11:49 p.m.
--- On Wed, 7/29/09, Mikus Grinbergs <mikus(a)bga.com> wrote:
From: Mikus Grinbergs <mikus(a)bga.com>
Subject: Re: F11 for the X01- shutdown
To: fedora-olpc-list(a)redhat.com
Cc: "Yioryos Asprobounitis" <mavrothal(a)yahoo.com>
Date: Wednesday, July 29, 2009, 8:25 PM
> Yet another log of random
shutdown.
I noticed from the /var/log/messages content enclosed in
your recent post that your processor is running at 497.488
MHz. My XO-1s are running at around 430.911 MHz.
Are you overclocking your XO-1 ?
Fro the last 1 year+ :)
The olpc.fth is in my Ubuntu/swap containing SDcard (follows)
_However_ both random shutdown and X problems are happening also without the SDcard when
the XO is using the original f110XO1/os2 olpc.fth
--------------------------------------------------
SDcard olpc.fth:
8:18 p.m.
New subject: F11 for the X01- shutdown
Yioryos, it seems you and I are the only ones to see this problem, and
there are others who are not seeing it.
I wonder if it is environmental ... that is it is triggered also by
something else in the environment.
Do you have the units associated with an access point?
What is the manufacturer and model of the access point, and what
software is it running?
Are there any other laptops nearby that are running older versions of
the wireless firmware? For instance I've a B2 on camera sentry duty in
another room running build 406.
--
James Cameron
http://quozl.linux.org.au/
11:49 p.m.
New subject: F11 for the X01- shutdown
--- On Thu, 7/30/09, James Cameron <quozl(a)laptop.org> wrote:
From: James Cameron <quozl(a)laptop.org>
Yioryos, it seems you and I are the
only ones to see this problem, and
there are others who are not seeing it.
This is good :)
I wonder if it is environmental ... that is it is triggered
also by
something else in the environment.
Do you have the units associated with an access point?
What is the manufacturer and model of the access point, and
what
software is it running?
So far a PHILIPS CGA5722N/TE with WEP security (the software version info will take some
time) and a public point (SpeedTouch-no other info available.Sorry)
Are there any other laptops nearby that are running older
versions of
the wireless firmware? For instance I've a B2 on
camera sentry duty in
another room running build 406.
There area other machines connected but no other XOs as far as I can tell.
Do you also get "eth0 no private ioctls" during boot? I do not remember that
with 767/802
12:15 a.m.
New subject: F11 for the X01- shutdown
james wrote:
Yioryos, it seems you and I are the only ones to see this problem,
and
there are others who are not seeing it.
reuben caron (at 1cc) mentioned that he'd seen the shutdown thing
too. (i believe he's set up logging in case it happens again.)
I wonder if it is environmental ... that is it is triggered also by
something else in the environment.
i think you said you'd strace'd init, and had seen the shutdown
signal, correct? i wonder if it would be worth stracing the most
likely (e.g., under heavy development) candidates: network
manager, X11, others? the goal, of course, being to find out who
sent the signal to init.
(while i'm clearly relieved that powerd is definitely not
involved in some of these shutdowns, it would have narrowed the
search a lot if it _was_ the likely culprit. :-)
paul
=---------------------
paul fox, pgf(a)laptop.org
2:24 a.m.
New subject: F11 for the X01- shutdown
I had another one in the middle of the session as soon as I connected to the WiFi and
tried to load a page.
So far in 6 incidences it never happened in GNOME (statistically it should have if it was
a Fedora issue) never without wifi and never with long established (>3min) wifi. The
frequency is about 1 out of 10 attempts to launch an activity (presence/collaboration?) or
load a web page in Browse (I also use Firefox,Midori under Sugar-though the frequency
might not warrant an event) within 1min after establishing a network connection.
So although NM is involved I do not think is responsible (see GNOME).
Just my 2 cents.
--- On Fri, 7/31/09, Paul Fox <pgf(a)laptop.org> wrote:
From: Paul Fox <pgf(a)laptop.org>
Subject: Re: F11 for the X01- shutdown
To: fedora-olpc-list(a)redhat.com
Date: Friday, July 31, 2009, 1:15 AM
james wrote:
> Yioryos, it seems you and I are the only ones to see
this problem, and
> there are others who are not seeing it.
reuben caron (at 1cc) mentioned that he'd seen the shutdown
thing
too. (i believe he's set up logging in case it
happens again.)
>
> I wonder if it is environmental ... that is it is
triggered also by
> something else in the environment.
i think you said you'd strace'd init, and had seen the
shutdown
signal, correct? i wonder if it would be worth
stracing the most
likely (e.g., under heavy development) candidates:
network
manager, X11, others? the goal, of course, being to
find out who
sent the signal to init.
(while i'm clearly relieved that powerd is definitely not
involved in some of these shutdowns, it would have narrowed
the
search a lot if it _was_ the likely culprit. :-)
paul
=---------------------
paul fox, pgf(a)laptop.org
_______________________________________________
Fedora-olpc-list mailing list
Fedora-olpc-list(a)redhat.com
https://www.redhat.com/mailman/listinfo/fedora-olpc-list
3:41 a.m.
New subject: F11 for the X01- shutdown
2009/7/31 Yioryos Asprobounitis <mavrothal(a)yahoo.com>:
I had another one in the middle of the session as soon as I connected to the WiFi and
tried to load a page.
