This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH v2 00/10] Undefined Behavior Sanitizer, this time with docs
On 10/08/2018 09:22 PM, Joel Brobecker wrote:
>> FWIW, I built GDB master today and ubsan (from LLVM, not GCC) flagged a
>> use of obstack_blank_fast() in minsyms.c with a negative offset (used to
>> shrink an obstack) when trying to do 'start' on /bin/ls:
>
> On my end, I am working on a resync of AdaCore's "head" version,
> which would include this change as well. My testing shows that
> it revealed some issues; I haven't had a chance to look into them yet
> (next on my list today, hopefully), but this is making me think
> there might be bona fide issues. Getting the location where things
> are happening is pretty nice :).
This also revealed PR 23743:
https://sourceware.org/bugzilla/show_bug.cgi?id=23743
I think the side-effects are blocking/disturbing enough (GDB dies) that I
think the best course of action is to temporarily disable the ubsan by
default until these issues are fixed. Then try re-enabling again, rinse/repeat.
Otherwise, as is, this is sort of imposing that the community prioritizes
fixing these ubsan-exposed bugs, which I think is unreasonable.
Thanks,
Pedro Alves