From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <45ED59ED.9080004@domain.hid> Date: Tue, 06 Mar 2007 13:09:17 +0100 From: Theo Veenker MIME-Version: 1.0 Subject: Re: [Xenomai-help] Re: high latency every 8 seconds References: <45E4646A.10205@domain.hid> <45E48E50.1030502@domain.hid> <45ED4351.2020905@domain.hid> <200703061234.51460.schwab@domain.hid> In-Reply-To: <200703061234.51460.schwab@domain.hid> Content-Type: text/plain; charset=us-ascii; 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: Ulrich Schwab Cc: xenomai@xenomai.org, Jan Kiszka Ulrich Schwab wrote: > On Tuesday 06 March 2007 11:32, Theo Veenker wrote: > >>Jan Kiszka wrote: >> >>>Theo Veenker wrote: >>> >>>>Jan Kiszka wrote: >>>> >>>>>Theo Veenker wrote: >>>>> >>>>>>Hi all, >>>>>> >>>>>>I'm installing xenomai 2.2.5 (kernel 2.6.17.14) on a number of >>>>> >>>>>When starting new, pick something new, i.e. 2.3 right now. Only if >>> >>>you >>> >>> >>>>>have good reasons, staying at older versions is recommended. >>>> >>>>I would love to, but I couldn't get my software running with 2.3. >>>>I had to convert (a month ago) my software from RTLinux to Xenomai >>>>having only a few days for it. I have a kernel module and an app >>>>that talks to it. I use the posix skin. But with 2.3 apparently >>>>I had to load posixio and something (forgot) manually, but I just >>>>couldn't get these loaded. Since time was running out I decided >>>>to stay with 2.2.5 for the time being (unfortunately). >>> >>>I do understand. But once your time constraints relaxed a bit, please >>>don't forget to post remaining issues here. Maybe they reveal some real >>>bug. >>> >>> >>>>>>systems (MSI 945P, Intel P4 3Ghz). My kernel runs fine on several >>>>>>other/older systems, but on the new systems there is something going >>>>>>on every 8 seconds. See output of latency test below. What could >>>>>>this be? Is this an SMI effect? >>>>> >>>>>Is CONFIG_XENO_HW_SMI_WORKAROUND selected? What does the kernel output >>>>>tell you about if it's working? Do you have an ICH7 chipset on your >>>>>board? Would be another reason to update Xenomai (2.3 takes care of >>>>>it, 2.2.x doesn't)... :) >> >>OK. I finally installed my clients. It took a while because although all >>systems are identical they didn't all behave the same. Apparently some >>MSI 945P boards have problems when installing an ubuntu 686 kernel image. >>Taking a 386 image as a starting point worked. >> >>Recompiled my kernel 2.6.17.14 with xenomai 2.2.5 and SMI WORKAROUND >>selected, but it didn't have any effect. I don't see any xenomai >>output in the kernel logs. Same thing without the workaround. So it >>doesn't even detect SMI. The mobo has a ICH7 chipset. > > ICH7 is not in the list of SMI chipsets in xenomai 2.2.5 > compare the file ksrc/arch/i386/smi.c between versions 2.3 and 2.2.5 > This is the only change needed for the SMI detection and workaround > to work for this chipset in 2.2.5, if You decide NOT to switch to 2.3 > You will need to enanble the SMI workaround in the configuration, of course. Thanks! I'll use that until I manage to use 2.3 for my app. Theo