So far in 6 incidences it never happened in GNOME (statistically it should have if it was
a Fedora issue) never without wifi and never with long established (>3min) wifi. The
frequency is about 1 out of 10 attempts to launch an activity (presence/collaboration?) or
load a web page in Browse (I also use Firefox,Midori under Sugar-though the frequency
might not warrant an event) within 1min after establishing a network connection.
So although NM is involved I do not think is responsible (see GNOME).
Just my 2 cents.
I have a theory on this bug which I am pretty confident about:
olpc-update recently grew code to detect when a laptop is stolen. When
the laptop is stolen, it will delete the lease and power off.
See /usr/bin/olpc-update-query, the code near:
# this machine has been reported stolen - get rid of
# any leases and poweroff
I suggest removing the call to halt and adding some logging instead,
to confirm this theory. If you could also log the value of
str(resp_map) and show us this then it will help us diagnose the
issue. Either activation.laptop.org is saying that your laptop is
stolen, or there is some bug on the client side where it is
misinterpreting the response and thinking that it has been stolen.
Thanks,
Daniel
5:31 a.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 02:26:13PM +0545, Daniel Drake wrote:
See /usr/bin/olpc-update-query, the code near:
# this machine has been reported stolen - get rid of
# any leases and poweroff
I think you're right. Here's why:
[root@localhost ~]# ls -l /security/
total 0
-rw-r--r-- 1 root root 0 Jul 31 06:02 update-attempt
[root@localhost ~]# /usr/sbin/olpc-update-query
Error setting OHM inhibit: org.freedesktop.DBus.Error.ServiceUnknown:
The name org.freedesktop.ohm was not provided by any .service files
Broadcast message from root(a)localhost.localdomain
(/dev/pts/0) at 6:23 ...
The system is going down for halt NOW!
[root@localhost ~]#
I suggest removing the call to halt and adding some logging instead,
to confirm this theory. If you could also log the value of
str(resp_map) and show us this then it will help us diagnose the
issue.
{'nonce': '6cb0c08d464152816a040ad071d4f46e', 'stolen':
'7df745eed2112c9fd3c356c35dda7cbdffa3c496d08e58641550fc0f651a12f4',
'time': '20090731T102917Z'}
{'nonce': '58aaf6558ab743a335bf5d5859a64a69', 'stolen':
'8038107e352154e6a0b3319cca4a20d884b095ce85268b9fc0e54eacde487cca',
'time': '20090731T103027Z'}
Either activation.laptop.org is saying that your laptop is
stolen, or there is some bug on the client side where it is
misinterpreting the response and thinking that it has been stolen.
Both units don't do this with build 802.
--
James Cameron
http://quozl.linux.org.au/
6:29 a.m.
New subject: F11 for the X01- shutdown
2009/7/31 James Cameron <quozl(a)laptop.org>:
On Fri, Jul 31, 2009 at 02:26:13PM +0545, Daniel Drake wrote:
> See /usr/bin/olpc-update-query, the code near:
> # this machine has been reported stolen - get rid of
> # any leases and poweroff
I think you're right. Here's why:
[root@localhost ~]# ls -l /security/
total 0
-rw-r--r-- 1 root root 0 Jul 31 06:02 update-attempt
[root@localhost ~]# /usr/sbin/olpc-update-query
Error setting OHM inhibit: org.freedesktop.DBus.Error.ServiceUnknown:
The name org.freedesktop.ohm was not provided by any .service files
Broadcast message from root(a)localhost.localdomain
(/dev/pts/0) at 6:23 ...
The system is going down for halt NOW!
[root@localhost ~]#
> I suggest removing the call to halt and adding some logging instead,
> to confirm this theory. If you could also log the value of
> str(resp_map) and show us this then it will help us diagnose the
> issue.
{'nonce': '6cb0c08d464152816a040ad071d4f46e', 'stolen':
'7df745eed2112c9fd3c356c35dda7cbdffa3c496d08e58641550fc0f651a12f4',
'time': '20090731T102917Z'}
{'nonce': '58aaf6558ab743a335bf5d5859a64a69', 'stolen':
'8038107e352154e6a0b3319cca4a20d884b095ce85268b9fc0e54eacde487cca',
'time': '20090731T103027Z'}
I think I see the problem. I think martin has misinterpreted the theft
deterrence protocol.
In check_stolen_hash(), please modify strings to be:
strings = ( uuid + ':' + nonce + ':STOLEN' )
i.e. remove the first entry, which actually indicates non-stolen.
Does that help?
Thanks,
Daniel
6:56 a.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 05:14:07PM +0545, Daniel Drake wrote:
I think I see the problem. I think martin has misinterpreted the
theft
deterrence protocol.
Per "rpm -qf /usr/sbin/olpc-update-query" one can see that it is from
package olpc-update-2.19-1.fc11.noarch
Is there an earlier version?
Where can I find the src.rpm and the upstream source?
In check_stolen_hash(), please modify strings to be:
strings = ( uuid + ':' + nonce + ':STOLEN' )
i.e. remove the first entry, which actually indicates non-stolen.
Does that help?
No.
Just to confirm the change you requested, here's the diff ...
