Re: [PATCH RFC 00/11] Add Memory Sanitizer support

[Ilya Leoshkevich <iii@linux.ibm.com>]

On Tue, 2023-02-07 at 20:05 +0100, Mark Wielaard wrote:
> Hi Ilya,
> 
> On Mon, Feb 06, 2023 at 11:25:02PM +0100, Ilya Leoshkevich via
> Elfutils-devel wrote:
> > This series adds minimalistic support for Memory Sanitizer (MSan)
> > [1].
> > MSan is compiler instrumentation for detecting accesses to
> > uninitialized memory.

[...]

> > - Minimal configuration of elfutils instrumented with MSan:
> > 
> >   autoreconf -i
> >   CC=clang ./configure --enable-maintainer-mode \
> >                        --enable-sanitize-memory --without-bzlib \
> >                        --without-lzma --without-zstd \
> >                        --disable-debuginfod --disable-libdebuginfod
> > \
> >                        --disable-demangler
> 
> Aren't there instrumented versions of bzip2, lzma/xz and/or zstd?
> 
> Can't debuginfod and libdebuginfod be instrumented?
> 
> Is the demangler disabled because you don't link against (an
> instrumented) libstdc++?

I think with some effort instrumenting the dependencies is possible.
bzlib and lzma are not particularly large, and zstd should support
this out of the box. Regarding C++, an instrumented LLVM's libc++
should also just work. With all this, it should be possible to test
elfutils with MSan without disabling the extra functionality.

But since you already test with valgrind, I figured it would be highly
unlikely that I find new bugs, and decided to limit the scope here.
For my current purposes - linking elfutils into libbpf - this proved
to be enough.

[...]

Best regards,
Ilya