#linuxcnc-devel | Logs for 2015-04-16

[00:23:08] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hy-vfd 2d74f91 06linuxcnc Merge remote-tracking branch 'bebro-huanyang-vfd/linuxcnc' into 2.7 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2d74f91
[00:23:08] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hy-vfd 5d59312 06linuxcnc 10(21 files in 4 dirs) move huanyang vfd driver to the proper place for linuxcnc * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=5d59312
[00:23:08] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hy-vfd cca0886 06linuxcnc 10src/hal/user_comps/huanyang-vfd/Makefile 10src/hal/user_comps/huanyang-vfd/hy_modbus.c 10src/hal/user_comps/huanyang-vfd/hy_modbus.h 10src/hal/user_comps/huanyang-vfd/hy_vfd.c fix perms on huanyang vfd driver * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=cca0886
[00:23:11] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hy-vfd eaeb72e 06linuxcnc 10src/hal/user_comps/huanyang-vfd/hy_modbus.c 10src/hal/user_comps/huanyang-vfd/hy_modbus.h hy_vfd: switch from DOS to Unix line endings * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=eaeb72e
[00:23:15] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hy-vfd 2f9b4ad 06linuxcnc 10src/Makefile 03src/hal/user_comps/huanyang-vfd/Submakefile 10src/hal/user_comps/huanyang-vfd/hy_modbus.h hy_vfd: add a Submakefile and some missing function declarations * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2f9b4ad
[00:23:20] <KGB-linuxcnc> 03Sebastian Kuzminsky 05hy-vfd 4b70e79 06linuxcnc 10src/hal/user_comps/huanyang-vfd/hy_modbus.c 10src/hal/user_comps/huanyang-vfd/hy_vfd.c hy_vfd: remove some dead code * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=4b70e79
[00:23:56] <seb_kuzminsky> the original author reimplemented (or maybe forked) libmodbus
[09:29:18] <pcw_home> skunkworks: do you remember testing that GUI hang a month or two ago?, I cant seem to find it in the logs
[10:04:53] <jepler> Hm, I was unaware of SO_BUSY_POLL: http://man7.org/linux/man-pages/man7/socket.7.html
[10:05:02] <jepler> Sets the approximate time in microseconds to busy poll on a
[10:05:05] <jepler> blocking receive when there is no data.
[10:06:54] <jepler> (via http://www.breakage.org/2013/08/30/oh-did-you-expect-the-cpu/)
[10:09:15] <pcw_home> ahh thats new since 3.11, sounds like hm2_eth should use that
[10:10:54] <jepler> the sysctls are available in the rt kernel I have handy -- 3.14-0.bpo.2-rt-amd64
[10:11:37] <jepler> but eyeballing hm2_7i80.0.read.time with busy_read=80 busy_poll=80 doesn't look any better than the default of 0
[10:13:25] <jepler> there may be other ideas on this guy's blog worth investigating. I arrived there trying to google an understanding of what nohz is
[10:14:54] <pcw_home> I think there was some funny business with the polling rate limitation in hm2-eth (it would hang with no delay)
[10:16:57] <pcw_home> even if the overall time is not changed it might improve the latency statistics (I need to hook-up hal-histogram and try)
[12:51:51] <skunkworks> pcw_home, The only gui hang I remember is when I was playing with revearse run...
[13:31:25] <PCW> there was something with the axis reload button I think
[13:32:07] <skunkworks> oh - yah.. it would hang for a bit. after a estop was it?
[13:32:55] <PCW> yeah someone on the forum has a possibly related problem
[13:37:17] <seb_kuzminsky> i've had axis go out to lunch for a handful of seconds sometimes, then come back and act like nothing happened
[13:37:26] <seb_kuzminsky> it doesn't happen often, and i dont know how to trigger it yet
[13:39:22] <PCW> There was a reliable way to trigger a hang (about 30 seconds) it but I cant remember, its somewhere in the linuxcnc-devel logs earlier this year
[13:45:13] <skunkworks> Bring up axis sim -> home -> run splash screen program -> hit estop -> then either hit the refresh or try to open another program.
[15:50:28] <skunkworks> zlog
[15:52:13] <seb_kuzminsky> cd