Add space redefinition test, early NV read test and early extend test.

Add space redefinition test, early NV read test and early extend test.
Fix Makefile.

[Luigi Semenzato, 2010-06-07 15:31:09]

Hi Gaurav, here are some more tests.

[gauravsh, 2010-06-08 07:53:43]

lgtm

http://codereview.chromium.org/2663002/diff/6001/7001
File src/testsuite/Makefile (right):

http://codereview.chromium.org/2663002/diff/6001/7001#newcode34
src/testsuite/Makefile:34: tpmtest_redefine \
nit: alpha order

http://codereview.chromium.org/2663002/diff/6001/7002
File src/testsuite/earlyextend.c (right):

http://codereview.chromium.org/2663002/diff/6001/7002#newcode33
src/testsuite/earlyextend.c:33: } while (result == TPM_E_DOING_SELFTEST ||
so this tests keep on calling extend until it succeeds. But since it's a test,
shouldn't it fail if it doesn't succeed the first time?

[Luigi Semenzato, 2010-06-09 16:23:05]

On 2010/06/08 07:53:43, gauravsh wrote:
> lgtm
> 
> http://codereview.chromium.org/2663002/diff/6001/7001
> File src/testsuite/Makefile (right):
> 
> http://codereview.chromium.org/2663002/diff/6001/7001#newcode34
> src/testsuite/Makefile:34: tpmtest_redefine \
> nit: alpha order

Thanks---I was using creation order, but alpha order is better.

> 
> http://codereview.chromium.org/2663002/diff/6001/7002
> File src/testsuite/earlyextend.c (right):
> 
> http://codereview.chromium.org/2663002/diff/6001/7002#newcode33
> src/testsuite/earlyextend.c:33: } while (result == TPM_E_DOING_SELFTEST ||
> so this tests keep on calling extend until it succeeds. But since it's a test,
> shouldn't it fail if it doesn't succeed the first time?

I wrote this test to get the timing.  As a test, it doesn't have to fail if
TPM_Extend isn't immediately available---only if it takes too long to be
available.  So, what's "too long"?  200ms?

[gauravsh, 2010-06-09 18:33:56]

On 2010/06/09 16:23:05, Luigi Semenzato wrote:
> On 2010/06/08 07:53:43, gauravsh wrote:
> > lgtm
> > 
> > http://codereview.chromium.org/2663002/diff/6001/7001
> > File src/testsuite/Makefile (right):
> > 
> > http://codereview.chromium.org/2663002/diff/6001/7001#newcode34
> > src/testsuite/Makefile:34: tpmtest_redefine \
> > nit: alpha order
> 
> Thanks---I was using creation order, but alpha order is better.
> 
> > 
> > http://codereview.chromium.org/2663002/diff/6001/7002
> > File src/testsuite/earlyextend.c (right):
> > 
> > http://codereview.chromium.org/2663002/diff/6001/7002#newcode33
> > src/testsuite/earlyextend.c:33: } while (result == TPM_E_DOING_SELFTEST ||
> > so this tests keep on calling extend until it succeeds. But since it's a
test,
> > shouldn't it fail if it doesn't succeed the first time?
> 
> I wrote this test to get the timing.  As a test, it doesn't have to fail if
> TPM_Extend isn't immediately available---only if it takes too long to be
> available.  So, what's "too long"?  200ms?

I prefer having an explicit notion of what constitutes a test success (since it
is a test for something). I can think of 2 options for this  - 

1) Make sure TPM_Extend() either returns success or self test running error -
but not another error. This ensures that the only time TPM_Extend fails is if
self test is running and we already have ways to measure how long that takes.

2) I prefer this option. Explicitly set the duration to a reasonable value - as
long as it's a parameter (200 ms is fine) which can be changed easily, it should
be ok. This will also be useful later if we want to make this test a part of the
hwqual autotest suite.

[Luigi Semenzato, 2010-06-09 19:45:37]

I will do both in an upcoming CL.

On Wed, Jun 9, 2010 at 11:33 AM,  <gauravsh@chromium.org> wrote:
> On 2010/06/09 16:23:05, Luigi Semenzato wrote:
>>
>> On 2010/06/08 07:53:43, gauravsh wrote:
>> > lgtm
>> >
>> > http://codereview.chromium.org/2663002/diff/6001/7001
>> > File src/testsuite/Makefile (right):
>> >
>> > http://codereview.chromium.org/2663002/diff/6001/7001#newcode34
>> > src/testsuite/Makefile:34: tpmtest_redefine \
>> > nit: alpha order
>
>> Thanks---I was using creation order, but alpha order is better.
>
>> >
>> > http://codereview.chromium.org/2663002/diff/6001/7002
>> > File src/testsuite/earlyextend.c (right):
>> >
>> > http://codereview.chromium.org/2663002/diff/6001/7002#newcode33
>> > src/testsuite/earlyextend.c:33: } while (result == TPM_E_DOING_SELFTEST
>> > ||
>> > so this tests keep on calling extend until it succeeds. But since it's a
>
> test,
>>
>> > shouldn't it fail if it doesn't succeed the first time?
>
>> I wrote this test to get the timing.  As a test, it doesn't have to fail
>> if
>> TPM_Extend isn't immediately available---only if it takes too long to be
>> available.  So, what's "too long"?  200ms?
>
> I prefer having an explicit notion of what constitutes a test success (since
> it
> is a test for something). I can think of 2 options for this  -
>
> 1) Make sure TPM_Extend() either returns success or self test running error
> -
> but not another error. This ensures that the only time TPM_Extend fails is
> if
> self test is running and we already have ways to measure how long that
> takes.
>
> 2) I prefer this option. Explicitly set the duration to a reasonable value -
> as
> long as it's a parameter (200 ms is fine) which can be changed easily, it
> should
> be ok. This will also be useful later if we want to make this test a part of
> the
> hwqual autotest suite.
>
>
> http://codereview.chromium.org/2663002/show
>