Re: [RFC PATCH] semihosting/arm-compat: remove heuristic softmmu SYS_HEAPINFO

[Peter Maydell]

On Tue, 1 Jun 2021 at 11:07, Alex Bennée <alex.bennee@linaro.org> wrote:
>
>
> Peter Maydell <peter.maydell@linaro.org> writes:
>
> > On Tue, 1 Jun 2021 at 10:12, Alex Bennée <alex.bennee@linaro.org> wrote:
> >>
> >> The previous numbers were a guess at best. While we could extract the
> >> information from a loaded ELF file via -kernel we could still get
> >> tripped up by self decompressing or relocating code. Besides sane
> >> library code has access to the same symbols in run time to make a
> >> determination of the location of the heap.
> >>
> >> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
> >> Cc: Andrew <astrauss11@gmail.com>
> >> ---
> >>  semihosting/arm-compat-semi.c | 19 ++++++++++---------
> >>  1 file changed, 10 insertions(+), 9 deletions(-)
> >
> > This seems like a pretty good candidate for breaking existing
> > working binaries. How much testing against different varieties of
> > guest-code-using-semihosting have you done ?
>
> None, which is why it's an RFC - but at least one user reported newlib
> attempts to use the numbers we gave it rather than falling back to
> numbers it knew from the build and getting it wrong. I suspect any code
> that doesn't have a fallback path is getting it right more by luck than
> judgement though. I'd be curious to hear of code that relies on the
> numbers it gets from QEMU.

Well, newlib is the main one I had in mind -- does it have a fallback
codepath?

-- PMM