Li, Aubrey wrote: > Comments below: > > 1. cpufreq.c > 1) Need to mention, I do not object to use event mode. > But from some platforms I tested like montevina, powertop > always show the highest frequency in event mode, while poll > mode can step the frequency down when idle.
I believe that's a bug in PAD, and that Eric has a fix for it. > 2) I raised this issue in the last comments: don't change > "Enable P-state" to "Enable PM", and if I recall correctly, you > agreed with that, but it's clobbered again. Yes, I'm fixing it in this new webrev. I'm moving this to the platform specific code. > 2. suggestion.c > The suggestion was originally designated to give power related > suggestion to the user to improve power efficiency, I don't think adding > the suggestion of freezing window is a good idea here. We can find > another way to let the user know that if it's necessary. For example, > showing "I", "F", "E" at the beginning in the status bar the same as "Q" and > "R". I replied to this comment before. If we show the options at the beginning, we'll be displaying them when they are off, what would we display when they have been activated? It also clobbers the overall display, I prefer only displaying it when it's activated. > 3. powertop.c > Freezing window feature needs more work here. It does not freeze the window > only, but changes the report behavior. Sometimes I saw 100% C0 residency and > it's unrecoverable. Sometimes I saw the screen like the following and can not > recover as well. > ============================================== > opensolaris powertop version 1.1 > > C-states (idle power) Avg residency P-states (frequencies) > C0 (cpu running) (6.2%) 800 Mhz 0.0% > C1 4.2ms (93.8%) 1600 Mhz 0.0% > 2533 Mhz 0.0% > 2534 Mhz 100.0% > > Wakeups-from-idle per second: 0.0 interval: 0.0s > no ACPI power usage estimate available > > Top causes for wakeups: > > > > > > > > > > > Sugguestions: you can freeze the 'I'dle, 'F'requency or 'E'vents subwindow > Q - Quit R - Refresh > ======================================================== > These issues are replicable when you keep pressing "I" per ptop interval. > Actually, when you close the cstate statistics, total event number is not > valid for the > subsequent report, and the whole report will be messed up. Ok, I'll have a look at this tomorrow morning. But the problem is not persistent, if you wait until all the keystrokes are processed, the report goes back to normal. Not sure there's a way to avoid this since it's the delay of stopping and restarting the DTrace program every time you freeze one of the windows. Rafael
