Linux Test Project

On 10/15/07, Ricardo Salveti de Araujo <rsalveti@...> wrote:
> Hi all,
>
> Based on the discussion at lkml, we ported the old remap_file_pages test case
> created by Ingo Molnar to LTP.
>
> Subrata, I guess that you can apply this version at LTP, I already tested it
> at x86, x86_64 and ppc and it worked fine on all systems.
>

I had reviewed this test case with Subrata and provided some comments,
I see that those have been implemented. Looks good to me

Balbir

On Mon, 2008-02-25 at 16:21 +0100, Renaud Lottiaux wrote:
> Hi !
> 
> This patch fix a concurrency issue due to the (false) sharing of
> file /dev/shm/cache.
> 
> This patch just create a different file for each process and unlink
> the file before exiting.
> 

Thanks Renaud for these concurrency fixes. In fact you are helping to do
lots'of cleanup in LTP. I was thinking whether i can enter in to some
agreement with you for this. Hah Hah. Just joking. Yes, we need lots of
cleanup everywhere for some reason or the other.

Can i expect that you would keep providing me fixes for these
concurrency issue(s) in LTP Test cases in days to come. May be by few
months from now, we can certify that LTP Test cases are SMP-Safe ??

Masatake Yamato,

Would you also like to investigate in to this area of concurrency fixes
for LTP Test Cases ?

Regards--
Subrata

> Regards.
> 
> R.
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________ Ltp-list mailing list Ltp-list@... https://lists.sourceforge.net/lists/listinfo/ltp-list

Hi Subrata !

> Thanks Renaud for these concurrency fixes. In fact you are helping to do
> lots'of cleanup in LTP. I was thinking whether i can enter in to some
> agreement with you for this. Hah Hah. Just joking. Yes, we need lots of
> cleanup everywhere for some reason or the other.

Happy to see that my fixes are of interrest !
We won't need anymore to maintain work-around patches to make LTP tests
running on our system. Great ! :)

> Can i expect that you would keep providing me fixes for these
> concurrency issue(s) in LTP Test cases in days to come.

Yes, definitly.
I still have some patches. But most of our work-around are external stuffs.
Like for instance, a script which make sure each test starts in a separated
directory, for tests not using the tst_tmpdir function, but doing conflicting
stuffs in local directories.

But I plan to fix LTP tests to remove our work-around script.

> May be by few months from now, we can certify that LTP Test cases are
> SMP-Safe ??

I hope so :)
But don't you think that "concurrency-safe" would be a better term. The
patches I send have nothing to do with SMP (at user level). I do not fix
locking issues or this kind of stuff. I just make tests able to run
concurrently to stress the kernel level SMP-safeness.

Regards.

R.

-- 
Renaud Lottiaux

Kerlabs
Bâtiment Germanium
80, avenue des buttes de Coësmes
35700 Rennes - France
Phone : (+33|0)6 80 89 19 34
Fax   : (+33|0)2 99 84 71 71
Email : Renaud.Lottiaux@...
Web   : http://www.kerlabs.com/

On Tue, 2008-02-26 at 09:42 +0100, Renaud Lottiaux wrote:
> Hi Subrata !
> 
> > Thanks Renaud for these concurrency fixes. In fact you are helping to do
> > lots'of cleanup in LTP. I was thinking whether i can enter in to some
> > agreement with you for this. Hah Hah. Just joking. Yes, we need lots of
> > cleanup everywhere for some reason or the other.
> 
> Happy to see that my fixes are of interrest !
> We won't need anymore to maintain work-around patches to make LTP tests
> running on our system. Great ! :)

Oh yes. It saves you additional burden of maintaining work-around
patches for yourself only. When you fix them in LTP directly, you
benefit and hence do others as well.

> 
> > Can i expect that you would keep providing me fixes for these
> > concurrency issue(s) in LTP Test cases in days to come.
> 
> Yes, definitly.
> I still have some patches. But most of our work-around are external stuffs.
> Like for instance, a script which make sure each test starts in a separated
> directory, for tests not using the tst_tmpdir function, but doing conflicting
> stuffs in local directories.
> 
> But I plan to fix LTP tests to remove our work-around script.
> 
> > May be by few months from now, we can certify that LTP Test cases are
> > SMP-Safe ??
> 
> I hope so :)
> But don't you think that "concurrency-safe" would be a better term. The
> patches I send have nothing to do with SMP (at user level). I do not fix
> locking issues or this kind of stuff. I just make tests able to run
> concurrently to stress the kernel level SMP-safeness.

Oh yes. You are correct. It should better be called as concurrency-safe
rather than SMP-safe. 
Actually we should be more thankful to you for the cleanups that you are
doing.

Mike/Ricardo,

What is your take on these fixes ??

Regards--
Subrata

> 
> Regards.
> 
> R.
>

On Monday 25 February 2008, Renaud Lottiaux wrote:
> This patch fix a concurrency issue due to the (false) sharing of
> file /dev/shm/cache.
>
> This patch just create a different file for each process and unlink
> the file before exiting.

since LTP isnt multithreaded, hopefully we can rely on the PID being unique, 
which means the patch looks fine to me.  otherwise you'll have to use 
mktemp() to generate /dev/shm/cache.XXXXXX ...
-mike

On Tuesday 26 February 2008 07:19:40 Subrata Modak wrote:
>Mike/Ricardo,
>
>What is your take on these fixes ??

The patch looks fine for me too, np.

-- 
Ricardo Salveti de Araujo