Red Hat Bugzilla – Bug 733851
UEFI system no longer boots after 3.1 kernel update
Last modified: 2013-01-10 02:41:05 EST
Created attachment 520174 [details]
output of lshw
Description of problem:
After installing F16's last kernel update my system no longer boots. Instead, the following is shown on screen:
Trying to allocate 1059 pages for VMLINUZ
Got pages at 0xd59d5000
[Linux-EFI, setup=0x10cc, size=0x422db0
From this point on nothing happens anymore, while the caps lock LED is blinking. The value for setup varies if booted with different kernel parameters. The system is a Thinkpad X220 (exakt model: 4290W1B, see attached lshw output) with BIOS compatibility module disabled. Fedora is installed on a GPT disk with EFI system partition and gets started from Grub legacy. Pretty much the standard Anaconda recommendation for UEFI systems.
With the 3.0 kernel from the initial installation the system starts fine. Naturally, there are no syslog entries from the failed boot attempts with the 3.1 kernel.
Version-Release number of selected component (if applicable):
$ rpm -q kernel
Steps to Reproduce:
1. Install 3.1 kernel from updates-testing repo
2. Reboot system
System hangs after hitting enter in Grub
System comes up like with previous 3.0 kernel
Output from lshw
dmesg output from successfull boot
Created attachment 520175 [details]
output of dmesg when booting with kernel 3.0
Matthew, any ideas?
I'll try to reproduce on Monday.
After I looked at it again today this is not kernel related at all. The problem seems to be grubby not creating a proper initrd entry for the new kernel in grub.conf. Argh. How could I have missed _that_?
(In reply to comment #4)
> After I looked at it again today this is not kernel related at all. The problem
> seems to be grubby not creating a proper initrd entry for the new kernel in
> grub.conf. Argh. How could I have missed _that_?
Thanks for letting us know. If you have logs that show errors from grubby, you might want to open a bug against it or probably dracut.
The missing initrd entry is probably bug 725185