From mboxrd@z Thu Jan 1 00:00:00 1970 From: "George Pontis" References: <04a701ce864f$91047590$b30d60b0$@z9.com> <51F182E4.70907@xenomai.org> In-Reply-To: <51F182E4.70907@xenomai.org> Date: Fri, 26 Jul 2013 09:14:26 -0700 Message-ID: <00d101ce8a1b$32cfeac0$986fc040$@z9.com> MIME-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Content-Language: en-us Subject: Re: [Xenomai] cannot open switchtest device ? List-Id: Discussions about the Xenomai project List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: 'Gilles Chanteperdrix' Cc: xenomai@xenomai.org > -----Original Message----- > From: Gilles Chanteperdrix [mailto:gilles.chanteperdrix@xenomai.org] > Sent: Thursday, July 25, 2013 12:56 PM > To: George Pontis > Cc: xenomai@xenomai.org > Subject: Re: [Xenomai] cannot open switchtest device ? >=20 > On 07/21/2013 10:19 PM, George Pontis wrote: >=20 > > I find a problem trying to run xeno-test when it goes to open the = switchtest > > device. Before this post I searched the web and found a thread = started by > > "raespi". > > We have a similar situation in which we are both using buildroot, = but I'm > > not sure that the underlying cause is the same. I am building a = kernel for > > 3.8.13 using > > the patches from git, and also the current git xenomai. ( Current as = of a > > month ago ). The target is an custom board based on the iMX53. > > > > To narrow the source of the problem, I modified switchtest.c to = report on > > any case where the open() returns a negative number. I also made a > temporary > > patch > > to not prepend the name with "/dev/". I might add that I'm using a = new RTDM > > driver named fpga_rtdm that can be opened and access successfully. > > Here's what happens: >=20 >=20 > This may look like a stupid suggestion, but that is the only = explanation > I see: you are probably using glibc's open instead of xenomai's open. > This may happen for instance if you are compiling xenomai libraries > statically and switchtest Makefile does not use the "wrap-link.sh" = script. > -- > = Gilles. Not stupid, that was a useful suggestion. Both are dynamically linked, = and I'm actually using uClibc for this embedded system. I tried to open the switchtest device from my application and it was = successful. One difference was that my application was opening these with a call to = rt_dev_open(), while the switchtest.c file used open(). Also, when building switchtest = the open() is resolved with open64() due to selecting large file support. A quick mod of the switchtest code to use rt_dev_open() instead of = open() would not build without more effort. George