Distribution: Gentoo with kernel from kernel.org Hardware Environment: Acer TM630, see http://www.odi.ch/prog/tm630/index.php Software Environment: Gentoo Linux, see http://www.odi.ch/prog/tm630/gentoo.php Problem Description: The RTC runs too slow. I get about 50 % of the speed that it should be. Maybe related to bug #3057 Steps to reproduce: date (wait 1 minute) date or sleep 10 (will sleep for 20 seconds)
with 2.6.8.1-mm1: [damir@Asteraceae /]$ date; sleep 1; date; sleep 2; date; sleep 3; date Thu Aug 19 14:42:34 CEST 2004 Thu Aug 19 14:42:35 CEST 2004 Thu Aug 19 14:42:37 CEST 2004 Thu Aug 19 14:42:40 CEST 2004 so it's either vanilla-specific, or gentoo-problem or Hardware to blame :-)
Damir, Of course the output of your test is consistent. But you have to compare the actual time with a real clock (use your wrist watch). All works fine when I use 2.6.6. But since I use 2.6.8.1 my clock is running at 50 % of real speed. I wouldn't mind if that meant that I could live twice as long :-) It's not about who to blame. It worked before, and now it's broken. All I changed is the kernel. If I change back to 2.6.6 the clock runs correct.
The `dmesg -s 1000000' output would be useful.
Created attachment 3532 [details] dmesg -s 1000000 File as requested. The machine has been running for some hours now.
See also Fedora bug: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=127235 Same thing but on an HP Pavillion ZE-4201 notebook.
The bug is still present in 2.6.9
Still seeing this in 2.6.10-rc3. clock=pmtmr does not help either. Is there anything I can do to help fixing this?
WFM: with 2.6.10 using CONFIG_X86_PM_TIMER it works correctly now.
Does this problem still exist or is it considered resolved? If it still exists, does acpi=off change the behaviour?
It works now even without CONFIG_X86_PM_TIMER in 2.6.13-rc6. I tried with 250 and 1000Hz settings, with and without ACPI. Thanks a lot!
Considered resovlved by the submitter. Reopen if necessary.