From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <4B7523E6.1080008@domain.hid> Date: Fri, 12 Feb 2010 10:48:22 +0100 From: Jan Kiszka MIME-Version: 1.0 References: <201001292104.42395.lux-integ@domain.hid> <4B657977.6050906@domain.hid> <1264944150.16492.72.camel@domain.hid> <4B6584DA.2010707@domain.hid> <1264944906.16492.74.camel@domain.hid> <4B6588AC.4070105@domain.hid> <1265280370.2348.130.camel@domain.hid> <4B743B4E.3060703@domain.hid> In-Reply-To: <4B743B4E.3060703@domain.hid> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: [Xenomai-help] help with xenomai-2.5.0 on pure64bit linux x86_64 List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Wolfgang Mauerer Cc: xenomai@xenomai.org Wolfgang Mauerer wrote: > Philippe Gerum wrote: >> On Sun, 2010-01-31 at 14:42 +0100, Jan Kiszka wrote: >>> Philippe Gerum wrote: >>>> On Sun, 2010-01-31 at 14:25 +0100, Jan Kiszka wrote: >>>>> Philippe Gerum wrote: >>>>>> On Sun, 2010-01-31 at 13:37 +0100, Jan Kiszka wrote: >>>>>>> lux-integ wrote: >>>>>>>> Greetings >>>>>>>> >>>>>>>> I am new to xenomai and I am attempting to install a RT-kernel on a pure >>>>>>>> 64-bit (not multilib) amd64 linux setup. The machine has a Sempron CPU and >>>>>>>> 2 GBytes of RAM. The distribution is cross-lfs compiled from sources. This >>>>>>>> is what I have done:- >>>>>>>> >>>>>>>> a) I fetched fresh kernel2.6.32.2 source-code and placed it >>>>>>>> in /usr/src/linux like so:- >>>>>>>> `linux' -> `linux-2.6.32.2' >>>>>>>> >>>>>>>> b) ..I then applied the xenomai patch as such:- >>>>>>>> $SOURCES/xenomai-2.5.0/scripts/prepare-kernel.sh --linux=/usr/src/linux --adeos=$SOURCES/xenomai-2.5.0/ksrc/arch/x86/patches/adeos-ipipe-2.6.32.2-x86-2.5-00.patch --arch=x86_64 >> >> >>>>>> This is likely our old friend CONFIG_SPARSE_IRQ which is at work. Just >>>>>> disable this, until we have introduced the required changes to support >>>>>> more than 1024 IRQs. >>>>> If that is true, then please include a simple "depends on !CONFIG_IPIPE" >>>>> in the x86 patch for the meantime. >>>>> >>>> No. Work to lift this limitation is in progress here. >>> That's even nicer (if 'lift' means removing any limitation of >>> CONFIG_NR_CPUS). >>> >>> Jan >>> >> Could someone gives this patch a go, with CONFIG_SPARSE_IRQ on (or some >> unreasonable number of CPUs in 64bit mode, whichever comes first)? >> >> http://download.gna.org/adeos/patches/v2.6/x86/adeos-ipipe-2.6.32.7-x86-2.6-00.patch > > boots (and seems to work fine after a rather short test) w/o Xenomai and > the attached .config (SPARSE_IRQ, NR_CPUS=512). With Xenomai is compiled > in, it reboots right after registering the Xenomai domain with ipipe - > see the attached console logs. Did not dig any deeper so far. > The test was done on qemu 0.10.50 (-cpu core2duo, KVM) Try to disable CONFIG_KPROBES_SANITY_TEST. I guess one problem is that Kprobes aren't prepared for use over I-pipe. I still have to stable 2.6.32 here over qemu (w/ or w/o kvm). Some setups even fail under -smp 1. :( Will try Wolfgang's config now. Jan -- Siemens AG, Corporate Technology, CT T DE IT 1 Corporate Competence Center Embedded Linux