From mboxrd@z Thu Jan 1 00:00:00 1970 From: Romain Naour Date: Mon, 10 Feb 2020 14:49:53 +0100 Subject: [Buildroot] [PATCH v4 6/9] board/qemu: add post-image script for gitlab qemu testing In-Reply-To: <20200210141326.289a52ff@windsurf> References: <20200209180327.455426-1-romain.naour@smile.fr> <20200209180327.455426-7-romain.naour@smile.fr> <20200210141326.289a52ff@windsurf> Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net Le 10/02/2020 ? 14:13, Thomas Petazzoni a ?crit?: > On Sun, 9 Feb 2020 19:03:24 +0100 > Romain Naour wrote: > >> This commit add the same post-image script for each qemu board >> in order to generate start-qemu.sh in BINARIES_DIR. >> The start-qemu.sh script can be used by Buildroot user to start >> Qemu or by a gitlab CI. >> >> To find the correct qemu command line in the readme.txt, we use >> the second post script argument which must contain "$(BR2_DEFCONFIG)" >> >> BR2_ROOTFS_POST_SCRIPT_ARGS="$(BR2_DEFCONFIG)" >> >> Doing a basename allow to retrieve the name of the defconfig >> file (ex: qemu_aarch64_virt_defconfig) that should match >> on on the "tag" previously introduced in the readme.txt. >> >> The script start-qemu.sh is able to modify the qemu command >> line if CI_JOB_NAME defined in the environment. This allow >> to disable the graphical output and redirect serial I/Os to >> console. >> >> Only sh4/sh4eb needs a special handling by adding >> "-serial stdio -display none" instead of "-display none". >> >> Signed-off-by: Romain Naour > > Is it really a good idea to automatically start Qemu at the end of the > build of those defconfigs ? While I understand how it makes sense for > Gitlab CI, it is not very practical/convenient for normal users. Qemu is not started automatically at the end of the build. The build process only create the script start-qemu.sh that must be called by the user (or by the gitlab-ci). > > Can't we start the test under Qemu by a separate command from > .gitlab-ci.yml instead? This is what this patch is doing :) Best regards, Romain > > Thanks, > > Thomas >