From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Hogan Subject: Re: [PATCH] metag: move EXPORT_SYMBOLs in libs to metag_ksyms.c Date: Wed, 3 Jul 2013 10:14:41 +0100 Message-ID: <51D3EB81.5040509@imgtec.com> References: <1372774074-18553-1-git-send-email-james.hogan@imgtec.com> <87bo6kfp32.fsf@rustcorp.com.au> <51D3E464.9050707@imgtec.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-kbuild-owner@vger.kernel.org To: Geert Uytterhoeven Cc: Rusty Russell , "linux-kernel@vger.kernel.org" , linux-arch , linux-kbuild List-Id: linux-arch.vger.kernel.org On 03/07/13 09:50, Geert Uytterhoeven wrote: > On Wed, Jul 3, 2013 at 10:44 AM, James Hogan wrote: >> (cc'ing linux-arch and linux-kbuild) >> >> On 03/07/13 01:45, Rusty Russell wrote: >>> James Hogan 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 >>>> Cc: Rusty Russell >>>> --- >>>> 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 From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from multi.imgtec.com ([194.200.65.239]:45301 "EHLO multi.imgtec.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932399Ab3GCJOz (ORCPT ); Wed, 3 Jul 2013 05:14:55 -0400 Message-ID: <51D3EB81.5040509@imgtec.com> Date: Wed, 3 Jul 2013 10:14:41 +0100 From: James Hogan MIME-Version: 1.0 Subject: Re: [PATCH] metag: move EXPORT_SYMBOLs in libs to metag_ksyms.c References: <1372774074-18553-1-git-send-email-james.hogan@imgtec.com> <87bo6kfp32.fsf@rustcorp.com.au> <51D3E464.9050707@imgtec.com> In-Reply-To: Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Sender: linux-arch-owner@vger.kernel.org List-ID: To: Geert Uytterhoeven Cc: Rusty Russell , "linux-kernel@vger.kernel.org" , linux-arch , linux-kbuild Message-ID: <20130703091441.1RcTtSPIFqrUgPYVf1ksVleswIqvzel5rGJQF53p8j0@z> On 03/07/13 09:50, Geert Uytterhoeven wrote: > On Wed, Jul 3, 2013 at 10:44 AM, James Hogan wrote: >> (cc'ing linux-arch and linux-kbuild) >> >> On 03/07/13 01:45, Rusty Russell wrote: >>> James Hogan 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 >>>> Cc: Rusty Russell >>>> --- >>>> 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