Eric Saxe wrote:
> Hi Rafael,
> 
> Rafael Vanoni wrote:
>> Forwarding Eric's code review with comments and attaching a patch with 
>> fixes. Please see inline.
>>
>>> battery.c
>>>    - battery_stat_snapshot() returns error which print_battery 
>>> ignores.  If print_battery() isn't
>>>       going to do anything, should battery_stat_snapshot() just be of 
>>> type static void?
>>>    - Perhaps it could instead call pt_error(), and emit some sort of 
>>> error message if something
>>>      goes wrong reading the battery kstats?
>> Fixed in the proposed patch.
>>
>>> cpufreq.c:
>>>     - I'm still dubious about having PowerTOP check, and enable cpupm 
>>> in the way that it does.
>> Yeah, I'm working on that. But if we remove it (since cpupm is on by 
>> default now), we should remove all the suggestions code. Which is also 
>> okay, since I'm working on something to allow suggestions through a 
>> proper interface. Does that sound reasonable ?
> The suggestions framework I think is fine (actually, isn't there a 
> suggestion to suggest running as root to get all the functionality?).
> Which interface are you looking to migrate to for enabling CPU power 
> management (if it's disabled)?

Yeah, you're right. The suggestions code will still be used, the back 
end would be different.

thanks
Rafael

>>>     - pt_cpufreq_snapshot() isn't very robust if something goes wrong 
>>> looking up the kstats. It
>>>       should probably check the return codes from the kstat 
>>> interfaces, and use pt_error, and
>>>       perhaps return either a fatal (or non-fatal) error as 
>>> appropriate if something goes wrong.
>> Also fixed in this patch.
>>
>>> display.c:
>>>     - event_bubblesort(): there's a qsort() library interface...can 
>>> we use that, or some other more
>>>        efficient sorting routine instead since bubble sort is O(n2)
>> Done.. that was easier than I thought - and should've been fixed a 
>> while ago :/
>>
>>>     - show_acpi_power_line(): 351: typo: "enery" should be "energy"? 
>>> Should this say something
>>>        like: "Critically low battery power"?
>> Yeah, "critical energy state" could be anything from a nuclear 
>> meltdown to low batt :) Changed in this patch too.
> Heh.
> 
>>> powertop.c
>>>      - 315: Do you have to run as root to get suggestions, or do you 
>>> need certain "privileges"?
>>>                Will PT allow you to run as non-root anyway?
>> The only existing suggestion requires root privileges, so we don't 
>> suggest anything if a regular user is running the tool. Non-root users 
>> can run PowerTOP as long as they have dtrace privileges.
> Thanks..although I think it's the case that nothing is suggested other 
> than suggesting that the user run as root (to get suggestions). :)
> 
>> Another change I'm suggesting with this patch is that we only display 
>> the Intel copyright banner as part of the usage message. The rationale 
>> for this one is that showing the copyright during execution might 
>> imply that the data being generated on a user's system is "owned" by 
>> Intel. I've also updated the copyright year to 2008. But I'll 
>> obviously defer to our friends at Intel on both of these changes ;)
>>
>> Please send your comments asap as we're putting PowerTOP back into 
>> snv100 for OpenSolaris 2008.11 - which closes on the 29th.
> Thanks,
> -Eric
> _______________________________________________
> tesla-dev mailing list
> tesla-dev at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/tesla-dev


Reply via email to