[PATCH v3] elf: Test dlopen (NULL, RTLD_LAZY) from an ELF constructor

[Sam James]

Florian Weimer <fweimer@redhat.com> writes:

> * Sam James:
>
>>> diff --git a/elf/tst-dlopen-constructor-null.c b/elf/tst-dlopen-constructor-null.c
>>> new file mode 100644
>>> index 0000000000..58c4660c7d
>>> --- /dev/null
>>> +++ b/elf/tst-dlopen-constructor-null.c
>>> @@ -0,0 +1,38 @@
>>> +/* Verify that dlopen (NULL, RTLD_LAZY) does not complete initialization.
>>> +   Copyright (C) 2024 Free Software Foundation, Inc.
>>> +   This file is part of the GNU C Library.
>>> +
>>> +   The GNU C Library is free software; you can redistribute it and/or
>>> +   modify it under the terms of the GNU Lesser General Public
>>> +   License as published by the Free Software Foundation; either
>>> +   version 2.1 of the License, or (at your option) any later version.
>>> +
>>> +   The GNU C Library is distributed in the hope that it will be useful,
>>> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
>>> +   Lesser General Public License for more details.
>>> +
>>> +   You should have received a copy of the GNU Lesser General Public
>>> +   License along with the GNU C Library; if not, see
>>> +   <https://www.gnu.org/licenses/>.  */
>>> +
>>> +/* This test mimics what the glvndSetupPthreads function in libglvnd
>>> +   does. */
>>
>> Have you confirmed that reverting the fixes breaks the test?
>
> If I cherry-pick the changes from commit 9897ced8e78db5d813166a7ccccfd
> (minus the elf/Makefile update for the tests, due to conflicts), I get
> this:
>
> FAIL: elf/tst-dlopen-constructor-null
> original exit status 1
> info: tst-dlopen-constructor-null-mod1.so constructor
> info: tst-dlopen-constructor-null-mod2.so constructor (mod1_status=0)error: mod1_status == 1 && mod2_status == 0 expected
>
> So I assume the test does exercise the bug.
>
>>> +#include <stdlib.h>
>>> +#include <support/check.h>
>>> +
>>> +/* Defined an initialized in the shared objects.  */
>>
>> "an initialized"?
>
> “and initialized”.  Fixed locally, thanks.

Thanks. You can consider this a review to unblock you given this
seems straightforward, or wait for someone else to look.

(Just giving the caveat as I don't want to mislead with my confidence on
it. I don't think there's anything wrong with the test.)