From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: MIME-Version: 1.0 In-Reply-To: <201201242024.40361.jbe@pengutronix.de> References: <4F1EFADC.7020308@gmail.com> <201201242024.40361.jbe@pengutronix.de> Date: Wed, 25 Jan 2012 10:29:21 +0100 Message-ID: From: Guillermo Rodriguez Garcia Subject: Re: [ptxdist] X.org on mini2440 Reply-To: ptxdist@pengutronix.de List-Id: PTXdist Development Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Sender: ptxdist-bounces@pengutronix.de Errors-To: ptxdist-bounces@pengutronix.de To: Juergen Beisert Cc: "Olbrich, Michael" , ptxdist@pengutronix.de Hi Juergen, 2012/1/24 Juergen Beisert : [...] >> Upon inspecting /usr/bin/startx I see that the following has leaked >> through: >> >> > mcookie=3D`/home/grodriguez/work/ptxdist/OSELAS.BSP-Pengutronix-Mini2440-= 2011.12.0/platform-mini2440/sysroot-host/bin/mcookie` > > This should help: > > $ ptxdist menuconfig > =A0Shell & Console Tools ---> > =A0 =A0 =A0 util-linux-ng -----> > =A0 =A0 =A0 =A0 =A0[*] mcookie No, this is already selected (it is automatically brought in when you select the xinit package, so the dependencies are setup properly). And in fact the mcookie command is available in the target, and works: root@mini2440:~ mcookie 1b82325e2a4a0187fe4f59ffaa118194 root@mini2440:~ which mcookie /usr/bin/mcookie The problem is that the startx script that is copied to the target's rootfs has a hardcoded path to the mcookie binary on the host. I tried "ptxconfig clean xorg-app-xinit; ptxconfig go", and the output from configure shows this: [...] checking for mcookie... /home/grodriguez/work/ptxdist/OSELAS.BSP-Pengutronix-Mini2440-2011.12.0/pla= tform-mini2440/sysroot-host/bin/mcookie [...] And indeed the generated /usr/bin/startx script uses this path: root@mini2440:~ /usr/bin/startx /usr/bin/startx: line 183: /home/grodriguez/work/ptxdist/OSELAS.BSP-Pengutronix-Mini2440-2011.12.0/pla= tform-mini2440/sysroot-host/bin/mcookie: not found Couldn't create cookie > >> Not sure whether this is a bug in the configure script in the xinit >> package, and/or something that can be fixed in ptxdist itself. In my cas= e, >> I fixed this by manually editing the startx script and replacing the abo= ve >> with a pregenerated mcookie value. > > Maybe a missing dependency from the xinit package into the util-linux-ng > package. > > @Michael: what do you think? > >> Next thing was that X.org would not start if the kernel is not compiled >> with support for virtual terminals. While obvious in retrospect, I wonder >> whether it is possible for ptxdist to detect this kind of dependencies >> (e.g. detect that a required kernel option is not set in the current ker= nel >> configuration). While this wouldn't help when ptxdist is used to generat= e a >> rootfs only, it would certainly be useful if ptxdist is also generating = the >> kernel for the target platform. Rebuilt and reflashed the kernel and the >> problem went away. > > No, PTXdist cannot detect it automatically. The Kernel is part of the > platform, Xorg is part of the userland. It's up to you as the developer to > get the configuration in a way that matches your requirements. OK. I thought that maybe ptxdist had a way to know since it has access to both the kernel config and the userland config. > > Our Mini2440 BSP can only act as a template for you. I did not configure = it > for maximum features, because I cannot test everything ;-) That's perfect :) It is nice the way it is. I'd rather start from minimal and add features incrementally than the opposite. > >> Finally I found that even when I had configured X.org to use the "void" >> driver for the keyboard, it would keep complaining about keymaps and >> keyboard initialisation until I added xkeyboard-config (presumably for >> /usr/lib/X11/xkb/rules/base) and xkbcomp (for the latter I had to enable >> the "Keyboard mappings" option in ptxdist). Not sure whether these two a= re >> actually required for all X.org installations but I could not find any >> other way around. If this is really a requirement perhaps these >> dependencies should be forced. > > Same here: @Michael: what do you think? > >> After this I could run X on the target device. Used fluxbox as a WM. >> >> I can provide more information on any of the above issues if required. > > If you know what dependencies should be met (if the user want them), patc= hes > are welcome at the PTXdist mailing list. > I cannot comment on this, as its long time ago, that I worked with Xorg o= n any > kind of embedded device. I was hoping that some X.org guru could confirm whether keyboard/keymap support is always required (and then it's just a matter of checking the required dependencies and get them into ptxdist) or whether it is just a configuration problem on my side. Thanks, Guillermo Rodriguez Garcia guille.rodriguez@gmail.com -- = ptxdist mailing list ptxdist@pengutronix.de