Re: HVM domains crash after upgrade from XEN 4.5.1 to 4.5.2

[Atom2 <ariel.atom2@web2web.at>]

[-- Attachment #1: Type: text/plain, Size: 2445 bytes --]

Am 20.11.15 um 00:53 schrieb Andrew Cooper:
> On 19/11/2015 20:02, Atom2 wrote:
>> Am 19.11.15 um 02:06 schrieb Andrew Cooper:
>>> Thanks! That is what I was looking for.
>>>
>>> Sadly, it is less useful than I was hoping.  The guest is not appearing
>>> to do anything interesting which causes the bad state; it is almost a
>>> full second between the previous action of note, and the crash.
>>>
>>> Can you email me the bad HVMLoader binary please?
>> Hi Andrew,
>> Sure I can email the binary. What version of the HVMLoader do you want
>> me to send to you? Should this be one with the debug option and debug
>> symbols or a standard build without any of those? I assume we are
>> talking about amd64 - right?
> hvmloader is always a 32bit binary, no matter what your host environment
> is.  I would like any of the ones which reliably crash for  you,
> preferably debug if that is available.
Hi Andrew,
appologies for my delay in responding. I have just mailed you the 
crashing hvmloader off-list using my main mail address. It is the debug 
version from xen-4.5.2 together with symbols (as splitdebug) using the 
non-working seabios-1.8.2 - all packed up in a compressed tar file using 
relative names and maintaining the directory structure required for 
splitdebug to work.

If you would also like me to send you the failing seabios binary 
separately, just let me know. My understanding, however, is that seabios 
is anyways emebedded in the hvmloader binary.
>> Given the recent findings do you still want me to boot with minimal
>> command line options to rule out an interaction with the options as
>> suggested in an earlier answer?
> No - this is unnecessary now.
>
>> I can certainly do this, but would need to reconfigure certain things
>> as both my HVM domUs require PCI passthrough devices (which you have
>> stripped off the suggested minimal command line) to be available.
>> Failing to see those devices puts those domUs into an unconfigured
>> state and they re-start their initial setup process.
> Don't worry - Jan figured this out from the disassembly, which I
> admittedly missed.
>
> It would be nice to see your vm's .cfg file to help me reproduce the issue.
Attached please find the config file for one of the HVM virtual machines 
that crashes with the hmvloader binary sent to you by mail. The vm, btw, 
is running the pfsense firewall using FreeBSD. Any more information 
required - just ask.

Many thanks Atom2

[-- Attachment #2: pfsense.cfg --]
[-- Type: text/plain, Size: 549 bytes --]

builder			= 'hvm'
cpus			= '2-7'
vcpus			= 2
cpu_weight		= 512
memory			= 640
name			= 'pfsense'

disk			= [ 'vdev=xvda,format=raw,access=rw,target=/etc/xen/guests/disk.d/pfsense.disk' ]

vif			= [ 'mac=00:16:3e:a1:64:01,bridge=xenbr0,type=vif,vifname=pfsense.0,script=vif-bridge.noTXoffload' ]

on_poweroff		= 'destroy'
on_reboot		= 'restart'
on_crash		= 'destroy'
localtime		= 0

boot			= 'c'

vnc			= 1
vnclisten		= '127.0.0.1'
vncpasswd		= ''
keymap			= 'de'
nographic		= 1
serial			= 'pty'
nx			= 1

pci			= [ '04:00.0', '0a:08.0', '0a:0b.0' ]

[-- Attachment #3: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel