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

Re: [win-pv-devel] Multiple save/restore failing with windows 8.1 64 bit domU



Il 24/04/2015 17:01, Paul Durrant ha scritto:
-----Original Message-----
From: Fabio Fantoni [mailto:fabio.fantoni@xxxxxxx]
Sent: 24 April 2015 15:46
To: Paul Durrant; win-pv-devel@xxxxxxxxxxxxxxxxxxxx
Subject: Re: Multiple save/restore failing with windows 8.1 64 bit domU

Il 02/04/2015 16:41, Paul Durrant ha scritto:
-----Original Message-----
From: Fabio Fantoni [mailto:fabio.fantoni@xxxxxxx]
Sent: 02 April 2015 14:58
To: Paul Durrant; win-pv-devel@xxxxxxxxxxxxxxxxxxxx
Subject: Re: Multiple save/restore failing with windows 8.1 64 bit domU

Il 01/04/2015 16:43, Fabio Fantoni ha scritto:
Il 01/04/2015 15:21, Paul Durrant ha scritto:
-----Original Message-----
From: Fabio Fantoni [mailto:fabio.fantoni@xxxxxxx]
Sent: 30 March 2015 12:00
To: win-pv-devel@xxxxxxxxxxxxxxxxxxxx
Cc: Paul Durrant
Subject: Multiple save/restore failing with windows 8.1 64 bit domU

I tried 2 times multiple save/restore (start domU, use it for some
minutes, do a save restore, use it, do another save/restore) and in
both
case second save failed:

The patch I posted this morning should fix this one for you.

     Paul
Thanks.

Today I tried windows 10 pro 64 bit technical preview, clean install,
pv drivers (today's build) install ok but after restore display
(connecting with spice) seems freezed, waited 4-5 minutes without
changes and after I tried also xl shudown but do nothing.
DomU have pv disk and network working, xl shutdown (before try
save/restore) is working.
I updated pv drivers to today build and first save/restore now seems was
working also in Windows 10 but still not the multiple (seem only partial
solved), xl shutdown this time worked but it tooks some minutes.
In attachments domU's qemu logs with trace enabled.

It looks like your 2nd save restore worked... certainly log.1 shows the VM
coming back from suspend and then going into suspend. The drivers basically
seem to be behaving as expected.
    Paul
Thanks for fix the first problem, the second remained seems related to
save/restore with qxl.
With windows 7 freeze for some seconds instead and make me crazy trying
to found the cause for 1-2 years without result :(
With windows 7 temp. freeze after latest xenvif changes something is
changed... after restore freeze of some seconds in qemu log show pv
drivers lines from:
xen_platform_log xen platform: XENVIF|PdoRequestEject:
FFFFFA8003067540 (0)
Hmm. That suggests that, when XENVIF is resuming, for some reason it has 
decided the backend has gone offline. Not sure why that would happen unless 
there's some sort of race with net attach.

And windows popup appair showing that Xen PV Network Device #0 was
removed (even if finished the second freeze network return working).
Is it normal?

I have setted fixed mac address in domU's xl cfg, so should not saw
different netcard after save/restore FWIK


qemu log (with various trace and debug active) in attachment, first
freeze of some seconds was before this line:
xen_platform_log xen platform: XENVIF|PdoRequestEject:
FFFFFA8003067540 (0)
And second freeze near was before this line:
xen_platform_log xen platform: XENNET|DriverEntry: XENNET 8.0.0 (18)
(24.03.2015)
And third freeze after xl shutdown and before this line:
xen_platform_log xen platform: XEN|SystemPowerStateCallback: <- S0
Is there a way to know if there are unexpected case in pv drivers on
restore that can cause the freeze or not?

 From that, not really. Do you know if the guest vcpus are busy during the 
freeze? Do you have any xentrace analysis suggesting what the guest is doing?

   Paul

I already debugged a crash with gdb but I'm unable to do useful debug with temporary freeze :( During domUs freeze I saw with xentop that domU's cpu sec increasing and cpu(%) at 200 (probably for both vcpus at 100%) I don't have knowledge for good xen debug to find on what instructions freeze and/or what is the unexpected case that cause the problem. Where I can find an howto for do more useful debugging for this case? (if exists)

Thanks for any reply and sorry for my bad english.

If you more informations/tests tell me and I'll post them.

Thanks for any reply and sorry for my bad english.

About save/restore on W8 I tried now with new pv build and with first
save/restore is all ok but with on second now xl save/restore
completes correctly but after restore freeze same as the W10 one
(since first save/restore).

If you more informations/tests tell me and I'll post them.