# diff -u /usr/sbin/olpc-update-query.orig /usr/sbin/olpc-update-query
--- /usr/sbin/olpc-update-query.orig 2009-07-31 07:51:27.000000000 -0400
+++ /usr/sbin/olpc-update-query 2009-07-31 07:51:51.000000000 -0400
@@ -125,8 +125,7 @@
def check_stolen_hash(hash, nonce, uuid):
- strings = ( uuid + ':' + nonce,
- uuid + ':' + nonce + ':STOLEN' )
+ strings = ( uuid + ':' + nonce + ':STOLEN' )
for msg in strings:
if hash == hashlib.sha256(msg).hexdigest():
return True
#
And the output still shows it would have halted and str(resp_map) shows
{'nonce': '55462871e50934b75f96ff311776adc5', 'stolen':
'19e8b87862c9d321f5aeec5de8d405b40de4b0a7f06c2dd6ab4a170aba85716e',
'time': '20090731T115207Z'}
--
James Cameron
http://quozl.linux.org.au/
7:54 a.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 5:56 AM, James Cameron<quozl(a)laptop.org> wrote:
And the output still shows it would have halted and str(resp_map)
shows
{'nonce': '55462871e50934b75f96ff311776adc5', 'stolen':
'19e8b87862c9d321f5aeec5de8d405b40de4b0a7f06c2dd6ab4a170aba85716e',
'time': '20090731T115207Z'}
It does sound like the server response says stolen. What does the
'data' variable -- which will hold the raw server response -- contain?
Arguably, if the machines are jailbroken, then olpc-update-query
shouldn't care. Clearly your machines are devkey'd.
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
10:39 a.m.
New subject: F11 for the X01- shutdown
--- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com> wrote:>
It does sound like the server response says stolen. What
does the
'data' variable -- which will hold the raw server response
-- contain?
Arguably, if the machines are jailbroken, then
olpc-update-query
shouldn't care. Clearly your machines are devkey'd.
Well,... I _believe_ I had security disabled in OFW long time ago but I did not have a
secyrity key installed in this build (I do now :) We'll see how it goes.
However if this is the problem why is not doing it all the time? Is it checking only once
per day? Come to think of it I _think_ I had one sudden shutdown per day. We'll
see...
BTW, wget is missing from the build
Also getting the key from laptop.org I get the response
Resolving activation.laptop.org... 18.85.46.43
Connecting to activation.laptop.org|18.85.46.43|:443... connected.
ERROR: cannot verify activation.laptop.org's certificate, issued by `/CN=One Laptop
Per Child Root Certification Authority/C=US/O=One Laptop Per Child Association,
Inc./ST=Massachusetts/L=Cambridge/emailAddress=ca(a)laptop.org':
Unable to locally verify the issuer's authority.
To connect to activation.laptop.org insecurely, use `--no-check-certificate'.
Unable to establish SSL connection.
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org
-- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working
code first
- http://wiki.laptop.org/go/User:Martinlanghoff
10:43 a.m.
New subject: F11 for the X01- shutdown
Yioryos -
activation.laptop.org does indeed have an expired security
certificate. We're working on fixing that, but it is a surprisingly
non-trivial task. In the meantime, follow the instructions and use
the --no-check-certificate option.
- Ed
On Jul 31, 2009, at 11:39 AM, Yioryos Asprobounitis wrote:
--- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com>
wrote:>
> It does sound like the server response says stolen. What
> does the
> 'data' variable -- which will hold the raw server response
> -- contain?
>
> Arguably, if the machines are jailbroken, then
> olpc-update-query
> shouldn't care. Clearly your machines are devkey'd.
>
Well,... I _believe_ I had security disabled in OFW long time ago
but I did not have a secyrity key installed in this build (I do
now :) We'll see how it goes.
However if this is the problem why is not doing it all the time? Is
it checking only once per day? Come to think of it I _think_ I had
one sudden shutdown per day. We'll see...
BTW, wget is missing from the build
Also getting the key from laptop.org I get the response
Resolving activation.laptop.org... 18.85.46.43
Connecting to activation.laptop.org|18.85.46.43|:443... connected.
ERROR: cannot verify activation.laptop.org's certificate, issued by
`/CN=One Laptop Per Child Root Certification Authority/C=US/O=One
Laptop Per Child Association,
Inc./ST=Massachusetts/L=Cambridge/emailAddress=ca(a)laptop.org
':
Unable to locally verify the issuer's authority.
To connect to activation.laptop.org insecurely, use `--no-check-
certificate'.
Unable to establish SSL connection.
>
>
> m
> --
> martin.langhoff(a)gmail.com
> martin(a)laptop.org
> -- School Server Architect
> - ask interesting questions
> - don't get distracted with shiny stuff - working
> code first
> - http://wiki.laptop.org/go/User:Martinlanghoff
>
_______________________________________________
Fedora-olpc-list mailing list
Fedora-olpc-list(a)redhat.com
https://www.redhat.com/mailman/listinfo/fedora-olpc-list
10:49 a.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 9:39 AM, Yioryos
Asprobounitis<mavrothal(a)yahoo.com> wrote:
--- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com> wrote:>
> It does sound like the server response says stolen. What
> does the
> 'data' variable -- which will hold the raw server response
> -- contain?
>
> Arguably, if the machines are jailbroken, then
> olpc-update-query
> shouldn't care. Clearly your machines are devkey'd.
