[Bug 211897] horrible time skew under 10.3-R-p7

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=211897

--- Comment #8 from Martin S. Weber <freebsd-bugzilla at ayaken.net> ---
(In reply to Konstantin Belousov from comment #4)

So, I can confirm that unpatched finds my CPU at 4GHz equals 4000 MHz.

Sadly there seems to be a random element in -p7 finding my CPU at 4 GHz equals
4400 MHz, as my most recent two -p7 boots could not reproduce; I've checked,
out of my previous 24 boots (4 of which being -p7) only once (a p7) my CPU was
found to be 4439.49-MHz, all the other times it's 4000.08 +/- 0.01 MHz

On a -p7 boot that does not see my CPU as 4.4 GHz I cannot see the massive
clock drift I have observed on the one p7 boot. I do not need to adjust my
eventtimers & timecounters - it just works. I'll have to wait and see when the
next boot gives me a 4.4 Ghz. Maybe it's windows (I dual boot between work OS
and gaming console) that left something programmed in the CPU (even though I
cold-boot between the OS as otherwise my sound & video do funny things).

Per Arthur Chance, I note that sysctl.dev.cpu.0.freq_levels outputs (on -p7
that sees my CPU as 4Ghz):
dev.cpu.0.freq_levels: 4001/88000 4000/88000 3800/81704 3500/72258 3300/66530
3100/60999 2900/55687 2600/48456 2400/43627 2200/39003 1900/32072 1700/27930
1500/23963 1300/20178 1000/15137 800/11779

Not sure where the PR is going, I clearly had the problem once, even though I
cannot reliably reproduce it now.

-- 
You are receiving this mail because:
You are the assignee for the bug.