[PATCH] Make resolve and failover test work with CK_FORK=no

[Sumit Bose]

Hi,

the resolver and failover unit test are using a leak checker which calls
'talloc_free(talloc_autofree_context());'. This is not recommended and
makes the test fail it CK_FORK is set to 'no'. The attached patch
introduces a global talloc context which should be used by the test
instead of NULL for root of the memory hierarchy.

bye,
Sumit

[Stephen Gallagher]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 02/03/2010 07:51 AM, Sumit Bose wrote:
>  Hi,
>  
>  the resolver and failover unit test are using a leak checker which calls
>  'talloc_free(talloc_autofree_context());'. This is not recommended and
>  makes the test fail it CK_FORK is set to 'no'. The attached patch
>  introduces a global talloc context which should be used by the test
>  instead of NULL for root of the memory hierarchy.
>  
>  bye,
>  Sumit
>   
Ack. Tests run successfully in both fork and nofork mode.

- -- 
Stephen Gallagher
RHCE 804006346421761

Delivering value year after year.
Red Hat ranks #1 in value among software vendors.
http://www.redhat.com/promo/vendor/
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.10 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAktpkUUACgkQeiVVYja6o6NFGQCgrRJbvJIxGhc21yEhLYCLxCIm
> DjQAnRI58H6c+vVpeRkwZyl7J4cyxAYl
> =F0E9
> -----END PGP SIGNATURE-----

[Stephen Gallagher]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 02/03/2010 10:07 AM, Stephen Gallagher wrote:
>  On 02/03/2010 07:51 AM, Sumit Bose wrote:
> > Hi,
>  
> > the resolver and failover unit test are using a leak checker which calls
> > 'talloc_free(talloc_autofree_context());'. This is not recommended and
> > makes the test fail it CK_FORK is set to 'no'. The attached patch
> > introduces a global talloc context which should be used by the test
> > instead of NULL for root of the memory hierarchy.
>  
> > bye,
> > Sumit
>  
>  
>  Ack. Tests run successfully in both fork and nofork mode.
>   
Pushed to master.

- -- 
Stephen Gallagher
RHCE 804006346421761

Delivering value year after year.
Red Hat ranks #1 in value among software vendors.
http://www.redhat.com/promo/vendor/
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.10 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAktpmAkACgkQeiVVYja6o6OjRgCgmc8G2hBXwft0LxRrp5DBYjfe
> XNUAoJ0xza4Qy901JQ8mA3Smq5pfLET7
> =7lsr
> -----END PGP SIGNATURE-----

[Martin Nagy]

Sumit Bose wrote:
>  Hi,
>  
>  the resolver and failover unit test are using a leak checker which calls
>  'talloc_free(talloc_autofree_context());'. This is not recommended and
>  makes the test fail it CK_FORK is set to 'no'. The attached patch
>  introduces a global talloc context which should be used by the test
>  instead of NULL for root of the memory hierarchy.
>  
>  bye,
>  Sumit 
Little too late, but NACK. There is one missing detail.
From server/tests/common.h:
+TALLOC_CTX *global_talloc_context;

This will cause every file that includes common.h to define its own
global_talloc_context. I'm not surprised the tests run well, however.
My recommendation is that we add this to common.c:
TALLOC_CTX *global_talloc_context = NULL;
and add the 'extern' storage class specifier in common.h.

Sumit, can you please make another patch that will fix this problem?
(since the first patch was already acked & pushed)

Martin