>
Well,... I _believe_ I had security disabled in OFW long time
Can you dump the contents of 'data' for us? If the server seems to
claim your XO is stolen, part of the question is why?
cheers,
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
11:08 a.m.
New subject: F11 for the X01- shutdown
--- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com> wrote:
From: Martin Langhoff <martin.langhoff(a)gmail.com>
Subject: Re: F11 for the X01- shutdown
To: "Yioryos Asprobounitis" <mavrothal(a)yahoo.com>
Cc: "James Cameron" <quozl(a)laptop.org>, "Daniel Drake"
<dsd(a)laptop.org>, fedora-olpc-list(a)redhat.com
Date: Friday, July 31, 2009, 11:49 AM
On Fri, Jul 31, 2009 at 9:39 AM,
Yioryos
Asprobounitis<mavrothal(a)yahoo.com>
wrote:
>
> --- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com>
wrote:>
>> It does sound like the server response says
stolen. What
>> does the
>> 'data' variable -- which will hold the raw server
response
>> -- contain?
>>
>> Arguably, if the machines are jailbroken, then
>> olpc-update-query
>> shouldn't care. Clearly your machines are
devkey'd.
>>
>
> Well,... I _believe_ I had security disabled in OFW
long time
Can you dump the contents of 'data' for us? If the server
seems to
claim your XO is stolen, part of the question is why?
If you tell me what exactly you want and how I get it, with pleasure. (I know is a dev
thread but I'm far from one :)
Also, by now I do have the devkey installed
cheers,
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org
-- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working
code first
- http://wiki.laptop.org/go/User:Martinlanghoff
11:24 a.m.
New subject: F11 for the X01- shutdown
Actually issuing
[root@localhost activity]# /usr/sbin/olpc-update-query
consistently shuts down my XO!
And I do have the devkey installed
--- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com> wrote:
From: Martin Langhoff <martin.langhoff(a)gmail.com>
Subject: Re: F11 for the X01- shutdown
To: "Yioryos Asprobounitis" <mavrothal(a)yahoo.com>
Cc: "James Cameron" <quozl(a)laptop.org>, "Daniel Drake"
<dsd(a)laptop.org>, fedora-olpc-list(a)redhat.com
Date: Friday, July 31, 2009, 11:49 AM
On Fri, Jul 31, 2009 at 9:39 AM,
Yioryos
Asprobounitis<mavrothal(a)yahoo.com>
wrote:
>
> --- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com>
wrote:>
>> It does sound like the server response says
stolen. What
>> does the
>> 'data' variable -- which will hold the raw server
response
>> -- contain?
>>
>> Arguably, if the machines are jailbroken, then
>> olpc-update-query
>> shouldn't care. Clearly your machines are
devkey'd.
>>
>
> Well,... I _believe_ I had security disabled in OFW
long time
Can you dump the contents of 'data' for us? If the server
seems to
claim your XO is stolen, part of the question is why?
cheers,
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org
-- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working
code first
- http://wiki.laptop.org/go/User:Martinlanghoff
1:57 p.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 10:24 AM, Yioryos
Asprobounitis<mavrothal(a)yahoo.com> wrote:
Actually issuing
[root@localhost activity]# /usr/sbin/olpc-update-query
consistently shuts down my XO!
And I do have the devkey installed
Yes, that's consistent with the diagnosis we've been making.
If you can edit your olpc-update-query script (make a backup!) find
the line that calls 'halt', comment it out (prefixing it with a #) and
add right under it, with the same indentation:
report(0, 'Server said stolen: %s' % data)
the whole block of code will look like
if isxo and 'stolen' in resp_map:
check_stolen_hash(resp_map['stolen'], uuid, nonce)
# this machine has been reported stolen - get rid of
# any leases and poweroff
if os.path.exists(LEASE_FILE):
os.unlink(LEASE_FILE)
#call(['/usr/bin/halt'])
report(0, 'Server said stolen: %s' % data)
exit()
cheers,
m
--- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com> wrote:
> From: Martin Langhoff <martin.langhoff(a)gmail.com>
> Subject: Re: F11 for the X01- shutdown
> To: "Yioryos Asprobounitis" <mavrothal(a)yahoo.com>
> Cc: "James Cameron" <quozl(a)laptop.org>, "Daniel Drake"
<dsd(a)laptop.org>, fedora-olpc-list(a)redhat.com
> Date: Friday, July 31, 2009, 11:49 AM
> On Fri, Jul 31, 2009 at 9:39 AM,
> Yioryos
> Asprobounitis<mavrothal(a)yahoo.com>
> wrote:
> >
> > --- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com>
> wrote:>
> >> It does sound like the server response says
> stolen. What
> >> does the
> >> 'data' variable -- which will hold the raw server
> response
> >> -- contain?
> >>
> >> Arguably, if the machines are jailbroken, then
> >> olpc-update-query
> >> shouldn't care. Clearly your machines are
> devkey'd.
> >>
> >
> > Well,... I _believe_ I had security disabled in OFW
> long time
>
> Can you dump the contents of 'data' for us? If the server
> seems to
> claim your XO is stolen, part of the question is why?
>
> cheers,
>
>
> m
> --
> martin.langhoff(a)gmail.com
> martin(a)laptop.org
> -- School Server Architect
> - ask interesting questions
> - don't get distracted with shiny stuff - working
> code first
> - http://wiki.laptop.org/go/User:Martinlanghoff
>
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
2:29 p.m.
