Re: [PATCH 17/18] objtool: Remove --lto and --vmlinux

From: Josh Poimboeuf <jpoimboe@redhat.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: x86@kernel.org, linux-kernel@vger.kernel.org,
	Miroslav Benes <mbenes@suse.cz>
Subject: Re: [PATCH 17/18] objtool: Remove --lto and --vmlinux
Date: Thu, 14 Apr 2022 19:18:53 -0700	[thread overview]
Message-ID: <20220415021853.vnavjq6u6ibgva2l@treble> (raw)
In-Reply-To: <YlfXvlvNmCYFIq0w@hirez.programming.kicks-ass.net>

On Thu, Apr 14, 2022 at 10:13:50AM +0200, Peter Zijlstra wrote:
> On Wed, Apr 13, 2022 at 04:19:52PM -0700, Josh Poimboeuf wrote:
> 
> > @@ -131,6 +131,12 @@ static inline u32 reloc_hash(struct reloc *reloc)
> >  	return sec_offset_hash(reloc->sec, reloc->offset);
> >  }
> >  
> > +/* is it a whole archive (vmlinux.o or module)? */
> > +static inline bool is_linked_object(struct elf *elf)
> > +{
> > +	return elf->num_files > 1;
> > +}
> > +
> >  struct elf *elf_open_read(const char *name, int flags);
> >  struct section *elf_create_section(struct elf *elf, const char *name, unsigned int sh_flags, size_t entsize, int nr);
> 
> Ooh, nice... yes, that was the entire point of --lto.

Urgh, some kernel modules only have a single source file, which
obviously breaks this.

We might need an explicit --link option after all.  But with different
semantics from --lto.  It would just mean "is a linked object".

-- 
Josh