Thanks for any reply and sorry for my bad english.



xl -vvv save W8 /mnt/vm/save/W8
libxl: debug: libxl.c:6705:libxl_retrieve_domain_configuration: no
vtpm from xenstore for domain 40
libxl: debug: libxl.c:6707:libxl_retrieve_domain_configuration: no pci
from xenstore for domain 40
Saving to /mnt/vm/save/W8 new xl format (info 0x1/0x0/1941)
libxl: debug: libxl.c:953:libxl_domain_suspend: ao 0x68d910: create:
how=(nil) callback=(nil) poller=0x68d970
libxl: debug: libxl_dom.c:1573:libxl__toolstack_save: domain=40
toolstack data size=49
libxl: debug: libxl.c:975:libxl_domain_suspend: ao 0x68d910:
inprogress: poller=0x68d970, flags=i
libxl-save-helper: debug: starting save: Success
xc: detail: xc_domain_save: starting save of domid 40
libxl: debug: libxl_dom.c:1278:domain_suspend_callback_common:
issuing
PVHVM suspend request via XenBus control node
libxl: debug: libxl_event.c:577:libxl__ev_xswatch_register: watch
w=0x68dae0 wpath=/local/domain/40/control/shutdown
token=3/0:
register
slotnum=3
libxl: debug: libxl_event.c:514:watchfd_callback: watch w=0x68dae0
wpath=/local/domain/40/control/shutdown token=3/0: event
epath=/local/domain/40/control/shutdown
libxl: debug: libxl_event.c:615:libxl__ev_xswatch_deregister: watch
w=0x68dae0 wpath=/local/domain/40/control/shutdown
token=3/0:
deregister slotnum=3
libxl: debug:

libxl_dom.c:1360:domain_suspend_common_pvcontrol_suspending:
guest
acknowledged suspend request
libxl: debug:
libxl_dom.c:1379:domain_suspend_common_wait_guest:
wait
for the guest to suspend
libxl: debug: libxl_event.c:577:libxl__ev_xswatch_register: watch
w=0x68daf8 wpath=@releaseDomain token=3/1: register slotnum=3
libxl: debug: libxl_event.c:514:watchfd_callback: watch w=0x68daf8
wpath=@releaseDomain token=3/1: event epath=@releaseDomain
libxl: error:
libxl_dom.c:1451:suspend_common_wait_guest_timeout:
guest did not suspend, timed out
libxl: debug: libxl_event.c:615:libxl__ev_xswatch_deregister: watch
w=0x68daf8 wpath=@releaseDomain token=3/1: deregister
slotnum=3
xc: error: Suspend request failed: Internal error
xc: error: Domain appears not to have suspended: Internal error
libxl-save-helper: xc_domain_save.c:2077: xc_domain_save:
Assertion
`rc' failed.
libxl: error: libxl_utils.c:430:libxl_read_exactly: file/stream
truncated reading ipc msg header from domain 40 save/restore
helper
stdout pipe
libxl: error: libxl_exec.c:129:libxl_report_child_exitstatus: domain
40 save/restore helper [-1] died due to fatal signal Aborted
libxl: error: libxl_dom.c:2035:remus_teardown_done: Remus: failed
to
teardown device for guest with domid 40, rc -3
libxl: debug: libxl_event.c:1766:libxl__ao_complete: ao 0x68d910:
complete, rc=-3
libxl: debug: libxl_event.c:1738:libxl__ao__destroy: ao 0x68d910:
destroy
Failed to save domain, resuming domain
libxl: debug: libxl.c:542:libxl_domain_resume: ao 0x68d910: create:
how=(nil) callback=(nil) poller=0x68d970
xc: error: Dom 40 not suspended: (shutdown 0, reason 255): Internal
error
libxl: error: libxl.c:511:libxl__domain_resume: xc_domain_resume
failed for domain 40: Invalid argument
libxl: debug: libxl_event.c:1766:libxl__ao_complete: ao 0x68d910:
complete, rc=-3
libxl: debug: libxl.c:545:libxl_domain_resume: ao 0x68d910:
inprogress: poller=0x68d970, flags=ic
libxl: debug: libxl_event.c:1738:libxl__ao__destroy: ao 0x68d910:
destroy
xc: debug: hypercall buffer: total allocations:29 total releases:29
xc: debug: hypercall buffer: current allocations:0 maximum
allocations:2
xc: debug: hypercall buffer: cache current size:2
xc: debug: hypercall buffer: cache hits:20 misses:2 toobig:7
If you need more informations/tests tell me and I'll post them.

Thanks for any reply and sorry for my bad english.


_______________________________________________
win-pv-devel mailing list
win-pv-devel@xxxxxxxxxxxxxxxxxxxx
http://lists.xenproject.org/cgi-bin/mailman/listinfo/win-pv-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.