Started Labbook 2015.
November 18, 2014
- The Linux-server mremote will be discontinued on December 15. Copied all my ZSB files from mremote to u033089.uwp.science.uva.nl (Jose). Other choices are
- Quax u030334
- Diertens u031061
- Jaap u022471
- u040291
- u040287 (could not login)
June 19, 2014
- With the DeLock converter I could play chess, both in nb-ros and u019961. Had to install my code on /tmp on u019961, because I only have 1Gb disk-quota.
- Casper could not access the daemon, because the umirtxsimulator in /opt/prac/robotics/bin/umisimulator still looks at the name_file at /home/rtxipc.
- Tried to make the simulator in /opt/prac/robotics/assistance/robotics, but fails in missing gtk/gtk.h (gtk-2.0 ?)
- Copied /usr/include/gtk-2.0 from nb-ros. Copied assistance/robotics to /tmp. Still many compilation errors (also on nb-ros).
- Compilation fails on GStaticMutex in gdkpixbuf.h, which according to the glib developers is deprecated.
- The simulator in /opt/prac/robotics/bin is from June 2006. In June 2009 I was able to compile the umirtxsimulator with gcc-2.95.3.
- There are three other versions of the simulator: /opt/prac/robotics/assistance/hgrepos/robotics/robotics/, /home/avisser1/onderwijs/ZSB/assistance/robotics/src/ and /home/avisser1/onderwijs/ZSB/assistance/repository/robotics/. Yet, the Makefiles all have the same size (in my home from June 2009 and in /opt/prac from June 2011). The src in my home was depended on the gcc-2.95.3 which is now available in /opt/prac/cs_ai/packages/. Updated the different dependencies files. ias complains about missing gtk/gtk.h. Removed ias.c from the makefile. Had to remake libumi, libarm and libxmlparser. Link fails on missing gtk-x11-2.0. Without this library, the make fails on missing main (and alter_slider). The library is present, yet only as dynamic library (not as libgtk-x11-2.0.a). Yet, I could not find a package which the static library. Now looking if I could make a dynamic library from the umirtx code. Succeeded with libumi.
- Made the simulator, only complains about missing ias.c functions.
- Tried to include a minimal set of gtk-definitions, but failed on definition of GInitiallyUnowned in gtkobject.h. Made the object an empty struct. Couldnot find guint32, including glib.h resulted in many new dependencies.
June 18, 2014
- Connected the pl2303 to u019961 (the other usb-rs232 connector has to carry too much weight). The rtxd in onderwijs/ZSB/assistance/repository/umi-drivers/ reports that it could restart.
- Called rtxsh in onderwijs/ZSB/assistance/repository/umi-drivers/ and executed arm; init_comms 1 2. No response, nothing new in the log.
- Debugged downloadlib in armraw.c. First switched off the localhost branch. With hton-conversion of the port-number, rtxsh makes a connection to the socket, but does not receive an acknowledgement. Without the conversion I get the error downloadlib via internet socket: Connection refused. I will check armraw.c at nb-ros.
- nb-ros uses the code which I just switched off. Turned it on again and checked it again with the debugger. Get nice warning: downloadlib has problems connecting to socket: Connection refused. Checked the permissions of /opt/prac/robotics/rtxipc/ports/rtx/rtx-socket, but I am owner and there are write permissions for the group. Try the same sequence on nb-ros.
- The shell on nb-ros is not the same, because downloadlib is called without explicit command from the prompt.
- Calling init_comms 1 2 gave a communication error. In the logfile I see CMD_TOGGLE_OFF vives wrong response. With command soak init_soak the robot initializes and starts moving its wrist. Yet, the command soak off gives Communication not initialised.
- Tried chess. Initializing goes ok, yet I get a lot of warnings on position of joints (something wrong with reading positions). Yet, the warnings are overrulled by Can't execute an empty list warnings. In the log only two RTX onetransactions are visible: FIRST and INIT.
- The No Piece in PICK_UP goes wrong in pathplan/move_piece.c. Put protection around the printfs to print only once.
- Executed chess again (without switching off the UmiRtx). Now I don't see the FIRST and INIT commands in the log. Yet, I receive several warnings: enable PWM output: not enough bytes in command, general_status: illegal command, position error with zed: Checksum error, position error with elb: not enough bytes in command, error with arm_go: No response.
- The PWM error is in umi_init on a call to arm_stop, after a succesfull arm_init_comms and arm_restart. The onetransaction is printed from the daemon if the debug_level is >1. The CMD_STOP is quite transparant followed by a rtx_command(). Via the virtual interface this is redirected to tty_arm_command, which redirects to rtx_stop. Seems that rtx_stop is called with a wrong parameter (mode < 0 || >= 4). arm_stop was called with FREE_OFF=3 (other choices DEAD_STOP, RAMP_STOP, FREE_STOP). Added a debug request to arm_stop. The command arm_stop is executed with the right parameter, yet this time the robot steers its gripper in its band. Several rtx_do are printed (check what the parameters mean).
- Increased the debug level in umi_init from 1 to 3, which should print the bits.
