[Solved] VFS: Unable to mount root fs on unknown-block(0,0)

piotr5 · Tux's lil' helper Joined: 14 Dec 2012 Posts: 117

I'm on linux-4.4.6 since linux-4.9.16 fails to boot with above kernel panic. I tested the new kernel with the old initramfs and it boots correctly (although not to the end since kernel-modules are missing.) then I tested the old kernel with the new initramfs and I get a kernel panic of unknown type. did I do something wrong in creating initramfs? are there any constraints except for cpio of newc type compressed by xz?

NeddySeagoon · Posted: Thu Apr 13, 2017 12:03 pm Post subject:

piotr5,

piotr5 · Tux's lil' helper Joined: 14 Dec 2012 Posts: 117

there isn't much to tell: I started "genkernel all --kernel-config=/etc/kernels/kernel-config-x86_64-4.4.6-gentoo", then extracted the resulting initrd, altered it and added some files, and packaged it up again with

NeddySeagoon · Posted: Fri Apr 14, 2017 8:20 am Post subject:

piotr5,

Any kernel version should work. I didn't intend to suggest otherwise.
You cannot mix one kernel with the initrd intended for another as one kernel will not load a different kernels modules.
The

piotr5 · Tux's lil' helper Joined: 14 Dec 2012 Posts: 117

I don't need to create any new initramfs, the one genkernel created works -- except that as it is unchanged I cannot mount root from there. the culprit is my creating a new initramfs. the altered one just boots till input-setup and then before entering any init scripts just complains it cannot mount unknown-block(0,0), i.e. it cannot mount initramfs or whatever virtual filesystem it's supposed to mount first. I know that for sure since my real linux-partition is far from any kind of (0,0) partition.

as for lvm and luks, I am using both, but they are in my config-file for genkernel, not on command-line. so "genkernel all" is by far not as simple as I can...

so, I did extract both, the original initramfs as created by genkernel and the altered version. and diff -r only shows the expected differences: one startup-file which didn't even get executed before crashing, and a few files it would have required. well, diff also complained about "dev/console", "dev/null", "dev/tty0", "dev/tty1", "dev/ttyS0" and "dev/zero" being character-devices. but that too is to be expected. so the only mistake I could have made is in the commandline for cpio or xz. so, what's wrong with

szatox · Advocate Joined: 27 Aug 2013 Posts: 3137

Maybe it's about your boot params rather than initramfs.
Try going with root=/dev/ram0 real_root=<your root device>.
Also, adding "debug" will drop you to shell once or twice during early boot, so you can have a look around.

And talking about building your custom initramfs: kernel's builtin decompressor for xz supports only a subset of xz compression options. It is not able to understand the output from xz's defaults. I don't know what the supported subset or what options you need to pass to xz. The easy mode is to go with gzip or bzip2 instead. I think gzip is even enabled by default, so you don't have to recompile kernel.

piotr5 · Tux's lil' helper Joined: 14 Dec 2012 Posts: 117

many thanks! that's the answer I were looking for! I made the mistake of believing the subset of xz supported by kernel would mean lower compression-rate. now it seems having checksums is a problem.