From wookey@wookware.org Fri Apr 23 13:54:05 2010 Received: from dream ([213.165.225.113] helo=dream.aleph1.co.uk) by stoneboat.aleph1.co.uk with esmtps (TLS1.0:RSA_AES_256_CBC_SHA1:32) (Exim 4.69) (envelope-from ) id 1O5IOP-00008K-3y; Fri, 23 Apr 2010 13:54:03 +0100 Received: from wookey by dream.aleph1.co.uk with local (Exim 4.69) (envelope-from ) id 1O5IdF-0005Us-OS; Fri, 23 Apr 2010 14:09:21 +0100 Date: Fri, 23 Apr 2010 14:09:19 +0100 From: Wookey To: Charles Manning Message-ID: <20100423130917.GQ7110@dream.aleph1.co.uk> Mail-Followup-To: Charles Manning , balloon@balloonboard.org References: <201004060933.54586.manningc2@actrix.gen.nz> <201004061623.24951.manningc2@actrix.gen.nz> <20100406130412.GH7110@dream.aleph1.co.uk> <201004071007.33147.manningc2@actrix.gen.nz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <201004071007.33147.manningc2@actrix.gen.nz> Organization: Aleph One Ltd User-Agent: Mutt/1.5.18 (2008-05-17) X-SA-Exim-Connect-IP: 213.165.225.113 X-SA-Exim-Mail-From: wookey@wookware.org X-Spam-Checker-Version: SpamAssassin 3.2.5 (2008-06-10) on stoneboat.aleph1.co.uk X-Spam-Level: X-Spam-Status: No, score=-2.6 required=4.5 tests=AWL,BAYES_00 autolearn=no version=3.2.5 X-SA-Exim-Version: 4.2.1 (built Wed, 25 Jun 2008 17:14:11 +0000) X-SA-Exim-Scanned: Yes (on stoneboat.aleph1.co.uk) Cc: balloon@balloonboard.org Subject: Re: [Balloon] Problem loading fs tarball into balloon X-BeenThere: balloon@balloonboard.org X-Mailman-Version: 2.1.11 Precedence: list List-Id: Balloon List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 23 Apr 2010 12:54:05 -0000 +++ Charles Manning [2010-04-07 10:07 +1200]: > On Wednesday 07 April 2010 01:04:14 Wookey wrote: > > +++ Charles Manning [2010-04-06 16:23 +1200]: > > > On Tuesday 06 April 2010 12:23:16 Wookey wrote: > > > > +++ Charles Manning [2010-04-06 09:33 +1200]: > > > > > Ballooners.. > > > > > > > > > > I have successfully used bbl to load a new bootloader, fpga binary > > > > > and kernel, but things came unstuck when I tried to load the tarball > > > > > off the usb flash due to missing kernel modules. > > > > > > On further looking... > > > Looks like the problem is in the module definitions meaning modprobe > > > looks in the wrong places. modprobe fails but manual insmod works. > > > > That suggests that the modules.dep file generation has gone wrong. > # pwd > /lib/modules/2.6.29.1-pxa270 > # cat modules.dep > kernel/drivers/usb/gadget/g_ether.ko: kernel/drivers/usb/gadget/pxa27x_udc.ko > kernel/drivers/usb/gadget/g_serial.ko: kernel/drivers/usb/gadget/pxa27x_udc.ko > kernel/drivers/usb/gadget/pxa27x_udc.ko: > kernel/drivers/usb/core/usbcore.ko: OK. I've reproduced this and made some progress on what's going wrong. Currently we do this to generate the depmod file 'offline' with /sbin/depmod -ae -b $(TARGETSKEL) -F $(BUILDDIR)/kernel/linux-$(KERNVER)/System.map $(KERNVER)-pxa270 which says 'run it in $TARGETSKEL but then strip that path from the entries that actually go in the modules.dep file, do it for kernel $(KERNVER)-pxa270 and check the symbols against $(BUILDDIR)/kernel/linux-$(KERNVER)/System.map' In a real run this expands to: /sbin/depmod -aev -b ../../build/initrd/buildroot/target/generic/target_skeleton_balloon3_installer -F ../../build/kernel/linux-2.6.29.1/System.map 2.6.29.1-pxa270 which works just fine on Lenny with modules-init-tools 3.4-1, producing a file like: /lib/modules/2.6.29.1-pxa270/kernel/drivers/usb/gadget/g_ether.ko: /lib/modules/2.6.29.1-pxa270/kernel/drivers/usb/gadget/pxa27x_udc.ko /lib/modules/2.6.29.1-pxa270/kernel/drivers/usb/gadget/g_serial.ko: /lib/modules/2.6.29.1-pxa270/kernel /drivers/usb/gadget/pxa27x_udc.ko but on squeeze with modules-init-tools 3.11-1 (or the current 3.12~pre2-3), it fails, missing off the /lib/modules/2.6.29.1-pxa270/ from the front of each modules.dep entry. As modprobe strips everything before the first / it starts looking for (for example) drivers/usb/gadget/g_ether.ko, which fails. Using realpath to get absolute instead of relative paths doesn't make any difference. Downgrading to 3.4-1 makes it work again, which I suggest is the best plan for the time being. I've reported this to Debian and will try and work up a patch when I get a mo, but I need to make Pootle work this afternoon. Wookey -- Principal hats: iEndian - Balloonboard - Toby Churchill - Emdebian http://wookware.org/