permissions or something like that.
>
> As long as they do that, then we're fine.

Yup.

Regarding 4.: making insecure misconfiguration harder might be worth
exploring.