Re: x86 - clang / objtool status

[Thomas Gleixner]

On Thu, 18 Jul 2019, Nick Desaulniers wrote:
> On Thu, Jul 18, 2019 at 1:40 PM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
> > after picking up Josh's objtool updates I gave clang a test ride again.
> 
> Thanks for testing and the reports; these are valuable and we
> appreciate the help debugging them.
> 
> > 2) debian distro config
> 
> Is this checked into the tree, or where can I find it?

See below.

> >   drivers/gpu/drm/radeon/evergreen_cs.o: warning: objtool: evergreen_cs_parse() falls through to next function evergreen_dma_cs_parse()
> 
> fall through warnings look new to me, but Linaro's KernelCI is
> currently screaming with tons of reports of -Wfallthrough throughout
> the kernel.  I assume they're related?

I don't think so. The compiler does not warn about a missing fallthrough.

> > 3) allmodconfig:
> >
> >   drivers/gpu/drm/i915/gem/i915_gem_execbuffer.o: warning: objtool: .altinstr_replacement+0x56: redundant UACCESS disable
> 
> Do you still have these object files laying around? Josh asked for
> them in a new thread (from the previous thread), not sure if it's ok
> to attach object files to emails to LKML? (html email is not allowed,
> are binary attachments?)

I've uploaded the configs and object files to:

     https://tglx.de/~tglx/clang.tar.bz2

contains:

clang/
clang/debian/
clang/debian/atom.o
clang/debian/platform.o
clang/debian/i915_gem_execbuffer.o
clang/debian/.config
clang/debian/evergreen_cs.o
clang/allmod/
clang/allmod/ubsan.o
clang/allmod/i915_gem_execbuffer.o
clang/allmod/.config
clang/allmod/sata_dwc_460ex.o
clang/allmod/common.o
clang/allmod/signal.o
clang/allmod/ia32_signal.o

i.e. the .config and the offending object files for the debian and
allmodconfig builds.

Thanks,

	tglx