Re: Segfault in pahole 1.18 when building kernel 5.9.1 for arm64

[Jiri Slaby]

On 20. 10. 20, 19:15, Andrii Nakryiko wrote:
> On Tue, Oct 20, 2020 at 3:51 AM Jiri Slaby <jirislaby@xxxxxxxxxx> wrote:
> > Hi,
> >
> > On 19. 10. 20, 1:18, Érico Rolim wrote:
> > > I'm trying to build kernel 5.9.1 for arm64, and my dotconfig has
> > > `CONFIG_DEBUG_INFO_BTF=y`, which requires pahole for building. However, pahole
> > > version 1.18 segfaults during the build, as can be seen below:
> > >
> > > PAHOLE: Error: Found symbol of zero size when encoding btf (sym:
> > > '__kvm_nvhe_arm64_ssbd_callback_required', cu:
> > > 'arch/arm64/kernel/cpu_errata.c').
> >
> > The symbol is an alias coming from arch/arm64/kernel/vmlinux.lds:
> > __kvm_nvhe_arm64_ssbd_callback_required = arm64_ssbd_callback_required;;
>
> What's readelf's output for that symbol? If it's legal for SST_OBJECT
> to have size zero, then we should just skip those in pahole. But it
> shouldn't crash in either case, of course. But as Arnaldo mentioned,
> that code changed significantly recently, so please check with latest
> pahole from tmp.libbtf_encoder branch.
...
> > Yeah, I observe the very same. I reported it at:
> > https://bugzilla.suse.com/show_bug.cgi?id=1177921

If you looked here, you would see:
> $ readelf -Ws vml |grep arm64_ssbd_callback_re
> 154271: ffff80001133e000     0 OBJECT  GLOBAL DEFAULT   22 
__kvm_nvhe_arm64_ssbd_callback_required
> 159609: ffff80001133e000     8 OBJECT  WEAK   DEFAULT   22 
arm64_ssbd_callback_required

Yes, its zero-sized. And yes, the error happens even with 
tmp.libbtf_encoder, but pahole doesn't crash and the build finishes fine.

thanks,
--
js