This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: Breakage on builder RHEL-s390x-m64, revision 7a6dbc2fdb2323c35e981f93236f323e9d7c0b24
On Wednesday, August 29 2018, Tom Tromey wrote:
>>>>>> "Sergio" == Sergio Durigan Junior <sergiodj@redhat.com> writes:
>
> Sergio> After my last commit, which updated GDB's local copy of gnulib, I'm
> Sergio> seeing a breakage on the RHEL 7.1 s390x builder (see the rest of the
> Sergio> quoted message). Unfortunately I don't have easy access to a similar
> Sergio> machine, so I would like to know if:
>
> Something fishy is going on that builder. I did a try run of my ubsan
> series and got many errors like this:
>
> (gdb) break check_vx
> /usr/include/c++/4.8.2/debug/vector:
> Error: attempt to subscript container with out-of-bounds index gdb:
> /home/dje/src/gcc-6-branch/libstdc++-v3/src/c++11/debug.cc:942: void
> {anonymous}::print_string({anonymous}::PrintContext&, const char*,
> const _Parameter*, std::size_t): Assertion `param_index <
> num_parameters' failed.
>
> However I think this should only happen if libstdc++ debug mode is
> enabled. (Is this correct? Maybe this is the error.)
Thanks, Tom. This may have something to do with the errors, I think.
> I did do some try runs in the past of libstdc++ debug mode. So, I am
> wondering if there could somehow be stale .o files around. I don't see
> how really -- I'd expect configure to make config.h, and for that to
> force a full rebuild. But it still looks suspicious.
Yeah, the BuildBot does a full rebuild for every commit. It deletes the
previous build directory before starting the next build.
--
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF 31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/