Bug 1350264

Summary:	no graphical display with xen pv guest on F24
Product:	[Fedora] Fedora	Reporter:	Michael Young <m.a.young>
Component:	qemu	Assignee:	Fedora Virtualization Maintainers <virt-maint>
Status:	CLOSED ERRATA	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	24	CC:	amit, amit.shah, berrange, cfergeau, crobinso, dwmw2, itamar, pbonzini, rjones, virt-maint
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	qemu-2.6.2-8.fc24	Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2017-05-12 19:22:26 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Michael Young 2016-06-26 21:07:41 UTC

I am still having problems getting a graphical screen booting a xen pv guest with graphics using /usr/bin/qemu-system-i386 with qemu-2.6.0-4.fc24 (as expected the fix for Bug 1340931 doesn't fix the root problem). I have been trying to work out what is happening, and although I haven't found the problem I have some more clues. The output from qemu via xen when this failure happens is

xen be core: xen be: watching backend path (backend/console/2) failed
xen be core: xen be: watching backend path (backend/vkbd/2) failed
xen be core: xen be: watching backend path (backend/vfb/2) failed
xen be core: xen be: watching backend path (backend/qdisk/2) failed
xen be core: xen be: watching backend path (backend/qnic/2) failed
qemu-system-i386: terminating on signal 1 from pid 8157

so the there seems to be a problem with qemu connecting with the xenstored socket. Also if I build qemu-system-i386 with the --disable-rbd option added then then I do get a graphic screen again, so it seems to be some interaction between xen and rbd.

Comment 1 Cole Robinson 2016-06-28 15:33:03 UTC

Have you reported this to xen-devel or similar? Devs there probably have a better chance of figuring this out...

Comment 2 Michael Young 2016-07-10 15:35:23 UTC

I have identified the problem. The xs_watch function (called from xenstore_scan in hw/xen/xen_backend.c) creates a read pthread with an initial stack size of 16384, which is okay for qemu without rbd, but not enough if rbd is enabled. It does work if I increase the stack size setting to 24576. Do you know why there is an increased stack size requirement for pthreads in qemu with rbd?

Comment 3 Paolo Bonzini 2016-07-14 21:58:00 UTC

No particular reason, just that rbd has pretty deep call stacks.

Comment 4 Cole Robinson 2017-03-19 20:31:28 UTC

Sounds like this was eventually fixed in qemu:

commit 66668d197fa40747e835e15617eda2f1bc80982f
Author: Peter Lieven <pl>
Date:   Fri Jul 15 11:45:11 2016 +0200

    vnc-tight: fix regression with libxenstore

Comment 5 Fedora Update System 2017-04-15 18:25:21 UTC

qemu-2.6.2-8.fc24 has been submitted as an update to Fedora 24. https://bodhi.fedoraproject.org/updates/FEDORA-2017-374389c196

Comment 6 Fedora Update System 2017-04-16 21:23:06 UTC

qemu-2.6.2-8.fc24 has been pushed to the Fedora 24 testing repository. If problems still persist, please make note of it in this bug report.
See https://fedoraproject.org/wiki/QA:Updates_Testing for
instructions on how to install test updates.
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2017-374389c196

Comment 7 Fedora Update System 2017-05-12 19:22:26 UTC

qemu-2.6.2-8.fc24 has been pushed to the Fedora 24 stable repository. If problems still persist, please make note of it in this bug report.