Linux Users

Hi 🙂

If the C examples that come with the driver work, then at least we know that it's installed correctly and it also correctly controls the hardware. That's good.

I see that your program is using the pthread library from the backtrace. You may be running into a limitation of the driver: it's not multi-thread safe [1] and calls to the driver can only be made from a single thread. If all of your DAQmx Base calls happen in the same thread, then you may want to read the KB about linking against the pthread library [2].

[1] DAQmx Base 3.3 Linux Readme
http://ftp.ni.com/support/softlib/multifunction_daq/nidaqmxbase/3.3/Linux/readme.txt

[2] Why Does my Application Segmentation Fault When Using the pthread Library?
http://digital.ni.com/public.nsf/allkb/EB202A78FB1D083186257117005CEF21?OpenDocument

thanks for reply!!

i will check the links you gave me...

meanwhile, i tried to write a simpler c++ program, and it runs just fine, so the problem must be before the fork....

i made some more tests with the code i posted yesterday, and i noticed some strange things:

- the DAQmxBaseCreateTask takes a long time to return (more or less 40 seconds)

- it looks for /etc/nss_mdns.conf file, that was missing

Jan 12 09:19:46 vds-dev ext_io[20063]: Starting ExtIO module... Jan 12 09:19:46 vds-dev ext_io[20063]: Loop count not specified. Using default... Jan 12 09:19:46 vds-dev ext_io[20063]: Init I/O card started... Jan 12 09:21:05 vds-dev ext_io[20063]: mdns: Couldn't open nss_mdns configuration file /etc/nss_mdns.conf, using default. Jan 12 09:21:25 vds-dev ext_io[20063]: Task created... Jan 12 09:21:25 vds-dev ext_io[20063]: Channel blocks created... Jan 12 09:21:25 vds-dev ext_io[20063]: Init I/O card completed successfully and task started!! Jan 12 09:21:25 vds-dev ext_io[20063]: ExtIO object started successfully...

so i created a link

[alberto@vds-dev user]$ ls -ls /etc/nss_* 0 lrwxrwxrwx  1 root root 15 Jan 12 09:38 /etc/nss_mdns.conf -> nss_nimdns.conf 0 lrwxrwxrwx  1 root root 36 Nov 23 10:09 /etc/nss_nimdns.conf -> /etc/natinst/.nicore/nss_nimdns.conf

but it changed nothing (it still takes 40 seconds to init)

Jan 12 09:39:01 vds-dev ext_io[21086]: Starting ExtIO module... Jan 12 09:39:01 vds-dev ext_io[21086]: Loop count not specified. Using default... Jan 12 09:39:01 vds-dev ext_io[21086]: Init I/O card started... Jan 12 09:40:41 vds-dev ext_io[21086]: Task created... Jan 12 09:40:41 vds-dev ext_io[21086]: Channel blocks created... Jan 12 09:40:41 vds-dev ext_io[21086]: Init I/O card completed successfully and task started!! Jan 12 09:40:41 vds-dev ext_io[21086]: ExtIO object started successfully...

i have also tried to run the program with the debugger and stop it during the wait time, and it's the output of backtrace:

Program received signal SIGINT, Interrupt. 0xb754e431 in pthread_cond_wait@@GLIBC_2.3.2 () from /lib/tls/libpthread.so.0 (gdb) backtrace #0  0xb754e431 in pthread_cond_wait@@GLIBC_2.3.2 () from /lib/tls/libpthread.so.0 #1  0xb7dc2fbb in __pthread_cond_wait (cond=0x8084744, mutex=0x8084774) at forward.c:138 #2  0xb7617e12 in EventLoggingLogUserBuffer () from /usr/local/lib/liblvrtdark.so.8.2 #3  0xb7617d02 in EventLoggingLogUserBuffer () from /usr/local/lib/liblvrtdark.so.8.2 #4  0xb7636fec in ThFreeSysUIThread () from /usr/local/lib/liblvrtdark.so.8.2 #5  0xb763600b in ThFreeSysUIThread () from /usr/local/lib/liblvrtdark.so.8.2 #6  0xb76e513e in WaitLVClientReady () from /usr/local/lib/liblvrtdark.so.8.2 #7  0xb76df6fa in ReserveEPDS () from /usr/local/lib/liblvrtdark.so.8.2 #8  0xb7ce33e7 in LV_DAQmxBaseCreateTask (taskName=0x8048fff "", taskHandle=0x8085a58) at nidaqmxbaselv.c:1770 #9  0xb7f40aca in DAQmxBaseCreateTask () from /usr/local/lib/libnidaqmxbase.so.3 #10 0x08048c20 in TExtIO::InitIOCard (this=0x8085a50) at ext_io.cpp:97 #11 0x08048b24 in TExtIO (this=0x8085a50, ext_io_parms=0xbfeab9ec) at ext_io.cpp:59 #12 0x08048a5e in main (argc=1, argv=0xbfeaba94) at ext_io.cpp:40

i have not tried to do the same with the sample code, but from the tests i've done, it takes more or less the same time for DAQmxBaseCreateTask to return.

thanks in advance for any further help you may provide me!!

