From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail.multimedia-labs.de ([82.149.226.172]) by linuxtogo.org with esmtp (Exim 4.72) (envelope-from ) id 1Pj1SE-0008DJ-0H for openembedded-devel@lists.openembedded.org; Sat, 29 Jan 2011 04:26:26 +0100 Received: from localhost (localhost [127.0.0.1]) by mail.multimedia-labs.de (Postfix) with ESMTP id 1904D314D513 for ; Sat, 29 Jan 2011 04:25:34 +0100 (CET) X-Virus-Scanned: Debian amavisd-new at mail.multimedia-labs.de Received: from mail.multimedia-labs.de ([127.0.0.1]) by localhost (mail.multimedia-labs.de [127.0.0.1]) (amavisd-new, port 10024) with LMTP id HMyiF6TI6QQ9 for ; Sat, 29 Jan 2011 04:25:28 +0100 (CET) Received: from [172.22.22.61] (ip-94-79-168-47.unitymediagroup.de [94.79.168.47]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mail.multimedia-labs.de (Postfix) with ESMTPSA id 97AEA314D363 for ; Sat, 29 Jan 2011 04:25:28 +0100 (CET) Message-ID: <4D4388AA.40208@opendreambox.org> Date: Sat, 29 Jan 2011 04:25:30 +0100 From: Andreas Oberritter User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.13) Gecko/20101210 Lightning/1.0b2 Thunderbird/3.1.7 MIME-Version: 1.0 To: openembedded-devel@lists.openembedded.org Subject: [PATCH/RFC] kernel module dependencies / my recent commits X-BeenThere: openembedded-devel@lists.openembedded.org X-Mailman-Version: 2.1.11 Precedence: list Reply-To: openembedded-devel@lists.openembedded.org List-Id: Using the OpenEmbedded metadata to build Distributions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 29 Jan 2011 03:26:26 -0000 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Hi all, it seems that my last commits broke the build of external kernel modules at least on clean builds, but maybe also for incremental builds. To quote from my commit message: * As a special case, kernel modules must depend on kernel-${KERNEL_VERSION}, which gets runtime provided by the kernel package. The reason is, that the package version can be different from ${KERNEL_VERSION}, because the latter depends on the kernel configuration, e.g. CONFIG_LOCALVERSION. This is also what kernel.bbclass does when calling do_split_packages(). Now it seems like there's a problem with using the variable KERNEL_VERSION as part of a recipe's RDEPENDS, when used in a package's name, because it can be empty before the kernel is built (and apparently it may also become empty when loaded from BitBake's cache). I've included a patch to kernel.bbclass to workaround or solve this problem below, but I'm not sure whether it's the right thing. If it's not and if there's no better solution, then please revert my commits using git revert 254193c8f65511a8f344559f7714b50d35afb22b... 2dac84b9a7b820466ed626420fec120093b56e16 I am sorry for the inconvenience I've caused. Regards, Andreas --- kernel.bbclass: Add kernel-* to PACKAGES_DYNAMIC * Use it as a hint for BitBake to know that packages inheriting kernel.bbclass do provide kernel-x.y.z-localversion. Signed-off-by: Andreas Oberritter diff --git a/classes/kernel.bbclass b/classes/kernel.bbclass index ea1daf6..0d1b4ad 100644 --- a/classes/kernel.bbclass +++ b/classes/kernel.bbclass @@ -32,6 +32,7 @@ INITRAMFS_TASK ?= "" inherit kernel-arch +PACKAGES_DYNAMIC += "kernel-*" PACKAGES_DYNAMIC += "kernel-module-*" PACKAGES_DYNAMIC += "kernel-image-*" PACKAGES_DYNAMIC += "kernel-firmware-*"