From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <441AE319.8060102@domain.hid> Date: Fri, 17 Mar 2006 17:26:01 +0100 From: Philippe Gerum MIME-Version: 1.0 Subject: Re: FW: [Xenomai-help] Xenomai thread runtime statistics References: <067C9A1F6AFEB643895EA4513E116884178FC4@domain.hid> <441ADC6B.5010909@domain.hid> <441ADFAA.9040602@domain.hid> <441AE047.6020505@domain.hid> <441AE25F.7040903@domain.hid> In-Reply-To: <441AE25F.7040903@domain.hid> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Jan Kiszka Cc: xenomai@xenomai.org, "Moser, Dan" Jan Kiszka wrote: > Philippe Gerum wrote: > >>Jan Kiszka wrote: >> >>>Philippe Gerum wrote: >>> >>> >>>>Moser, Dan wrote: >>>> >>>> >>>>>Hi, >>>>> >>>>>I am still searching for ways (hopefully via some API(s)) to get >>>>>Xenomai >>>>>run-time thread execution info (my original post is below). >>>>> >>>>>Being a newbie, I apologize if the answer "should be obvious". I am >>>>>still getting up to speed with both Linux and Xenomai. >>>>> >>>>>Is there a better list on which I should post this question? >>>>> >>>> >>>>This question is perfectly relevant to this list, the thing is that >>>>there's no simple, built-in way of getting execution stats for real-time >>>>threads yet. The best way to find out the execution time of your work >>>>cycle would be to measure it using the high precision clock information. >>> >>> >>>Hmm, wouldn't it be feasible to add such statistics to I-pipe level? I'm >>>thinking of something like time spent in domain X, Y, Z vs. the overall >>>runtime, preferably converted to percentage usage. >>> >> >>This would be nice, but only part of the issue, since you might have >>several real-time context switches occurring in a single Xenomai domain >>activation. > > > You mean per-thread statistics? Yes, like in Dan's case, I guess. For sure, but a very common question is: > "How much remains for Linux?" And this feature could give a general hint > (although detailed benchmarking will typically be required as well, as > the effect of Xenomai+RT-load on application-specific Linux service may > have to be analysed more thoroughly). > Also true. > Jan > -- Philippe.