[Testing] Testing Summary, Auckland - 28 April 2012

Tom Parker tom at carrott.org
Sat Apr 28 18:17:06 EDT 2012


Testing Summary, Auckland - 28 April 2012
Who: Tom

We tested 12.1.0 for XO 1.5, customized (build 8) on XO-1.5 and XO-1.75.

Summary, details below:
Old prototype XO-1.5 hardware doesn’t boot.
One XO had very very small fonts.
One XO-1.5 had suspend-resume problems.
Browse works better on XO-1.5 than XO-1.75 but Google Docs still doesn’t 
work.
Open Firmware power log collection doesn’t work on Q3B22, works on Q3C05
Measure’s trigger and documenation on aslo are broken.
Chat stops receiving messages, claims nothing is wrong.
Laptops leave and re-join the Chat for no obvious reason.
Something corrupts the neighbourhood, duplicating XOs.


Details:
One of our XO-1.5s refused to fs-update os8, complaining about something 
to do with the first block and erasing or something. Unfortunately I 
observed this a few days ago and then when I came to re-visit, os8 just 
installed on all the XOs, so I don’t know what happened there.


poppy-copy (XO-1.5 B2) failed to boot after installing os8 complaining 
about read only filesystems. re-flashing did not help. I haven’t looked 
in to this in any detail.


rosella (XO-1.5) has very very small fonts on os8, even after a reboot


Ivy (XO-1.5) had a very bad resume from suspend, the laptop locked up 
for so long I thought it had crashed. I was about to hard reset it when 
processed some of my keyboard input and locked again. Eventually it 
started behaving normally but there was no eth0 for a few more minutes. 
After this it appeared to happy. I have the whole /var/log/messages if 
anyone wants it, relevant excerpt here:
Apr 27 23:28:54 xo-a7-4a-02 kernel: [  560.512719] Restarting tasks ...
Apr 27 23:28:54 xo-a7-4a-02 kernel: [  560.545299] olpc-dcon: The CPU 
has control
Apr 27 23:28:54 xo-a7-4a-02 kernel: [  560.563269] done.
Apr 27 23:28:54 xo-a7-4a-02 kernel: [  561.414231] dcon_freeze_store: 1
Apr 27 23:28:54 xo-a7-4a-02 kernel: [  561.417558] dcon_source_switch to 
DCON
Apr 27 23:28:54 xo-a7-4a-02 kernel: [  561.466091] olpc-dcon: The DCON 
has control
Apr 27 23:28:55 xo-a7-4a-02 kernel: [  561.819518] dcon_freeze_store: 1
Apr 27 23:28:55 xo-a7-4a-02 systemd[1]: Cannot add dependency job for 
unit fedora-storage-init.se
rvice, ignoring: Unit fedora-storage-init.service is masked.
Apr 27 23:28:55 xo-a7-4a-02 systemd[1]: Cannot add dependency job for 
unit fedora-storage-init-la
te.service, ignoring: Unit fedora-storage-init-late.service is masked.
Apr 27 23:28:55 xo-a7-4a-02 systemd[1]: Cannot add dependency job for 
unit plymouth-start.service
, ignoring: Unit plymouth-start.service is masked.
Apr 27 23:28:55 xo-a7-4a-02 systemd[1]: Cannot add dependency job for 
unit fedora-autorelabel-mar
k.service, ignoring: Unit fedora-autorelabel-mark.service is masked.
Apr 27 23:28:55 xo-a7-4a-02 systemd[1]: Cannot add dependency job for 
unit fedora-autorelabel.ser
vice, ignoring: Unit fedora-autorelabel.service is masked.
Apr 27 23:28:55 xo-a7-4a-02 systemd[1]: Cannot add dependency job for 
unit fedora-loadmodules.ser
vice, ignoring: Unit fedora-loadmodules.service is masked.
Apr 27 23:28:56 xo-a7-4a-02 kernel: [  563.440109] libertas_sdio 
mmc1:0001:1: eth0: command 0x000
6 timed out
Apr 27 23:28:56 xo-a7-4a-02 kernel: [  563.447104] libertas_sdio 
mmc1:0001:1: eth0: Timeout submi
tting command 0x0006
Apr 27 23:28:56 xo-a7-4a-02 kernel: [  563.455070] libertas_sdio 
mmc1:0001:1: eth0: PREP_CMD: command 0x0006 failed: -110
Apr 27 23:28:59 xo-a7-4a-02 kernel: [  566.470106] libertas_sdio 
mmc1:0001:1: eth0: command 0x0024 timed out
Apr 27 23:28:59 xo-a7-4a-02 kernel: [  566.477108] libertas_sdio 
mmc1:0001:1: eth0: Timeout submitting command 0x0024
Apr 27 23:28:59 xo-a7-4a-02 kernel: [  566.485406] libertas_sdio 
mmc1:0001:1: eth0: PREP_CMD: command 0x0024 failed: -110


Browse on os8 on XO-1.5 is much more stable than on XO-1.75. However 
Google Docs on XO-1.5 does nothing when you try to create or open a 
document -- browse continues to work but no document opens. This error 
is logged a lot of times:

** Message: console message:  @0: Unable to post message to 
https://3.docs.google.com. Recipient has origin https://docs.google.com.

which might be related.

