From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nicholas Piggin Subject: Re: [PATCH] kbuild: provide THIN_ARCHIVES option for all architectures Date: Tue, 30 May 2017 09:13:43 +1000 Message-ID: <20170530091343.1f98b24b@roar.ozlabs.ibm.com> References: <20170529081103.29999-1-npiggin@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-kbuild-owner@vger.kernel.org To: Linus Torvalds Cc: "linux-arch@vger.kernel.org" , Linux Kbuild mailing list , Stephen Rothwell List-Id: linux-arch.vger.kernel.org On Mon, 29 May 2017 10:14:50 -0700 Linus Torvalds wrote: > On Mon, May 29, 2017 at 10:03 AM, Linus Torvalds > wrote: > > > > So either thin archives work or they don't. Don't make it some painful > > drawn-out process that actually makes it harder to test rather than > > easier. > > Side note: what would make them not work? Are there known bugs in 'ar' > wrt thin archives, versioning issues, or other issues? We haven't run into any bugs AFAIK. It's been pretty solid. *thin* archives does require binutils 2.19. That's nearly 10 years old, but we advertise 2.12 minimum at the moment. We'd have make "T" an optional argument if we turn it on unconditionally. Regular archives shouldn't be significantly worse than ld -r though, so yeah maybe we could do that. powerpc required a build fix: 43c9127d94d -override AR := GNUTARGET=elf$(BITS)-$(GNUTARGET) $(AR) +KBUILD_ARFLAGS += --target=elf$(BITS)-$(GNUTARGET) So there could be a couple of small things like this. Thanks, Nick From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pf0-f193.google.com ([209.85.192.193]:33612 "EHLO mail-pf0-f193.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750994AbdE2XN6 (ORCPT ); Mon, 29 May 2017 19:13:58 -0400 Date: Tue, 30 May 2017 09:13:43 +1000 From: Nicholas Piggin Subject: Re: [PATCH] kbuild: provide THIN_ARCHIVES option for all architectures Message-ID: <20170530091343.1f98b24b@roar.ozlabs.ibm.com> In-Reply-To: References: <20170529081103.29999-1-npiggin@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-arch-owner@vger.kernel.org List-ID: To: Linus Torvalds Cc: "linux-arch@vger.kernel.org" , Linux Kbuild mailing list , Stephen Rothwell Message-ID: <20170529231343.MYtsd8jhtcyq7A1dS5Fz62WLKQXv19upYixooJ_IoL8@z> On Mon, 29 May 2017 10:14:50 -0700 Linus Torvalds wrote: > On Mon, May 29, 2017 at 10:03 AM, Linus Torvalds > wrote: > > > > So either thin archives work or they don't. Don't make it some painful > > drawn-out process that actually makes it harder to test rather than > > easier. > > Side note: what would make them not work? Are there known bugs in 'ar' > wrt thin archives, versioning issues, or other issues? We haven't run into any bugs AFAIK. It's been pretty solid. *thin* archives does require binutils 2.19. That's nearly 10 years old, but we advertise 2.12 minimum at the moment. We'd have make "T" an optional argument if we turn it on unconditionally. Regular archives shouldn't be significantly worse than ld -r though, so yeah maybe we could do that. powerpc required a build fix: 43c9127d94d -override AR := GNUTARGET=elf$(BITS)-$(GNUTARGET) $(AR) +KBUILD_ARFLAGS += --target=elf$(BITS)-$(GNUTARGET) So there could be a couple of small things like this. Thanks, Nick