EDIT (2 posts in a row are already more than enough )

i run examples/dio/readDigMultiplePorts sample program with the debugger too, to see if there are any differences on the function calls while the program is calling DAQmxBaseCreateTask, but it looks the same:

Program received signal SIGINT, Interrupt. 0xb7661431 in pthread_cond_wait@@GLIBC_2.3.2 () from /lib/tls/libpthread.so.0 (gdb) backtrace #0  0xb7661431 in pthread_cond_wait@@GLIBC_2.3.2 () from /lib/tls/libpthread.so.0 #1  0xb7ed5fbb in __pthread_cond_wait (cond=0x8084744, mutex=0x8084774) at forward.c:138 #2  0xb7729e12 in EventLoggingLogUserBuffer () from /usr/local/lib/liblvrtdark.so.8.2 #3  0xb7729d02 in EventLoggingLogUserBuffer () from /usr/local/lib/liblvrtdark.so.8.2 #4  0xb7748fec in ThFreeSysUIThread () from /usr/local/lib/liblvrtdark.so.8.2 #5  0xb774800b in ThFreeSysUIThread () from /usr/local/lib/liblvrtdark.so.8.2 #6  0xb77f713e in WaitLVClientReady () from /usr/local/lib/liblvrtdark.so.8.2 #7  0xb77f16fa in ReserveEPDS () from /usr/local/lib/liblvrtdark.so.8.2 #8  0xb7df63e7 in LV_DAQmxBaseCreateTask (taskName=0x8049139 "", taskHandle=0xbfa96d48) at nidaqmxbaselv.c:1770 #9  0xb7f3baca in DAQmxBaseCreateTask () from /usr/local/lib/libnidaqmxbase.so.3 #10 0x08048b66 in main (argc=1, argv=0xbfa97624) at readDigMultiplePorts.c:51

and, of course, it takes more or less the same time to complete.

Messaggio modificato da Yawgmoth

So even the straight-C example programs that install with DAQmx Base stall for 40 seconds when they go to create the task? That's very unusual and shouldn't be happening. Do you also have LabVIEW? I'm curious if the LabVIEW examples are equally slow.

JoeFriedchicken ha scritto: So even the straight-C example programs that install with DAQmx Base stall for 40 seconds when they go to create the task? That's very unusual and shouldn't be happening.

yes, it hangs for more or less 40 seconds on DAQmxBaseCreateTask too....

JoeFriedchicken ha scritto: Do you also have LabVIEW? I'm curious if the LabVIEW examples are equally slow.

not yet... i'll set it up and check its behaviour...

thanks again for reply!!

While running the driver's manual and automated test suites over the last few releases, I have never encountered a 40 second delay in DAQmxBaseCreateTask. Are there any atypical aspects of your development system? Any news along the LabVIEW testing?

the problem is that there are quite a lot of atypical aspects on my development system....

and i had no time to work on that the last days 'cos i have to fix some bugs on our software.....

i will let you know as soon as i can restart testing... thanks a lot for help!!

i tried to look for the evaluation version of LabVIEW, but it seems it's not available for linux....

btw, i had to rebuild the kernel, and after that i rebuilt the ni driver too.... now the demo programs starts in more or less 10 seconds....

it seems to point to the kernel, which on one side it good 'cos it helps to realize where the problem is, but on the oether side i'm not sure i'm skilled enough to debug that at kernel level....

do you know of any kernel options that may impact on the driver??

I respond to this only:

"i tried to look for the evaluation version of LabVIEW, but it seems it's not available for linux.... "

My university has a site license which came with both the Linux and the Windows versions.

My university does not distribute the Linux version because it lacks any protection against license violation (which MAY be why N.I. does not distribute it either; this is a plausible, yet to be proven, conjecture).

By unique exception, because of my age and records, I have been allowed to give the Linux version a short try, and I swiftly returned it after finding out that beyond the protection it also lacks the most important tool required to build VI's.

Just open http://www.ni.com/labview/how_to_buy.htm  , click Data Acquisition Functions and Wizards  and enjoy not having wasted the time to download, had you found it.

Naturally, nor does it (LabVIEW Linux) support USB devices ( ni.com/info code rddqld )..

So, if you would have found a Linux demo version, you would just have added one or more or while my choice would be    (shocked)

Jacques Goldberg, Prof.