Messages in this thread | | | From | "Pan, Harry" <> | Subject | Re: [PATCH 3/3] perf/x86/rapl: Enable Baytrail/Braswell RAPL support | Date | Tue, 13 Sep 2016 17:30:43 +0000 |
| |
On Tue, 2016-09-13 at 15:41 +0200, Thomas Gleixner wrote: > On Sun, 11 Sep 2016, Harry Pan wrote: > > This patch also enables multiple quirks. > > This patch adds a single quirk for Baytrail. > > Please stop sending out patches 5 seconds after a review. Take your time Definitely I take this seriously because I felt awkward as well.
> > + /* > > + * Some Atom processors (BYT/BSW) have 2^ESU microjoules > > + * increment, refer to Software Developers' Manual, Vol. 3C, > > + * Order No. 325384, Table 35-8 of MSR_RAPL_POWER_UNIT. > > + * > > + * TODO: In order to fit BYT/BSW quirk model, here remind > > + * this generates timer rate in 80ms; by default > > + * ESU of BYT/BSW is 5, so it leads (1000/200)*2^4. > > This sentence is not a sentence and I can't make any sense of it at > all. > > What's the TODO here? And why is that TODO not addressed in this patch? > I reviewed my sentence and agreed your comment; yes, it is incorrect to be a "TODO" tag since no decent suggestion/option.
This things is because of the Baytrail/Braswell quirk breaks original assumption of perf RAPL polling timer rate calculation regarding of counter overflow case based on 200W; in short, it leads every 80ms system triggers an event to read counters, and this is concern I want to comment (wrong tag?) because I could no assess any side effect. Perhaps I should revise it as "remark" or "caveat" because I do not have decent suggestion (fulfill "TODO" tag) so far.
Alternately, it shall not affect functionality since I compared w/ powercap driver through sysfs nodes during experiment, yet I am humble to take any advice to make this patch better.
Sincerely, Harry
| |