GSOC 2017: Project Proposal

[sandhya bankar]

On Fri, Apr 7, 2017 at 7:27 AM, Dmitry V. Levin <ldv at altlinux.org> wrote:
> On Wed, Apr 05, 2017 at 10:07:22AM +0530, sandhya bankar wrote:
>> On Mon, Apr 3, 2017 at 9:43 PM, Eugene Syromiatnikov <esyr at redhat.com> wrote:
>> > On Mon, Apr 03, 2017 at 09:26:59PM +0530, sandhya bankar wrote:
>> >> On Mon, Apr 3, 2017 at 9:10 PM, Eugene Syromiatnikov <esyr at redhat.com> wrote:
>> >> > On Mon, Apr 03, 2017 at 08:51:34PM +0530, sandhya bankar wrote:
>> >> >> To deal with code coverage, I will run individual test cases to find
>> >> >> out why the coverage is below 100%.
>> >> >
>> >> > OK, let's be more specific: what are your suggestions in regards of
>> >> > increasing coverage of dm.c up to 100%?
>> >>
>> >> First I will check  https://codecov.io/gh/strace/strace/src/master/dm.c
>> >>
>> >> This will give us information about which lines are hit, lines are
>> >> missed  by testcase. So according to this data, we can add more tests
>> >> to hit all missed lines. On dashboard this lines highlighted by Red
>> >> color so It is easy to write test cases around those lines.
>> > So, how do you expect to write test that checks dm_decode_dm_target_deps
>> > function, for example?
>>
>> Hi,
>>
>>  I have checked the test case. The test case to check
>> dm_decode_dm_target_deps is already present in ioctl_dm.c.
>>
>> I am suspecting that, it seems there is problem in below case
>>
>> case DM_TABLE_DEPS:
>>                 if (entering(tcp) || syserror(tcp))
>>                       break;
>>                 dm_decode_dm_target_deps(tcp, arg, ioc);
>>                 break;
>>
>> When we come to check the dm_decode_dm_target_deps(). The switch case
>> is breaking before calling it.
>>
>> I have not debug it completely but both the cases of the if()
>> statement somehow becoming true and that's why we are not able to call
>> the dm_decode_dm_target_deps(). (I am not sure whether It is correct
>> behavior )
>>
>> I did one hack to traverse the dm_decode_dm_target_deps() code path. I
>> have commented the both lines of switch case as below. Now the code
>> can check the dm_decode_dm_target_deps().
>>
>> I know it is not the right way to fix this problem, but after more
>> debugging I can find the reason and fix it.
>
> I cannot imagine anything as remote as this.  Very much surprised to hear
> this from a person with experience of patching kernel drivers.


Hi,

 What I understood from last question is that how we should improve
code coverage of the dm_decode_dm_target_spec() by the test case. I
had checked the test ioctl_dm.* and found that the test case is
present there. So I just figured out why the function is not getting
traverse. I am not proposing any solution here. I was just found the
reason why it is not getting traversed by doing this little
experiment.

As you suggested after setting the proper dm setup on virtual machine
same test will get pass. In that case the entering() and syserror()
will became false and we will eventually call the
dm_decode_dm_target_deps().

I was not aware of this virtual machine thing, It is matter of
experience and knowledge of the specific subsystem.




Thanks,
Sandhya
>
>> diff --git a/dm.c b/dm.c
>> index a69077b..7e3fb53 100644
>> --- a/dm.c
>> +++ b/dm.c
>> @@ -516,8 +518,8 @@ dm_known_ioctl(struct tcb *const tcp, const
>> unsigned int code,
>>                 dm_decode_dm_target_spec(tcp, arg, ioc);
>>                 break;
>>         case DM_TABLE_DEPS:
>> -               if (entering(tcp) || syserror(tcp))
>> -                       break;
>> +               //if (entering(tcp) || syserror(tcp))
>> +               //      break;
>>                 dm_decode_dm_target_deps(tcp, arg, ioc);
>>                 break;
>>         case DM_LIST_DEVICES:
>>
>> Please suggest on it.
>
> The most simple way to test the exiting part of DM_TABLE_DEPS
> is to run a virtual machine with a proper dm setup inside.
>
>
> --
> ldv
>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Strace-devel mailing list
> Strace-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/strace-devel
>