[PATCH v2 03/12] sandbox: Migrate getopt section to linker list

[Andrew Scull]

Looking at this again with a recent fetch of master, LTO is deleting
the linker list entries for the getopt declarations. Both clang and
gcc do this. I've not got an idea why at the moment and would love to
hear any suggestions.

On Mon, 2 May 2022 at 17:27, Tom Rini <trini at konsulko.com> wrote:
>
> On Mon, May 02, 2022 at 09:24:52AM -0700, Andrew Scull wrote:
> > On Fri, 29 Apr 2022 at 08:11, Tom Rini <trini at konsulko.com> wrote:
> > >
> > > On Thu, Apr 14, 2022 at 01:59:32PM +0000, Andrew Scull wrote:
> > >
> > > > Use the common infrastructure to create a linker list of the sandbox
> > > > command line flags rather than using a custom method.
> > > >
> > > > The list is changed from containing pointers to containing structs and
> > > > the uses are updated accordingly.
> > > >
> > > > Signed-off-by: Andrew Scull <ascull at google.com>
> > > > Reviewed-by: Simon Glass <sjg at chromium.org>
> > > > ---
> > > >  arch/sandbox/cpu/os.c               | 21 ++++++++++-----------
> > > >  arch/sandbox/cpu/start.c            | 10 +++++-----
> > > >  arch/sandbox/cpu/u-boot-spl.lds     |  6 ------
> > > >  arch/sandbox/cpu/u-boot.lds         |  6 ------
> > > >  arch/sandbox/include/asm/getopt.h   | 19 ++++++++++++-------
> > > >  arch/sandbox/include/asm/sections.h | 25 -------------------------
> > > >  6 files changed, 27 insertions(+), 60 deletions(-)
> > >
> > > This particular test causes all of the pytest infrastructure to fail.  I
> > > believe "make qcheck" will show this, even, for a possibly easier
> > > reproducer.
> >
> > I didn't run this series through the CI so I hadn't been running the
> > pytest suite. `make qcheck` does run the pytest suite and shows a
> > bunch more ASAN failures. Some are simple but others are less obvious
> > and are going to need some time to dig into.
> >
> > The issues being found by ASAN here are in the main u-boot code, not
> > just the test code, and a couple of them are in crypto components so
> > there might be some more interesting bugs to discover.
>
> OK, yeah, ASAN failures within the test suite are less important.  I
> forgot to mention maybe I disabled ASAN in sandbox and still got the
> basic CI failure I mentioned above, bisect'd down to here.  Please make
> sure CI does pass on this for the next iteration and it's OK to go with
> documenting how to enable ASAN afterwards and that tests need further
> clean-up.
>
> --
> Tom