Hi,
isn't that a bug in e2fsck anyway? There is accept_time_fudge which
defaults to true and should take care of this situation. Even when it
wouldn't default to true, my e2fsck.conf already had its alias
buggy_init_scripts set to 1. Nevertheless, I'm briefly seeing an fsck
running now on every boot (CMOS clock is ticking CET localtime).
BTW, no longer fixing up system time before fsck is one thing, but
what about mounting? Ain't that prone to create file timestamps in the
future for everything that is touched before the init system has a
chance to run a correctly parameterized hwclock?
Wouldn't it make more sense to have TZ correction an optional module
of initramfs so the user can decide whether they want the rock or the
hard place? People running their CMOS in localtime or never ever
changing their TZ in decades (like me) might like to have it in there,
while highly mobile others who nevertheless want their clocks to
follow the sun might opt for leaving it out and living with the need
to disable e2fsck's detection of an entire error class and an
occasional warning from tar and thelikes about files from the future.
Let alone those without a dual-boot game console installed, who would
run their CMOS clocks in UTC anyway. I probably don't grok the complete
problem space, though...
Ideally, there would be some way of passing a small set of crucial
configuration AV pairs into initramfs without the need for a full
rebuild, and TZ and UTC would be two of them...
Andre.
--
Cool .signatures are so 90s...
-> Andre Beck +++ ABP-RIPE +++ IBH IT-Service GmbH, Dresden <-
--
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
