Re: [PATCH v3 0/9] klp-convert livepatch build tooling

[Miroslav Benes <mbenes@suse.cz>]

On Tue, 16 Apr 2019, Joe Lawrence wrote:

> On 4/10/19 11:50 AM, Joe Lawrence wrote:
> > Hi folks,
> > 
> > This is the third installment of the klp-convert tool for generating and
> > processing livepatch symbols for livepatch module builds.  For those
> > following along at home, archive links to previous versions:
> > 
> > RFC:
> >    https://lore.kernel.org/lkml/cover.1477578530.git.jpoimboe@redhat.com/
> > v2:
> >    https://lore.kernel.org/lkml/f52d29f7-7d1b-ad3d-050b-a9fa8878faf2@redhat.com/
> > 
> > (Note that I don't see v2 archived at lore, but that is a link to the
> > most recent subthread that lore did catch.)
> > 
> > 
> > Livepatches may use symbols which are not contained in its own scope,
> > and, because of that, may end up compiled with relocations that will
> > only be resolved during module load. Yet, when the referenced symbols are
> > not exported, solving this relocation requires information on the object
> > that holds the symbol (either vmlinux or modules) and its position inside
> > the object, as an object may contain multiple symbols with the same name.
> > Providing such information must be done accordingly to what is specified
> > in Documentation/livepatch/module-elf-format.txt.
> 
> Hi Miroslav,
> 
> I noticed that some binutils programs like gdb, objdump, etc. don't like the
> .ko kernel objects that we're generating from this patchset, specifically
> those with the additional '.klp.rela.<obj>..text' livepatch symbol relocation
> sections.
> 
> For reference, I opened a new bugzilla with more details here:
> https://sourceware.org/bugzilla/show_bug.cgi?id=24456

Another great catch.
 
> And was about to ping the binutils mailing list about the assertion that is
> tripping in bfd/elf.c.  The thought occurred to me that you guys might already
> be carrying a patch to workaround this issue?

No, unfortunately we don't. At least I don't know about anything and a 
quick test on openSUSE Leap 15.0 confirms it (objdump -D gives "bad 
value").

Thanks
Miroslav