Re: linux-next: build warnings from the build of Linus' tree

From: Masami Hiramatsu
Date: Mon Aug 27 2018 - 20:43:48 EST


Hi Stephen,

On Tue, 28 Aug 2018 07:23:33 +1000
Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:

> Hi Masami,
>
> On Sun, 26 Aug 2018 17:10:58 +0900 Masami Hiramatsu <mhiramat@xxxxxxxxxx> wrote:
> >
> > Hmm, unfortunately, I couldn't reproduce the ".data..LPBX1" error with
> > my gcc-8.1.0 + ld-2.30 distributed at https://cdn.kernel.org/pub/tools/crosstool/.
> > Could you also try to enable CONFIG_GCOV_PROFILE_ALL=y and build the kernel?
> > If it causes similar error, it should be caused by gcov support.
>
> Using the allyesconfig as a base and disabling CONFIG_COMPILE_TESTING
> allowed me to enable COMFIG_GCOV_PROFILE_ALL, but that just generated a
> lot more messages. Some about data..Lusan sections (I think).

OK, so it sounds like GCOV kernel itself is broken with your toolchain
(native gcc-8.2.0 on ppc64?) as same as arm32.

I added Peter in loop, he is the maintainer of GCOV kernel.

Hi Peter,
I recently added a gcov profiling for ftrace, following Documentation/dev-tools/gcov.rst.
6b7dca401cb1 ("tracing: Allow gcov profiling on only ftrace subsystem")

and it caused may orphan section warnings/errors on arm and powerpc64.
We also found same error happens when CONFIG_GCOV_PROFILE_ALL=y.
So I guess GCOV kernel subsystem broken in some environment.

(cross build)
gcc-7.3.0 + ld-2.30 => arm:NG, arm64:OK, powerpc64:OK
gcc-8.1.0 + ld-2.30 => arm:NG, arm64:OK, powerpc64:OK
(native build?)
gcc-8.2.0 + ld 2.31.1 => powerpc64:NG
gcc-7.2.1 + ld 2.27 => x86-64: OK

ppc64 NG report:
https://lkml.org/lkml/2018/8/24/72

arm NG report:
https://lkml.org/lkml/2018/8/24/345

Would you know how we can fix this issue?

Thank you,

--
Masami Hiramatsu <mhiramat@xxxxxxxxxx>