From owner-freebsd-arm@freebsd.org Fri Oct 13 17:59:36 2017 Return-Path: Delivered-To: freebsd-arm@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id 076A2E28AB3 for ; Fri, 13 Oct 2017 17:59:36 +0000 (UTC) (envelope-from fbsd@www.zefox.net) Received: from www.zefox.net (www.zefox.net [69.239.235.194]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client CN "www.zefox.org", Issuer "www.zefox.org" (not verified)) by mx1.freebsd.org (Postfix) with ESMTPS id C43406F14B for ; Fri, 13 Oct 2017 17:59:35 +0000 (UTC) (envelope-from fbsd@www.zefox.net) Received: from www.zefox.net (localhost [127.0.0.1]) by www.zefox.net (8.15.2/8.15.2) with ESMTPS id v9DHxiQM074258 (version=TLSv1.2 cipher=DHE-RSA-AES256-GCM-SHA384 bits=256 verify=NO); Fri, 13 Oct 2017 10:59:44 -0700 (PDT) (envelope-from fbsd@www.zefox.net) Received: (from fbsd@localhost) by www.zefox.net (8.15.2/8.15.2/Submit) id v9DHxheV074257; Fri, 13 Oct 2017 10:59:43 -0700 (PDT) (envelope-from fbsd) Date: Fri, 13 Oct 2017 10:59:43 -0700 From: bob prohaska To: freebsd-arm@freebsd.org Subject: Re: Difficulty with armv6 to v7 transition. Message-ID: <20171013175943.GA74121@www.zefox.net> References: <20171009175216.GA52497@www.zefox.net> <1507573171.84167.9.camel@freebsd.org> <20171011023356.GA57571@www.zefox.net> <20171011030021.GB57571@www.zefox.net> <20171013020604.GA70845@www.zefox.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20171013020604.GA70845@www.zefox.net> User-Agent: Mutt/1.5.24 (2015-08-30) X-BeenThere: freebsd-arm@freebsd.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: "Porting FreeBSD to ARM processors." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 13 Oct 2017 17:59:36 -0000 It turns out that simply commenting out lines 447-452 in /usr/src/Makefile.inc1 allows buildworld to run, even with no /etc/make.conf in place. This seems like a fairly reckless experiment, but having no better ideas, would anybody hazard a guess as to odds that the resulting userland might be expected to run? Are there any pre-install tests to be tried? At this point the kernel is at r324562, /usr/src is at r324599. Uname -p reports armv7. Thanks for reading, bob prohaska On Thu, Oct 12, 2017 at 07:06:04PM -0700, bob prohaska wrote: > It's starting to look as if the trouble might have little or nothing > to do with the armv6-armv7 transition and instead has some other cause. > > At this point /etc/make.conf contains > KERNCONF=RPI2 > TARGET=arm > TARGET_ARCH=armv7 > DESTDIR=/ > > Buildkernel works, installkernel demanded a DESTDIR and worked once it > was added, so /etc/make.conf is being read and acted upon. > > If make buildworld is invoked, using the make.conf file above, make reports > make[1]: "/usr/src/Makefile.inc1" line 450: To cross-build, set TARGET_ARCH. > *** [buildworld] Error code 1 > > I had to reconstruct all of /usr after mistakenly deleting it during an > attempted housecleaning. The restored /usr seems to boot normally and > builds kernels just fine, but does not allow su to root, so permissions > (or something) are not entirely correct. > > Might this be related to the failure to recognize or act upon the > TARGET_ARCH=armv7 setting? /usr/src is at 324562, along with the > kernel. Userland dates from late June. > > Thanks for reading, and any ideas. > > bob prohaska > > On Tue, Oct 10, 2017 at 08:00:21PM -0700, bob prohaska wrote: > > On Tue, Oct 10, 2017 at 08:41:17PM -0600, Warner Losh wrote: > > > On Tue, Oct 10, 2017 at 8:39 PM, Warner Losh wrote: > > > > > > > > > > > > > > > On Tue, Oct 10, 2017 at 8:33 PM, bob prohaska wrote: > > > > > > > >> On Mon, Oct 09, 2017 at 12:19:31PM -0600, Ian Lepore wrote: > > > >> > On Mon, 2017-10-09 at 10:52 -0700, bob prohaska wrote: > > > >> > > On an RPI2 model B, invoking? > > > >> > > make -j4 -DNO_CLEAN MACHINE_ARCH=armv7 buildworld > buildworld.log > > > >> > > > > >> > Never set MACHINE_ARCH when building, use TARGET_ARCH. ?Be sure to set > > > >> > TARGET_ARCH when installing as well. > > > >> > > > >> Tried it, like so: > > > >> root@www:/usr/src # make -j4 buildworld TARGET_ARCH=armv7 > > > > >> buildworld.log & [1] 1006 > > > >> root@www:/usr/src # 1 error > > > >> > > > >> [1] Exit 2 make -j4 buildworld > > > >> TARGET_ARCH=armv7 > buildworld.log > > > >> root@www:/usr/src # more *.log > > > >> --- buildworld --- > > > >> make[1]: "/usr/src/Makefile.inc1" line 450: To cross-build, set > > > >> TARGET_ARCH. > > > >> *** [buildworld] Error code 1 > > > >> > > > >> make: stopped in /usr/src > > > >> > > > >> I also tried setting TARGET=arm and WITHOUT_SYSTEM_COMPILER=yes in various > > > >> iterations. Should the variables be set somewhere else, in a config file? > > > >> > > > >> At this point the kernel is at r324499, along with the sources. Userland > > > >> dates from late June (operator error). Kernels build, but could that make > > > >> the trouble I'm seeing? Clang -v reports > > > >> > > > >> FreeBSD clang version 4.0.0 (tags/RELEASE_400/final 297347) (based on > > > >> LLVM 4.0.0) > > > >> Target: armv6-unknown-freebsd12.0-gnueabihf > > > >> Thread model: posix > > > >> InstalledDir: /usr/bin > > > >> > > > >> Thanks again! > > > > > > > > > > > > uname -a says what? > > > > > > > > > root@www:/usr/src # > > root@www:/usr/src # uname -a > > FreeBSD www.zefox.com 12.0-CURRENT FreeBSD 12.0-CURRENT #0 r324499: Tue Oct 10 18:06:39 PDT 2017 root@www.zefox.com:/usr/obj/usr/src/sys/RPI2 arm > > > > > What happens if you do a build with TARGET_ARCH=armv7? > > > > > > > root@www:/usr/src # make TARGET_ARCH=armv7 buildworld > buildworld.log > > make[1]: "/usr/src/Makefile.inc1" line 450: To cross-build, set TARGET_ARCH. > > root@www:/usr/src # > > > > FWIW, /etc/make.conf does not exist. Seemingly the variable isn't recognized. > > Maybe a config error someplace? > > > > Thanks again, > > > > bob prohaska > >