Description of problem: I see this messages in dmesg in the boot process: [ 3.936279] Write protecting the kernel read-only data: 3576k [ 4.000260] Clocksource tsc unstable (delta = -77010480 ns) [ 4.219021] end_request: I/O error, dev fd0, sector 0 -> Here I need to enter my password for disk decryption. [ 25.763030] end_request: I/O error, dev fd0, sector 0 [ 26.298030] end_request: I/O error, dev fd0, sector 0 [ 27.521031] end_request: I/O error, dev fd0, sector 0 [ 28.064809] EXT4-fs: barriers enabled [ 28.067738] kjournald2 starting: pid 1271, dev dm-0:8, commit interval 5 seconds [ 28.067756] EXT4-fs: delayed allocation enabled [ 28.067761] EXT4-fs: file extents enabled [ 28.069202] EXT4-fs: mballoc enabled [ 28.069208] EXT4-fs: mounted filesystem dm-0 with ordered data mode I'm not sure who's responsible for this (senseless?) accesses, but these seems to completely happen in initrd, i.e. before mounting real root. Is this a bug or feature? Version-Release number of selected component (if applicable): udev.x86_64 139-2.fc11 nash.x86_64 6.0.81-1.fc11 busybox.x86_64 1:1.13.2-2.fc11 How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Created attachment 335939 [details] init from initrd
if this is happening in initrd, then nash is to be blamed.
Thomas, Do you by any chance have an IDE floppy drive? Or do you have built your own kernel? The classic floppy driver is a module and thus is not present when the initrd gets executed, so no way these messages can come from the initrd with a Fedora provided kernel.
> Do you have built your own kernel? Yes. Indeed, I do. With build-in floppy driver, no module. So I guess the above described behaviour is a feature ;-)
(In reply to comment #4) > > Do you have built your own kernel? > > Yes. Indeed, I do. With build-in floppy driver, no module. > > So I guess the above described behaviour is a feature ;-) Yes, could be caused by lvm or by specifying root (in the kernel cmdline, through grub.conf) as UUID= or LABEL=, lvm will scan all block devices (although it may skip floppies) for physical volumes and specifying root as UUID= / LABEL= will cause nash to scan all block devices to find the UUID / LABEL. You could try working around this by specifying your root= as /dev/.... in grub.conf Closing this as not a bug.