New subject: F11 for the X01- shutdown
--- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com> wrote:
From: Martin Langhoff <martin.langhoff(a)gmail.com>
Subject: Re: F11 for the X01- shutdown
To: "Yioryos Asprobounitis" <mavrothal(a)yahoo.com>
Cc: "James Cameron" <quozl(a)laptop.org>, "Daniel Drake"
<dsd(a)laptop.org>, fedora-olpc-list(a)redhat.com
Date: Friday, July 31, 2009, 2:57 PM
On Fri, Jul 31, 2009 at 10:24 AM,
Yioryos
Asprobounitis<mavrothal(a)yahoo.com>
wrote:
>
> Actually issuing
> [root@localhost activity]#
/usr/sbin/olpc-update-query
> consistently shuts down my XO!
> And I do have the devkey installed
Yes, that's consistent with the diagnosis we've been
making.
If you can edit your olpc-update-query script (make a
backup!) find
the line that calls 'halt', comment it out (prefixing it
with a #) and
add right under it, with the same indentation:
report(0, 'Server said stolen: %s'
% data)
the whole block of code will look like
if isxo and 'stolen' in resp_map:
check_stolen_hash(resp_map['stolen'], uuid, nonce)
# this machine has been
reported stolen - get rid of
# any leases and poweroff
if os.path.exists(LEASE_FILE):
os.unlink(LEASE_FILE)
#call(['/usr/bin/halt'])
report(0, 'Server said stolen:
%s' % data)
exit()
This works too in NOT crashing my XO upon /usr/sbin/olpc-update-query.
Which one is the best theft deterrent is up to you guys.
cheers,
m
>
> --- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com>
wrote:
>
>> From: Martin Langhoff <martin.langhoff(a)gmail.com>
>> Subject: Re: F11 for the X01- shutdown
>> To: "Yioryos Asprobounitis" <mavrothal(a)yahoo.com>
>> Cc: "James Cameron" <quozl(a)laptop.org>,
"Daniel Drake" <dsd(a)laptop.org>,
fedora-olpc-list(a)redhat.com
>> Date: Friday, July 31, 2009, 11:49 AM
>> On Fri, Jul 31, 2009 at 9:39 AM,
>> Yioryos
>> Asprobounitis<mavrothal(a)yahoo.com>
>> wrote:
>> >
>> > --- On Fri, 7/31/09, Martin Langhoff <martin.langhoff(a)gmail.com>
>> wrote:>
>> >> It does sound like the server response
says
>> stolen. What
>> >> does the
>> >> 'data' variable -- which will hold the
raw server
>> response
>> >> -- contain?
>> >>
>> >> Arguably, if the machines are jailbroken,
then
>> >> olpc-update-query
>> >> shouldn't care. Clearly your machines
are
>> devkey'd.
>> >>
>> >
>> > Well,... I _believe_ I had security disabled
in OFW
>> long time
>>
>> Can you dump the contents of 'data' for us? If the
server
>> seems to
>> claim your XO is stolen, part of the question is
why?
>>
>> cheers,
>>
>>
>> m
>> --
>> martin.langhoff(a)gmail.com
>> martin(a)laptop.org
>> -- School Server Architect
>> - ask interesting questions
>> - don't get distracted with shiny stuff -
working
>> code first
>> - http://wiki.laptop.org/go/User:Martinlanghoff
>>
>
>
>
>
--
martin.langhoff(a)gmail.com
martin(a)laptop.org
-- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working
code first
- http://wiki.laptop.org/go/User:Martinlanghoff
6:11 p.m.
New subject: F11 for the X01- shutdown
Martin wrote:
Can you dump the contents of 'data' for us? If the server
seems to
claim your XO is stolen, part of the question is why?
It seems the other poster didn't answer this, so here's what I get.
Server said stolen: ['oatc-resp', 1, {'nonce':
'e2e89084ce657bb521f671ef9772670b', 'stolen':
'dbd1f60dfa73f5ca775c9a2dd0f11a63db76d0aa499a52bf5188257552305473',
'time': '20090731T230912Z'}]
--
James Cameron
http://quozl.linux.org.au/
7:23 p.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 5:11 PM, James Cameron<quozl(a)laptop.org> wrote:
Martin wrote:
> Can you dump the contents of 'data' for us? If the server seems to
> claim your XO is stolen, part of the question is why?
It seems the other poster didn't answer this, so here's what I get.
Server said stolen: ['oatc-resp', 1, {'nonce':
'e2e89084ce657bb521f671ef9772670b', 'stolen':
'dbd1f60dfa73f5ca775c9a2dd0f11a63db76d0aa499a52bf5188257552305473',
'time': '20090731T230912Z'}]
This is _so damn weird_. The server at OLPC claims your XO is stolen.
CC'in Reuben as I will need access to that server to see WTH is going
on there.
Look James, it's ok, you can tell us now; where _did_ you get that XO from?
BTW, I can't check of the hash 'checks out' with what Bitfrost
expects, which would be nice to know. Can you test that for me?
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
7:46 p.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 06:23:20PM -0600, Martin Langhoff wrote:
On Fri, Jul 31, 2009 at 5:11 PM, James
Cameron<quozl(a)laptop.org> wrote:
> It seems the other poster didn't answer this, so here's what I get.
