qemu-devel
[Top][All Lists]
Advanced

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

Re: [Qemu-devel] [PATCH] migration: introduce decompress-error-check


From: Dr. David Alan Gilbert
Subject: Re: [Qemu-devel] [PATCH] migration: introduce decompress-error-check
Date: Fri, 27 Apr 2018 12:29:26 +0100
User-agent: Mutt/1.9.5 (2018-04-13)

* Xiao Guangrong (address@hidden) wrote:
> 
> 
> On 04/26/2018 10:01 PM, Eric Blake wrote:
> > On 04/26/2018 04:15 AM, address@hidden wrote:
> > > From: Xiao Guangrong <address@hidden>
> > > 
> > > QEMU 2.13 enables strict check for compression & decompression to
> > > make the migration more robuster, that depends on the source to fix
> > 
> > s/robuster/robust/
> > 
> 
> Will fix, thank you for pointing it out.
> 
> > > the internal design which triggers the unexpected error conditions
> > 
> > 2.13 hasn't been released yet.  Why do we need a knob to explicitly turn
> > off strict checking?  Can we not instead make 2.13 automatically smart
> > enough to tell if the incoming stream is coming from an older qemu
> > (which might fail if the strict checks are enabled) vs. a newer qemu
> > (the sender gave us what we need to ensure the strict checks are
> > worthwhile)?
> > 
> 
> Really smart.
> 
> How about introduce a new command, MIG_CMD_DECOMPRESS_ERR_CHECK,
> the destination will do strict check if got this command (i.e, new
> QEMU is running on the source), otherwise, turn the check off.

The problem is that if it's received by an old qemu it will fail,
so that breaks backwards migration which is unfortunate.

> > > 
> > > To make it work for migrating old version QEMU to 2.13 QEMU, we
> > > introduce this parameter to disable the error check on the
> > > destination
> > > 
> > > Signed-off-by: Xiao Guangrong <address@hidden>
> > > ---
> > 
> > > +++ b/qapi/migration.json
> > > @@ -455,6 +455,17 @@
> > >   #          compression, so set the decompress-threads to the number 
> > > about 1/4
> > >   #          of compress-threads is adequate.
> > >   #
> > > +# @decompress-error-check: check decompression errors. When false, the 
> > > errors
> > > +#                          triggered by memory decompression are ignored.
> > 
> > What are the consequences of such an error?  Is it a security hole to
> > leave this at false, when a malicious migration stream can cause us to
> > misbehave by ignoring the errors?
> 
> The issue fixed by strict error check is avoiding VM corruption if zlib failed
> to compress & decompress the memory, i.e, catch error conditions returned by
> zlib API.
> 
> > 
> > > +#                          When true, migration is aborted if the errors 
> > > are
> > > +#                          detected. For the old QEMU versions (< 2.13) 
> > > the
> > > +#                          internal design will cause decompression to 
> > > fail
> > > +#                          so the destination should completely ignore 
> > > the
> > > +#                          error conditions, i.e, make it be false if 
> > > these
> > > +#                          QEMUs are going to be migrated. Since 2.13, 
> > > this
> > > +#                          design is fixed, make it be true to avoid 
> > > corrupting
> > > +#                          the VM silently (Since 2.13)
> > 
> > Rather wordy; I'd suggest:
> > 
> > @decompress-error-check: Set to true to abort the migration if
> >          decompression errors are detected at the destination. Should be
> >          left at false (default) for qemu older than 2.13, since only
> >          newer qemu sends streams that do not trigger spurious
> >          decompression errors. (Since 2.13)
> > 
> 
> Yup, much better.
> 
> > But that's if we even need it (it SHOULD be possible to design something
> > into the migration stream so that you can detect this property
> > automatically instead of relying on the user to set the property).
> > 
> 
> Yes, can not agree with you more. :)

The challenge is how to put something into the stream without breaking
an old version of QEMU that's receiving the stream.

Dave
--
Dr. David Alan Gilbert / address@hidden / Manchester, UK



reply via email to

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