- Had to get the power from the umi, because of the self-collision. Get many warnings in log about another ip expected. Those are related to many CMD_WRITE during the initialization movements. After that the general_status looks like using the right ip, but the error from previous request seems not be reset.
June 16, 2014
- Trying to get the chess_demo working on nb-ros. pl2303 is nicely attached to ttyUSB0, the rtxd in assistance/repository nicely restarts. Yet, chess_demo in assistance/repository fails with the error message: "The IP was restarted. restart", which is an error message from armlib: ARM_RESTARTED.
- This error message could be generated at two places: rtxcmds.c in daemon and rtxlib (the file in daemon is a symbolic link). The error comes from rtx_raw. rtx_init_comms should give some debug info (switched on in putstatus?). The corresponding flag is RESP_IP_RESTART (0x0f).
- Added some debug statements to raw.c, but Makefile is still sun3. daemon doesn't link to raw.c, only rtxcmds and libhost.a. Also chess does not load rtxlib, only libarm.a
- Now the chess program works (could be the check of the cables), yet I got complains about Illegal command and not enough bytes in the list (after I spooled by errors about Can't execute an empty list, No piece in PICK_UP).
- The Illegal command is arm_error ARM_COMMAND_OOR or ARM_NOT_SUPPORTED, while the Not enough bytes is associated with ARM_FRAME_TIMEOUT. The first corresponds to RESP_CMD_NOT_SUP or RESP_CMD_OOR, the later corresponds to RESP_FRAME_OVERRUN.
June 3, 2014
- Could not directly reach u019961.fnwi.uva.nl from nb-ros, I had to go through acheron.fnwi.uva.nl.
- Looked at the daemon code in src/bewegen/umi-drivers/daemon. Seems to be only depend on ../hostlib/libhost.a. The daemon goes wrong after converting theio with htonl (line 743 of rtxd.c).
- Debugging failed (main was in alive.c, onetransaction was in seperate thread.
- Strange, main is in rtxd.c, not in alive.c (what has gdb loaded?).
- Added extra logline at lube 743. rtxsh isn't compiled yet, complains about GLIBC_2.14. chess gives same warning. recompiled rtxsh, no connection to rtxd (right config?).
- The rtxsh in onderwijs/ZSB/assistance/repository/umi-drivers/ gives the command line interface, yet soak init gives "Communication not initialised".
- The rtxsh in src hangs in initlevel->toplevel_init->docommand(,"init",1)
- This jumps to doinit in misc.c. This function does arm_init_comms, which doesn't come back. This function calls arm_interrupt(RTX_CMD_INIT,), which hangs in downloadlib (in addition, armraw could not be loaded by gdb, while position in file was known). Downloadlib fails on connect to lib_socket. The sun_path points to /opt/prac/robotics/rtxipc/ports/rtx-socket. Changed the rtx-socket from rtxipc group to dialout. Now shell could continue, although I get the warning: downloadlib has problems connecting to socket, Connection refused. This is the same error as October 7, 2013.
- Tried to control the robot from nb-usb, but it fails on the installing the usb-connection:
new full-speed USB device number 10 using xhci_hcd
Device not responding to set address.
unable to enumerate USB device on port 2
- With DeLock /dev/ttyUSB0 is created. rtxd started, chess_demo failed (init connections : No response). The rtxlog gave a warning about Unknown size msg, 0 instead of 32. Second try (couldn't sent acknowledgement).
May 25, 2014
- Finally, we are added to the group 'dailout' at U019961 in room G0.25.
- Pluging in the usb-serial converter gave the message usb 2-1.1.3: pl2303 converter now attached to ttyUSB0 on command dmesg | grep tty.
- Starting the daemon in src/bewegen/umi-drivers/daemon with command rtxd /dev/ttyUSB0 gives the response (in the rtxlog):
May 26 16:55:20 (RTXD) Restarting, listening to port 38838 on u019961.fnwi.uva.nl
- Tried chessdemo, but got response (in the rtxlog):
May 26 16:58:50 (7) New Connection (read/write)
May 26 16:58:50 (7) New client (read/write), user avisser1
May 26 16:59:05 (7) couldn't sent acknowledgement
- I made this version of chess on 7 Oct 2013 (same day as daemon). On that day, I linked back to June 6, 2011, for the same acknowledge error.
- Shell is from June 3, 2013. Gives error GLIBC_2.14 not found.
- The shell in linux/x86 (same date) starts up, but fails on downloadlib. This exectutable is linked to /lib64/libc.so.6, as ../../rtxsh. Difference is in libreadline (5 linux/86, 6 ./., libtermcap (only linux/86))
- Switched back to nb-ros. There the chess_demo worked, although the first move had an overshoot in the first z-move. The number of libraries of chess were less: libm, libc and linux-vdso.
- Copied chess_demo from nb-ros, but missing some higher level Makefiles. Remade chess, still same error (acknowledgent and May 26 17:46:26 (7) Unknown size msg, 0 instead of 32).
- Chess_demo is the same, difference is in libarm library or rtxd.
- Tried the version in onderwijs/ZSB/assistance/repository/umi-drivers/daemon (June 19, 2013). I see the daemon restarting, but both chess_demo programs get no response (also no error in rtxlog).
Previous Labbooks