[autotest] Add test for the re-taking of ownership after the owner key is lost

[autotest] Add test for the re-taking of ownership after the owner key is lost


BUG=chromium-os:11193
TEST=run this test

Change-Id: I95b402d47ec430a8556d012d9a2ee2614507a1e4

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=105706e

[Chris Masone, 2011-04-28 21:39:24]

PTAL

[petkov, 2011-04-28 22:16:25]

http://codereview.chromium.org/6880288/diff/2001/client/cros/constants.py
File client/cros/constants.py (right):

http://codereview.chromium.org/6880288/diff/2001/client/cros/constants.py#new...
client/cros/constants.py:1: # Copyright (c) 2010 The Chromium OS Authors. All
rights reserved.
2011

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py
File client/cros/ownership.py (right):

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
client/cros/ownership.py:104: def use_known_ownerkeys():
where is this routine used?

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
client/cros/ownership.py:122: return utils.read_file(os.path.join(dirname,
'mock_owner_private.key'))
mock_owner_private.key used in a couple of place -- define in a constant?

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
client/cros/ownership.py:129: return
cert_extract_pubkey_der(os.path.join(dirname, 'mock_owner_cert.pem'))
mock_owner_cert.pem in a constant?

[Chris Masone, 2011-04-28 22:43:51]

http://codereview.chromium.org/6880288/diff/2001/client/cros/constants.py
File client/cros/constants.py (right):

http://codereview.chromium.org/6880288/diff/2001/client/cros/constants.py#new...
client/cros/constants.py:1: # Copyright (c) 2010 The Chromium OS Authors. All
rights reserved.
On 2011/04/28 22:16:25, petkov wrote:
> 2011

Done.

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py
File client/cros/ownership.py (right):

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
client/cros/ownership.py:104: def use_known_ownerkeys():
On 2011/04/28 22:16:25, petkov wrote:
> where is this routine used?

login_OwnershipApi.py

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
client/cros/ownership.py:122: return utils.read_file(os.path.join(dirname,
'mock_owner_private.key'))
On 2011/04/28 22:16:25, petkov wrote:
> mock_owner_private.key used in a couple of place -- define in a constant?

Done.

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
client/cros/ownership.py:129: return
cert_extract_pubkey_der(os.path.join(dirname, 'mock_owner_cert.pem'))
On 2011/04/28 22:16:25, petkov wrote:
> mock_owner_cert.pem in a constant?

Done.

[petkov, 2011-04-28 22:48:47]

LGTM if my assumption about test order is correct.

It just looks that if you run OwnershipRetaken in isolation it will always
succeed because you never install the known keys (i.e., the current keys will
always be different than the known keys).

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py
File client/cros/ownership.py (right):

http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
client/cros/ownership.py:104: def use_known_ownerkeys():
On 2011/04/28 22:43:51, Chris Masone wrote:
> On 2011/04/28 22:16:25, petkov wrote:
> > where is this routine used?
> 
> login_OwnershipApi.py

So you rely on these test running in a specific sequence?

[Chris Masone, 2011-04-28 23:08:26]

Ah, I see the problem.

OwnershipRetaken is a UITest, and that class sets up all the fake keys
during initialization by default, unless you explicitly tell it not to.
 OwnershipApi is an OwnershipTest, so it manages all the keys on its own.

On Thu, Apr 28, 2011 at 3:48 PM, <petkov@chromium.org> wrote:

> LGTM if my assumption about test order is correct.
>
> It just looks that if you run OwnershipRetaken in isolation it will always
> succeed because you never install the known keys (i.e., the current keys
> will
> always be different than the known keys).
>
>
>
>
> http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py
> File client/cros/ownership.py (right):
>
>
>
http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
> client/cros/ownership.py:104: def use_known_ownerkeys():
> On 2011/04/28 22:43:51, Chris Masone wrote:
>
>> On 2011/04/28 22:16:25, petkov wrote:
>> > where is this routine used?
>>
>
>  login_OwnershipApi.py
>>
>
> So you rely on these test running in a specific sequence?
>
>
> http://codereview.chromium.org/6880288/
>

[petkov, 2011-04-28 23:10:03]

On 2011/04/28 23:08:26, Chris Masone wrote:
> Ah, I see the problem.
> 
> OwnershipRetaken is a UITest, and that class sets up all the fake keys
> during initialization by default, unless you explicitly tell it not to.
>  OwnershipApi is an OwnershipTest, so it manages all the keys on its own.


Got it -- refactoring and new code in a single CL confused me :-)

LGTM

> 
> On Thu, Apr 28, 2011 at 3:48 PM, <mailto:petkov@chromium.org> wrote:
> 
> > LGTM if my assumption about test order is correct.
> >
> > It just looks that if you run OwnershipRetaken in isolation it will always
> > succeed because you never install the known keys (i.e., the current keys
> > will
> > always be different than the known keys).
> >
> >
> >
> >
> > http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py
> > File client/cros/ownership.py (right):
> >
> >
> >
>
http://codereview.chromium.org/6880288/diff/2001/client/cros/ownership.py#new...
> > client/cros/ownership.py:104: def use_known_ownerkeys():
> > On 2011/04/28 22:43:51, Chris Masone wrote:
> >
> >> On 2011/04/28 22:16:25, petkov wrote:
> >> > where is this routine used?
> >>
> >
> >  login_OwnershipApi.py
> >>
> >
> > So you rely on these test running in a specific sequence?
> >
> >
> > http://codereview.chromium.org/6880288/
> >

[Chris Masone, 2011-04-29 19:39:06]

So, I thought about this more last night and realized that i _should_ have done
it as an OwnershipTest subclass, like you initially assumed.

This nontrivial refactor does that.  So...PTAL :-)

Sorry I wasted your time with the first revision :-/

[petkov, 2011-04-29 21:26:02]

LGTM