It appears my hardware clock just failed, so Gentoo Linux is no longer able to read it. (100% CPU, read/write from/to hwclock times out.) When it failed, it just kinda gave up, telling me to set it, but not giving me a chance before every other service started up and freaked out. Reproducible: Always Steps to Reproduce: 1. Kill your hardware clock. 2. Turn your computer on. Actual Results: Your time is no longer accurate. Expected Results: It should ask you what time it is, and set your system clock according to that. I'm attaching a patch which I've tested on my system, and appears to work fine. I'm at a point in my bash scripting skills where I know enough to be effective, but little enough to be dangerous. Please review it before applying. ;)
Created attachment 178175 [details, diff] Manual time input
I don't think a user should be prompted for date input in the middle of a boot. Yur patch reveals that it's already printing error for the user. Also please post emerge --info output.
yeah, as Serkan says, we arent going to make things interactive if you want to recover the clock, use ntp or similar service. does your system not boot if the hwclock is set to something wrong ? we can (should) fix that, but that's about it ...
Well, it's not that it doesn't boot (at least, I never got into the default bootlevel - I had hit Ctrl+Alt+Del by that point), it's that I simply don't know what was time-sensitive and could've been messed up by a bad clock. You could ask the user whether or not they want to enter the time manually, and have that prompt time out after a little bit. That would give you a non-interactive shell while letting "normal" desktop users set their clock before something bad happens.
messed up clocks wont cause irreparable damage. just log in, fix the clock, and reboot. at this time we arent adding sanity checks and autofixing things for the user. that isnt a goal and i'm not sure if it's something we want (due to the additional overhead for negligible gain).