On XO-1.75 Browse crashes before it renders the list of documents in 
google docs.

alt-left does not go back, is there a list of keyboard shortcuts somewhere?


power log collection does not work on rosella with Q3B22 and au215, it 
fails with <buffer at ....>:194: Can’t open directory. After updating to 
Q3C05 without reflashing the operating system, the power log collector 
works. Rolling back to Q3B22 restores the original error. This strongly 
suggests it is firmware related.


Measure’s trigger doesn’t seem to work at all now. Also, if you reduce 
either gain to zero the display stops updating until you turn off the 
trigger. Restoring the gain does not help. Measure uses about 15% cpu 
even when it is hidden, maybe it should stop processing when in the 
background?

The record feature doesn’t seem to do anything.

Measure’s page on aslo points to http://wiki.laptop.org/go/Measure which 
claims to be out of date and points back to aslo. Is there any 
up-to-date documentation?


Had no problems with the neighbourhood view putting 10 XO-1.75s and 1 
XO-1.5 onto a WPA-2 access point. The access point is a TP-Link running 
openwrt 10.03.1 r29592).


Laptops don’t wake up to multicast traffic so turned off power saving on 
all of them.


Joined 10 or so XO-1.75s to an instance of Chat, left them idle for an 
hour or so. On most laptops the chat interface shows a number of laptops 
left and joined the chat without any user interaction. One left and 
didn’t come back. That laptop has a bunch or errors in it’s log:

/home/olpc/Activities/Chat.activity/activity.py:90: Warning: 
g_object_set_qdata: assertion `G_IS_OBJECT (object)' failed
   toolbar_box.toolbar.insert(StopButton(self), -1)
Traceback (most recent call last):
   File "/home/olpc/Activities/Chat.activity/activity.py", line 309, in 
entry_activate_cb
	self.text_channel.send(text)
   File "/home/olpc/Activities/Chat.activity/activity.py", line 373, in send
	CHANNEL_TEXT_MESSAGE_TYPE_NORMAL, text)
   File "/usr/lib/python2.7/site-packages/dbus/proxies.py", line 140, in 
__call__
	**keywords)
   File "/usr/lib/python2.7/site-packages/dbus/connection.py", line 630, 
in call_blocking
	message, timeout)
dbus.exceptions.DBusException: 
org.freedesktop.DBus.Error.ServiceUnknown: The name :1.32 was not 
provided by any .service files

Looking with tcpdump we see that there is a lot of multicast traffic. 
The laptop that left the chat sees mdns traffic but does not see the 
chat traffic, the laptops still in the chat do see the chat traffic. 
This is likely caused by chat closing it’s socket, since tcpdump does 
not work in promiscuous mode, we only see multicast traffic that is 
destined for an open socket. It took me a while to discover this, so I 
was not able to confirm that the error above is consistent with chat 
closing the socket and so stopping the traffic showing up in tcpdump.

The traffic observed with tcpump on the access point itself is 
consistent with that observed by a working laptop in the chat. The 
access point does support promiscuous mode with tcpdump.

Leaving the network and rejoining it does not cause the laptop to start 
receiving the chat traffic but the chat log on the the other laptops in 
the chat show the broken one leave and then re-join.

The neighbourhood is behaving very strangely, the circle of XOs around 
the chat gets denser and a gap appears. One XO sees 12 XOs in a half 
circle around the chat and two more not joined to the chat and itself 
for a total of 15 XOs. A number are duplicated! Each XO’s neighbourhood 
shows a different number of duplicate XOs. We reported something similar 
in the olpc australia build where one laptop appeared many times in a 
collaboration, maybe this is the same problem?

Restarting the laptop that fell out of the chat was interesting. When it 
first joined the wifi network it did not see any other XOs. It is seeing 
a bunch of multicast mdns traffic, but not the chat traffic and the 
neighbourhood was empty of XOs. About 5 minutes later they all appear 
and with the normal number of laptops in the chat. There are a lot of 
errors like this in shell.log:

1335581697.226971 ERROR dbus.connection: Exception in handler for D-Bus 
signal:
Traceback (most recent call last):
   File "/usr/lib/python2.7/site-packages/dbus/connection.py", line 214, 
in maybe_handle_message
	self._handler(*args, **kwargs)
   File "/usr/lib/python2.7/site-packages/jarabe/model/neighborhood.py", 
line 430, in __buddy_info_updated_cb
	self.emit('buddy-updated', self._buddy_handles[handle], properties)
KeyError: dbus.UInt32(5L)

On the laptops which reported other laptops leaving and joining the 
chat, and on the laptops that reportedly left and joined the chat, I 
don’t see anything in any of the logs to suggest the reason for leaving 
and joining. Specifically there is no evidence that they fell off the 
wifi network and re-joined.

So, some bugs:

Something causes Chat to close it’s socket and mislead the user into 
thinking everything is ok.
Something causes XOs to leave and join the chat for no obvious reason.
Something (perhaps related to the leaving and joining?) corrupts the 
neighbourhood view.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: power-logs.tar.bz2
Type: application/x-bzip
Size: 20927 bytes
Desc: not available
URL: <http://lists.laptop.org/pipermail/testing/attachments/20120429/5fe5edb6/attachment-0001.bin>


More information about the Testing mailing list