>
> Server said stolen: ['oatc-resp', 1, {'nonce':
> 'e2e89084ce657bb521f671ef9772670b', 'stolen':
> 'dbd1f60dfa73f5ca775c9a2dd0f11a63db76d0aa499a52bf5188257552305473',
> 'time': '20090731T230912Z'}]
This is _so damn weird_. The server at OLPC claims your XO is stolen.
CC'in Reuben as I will need access to that server to see WTH is going
on there.
Look James, it's ok, you can tell us now; where _did_ you get that XO from?
Most recent unit is QUOZL E, which arrived last week from Adam Holt and
John Wadlington at OLPC. Reproduced now on three units, all of which
were issued under the contributors programme. Serial numbers and my
labels are
CSN75001153 QUOZL E C2
CSN74800DFD QUOZL A C2
CSN75000153 QUOZL 9 C2
olpc-update-query has not been successful on any of my units with this
build, and all have developer keys and /ofw mounted.
The previous version of olpc-update-query in build 802 does not generate
this condition.
BTW, I can't check of the hash 'checks out' with what
Bitfrost
expects, which would be nice to know. Can you test that for me?
No idea, sorry.
--
James Cameron
http://quozl.linux.org.au/
7:55 p.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 6:46 PM, James Cameron<quozl(a)laptop.org> wrote:
> Look James, it's ok, you can tell us now; where _did_ you get
that XO from?
Most recent unit is QUOZL E, which arrived last week from Adam Holt and
John Wadlington at OLPC. Reproduced now on three units, all of which
were issued under the contributors programme. Serial numbers and my
labels are
CSN75001153 QUOZL E C2
CSN74800DFD QUOZL A C2
CSN75000153 QUOZL 9 C2
I was joking but excellent that you've given us the SNs so I can poke
at the database on antitheft.laptop.org as soon as Reuben gets me
access there.
olpc-update-query has not been successful on any of my units with
this
build, and all have developer keys and /ofw mounted.
Well, it _is_ successful in the sense that the server states the XO is
stolen, and it removes any lease the XO may have, and then shuts down.
The previous version of olpc-update-query in build 802 does not
generate
this condition.
Correct. The old build was happy to ignore message saying "you're
stolen". Big chunks of the antitheft sw were not in place.
Now that they are, it'd arguably be a sane thing to skip certain
actions if the machine is devkey'd.
> BTW, I can't check of the hash 'checks out' with what
Bitfrost
> expects, which would be nice to know. Can you test that for me?
No idea, sorry.
No prob. Once I have access to antitheft.l.o I will try that out myself.
cheers,
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
1:15 p.m.
New subject: F11 for the X01- shutdown
2009/7/31 James Cameron <quozl(a)laptop.org>:
On Fri, Jul 31, 2009 at 05:14:07PM +0545, Daniel Drake wrote:
> I think I see the problem. I think martin has misinterpreted the theft
> deterrence protocol.
Per "rpm -qf /usr/sbin/olpc-update-query" one can see that it is from
package olpc-update-2.19-1.fc11.noarch
Is there an earlier version?
Where can I find the src.rpm and the upstream source?
It's a standard Fedora package.
> In check_stolen_hash(), please modify strings to be:
> strings = ( uuid + ':' + nonce + ':STOLEN' )
> i.e. remove the first entry, which actually indicates non-stolen.
>
> Does that help?
No.
Your diff is correct but I see a 2nd bug too: the return value of
check_stolen_hash is never checked.
handmade diff for you to test:
- if isxo and 'stolen' in resp_map:
- check_stolen_hash(resp_map['stolen'], uuid, nonce)
+ if isxo and 'stolen' in resp_map and
check_stolen_hash(resp_map['stolen'], uuid, nonce):
# this machine has been reported stolen - get rid of
let me know if you need more details...
1:54 p.m.
New subject: F11 for the X01- shutdown
--- On Fri, 7/31/09, Daniel Drake <dsd(a)laptop.org> wrote:
From: Daniel Drake <dsd(a)laptop.org>
Subject: Re: F11 for the X01- shutdown
To: "James Cameron" <quozl(a)laptop.org>
Cc: "Yioryos Asprobounitis" <mavrothal(a)yahoo.com>,
fedora-olpc-list(a)redhat.com, "Martin Langhoff"
<martin.langhoff(a)gmail.com>
Date: Friday, July 31, 2009, 2:15 PM
2009/7/31 James Cameron <quozl(a)laptop.org>:
> On Fri, Jul 31, 2009 at 05:14:07PM +0545, Daniel Drake
wrote:
>> I think I see the problem. I think martin has
misinterpreted the theft
>> deterrence protocol.
>
> Per "rpm -qf /usr/sbin/olpc-update-query" one can see
that it is from
> package olpc-update-2.19-1.fc11.noarch
>
> Is there an earlier version?
>
> Where can I find the src.rpm and the upstream source?
It's a standard Fedora package.
>> In check_stolen_hash(), please modify strings to
be:
>> strings = ( uuid + ':' + nonce + ':STOLEN'
)
>> i.e. remove the first entry, which actually
indicates non-stolen.
>>
>> Does that help?
>
> No.
Your diff is correct but I see a 2nd bug too: the return
value of
check_stolen_hash is never checked.
handmade diff for you to test:
- if isxo and 'stolen' in resp_map:
-
check_stolen_hash(resp_map['stolen'],
uuid, nonce)
+ if isxo and 'stolen' in resp_map and
check_stolen_hash(resp_map['stolen'], uuid, nonce):
# this machine has been reported stolen - get rid of
let me know if you need more details...
