linux-kernel - Re: [Xen-devel] [PATCH 2/4] xenbus: limit when state is forced to closed

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 9 Dec 2019 18:17:57 +0100
From:   Roger Pau Monné <roger.pau@...rix.com>
To:     "Durrant, Paul" <pdurrant@...zon.com>
CC:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "xen-devel@...ts.xenproject.org" <xen-devel@...ts.xenproject.org>,
        "Juergen Gross" <jgross@...e.com>,
        Stefano Stabellini <sstabellini@...nel.org>,
        "Boris Ostrovsky" <boris.ostrovsky@...cle.com>
Subject: Re: [Xen-devel] [PATCH 2/4] xenbus: limit when state is forced to
 closed

On Mon, Dec 09, 2019 at 04:26:15PM +0000, Durrant, Paul wrote:
> > > If you want unbind to actually do a proper unplug then that's extra work
> > and not really something I want to tackle (and re-bind would still need to
> > be toolstack initiated as something would have to re-create the xenstore
> > area).
> > 
> > Why do you say the xenstore area would need to be recreated?
> > 
> > Setting state to closed shouldn't cause any cleanup of the xenstore
> > area, as that should already happen for example when using pvgrub
> > since in that case grub itself disconnects and already causes a
> > transition to closed and a re-attachment afterwards by the guest
> > kernel.
> > 
> 
> For some reason, when I originally tested, the xenstore area disappeared. I checked again and it did not this time. I just ended up with a frontend stuck in state 5 (because it is the system disk and won't go offline) trying to talk to a non-existent backend. Upon re-bind the backend goes into state 5 (because it sees the 5 in the frontend) and leaves the guest wedged.

Likely blkfront should go back to init state, but anyway, that's not
something that needs fixing as part of this series.

Thanks, Roger.