[SSSD] [PATCH] Fix segmentation fault in test_io.

Jakub Hrozek jhrozek at redhat.com
Tue Apr 16 10:49:52 UTC 2013


On Mon, Apr 15, 2013 at 05:46:25PM +0530, Abhishek Singh wrote:
> Hi Jakub,
> 
> I am bit puzzled here, since all the test files get created/destroyed under
> TEST_DIR and runs well(absolute paths are already provided), so why do we
> need to change current working dir to TEST_DIR (use of tests_set_cwd)
> before running test?
> 

Mostly as an additional precation or in case the tests had some kind of
side-effect.

> 
> On Mon, Apr 15, 2013 at 3:58 PM, Jakub Hrozek <jhrozek at redhat.com> wrote:
> 
> > On Sat, Apr 13, 2013 at 02:38:45AM +0530, Abhishek Singh wrote:
> > > Please check this patch, previous patch contained extra changes which is
> > > not required.
> > >
> >
> > I would prefer if the src/conf_macros.m4 and src/tests/common.c changes
> > were a separate patch (with attribution set to Lukas) But more importantly:
> >
> > @@ -113,8 +121,8 @@ void test_sss_openat_cloexec_success(void **state)
> >      char path[PATH_MAX] = {'\0'};
> >      const char *relativepath;
> >
> > -    relativepath = strchr(get_filepath(path), 't');
> >      dir_fd = dirfd((DIR *)*state);
> > +    relativepath = get_filepath(path) + strlen(TEST_DIR) + 1;
> >
> > This is still guesswork, can you simply use basename() here? See man 3
> > basename,
> > I think you could easily call basename() on the return value of
> > get_filepath().
> >
> > Also the test still doesn't call tests_set_cwd(). It should call it right
> > before
> > calling run_tests, I think.
> >



More information about the sssd-devel mailing list