Re: The trouble with __weak and objtool got worse

[Peter Zijlstra <peterz@infradead.org>]

On Fri, Apr 15, 2022 at 01:22:29PM -0500, Segher Boessenkool wrote:
> Hi!
> 
> On Fri, Apr 15, 2022 at 01:19:04PM +0200, Peter Zijlstra wrote:
> [ huge snip ]
> 
> > This seems to happen when objtool cannot find a section symbol, in which
> > case it falls back to any other symbol to key off of, however in this
> > case that goes terribly wrong!
> 
> That is an objtool problem then.  Fix objtool so it looks at symbols
> directly?

I'm not sure what you're saying. It *does* look at the symbols, but at
the time we run objtool (pre link) there's multiple copies of a symbol
(weak and non-weak), and if we then generate symbol relative relocs, we
end up, post-link, with the relocs resolved wrong.

The weak symbol relocs get applied to the non-weak symbol, and fail
happens.

> > The other option seems to be to have objtool add section symbols it
> > needs, however due to ELF being a total PITA and requiring all LOCAL
> > symbols to be before GLOBAL symbols, this would mean re-ordering the
> > whole symbol table (I have the code somewhere :-/).
> 
> ELF requires no such thing?  Where do you see this?

The first google hit:

  https://docs.oracle.com/cd/E23824_01/html/819-0690/chapter6-79797.html

which has:

  "In each symbol table, all symbols with STB_LOCAL binding precede the
  weak symbols and global symbols. As Sections describes, a symbol table
  section's sh_info section header member holds the symbol table index for
  the first non-local symbol."

I ran into that when I violated that contraint at some point and
the various elf tools screamed at me.