On 08/29/13 18:44, Adrian Chadd wrote:
Hi!


Let's not finish this just yet! I'd like to try and nail down exactly
what's going on so PRs can be filed.

Absolutely. (I was thinking to do this off-list to minimize spam.)

* Is this with 9.2-RC2?
Yes this is 9.2-RC2. I think the problem is that the simple install path ends up with out-of-date components, rather than anything wrong with the components that should be used.

Are you wiling to try out a -HEAD snapshot to
make sure this stuff in -10 is also going to work?
Yes, but probably not until the weekend.

* Ok, so if you have no VESA option, does suspend/resume work in console
mode?
No. The system resumes (can ssh in), but the backlight doesn't come on (using a flashlight, I don't see any video mode running either). See http://www.sics.se/~lmfeeney/9.2-RC2/messages_console.

* Try manually setting the cpu frequency low and high (sysctl dev.0.cpu
- you can change 'freq') - see if that fixes your speed issues. Someone
else has reported this.

Before and after resume, dev.cpu.0.freq: 1700. Setting to 100 and back to 1700 didn't seem to have much effect.

Basically, what I see is that after resume, it seems to be easier to increase the load on the cpu (i.e. looking at top) by e.g. scrolling quickly.

My "test" sequence was: boot,
'startx' (so twm and a couple xterms)
'top (in one xterm)
'xterm -sb -geometry 100x65' (make xterm w/ scrollbar),
'cat /var/log/messages' (in the new xterm)
scroll very rapidly up and down for ~20sec
look at top output

suspend/resume and repeat the scrolling. The load seems to go quite a bit higher after resuming (CPU from ~10% to ~20%). But this is definitely not a scientific test AND I was primed by Gleb Smirnoff's comments to look for a performance hit.

For sensible use (i.e. not madly scrolling), I don't sense slowdown. But the machine is completely minimal: portmaster and xorg are the only ports installed. It could be that KDE + web + mail + applications managing lots of things on the screen would show a big slowdown, as reported by others.

It would probably make sense to define some metric that people who have reported problems can compare before and after suspsend/resume.

Best,
Laura
_______________________________________________
freebsd-acpi@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-acpi
To unsubscribe, send any mail to "freebsd-acpi-unsubscr...@freebsd.org"

Reply via email to