I do not know if with this patch the XO is still protected but certainly
/usr/sbin/olpc-update-query in NOT crashing my XO anymore
6:06 p.m.
New subject: F11 for the X01- shutdown
On Sat, Aug 01, 2009 at 12:00:19AM +0545, Daniel Drake wrote:
2009/7/31 James Cameron <quozl(a)laptop.org>:
> Where can I find the src.rpm and the upstream source?
It's a standard Fedora package.
Bugger. I always get that answer, and it never satisfies. I'm stalled,
unable to find the src.rpm and the upstream source, I ask, and someone
says "it's a Fedora package", as if to say "you ought to know and
I'm
not going to tell you what you need to know to proceed." ;-)
Your diff is correct but I see a 2nd bug too: the return value of
check_stolen_hash is never checked.
handmade diff for you to test:
- if isxo and 'stolen' in resp_map:
- check_stolen_hash(resp_map['stolen'], uuid, nonce)
+ if isxo and 'stolen' in resp_map and
check_stolen_hash(resp_map['stolen'], uuid, nonce):
# this machine has been reported stolen - get rid of
let me know if you need more details...
That fixed it. Running olpc-update-query manually no longer reports the
machine as stolen.
The change was:
--- olpc-update-query.orig 2009-07-31 23:02:51.000000000 -0400
+++ olpc-update-query 2009-07-31 23:02:31.000000000 -0400
@@ -190,8 +190,8 @@
if 'nonce' not in resp_map or nonce != resp_map['nonce']:
raise RuntimeError('bad nonce in reply')
- if isxo and 'stolen' in resp_map:
- check_stolen_hash(resp_map['stolen'], uuid, nonce)
+ if isxo and 'stolen' in resp_map and \
+ check_stolen_hash(resp_map['stolen'], uuid, nonce):
# this machine has been reported stolen - get rid of
# any leases and poweroff
if os.path.exists(LEASE_FILE):
--
James Cameron
http://quozl.linux.org.au/
6:13 p.m.
New subject: F11 for the X01- shutdown
Hi,
Bugger. I always get that answer, and it never satisfies. I'm
stalled, unable to find the src.rpm and the upstream source, I
ask, and someone says "it's a Fedora package", as if to say "you
ought to know and I'm not going to tell you what you need to know
to proceed." ;-)
Well, hey, this is fedora-olpc-list. Expecting that people will know
how to find source for a Fedora package or be willing to perform a web
search to find out can't be *that* unreasonable. :) Here's the answer:
Go to http://koji.fedoraproject.org/, type "olpc-update" into the
top-right search field, hit search, click on the build you want,
and you can download the source and binary RPMs.
If you're running a Fedora machine, or have access to one (such as
an XO), it's easier:
pullcord:cjb~ % yumdownloader --source olpc-utils
olpc-utils-1.0.3-2.fc12.src.rpm | 55 kB 00:00
pullcord:cjb~ %
- Chris.
--
Chris Ball <cjb(a)laptop.org>
6:22 p.m.
New subject: F11 for the X01- shutdown
Fantastic, thanks!
--
James Cameron
http://quozl.linux.org.au/
7:19 p.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 5:06 PM, James Cameron<quozl(a)laptop.org> wrote:
On Sat, Aug 01, 2009 at 12:00:19AM +0545, Daniel Drake wrote:
> 2009/7/31 James Cameron <quozl(a)laptop.org>:
> > Where can I find the src.rpm and the upstream source?
>
> It's a standard Fedora package.
Bugger. I always get that answer, and it never satisfies. I'm stalled,
unable to find the src.rpm and the upstream source, I ask, and someone
says "it's a Fedora package", as if to say "you ought to know and
I'm
not going to tell you what you need to know to proceed." ;-)
As all of my work (and all of OLPC's work) it's published in our git
repo. And it's straight from the master branch:
http://dev.laptop.org/git/projects/olpc-update/
> Your diff is correct but I see a 2nd bug too: the return value
of
> check_stolen_hash is never checked.
Good catch.
That fixed it. Running olpc-update-query manually no longer reports
the
machine as stolen.
But! The server response _does_ seem to say 'stolen'.
Am I missing anything? [I may well be, I am in the middle of a
training week where I am giving a "how to setup and run an XS/XO
deployment", been talking all day about configuration and
nandblasting, and now I am away from my dev/test environment. :-/ ]
cheers,
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
11:47 p.m.
New subject: F11 for the X01- shutdown
2009/8/1 Martin Langhoff <martin.langhoff(a)gmail.com>:
But! The server response _does_ seem to say 'stolen'.
Am I missing anything?
Yes.
Read http://wiki.laptop.org/go/Theft_deterrence_protocol carefully.
The wording is not blindingly clear but it does explain.
It advises that a server *always* sends a 'stolen' tag in the message,
so that a man-in-the-middle cannot simply filter out messages
including "stolen"
This is also why a hash is used, in order to make the messages always
the same length and the contents non-obvious.
Of the two possible values of stolen:
sha256(<uuid>:<nonce>) means NOT stolen
sha256(<uuid>:<nonce>:STOLEN) means stolen
Judging from James's feedback (i.e. the patch that he tested) the
server is only sending the "not stolen" tag. I would imagine this is
true for everyone seeing this issue.
