On Thursday, March 26, 2020 7:50:29 AM CET andrey....@gmail.com wrote: > On Wed, Mar 25, 2020 at 8:23 AM J. Roeleveld <jo...@antarean.org> wrote: > > On Tuesday, March 24, 2020 10:57:59 PM CET andrey....@gmail.com wrote: > > > On Tue, Mar 24, 2020 at 2:45 AM J. Roeleveld <jo...@antarean.org> wrote: > > > > On Tuesday, March 24, 2020 4:55:49 AM CET andrey....@gmail.com wrote: > > > > > I've somehow built a kernel that won't accept any initramfs. To > > > > > elaborate, it hangs in shell in dracut and panicks with a minimal > > > > > initramfs as if it rejects the initramfs entirely and tries to boot > > > > > an > > > > > lvm volume, which of course, fails. > > > > > > > > > > Kernel config: > > > > > http://sprunge.us/tqz3BM > > > > > > > > > > The minimal initramfs is described by the file listing > > > > > http://sprunge.us/GQVbqc > > > > > and init script http://sprunge.us/YHWY9Z , though this script is of > > > > > limited use since even with `exec sh` in the first non-comment line, > > > > > the kernel doesn't follow through. > > > > > > > > > > Thoughts? > > > > > > > > Please provide the info as attachments instead of temporary links. > > > > Few thoughts: > > > > 1) Are you certain your initramfs includes everything? My "config" file is > > a lot longer. > > Yes, and it is the absolute bare minimum necessary.
Ok, I don't use static binaries, simply use a ldd-script to generate the shared libs-list for the config. > > 2) How are you generating the initramfs? > > /usr/src/linux/usr/gen_init_cpio /usr/src/initramfs/files.list | xz -z Have you checked all files are included after this? > > /boot/intramfs-....img > > > > 3) how are you adding it to the bootloader? > > Grub's default form. To elaborate, the usual method grub does when > generating a linux entry in grub-mkconfig. Did you manage to find out if the initramfs file was actually added to the correct entry? > > 4) Have you tried embedding the initramfs into the kernel? > > Yes, and that made it work. Weird how it didn't work not embedded into > the kernel. > On further thought, it worked after a couple of debugging iterations. > Could it have been the bugs in the script at fault? Perhaps, but I suggested this as it's the only way I could make it work on my old laptop. "grub-mkconfig" only matches kernels and initramfs files together if they follow a specific naming convention. > > Also, can you provide the kernel messages leading to and including the > > panic? (Either full text or a clear photo taken with a > > camera/phone/tablet) > Well, the two panics were either trying to kill init or the familiar > "VFS: Unable to mount root fs on unknown-block(0,0)". Kernel > framebuffer load reset grub's high res console and the panic took over > the screen so messages right before would've been impossible to > extract without something like a serial console. I tend to disable the framebuffer and screen cleaning until I have a working boot. And when it boots, I tend to forget enabling all that as I have more important things to do. -- Joost
