linux-kernel - Re: [PATCH] metag: move EXPORT_SYMBOLs in libs to metag

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-prev] [day] [month] [year] [list]

Date:	Wed, 3 Jul 2013 10:14:41 +0100
From:	James Hogan <james.hogan@...tec.com>
To:	Geert Uytterhoeven <geert@...ux-m68k.org>
CC:	Rusty Russell <rusty@...tcorp.com.au>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	linux-arch <linux-arch@...r.kernel.org>,
	linux-kbuild <linux-kbuild@...r.kernel.org>
Subject: Re: [PATCH] metag: move EXPORT_SYMBOLs in libs to metag_ksyms.c

On 03/07/13 09:50, Geert Uytterhoeven wrote:
> On Wed, Jul 3, 2013 at 10:44 AM, James Hogan <james.hogan@...tec.com> wrote:
>> (cc'ing linux-arch and linux-kbuild)
>>
>> On 03/07/13 01:45, Rusty Russell wrote:
>>> James Hogan <james.hogan@...tec.com> writes:
>>>> Move all EXPORT_SYMBOLs in libs-y directories (arch/metag/{lib,tbx})
>>>> into metag_ksyms.c so that they don't get omitted by the static linker
>>>> if they're not used by any other statically linked code, which can
>>>> result in undefined symbols when building modules.
>>>>
>>>> For example a randconfig caused the following error:
>>>> ERROR: "csum_partial" [fs/reiserfs/reiserfs.ko] undefined!
>>>>
>>>> Signed-off-by: James Hogan <james.hogan@...tec.com>
>>>> Cc: Rusty Russell <rusty@...tcorp.com.au>
>>>> ---
>>>> Unless there are any objections I'll apply this patch for v3.11. I'm
>>>> assuming *_ksyms.c is the way that this problem is normally worked
>>>> around.
>>>
>>> Actually, normally those files are obj-y not lib-y.  See the top-level
>>> lib/.
>>>
>>> Subtle, I know.
>>
>> Hmm, subtle indeed, I hadn't noticed that.
>>
>> grepping for EXPORT_SYMBOL(csum_partial) gives:
>>
>> ./arch/avr32/kernel/avr32_ksyms.c
>> ./arch/sparc/lib/ksyms.c
>> ./arch/sh/kernel/sh_ksyms_32.c
>> ./arch/powerpc/kernel/ppc_ksyms.c
>> ./arch/unicore32/kernel/ksyms.c
>> ./arch/x86/um/ksyms.c
>> ./arch/x86/kernel/x8664_ksyms_64.c
>> ./arch/x86/kernel/i386_ksyms_32.c
>> ./arch/cris/kernel/crisksyms.c
>> ./arch/mips/kernel/mips_ksyms.c
>> ./arch/xtensa/kernel/xtensa_ksyms.c
>> ./arch/arm/kernel/armksyms.c
>>
>> ./arch/ia64/lib/checksum.c
>> ./arch/alpha/lib/checksum.c
>> ./arch/c6x/lib/checksum.c
>> ./arch/mn10300/lib/checksum.c
>> ./arch/cris/arch-v10/lib/old_checksum.c
>> ./arch/h8300/lib/checksum.c
>> ./arch/parisc/lib/checksum.c
>> ./arch/m32r/lib/csum_partial_copy.c
>> ./arch/m68k/lib/checksum.c
>> ./arch/metag/lib/checksum.c
>> ./arch/frv/lib/checksum.c
>>
>> ./lib/checksum.c
>>
>> All of the above arches have *c*sum.o in lib-y. Should they all be
>> switched to obj-y or the remaining EXPORT_SYMBOLs moved to *ksyms.c if
>> this problem is noticed?
> 
> There are more, cfr. https://lkml.org/lkml/2013/5/29/76 ("[PATCH/RFC] sh: Always
> link in helper functions extracted from libgcc"), which no one replied to.

Moving the EXPORT_SYMBOL into */ksyms.c would seem to make the
inclusion/exclusion of unused lib-y objects automatic:
With CONFIG_MODULES=y the EXPORT_SYMBOL in *ksyms.c would ensure it is
included.
With CONFIG_MODULES=n the EXPORT_SYMBOL would be preprocessed out and it
could be unused and excluded.

(although it is of course uglier to separate it from the definition).

Cheers
James

--
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/