This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Split self-dlopen tests from elf/tst-dlopen-aout
- From: Zack Weinberg <zackw at panix dot com>
- To: Florian Weimer <fweimer at redhat dot com>
- Cc: GNU C Library <libc-alpha at sourceware dot org>
- Date: Wed, 16 Oct 2019 13:10:31 -0400
- Subject: Re: [PATCH] Split self-dlopen tests from elf/tst-dlopen-aout
- References: <87v9so20im.fsf@oldenburg2.str.redhat.com>
On Wed, Oct 16, 2019 at 11:06 AM Florian Weimer <fweimer@redhat.com> wrote:
>
> From the beginning, elf/tst-dlopen-aout has exercised two different
> bugs: (a) failure to report errors for a dlopen of the executable
> itself in some cases (bug 24900) and (b) incorrect rollback of the
> TLS modid allocation in case of a dlopen failure (bug 16634).
>
> This patch retains elf/tst-dlopen-aout for (b) and introduces a new
> set of tests, elf/tst-dlopen-self, for (a). The elf/tst-dlopen-aout
> tests use the elf/tst-dlopen-self binaries (or iconv), so they are
> no longer self-dlopen tests.
I don't know the dynamic linker well enough to review this change in
detail. I like the idea of splitting up these two tests. I want to
suggest that the (b) test should be renamed to something more
descriptive at the same time as (a) is moved to its own binary.
tst-dlopen-aout sounds like it tests something to do with an ELF
executable trying to dlopen() an a.out-format shared library. (I
don't know if that was ever a thing that worked, but I could easily
believe that some ancient attempt to make it work had backward
compatibility implications that we're still stuck with, at least on
the older ABIs.)
zw