From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <4C2C3A74.6010608@domain.hid> Date: Thu, 01 Jul 2010 08:49:24 +0200 From: Gilles Chanteperdrix MIME-Version: 1.0 References: <22150A2C94CC924695BE248F94CD11F2CD0F@EXDB4.ug.kth.se> <4C2A518F.8030302@domain.hid> <22150A2C94CC924695BE248F94CD11F2CDC2@domain.hid> <4C2B0043.1070108@domain.hid> In-Reply-To: Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Xenomai-help] Xenomai and Gumstix Overo List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: =?UTF-8?B?RmVsaXBlIEJyYW5kw6NvIENhdmFsY2FudGk=?= Cc: "xenomai@xenomai.org" Felipe Brand=C3=A3o Cavalcanti wrote: > I am actually trying to do the exact same thing. >=20 > On Wed, Jun 30, 2010 at 5:28 AM, Gilles Chanteperdrix < > gilles.chanteperdrix@xenomai.org> wrote: >=20 >> Fredrik Asplund wrote: >>>> Xenomai 2.5.3 was tested on OMAP3 with the mainline kernel. If you >>>> absolutely want to use this vendor branch, you will have to fix the >>>> patch rejects by hand. >>> Ok, yes, I thought so. I guess I was hoping someone had done the same= >>> thing and had some pointers on what to do in this specific case (i.e.= >>> applying Xenomai using Bitbake on a vendor branch). >>> >>> I will try to apply the patch rejects (I want to deviate from the >>> original setup as little as possible, at least to begin with) >> If you do not want to deviate from Xenomai usual setup, use the mainli= ne >> kernel. If I were you, I would have serious doubts on a vendor kernel >> which touches generic files such as include/asm/cacheflush.h in order = to >> add some board-specific code. >> >> > My first choice was to run a vanilla kernel, but no luck - didn't find = any > suitable config files, or any reports of success. I am on the same syst= em - > Gumstix Overo, kernel 2.6.33, Xenomai 2.5.3. The 2.6.33 kernel has an overo_defconfig. This defconfig sets CONFIG_MACH_OVERO. arch/arm/mach-omap2/Kconfig contains the following lin= es: config MACH_OVERO bool "Gumstix Overo board" depends on ARCH_OMAP3 && ARCH_OMAP34XX select OMAP_PACKAGE_CBB So, it looks to me like the vanilla kernel supports Gumstix overo. >=20 > Anyway, I've managed to change the files manually - the changes seem ve= ry > minor (the changes in include/asm/cacheflush.h just move a function > elsewhere, no code changes at all). That is not the point. The point is that to add support for a board, you should not mess with generic code. So, it is probably safer to get the vanilla support working. The I-pipe patch modifies only SOC-specific files, so, any well written board support which adds board-specific files should merge with the I-pipe patch without any reject. >=20 > However, I am having problems with the kernel - any user-space Xenomai > program gives me the following error: > Xenomai or CONFIG_OPT_PERVASIVE disabled. > (modprobe xeno_nucleus?) is CONFIG_XENO_OPT_PERVASIVE in your configuration? > All the logs seem normal to me (I-pipe reports normal starup, Xenomai > reports successful start). I am starting to think that one of my change= s to > the kernel might be making the system to fail silently. I use Xenomai 2.5.3 on an OMAP3 based board without any problems. So, yes, either you are mis-configured (in order to check that, just use Xenomai default configuration), or your kernel is broken. >=20 > So, any clues? I can send my changes and logs tomorrow morning, if they= are > of any help. >=20 > Another option which I am considering is trying Xenomai with the 2.6.31= > kernel, or maybe even 2.6.30 or 2.6.28. I am sorry, but I am not willing to debug this, I am afraid you are on your own. --=20 Gilles.