http://bugzilla.kernel.org/show_bug.cgi?id=10561
------- Comment #9 from [EMAIL PROTECTED] 2008-04-29 03:44 ------- Created an attachment (id=15974) --> (http://bugzilla.kernel.org/attachment.cgi?id=15974&action=view) Info about system time being slowed down. | Can you reproduce this when the p4_clockmod driver is not used? No, I tried that in the past and again now -- I waited a long time and also run X and everything seems fine. As I mentioned at the beginning, I *can* remove p4-clockmod (and speedstep_lib) and the lockups will continue, but it seems to trigger something that causes the lockups... | (and why are you using p4_clockmod on a modern processor in the first place instead of acpi_cpufreq?) acpi-cpufreq won't load on this machine: "no such device". Another thing I tried is, after it had already locked up once, I echoed acpi_pm into /sys/devices/system/clocksource/clocksource0/current_clocksource, which caused a message to appear about acpi_pm being enabled (so it worked ok). However, it still locked up afterwards... | Another thing to try is to load the processor module | with processor.max_cstate=1 (via /etc/modprobe.conf) You mean "option processor max_cstate=1" (the other notation is for builtins). When probing processor I got a message about it being limited to max C-state 1. It also caused the time/tsc problem (below) to go away so I got no "tsc unstable" message and it didn't lock up. However, unlike booting with hpet=disable, in this case it kept using tsc, rather than use acpi_pm. If I'm not mistaken, it also means the processor can't sleep... so it's not very power-efficient. I discovered something about the "Clocksource tsc unstable" messages: When I boot (with hpet enabled or disabled), tsc is used at first. Then I modprobe processor and there's a period of from a few seconds to a few minutes before that message appears (after which the clocksource used changes to hpet or acpi_pm, depending on how I booted). Now, during that period, is seems like time pretty much comes to a halt! For every 5 seconds that pass in the real world, the system time changes by about 20 ms. I've gathered some data that shows how it develops and am attaching it (more info inside). Note that this is the same whether I boot with hpet enabled or disabled, as hpet/acpi_pm aren't used yet at this stage. -- Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee. ------------------------------------------------------------------------- This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone _______________________________________________ acpi-bugzilla mailing list acpi-bugzilla@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/acpi-bugzilla
