linux-kernel - module version magic and arches with symbol prefixes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 18 Jun 2009 11:24:44 -0400
From:	Mike Frysinger <vapier.adi@...il.com>
To:	Rusty Russell <rusty@...tcorp.com.au>
Cc:	Linux kernel mailing list <linux-kernel@...r.kernel.org>,
	Robin Getz <rgetz@...ckfin.uclinux.org>
Subject: module version magic and arches with symbol prefixes

the current check_modstruct_version() does this:
{
    const unsigned long *crc;

    if (!find_symbol("module_layout", NULL, &crc, true, false))
        BUG();
    return check_version(sechdrs, versindex, "module_layout", mod, crc);
}
the trouble here is that it looks for a literal "module_layout" symbol
and for ports that have symbol prefixes (a quick check shows Blackfin
& h8300), this aint going to work.

i tried to hack it in the Blackfin port by add this to the linker script:
module_layout = _module_layout
but that didnt seem to work.  maybe kallsysms couldnt find it or i
need to hack a different name ...

we could add a new function to asm-generic/sections.h:
#ifndef arch_symbol_name
#define arch_symbol_name(sym) sym
#endif
and in the case of Blackfin systems, we'd do:
#define arch_symbol_name(sym) "_" sym

no other consumer of find_symbol() has a problem because they're all
dynamic -- they scan the modules for required symbols and then scan
the kernel for those, so all the symbol prefixes line up.

that would fix the find_symbol() invocation, and check_version()
wouldnt need changing because in that case, it's look for the literal
symbol that scripts/mod/modpost.c is adding -- "module_layout" in this
case.

also, using BUG() here seems pretty damn harsh.  wouldnt it make more
sense to do something like:
    if (WARN_ON(!find_symbol("module_layout", NULL, &crc, true, false)))
        return 0;
this way the module is simply not loaded rather than killing the kernel
-mike
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/