I'm trying to install an image from a hddimg to harddisk by selecting "install" from the bootloader menu. This works with genericx86-64 but fails with intel-corei7-64 on my NUCs. The boot process looks identical to the point where the installer should say something like "Searching for hard drives..." and list the options. On intel-corei7-64 this does not happen at all. I asked Maxin to try this as well as it seemed a bit weird and he got the same results.
Can you provide some additional details such as which image you were trying to install, from USB. Which NUC exactly do you have? Any other build switches or just MACHINE=intel-corei7-64?
(In reply to comment #1) > Can you provide some additional details such as which image you were trying > to install, from USB. ah sorry: it's core-image-sato and I used the EFI boot to install. > Which NUC exactly do you have? I've tested both NUC6i5SYH & D54250WYK with Same results. > Any other build switches or just MACHINE=intel-corei7-64? Nothing else and nothing special in my configuration as far as I can see: TMPDIR = "/mnt/extra-ssd/tmp" DISTRO ?= "poky" PACKAGE_CLASSES ?= "package_rpm" EXTRA_IMAGE_FEATURES = "debug-tweaks" EXTRA_IMAGE_FEATURES += "ssh-server-openssh" USER_CLASSES ?= "buildstats image-mklibs" OE_TERMINAL = "screen" PATCHRESOLVE = "noop" BB_DISKMON_DIRS = "\ STOPTASKS,${TMPDIR},1G,100K \ STOPTASKS,${DL_DIR},1G,100K \ STOPTASKS,${SSTATE_DIR},1G,100K \ ABORT,${TMPDIR},100M,1K \ ABORT,${DL_DIR},100M,1K \ ABORT,${SSTATE_DIR},100M,1K" PACKAGECONFIG_append_pn-qemu-native = " sdl" PACKAGECONFIG_append_pn-nativesdk-qemu = " sdl" ASSUME_PROVIDED += "libsdl-native" CONF_VERSION = "1" ICECC_PATH = "/usr/bin/icecc" ICECC_PARALLEL_MAKE = "-j 64" BB_NUMBER_THREADS = "16" INHERIT += "buildhistory" BUILDHISTORY_COMMIT = "1" INHERIT += "distrodata"
Cal suggested dropping "console=ttyS0,115200" parameter from the kernel command line: that indeed makes the installer prompts show up and input works. So the installer IO is presumably going to the serial console by default.
I was assigned to this bug because BSP team wanted me to provide a comment from RMC perspective then close it. We already have RMC examples which gets rid of hard coded console setting that caused error. But based on a talk over IRC, we know exactly why a kernel serial console setting steals the virtual console for installer program. And when people don't have RMC or disable it, we will have same issue again. I think I should root cause it with more investigation.
There is another bug that will address this issue