Re: [MODULES] fix weak symbol handling

Rusty Russell (rusty@rustcorp.com.au)
Tue, 14 Jan 2003 14:11:30 +1100


In message <20030113110036.A873@twiddle.net> you write:
> I discovered this while working on oprofile for Alpha. I thought
> I'd avoid a whole series of nested ifdefs by marking some symbols
> weak, and so let them go undefined and resolve to null. Except
> that we don't handle that in the new module loader.
>
> Fixed thus.

That part looks OK. The second loop was only there so we placed
common symbols first: now we don't do that, your cleanup is a good
idea. I'll extract and test that part.

>
> I also correct a misconception in simplify_symbol. It is pointless
> to lookup an undefined symbol in the module in which it is undefined.

<sigh>. I don't think so.

PPC64 (not in tree yet):

+ /* REL24 references to (external) .function won't
+ resolve; deal with that below */
+ if (!sym->st_value
+ && ELF64_R_TYPE(rela[i].r_info) != R_PPC_REL24) {
+ printk("%s: Unknown symbol %s (index %u)\n",
+ me->name, strtab + sym->st_name,
+ sym->st_shndx);
+ return -ENOENT;
+ }

That's *why* find_symbol_internal() is not static, and why we don't
fail in simplify_symbol() 8(

Hope that clarifies,
Rusty.

--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/