From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <511DF744.7090402@xenomai.org> Date: Fri, 15 Feb 2013 09:52:20 +0100 From: Philippe Gerum MIME-Version: 1.0 References: <511BEACC.5030705@xenomai.org> <511D3366.6020504@xenomai.org> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Xenomai] Segfault error 4 when running latency test (and all userspace tests) List-Id: Discussions about the Xenomai project List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Kris Ellis Cc: xenomai@xenomai.org On 02/14/2013 08:32 PM, Kris Ellis wrote: > Thanks, I'll look into your buldroot generator. > Well, actually this was a wrong advice, at least with respect to using buildroot: that generator does not use BR2, only the one I'm using locally does. But I had to hack the Xenomai support in BR2's 2010.02 release, which you do have built-in now in the latest one. > I have not tried using gdb yet, so I have no backtrace that I can provide. > This would help. > As far as the --enable-x86-sep flag is concerned, is there a place in > buildroot to turn this on, or do I have to go and manually modify the > xenomai.mk file? > You need to fix up XENOMAI_CONF_OPT manually. I suspect that you might have a kernel compiled for a CPU which does not provide SEP, with the userland code configured by default with --enable-x86-sep. So I would try check this first, prior to pulling more hairs: First, log in your target box, and check whether /proc/cpuinfo mentions "sep" as an available feature for your processor. If not, then either: - fix your i386 CPU configuration for the kernel and rebuild it, assuming your CPU does have sep actually (this is the most likely option), leaving userland as it is. - Fix up XENOMAI_CONF_OPT to add --disable-x86-sep and rebuild userland, leaving the kernel config as it is. To speed up things, I would suggest to try out these tests with no BR2 in the picture for now, using a plain configure+build sequence for Xenomai. -- Philippe.