[Rtai] Problems with COMEDI buffered input/output commands from a realtime process under RTAI/LXRT

[LAMY Xavier 215881 Thésard]

Hello, 

  I would like to use COMEDI buffered input/output commands from a 
realtime process under RTAI/LXRT. This seems possible thanks to the 
porting done by "The Comedi Players" : 
 <http://cvs.gna.org/cvsweb/vulcano/addons/comedi/?cvsroot=rtai> http://cvs.gna.org/cvsweb/vulcano/addons/comedi/?cvsroot=rtai 

  I already succeed using buffered input/output commands with classic 
COMEDI, and also doing simple input/output through its RTAI/LXRT 
porting. But I still fail doing buffered input/output commands through 
the RTAI/LXRT porting. 

  I have tried the example provided in a old RTAI version : 
file testa.c 
in  <http://cvs.gna.org/cvsweb/stromboli/comedi_lxrt/?cvsroot=rtai> http://cvs.gna.org/cvsweb/stromboli/comedi_lxrt/?cvsroot=rtai 

  I've just adapted the include statements and add some debugging 
instructions to track down what is going wrong. 

  Here is my changes to the original example : 

************* 
$ diff -u0 testa.c.org testa.c 
-- testa.c.org 2008-04-28 13:36:00.000000000 +0200 
+++ testa.c     2008-04-28 13:49:18.000000000 +0200 
@@ -28,3 +28 @@ 
-#include <rtai_lxrt_user.h> 
-#include <rtai_lxrt.h> 
-#include <rtai_comedi_lxrt.h> 
+#include <rtai_comedi.h> 
@@ -105,0 +104,12 @@ 
+ 
+        printf("data:       @%d\n", 
+               cmd->data); 
+ 
+        printf("data_len:    %d\n", 
+               cmd->data_len); 
+ 
+        printf("chanlist:   @%d\n", 
+               cmd->chanlist); 
+ 
+        printf("chanlist_len:%d\n", 
+               cmd->chanlist_len); 
@@ -209,0 +220,4 @@ 
+       for (n = 0; n < BUFSZ/2; n++) { 
+               buf[n] = n; 
+       } 
+ 
@@ -212 +226 @@ 
-//     rt_make_hard_real_time(); 
+       rt_make_hard_real_time(); 

****************** 

  And here is the output of the new testa program : 

****************** 
$ sudo ./testa 

 OVERALL INFO: 
   Version code : 0x00074c 
   Board name   : pci-6259 
   Driver name  : ni_pcimio 
   Number of subdevices : 14 

 Subdevice : 0 
 Type : 1 (analog input) 
 Number of channels : 32 
 Maxdata : 65535 
 Number of ranges : 7 

 Subdevice : 1 
 Type : 2 (analog output) 
 Number of channels : 4 
 Maxdata : 65535 
 Number of ranges : 3 

 Subdevice : 2 
 Type : 5 (digital I/O) 
 Number of channels : 32 
 Maxdata : 1 
 Number of ranges : 1 

 Asynchronous test (with comedi commands). 

  command before testing : 
start:      now      0 
scan_begin: timer    100000 
convert:    timer    1 
scan_end:   count    1 
stop:       count    10 
data:       @-1209081772 
data_len:    40 
chanlist:   @-1209081776 
chanlist_len:1 

  First test returned 3 (invalid argument) 
start:      now      0 
scan_begin: timer    100000 
convert:    timer    800 
scan_end:   count    1 
stop:       count    10 
data:       @-826179500 
data_len:    40 
chanlist:   @-826179504 
chanlist_len:1 

  Second test returned 0 (success) 

#: 0, #SMPV: 0, MASK: 5, SEMCNT: 0 (TRUE COUNT) 
#: 1, #SMPV: 1, MASK: 5, SEMCNT: 2 (TRUE COUNT) 
#: 2, #SMPV: 2, MASK: 5, SEMCNT: 2 (TRUE COUNT) 
#: 3, #SMPV: 3, MASK: 5, SEMCNT: 3 (TRUE COUNT) 
#: 4, #SMPV: 4, MASK: 5, SEMCNT: 3 (TRUE COUNT) 
#: 5, #SMPV: 5, MASK: 5, SEMCNT: 3 (TRUE COUNT) 
#: 6, #SMPV: 6, MASK: 5, SEMCNT: 3 (TRUE COUNT) 
#: 7, #SMPV: 7, MASK: 7, SEMCNT: 3 (TRUE COUNT) 
#: 8, #SMPV: 8, MASK: 7, SEMCNT: 2 (TRUE COUNT) 
#: 9, #SMPV: 9, MASK: 7, SEMCNT: 1 (TRUE COUNT) 

***************** 

  What we can see is that : 
1) the buffer is not updated at all after executing the COMEDI command 
(field #SMPV should be updated with the samples values) 
2) the comedi_command_test change the data buffer and chanlist address 
(this is not expected!) 

  Here is a message posted 2 years ago by someone having similar 
problems : 

 <http://marc.info/?l=comedi&m=116291668731840&w=2> http://marc.info/?l=comedi&m=116291668731840&w=2 

  I am using : comedi-0.7.76 
               comedilib-0.8.1 
               linux kernel 2.6.17.9 (patched with RTAI 3.5) 

  Any advices on this problem will be very helpfull, thanks in advance 
for your help! 

 -- 

Xavier LAMY

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.rtai.org/pipermail/rtai/attachments/20080514/af5a0624/attachment.htm