Re: [bcachefs:header_cleanup 21/51] /bin/bash: line 1: 19420 Segmentation fault LLVM_OBJCOPY="llvm-objcopy" pahole -J --btf_gen_floats -j --lang_exclude=rust --skip_encoding_btf_inconsistent_proto --btf_gen_optimized --btf_base vmlinux drivers/misc/eeprom/at24.ko

[Mark Wielaard <mark@klomp.org>]

Hi Arnaldo,

On Tue, Jan 02, 2024 at 01:30:02PM -0300, Arnaldo Carvalho de Melo wrote:
> Em Thu, Dec 28, 2023 at 10:34:04AM -0700, Nathan Chancellor escreveu:
> > On Thu, Dec 28, 2023 at 11:21:06AM -0300, Arnaldo Carvalho de Melo wrote:
> > > Em Wed, Dec 27, 2023 at 03:43:38PM -0700, Nathan Chancellor escreveu:
> > > > I do not get a segfault personally but there are definitely a lot of
> > > > errors from pahole at the end. Arnaldo, is pahole not handling something
> > > > that it should with regards to clang's debug info?
> 
> > > Possibly, would it be possible to provide the vmlinux file where pahole
> > > is segfaulting when trying to encode BTF from DWARF?
> 
> > > These two files:
> 
> > > vmlinux drivers/misc/eeprom/at24.ko
> 
> > > So that I can run:
> > > LLVM_OBJCOPY="llvm-objcopy" pahole -J --btf_gen_floats -j --lang_exclude=rust --skip_encoding_btf_inconsistent_proto --btf_gen_optimized --btf_base vmlinux drivers/misc/eeprom/at24.ko
>  
> > Sure, here you go:
>  
> > https://1drv.ms/u/s!AsQNYeB-IEbqqAilUuGHOU8BMLbf
>  
> > If you need more objects, just let me know. That series of commands in
> > the previous email should also get them for you locally if you need all
> > of them or want to investigate more.
> 
> llvm-dwarfdump is the only that is managing to process this file on a
> fedora:39 system, more work needed to figure out if this is something
> elfutils is planning to support, what amount of work would be necessary,
> etc.
> 
> Mark, do you know about work on elfutils to support:
> 
> ⬢[acme@toolbox hexagon-randconfig-r005-20220913-pahole-crash]$ llvm-dwarfdump at24.ko | head -22
> at24.ko:        file format elf32-hexagon

That seems to identify itself as an EM_QDSP6 (QUALCOMM DSP6) 32bit ELF
file. Neither binutils not elfutils seems to know how to resolve
EM_QDSP6 specific relocations. Normally that wouldn't be necessary,
but sadly kernel modules are still ET_REL files, so eu-readelf/readelf
needs relocations resolved to process the DWARF .debug sections.

This seems to need Qualcomm to upstream support for this processor
type to bintuils and elfutils.

Cheers,

Mark