Re: [Qemu-devel] [PULL 00/26] Leak patches

[Marc-André Lureau]

Hi

----- Original Message -----
> On 6 September 2016 at 13:26, Marc-André Lureau
> <address@hidden> wrote:
> > The following changes since commit
> > 1fd66154fdf8305e6668a96046a22b863b4d7320:
> >
> >   Merge remote-tracking branch 'remotes/sstabellini/tags/xen-20160905' into
> >   staging (2016-09-06 11:43:18 +0100)
> >
> > are available in the git repository at:
> >
> >   address@hidden:elmarco/qemu.git tags/leak-pull-request
> >
> > for you to fetch changes up to 58d104a7f2698ffff3ee5976f96ebcff111be63b:
> >
> >   tests: fix postcopy-test leaks (2016-09-06 16:20:27 +0400)
> >
> > ----------------------------------------------------------------
> 
> I'm afraid this doesn't build with our minimum glib version:
> 
> /Users/pm215/src/qemu-for-merges/tests/libqtest.c:771:42: error: expected ')'
>                       (GTestFixtureFunc) fn, (GTestFixtureFunc)
>                       data_free_func);
>                                          ^
> /Users/pm215/src/qemu-for-merges/tests/libqtest.c:770:22: note: to
> match this '('
>     g_test_add_vtable(path, 0, data, NULL,
>                      ^
> 1 error generated.
> 
> The GTestFixtureFunc typedef was only introduced in glib 2.28, and our
> minimum is 2.22.

Argh,..

> 
> Also, g_test_add_vtable() in glib 2.22 has this prototype:
> 
> void    g_test_add_vtable               (const char     *testpath,
>                                          gsize           data_size,
>                                          gconstpointer   test_data,
>                                          void          (*data_setup)
>                                          (void),
>                                          void          (*data_test)
>                                          (void),
>                                          void          (*data_teardown)
>                                          (void));
> 
> but GTestFixtureFunc is typedefed in newer glib as
> void (*GTestFixtureFunc) (gpointer fixture, gconstpointer user_data);
> 
> so it looks like this function has changed signature somewhere
> between glib versions, which makes me a bit nervous about using it.

Perhaps we should get back to the simpler version, only using 
g_test_add_data_func_full() with 2.34:
https://patchwork.kernel.org/patch/9251373/

I can update the patch that way with a comment about expected leaks < 2.34.

there is no discussions about bumping the glib requirements btw?