Re: kern/57560: xennetback_xenbus.c rev 1.109 broke guests, causes DOM0 reboot

[Brad Spencer <brad%anduin.eldar.org@localhost>]

The following reply was made to PR port-xen/57560; it has been noted by GNATS.

From: Brad Spencer <brad%anduin.eldar.org@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: port-xen-maintainer%netbsd.org@localhost, gnats-admin%netbsd.org@localhost,
        netbsd-bugs%netbsd.org@localhost
Subject: Re: kern/57560: xennetback_xenbus.c rev 1.109 broke guests, causes
 DOM0 reboot
Date: Fri, 04 Aug 2023 13:10:49 -0400

 Taylor R Campbell <riastradh%NetBSD.org@localhost> writes:
 
 > The following reply was made to PR port-xen/57560; it has been noted by GNATS.
 >
 > From: Taylor R Campbell <riastradh%NetBSD.org@localhost>
 > To: brad%anduin.eldar.org@localhost
 > Cc: gnats-bugs%NetBSD.org@localhost
 > Subject: Re: kern/57560: xennetback_xenbus.c rev 1.109 broke guests, causes DOM0 reboot
 > Date: Fri, 4 Aug 2023 07:21:10 +0000
 >
 >  This is a multi-part message in MIME format.
 >  --=_usKIEqQi97a3RY2xFPoqMtrNfJcy98qu
 >  
 >  Can you please try the attached patch?
 
 Thanks for the quick response.  My target system for this all to
 function correctly is really NetBSD 10.0, but the results are likely to
 apply to -current just as well.  The Xen xennet backend is identical
 between the two right now (in as much as I can tell right now).
 
 It went as follows:
 
 The source tree was patched with the provided patch, a kernel built, and
 the DOM0 booted with it.  The test DOMU that I had been using did start
 and I was able to log into it.  I proceeded to start the production DOMU
 guests and all seemed to be well for a while.  However, I would say
 after 5 to 10 minutes, the DOM0 panic'ed / rebooted as before when the
 patch was not installed.  No DDB, no traceback... etc..  same thing,
 just a little later (the load on the DOMUs would have been ramping up a
 lot at this point, as well).  When the DOM0 had stablized, I booted it
 with the same source tree, except with xennetback_xenbus.c reverted back
 to 1.108.  I started all of the guests and the DOM0 is still running.
 This is all a pretty disruptive operation to perform, so I could not do
 anything more at the moment, but I would say that the patch improved the
 situation, but did not fully address the problem.  I will be watching
 the DOMU guests closely and the DOM0 to see if anything undesirable
 develops as there might be something else wrong, but for right now, the
 environment is stable.