From owner-freebsd-arm@freebsd.org Mon Oct 19 15:05:02 2015 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 7169BA19E1B for ; Mon, 19 Oct 2015 15:05:02 +0000 (UTC) (envelope-from mike@skew.org) Received: from chilled.skew.org (chilled.skew.org [70.90.116.205]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client CN "*.skew.org", Issuer "AlphaSSL CA - SHA256 - G2" (not verified)) by mx1.freebsd.org (Postfix) with ESMTPS id 4CE5F184 for ; Mon, 19 Oct 2015 15:05:02 +0000 (UTC) (envelope-from mike@skew.org) Received: from chilled.skew.org (localhost [127.0.0.1]) by chilled.skew.org (8.14.7/8.14.7) with ESMTP id t9JF50wo084842 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-GCM-SHA384 bits=256 verify=NO) for ; Mon, 19 Oct 2015 09:05:00 -0600 (MDT) (envelope-from mike@chilled.skew.org) Received: (from mike@localhost) by chilled.skew.org (8.14.7/8.14.7/Submit) id t9JF50PD084841 for freebsd-arm@freebsd.org; Mon, 19 Oct 2015 09:05:00 -0600 (MDT) (envelope-from mike) Date: Mon, 19 Oct 2015 09:04:59 -0600 From: Mike Brown To: freebsd-arm@freebsd.org Subject: Re: BeagleBone self-built 10-STABLE hangs, segfaults Message-ID: <20151019150459.GA57293@chilled.skew.org> References: <20151018162950.GA52948@chilled.skew.org> <1445187147.71631.59.camel@freebsd.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: X-Whoa: whoa. User-Agent: Mutt/1.5.23 (2014-03-12) X-BeenThere: freebsd-arm@freebsd.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: "Porting FreeBSD to ARM processors." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 19 Oct 2015 15:05:02 -0000 On Sun, Oct 18, 2015 at 11:27:48PM +0300, Andrey Fesenko wrote: > > I've seen those unknown attribute warnings before, I don't think > > they're anything to worry about. I haven't tried running a self-hosted > > build on any armv6 for a long time, it's possible there's something > > ugly like a code-generation bug that happens when running native but > > not on a cross build. Things like that are pretty rare, though. > > > > Confirm both Rpi and bb "Segmentation fault", i'm build with crochet > and CURRENT-r289459 (builded 289522) > > last lines sh -x crochet > + _DTBINTERMEDIATEDIR=/usr/obj/fdt/fdt.kUEP6Z > + _DTSIN=/usr/src/sys/boot/fdt/dts/arm/beaglebone.dts > + _DTBINTERMEDIATE=/usr/obj/fdt/fdt.kUEP6Z > + echo /usr/src/sys/tools/fdt/make_dtb.sh /usr/src/sys > /usr/src/sys/boot/fdt/dts/arm/beaglebone.dts /usr/obj/fdt/fdt.kUEP6Z > + cd /usr/src > + make 'TARGET_ARCH=armv6' buildenv > + _DTSOUT=bbone.dts > + dtc -I dtb -O dts /usr/obj/fdt/fdt.kUEP6Z/beaglebone.dtb > Segmentation fault > root@des:/usr/src # pkg info | grep eabi > arm-none-eabi-binutils-2.25.1 GNU binutils for bare metal arm cross-development > arm-none-eabi-gcc492-4.9.2_1 GNU Compiler Collection for bare metal > arm cross-development Does this happen at the same place very time? My problems on 10-STABLE seem to happen at random, as if it were bad RAM. In successive builds of the 'pkg' port, for example, I get different results each time: a segmentation fault during 'configure', then no errors, then a syntax error in some file, then a segmentation fault during 'cc' of something small, then no errors. I didn't see any such weirdness before the native build, though. How can I diagnose this further?