Daniel
11:55 p.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 10:47 PM, Daniel Drake<dsd(a)laptop.org> wrote:
Read http://wiki.laptop.org/go/Theft_deterrence_protocol carefully.
The wording is not blindingly clear but it does explain.
You are right. The wording is clear actually, but it's a few
paragraphs down. I see what's happening.
Judging from James's feedback (i.e. the patch that he tested)
the
server is only sending the "not stolen" tag. I would imagine this is
true for everyone seeing this issue.
And the missing test for the result of the hash check is what is killing us.
Ah, grumble...
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
5:16 a.m.
New subject: F11 for the X01- shutdown
On Fri, Jul 31, 2009 at 06:19:16PM -0600, Martin Langhoff wrote:
As all of my work (and all of OLPC's work) it's published in
our git
repo. And it's straight from the master branch:
http://dev.laptop.org/git/projects/olpc-update/
Thanks. I've checked it out and reviewed it. I've a trivial trailing
whitespace fix, but I'm not a member of the olpc-update group, so I
could not commit directly to the master branch. Please pull from
http://dev.laptop.org/~quozl/olpc-update.git/
Looking at the changes since C. Scott Ananian last worked on it, there's
quite a bit of change, but no obvious sign of test coverage ... so I'll
keep an eye out.
--
James Cameron
http://quozl.linux.org.au/
1:38 a.m.
New subject: F11 for the X01- shutdown
2009/8/1 James Cameron <quozl(a)laptop.org>:
That fixed it. Running olpc-update-query manually no longer reports
the
machine as stolen.
Thanks! one final request..please test this RPM update before I push
it to Fedora
olpc-update-2.20-1.fc11.noarch.rpm from
http://koji.fedoraproject.org/koji/taskinfo?taskID=1571473
cheers
Daniel
3:55 a.m.
New subject: F11 for the X01- shutdown
On Sat, Aug 01, 2009 at 12:23:17PM +0545, Daniel Drake wrote:
olpc-update-2.20-1.fc11.noarch.rpm from
http://koji.fedoraproject.org/koji/taskinfo?taskID=1571473
Tested, fixes the problem, it no longer halts:
[root@localhost ~]# olpc-update-query
Error setting OHM inhibit: org.freedesktop.DBus.Error.ServiceUnknown:
The name org.freedesktop.ohm was not provided by any .service files
[root@localhost ~]# md5sum /usr/sbin/olpc-update-query
252f69b36898458617ccc3320f39ca9f /usr/sbin/olpc-update-query
[root@localhost ~]# rpm -q olpc-update
olpc-update-2.20-1.fc11.noarch
[root@localhost ~]#
Now, just to satisfy my curiosity ... how did you do that ... that is,
how did you produce an rpm for download?
--
James Cameron
http://quozl.linux.org.au/
8:52 a.m.
New subject: F11 for the X01- shutdown
2009/8/1 James Cameron <quozl(a)laptop.org>:
On Sat, Aug 01, 2009 at 12:23:17PM +0545, Daniel Drake wrote:
> olpc-update-2.20-1.fc11.noarch.rpm from
> http://koji.fedoraproject.org/koji/taskinfo?taskID=1571473
Tested, fixes the problem, it no longer halts:
[root@localhost ~]# olpc-update-query
Error setting OHM inhibit: org.freedesktop.DBus.Error.ServiceUnknown:
The name org.freedesktop.ohm was not provided by any .service files
[root@localhost ~]# md5sum /usr/sbin/olpc-update-query
252f69b36898458617ccc3320f39ca9f /usr/sbin/olpc-update-query
[root@localhost ~]# rpm -q olpc-update
olpc-update-2.20-1.fc11.noarch
[root@localhost ~]#
Now, just to satisfy my curiosity ... how did you do that ... that is,
how did you produce an rpm for download?
Thanks for testing. I checked out the olpc-update repository from
Fedora cvs, bumped the spec file to point at the new version
(including make new-sources) locally, made a SRPM from the modified
spec, then created a scratch build using the koji command line client
and gave you the URL.
Now I can commit the changes, request a "real" koji build, and push it
as a F11 update.
http://fedoraproject.org/wiki/PackageMaintainers/UpdatingPackageHowTo
http://fedoraproject.org/wiki/Koji
thanks,
Daniel
8:15 a.m.
New subject: F11 for the X01- shutdown
On Sat, Aug 1, 2009 at 12:38 AM, Daniel Drake<dsd(a)laptop.org> wrote:
2009/8/1 James Cameron <quozl(a)laptop.org>:
> That fixed it. Running olpc-update-query manually no longer reports the
> machine as stolen.
Thanks! one final request..please test this RPM update before I push
it to Fedora
Thanks! Patch makes sense.
m
--
martin.langhoff(a)gmail.com
martin(a)laptop.org -- School Server Architect
- ask interesting questions
- don't get distracted with shiny stuff - working code first
- http://wiki.laptop.org/go/User:Martinlanghoff
5397
days inactive
5399
days old
33 comments
7 participants
participants (7)
-
Chris Ball -
Daniel Drake -
Ed McNierney -
James Cameron -
Martin Langhoff -
Paul Fox -
Yioryos Asprobounitis