#linuxcnc-devel Logs

Feb 18 2023

12:21 AM linuxcnc-build2: Build [#444](http://buildbot2.highlab.com/buildbot/#builders/14/builds/444) of `10-rip.debian-10-buster-rtpreempt-amd64` 8completed with warnings.
12:29 AM linuxcnc-build2: Build [#444](http://buildbot2.highlab.com/buildbot/#builders/7/builds/444) of `10-rip.debian-11-bullseye-rtpreempt-amd64` 8completed with warnings.
12:40 AM linuxcnc-build2: Build [#447](http://buildbot2.highlab.com/buildbot/#builders/11/builds/447) of `10-rip.debian-12-bookworm-rtpreempt-amd64` 8completed with warnings.
03:28 AM pere: andypugh_: how far are from the 2.9 release? any gut feeling?
03:37 AM pere: quite nice to see 34 machines used linuxcnc last week, <URL: https://qa.debian.org/popcon.php?package=linuxcnc > :)
07:02 AM -!- #linuxcnc-devel mode set to +v by ChanServ
11:55 AM linuxcnc-build2: Build [#449](http://buildbot2.highlab.com/buildbot/#builders/11/builds/449) of `10-rip.debian-12-bookworm-rtpreempt-amd64` 4failed.
01:06 PM linuxcnc-build2: Build [#446](http://buildbot2.highlab.com/buildbot/#builders/7/builds/446) of `10-rip.debian-11-bullseye-rtpreempt-amd64` 4failed.
01:08 PM linuxcnc-build2: Build [#446](http://buildbot2.highlab.com/buildbot/#builders/14/builds/446) of `10-rip.debian-10-buster-rtpreempt-amd64` 4failed.
01:20 PM linuxcnc-build2: Build [#447](http://buildbot2.highlab.com/buildbot/#builders/14/builds/447) of `10-rip.debian-10-buster-rtpreempt-amd64` 8completed with warnings.
01:21 PM linuxcnc-build2: Build [#447](http://buildbot2.highlab.com/buildbot/#builders/7/builds/447) of `10-rip.debian-11-bullseye-rtpreempt-amd64` 8completed with warnings.
01:38 PM rmu: So I continue puzzling on with smart serial, this time it involves float parameters
01:39 PM linuxcnc-build2: Build [#450](http://buildbot2.highlab.com/buildbot/#builders/11/builds/450) of `10-rip.debian-12-bookworm-rtpreempt-amd64` 8completed with warnings.
01:40 PM rmu: float value "10.0" is read by linuxcnc as "5.398241e-315". that is outside of float32. representation of 10.0 is 0x41200000. 5.398241e-315 (as double) is 0x0000000041200000. strange.
01:40 PM rmu: really strange thing is: "0.2" is read back correctly.
01:41 PM rmu: representation of 0.2 is 0x3e4ccccd
01:43 PM rmu: also, if it try to set parameter to 0.2, it works, but setting to 10 results in parameter not being set at all
01:48 PM rmu: can somebody tell me whats going on, feeling kinda stupid
02:13 PM rmu: https://paste.debian.net/1271233 output of hostmot2 debug info with descriptors of parameters.. datalength = 0x20 should be ok for 32 bit float?
02:38 PM pcw--home: Probably the driver promotes the floats to 64 bit floats
02:39 PM pcw--home: not sure if the crazy scaling numbers are causing an issue
02:41 PM pcw--home: Might help to look at how the driver deals with floats/scaling
02:43 PM rmu: what bugs me is that it seems to depend on the content of the float. scaling number doesn
02:43 PM rmu: t make a difference.
02:44 PM rmu: and it seems i don't get HM2_DBG messages from sserial.c
02:48 PM pcw--home: Not sure about that but you like need to trace the number through the code
02:50 PM pcw--home: We don't use any floats except for scaling (I think STMBL may)
02:50 PM rmu: STMBL uses floats in process data but not in parameters AFAIK
02:52 PM pcw--home: might be a driver bug since you may be the first to use a float parameter
02:58 PM rmu: pretty sure that is buggy. this "shift" business here https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/drivers/mesa-hostmot2/sserial.c#L1541 looks very suspicious... :-)
03:03 PM pcw--home: shifting a float? that would be bad...
03:07 PM pcw--home: Does the driver do something different with process data? (since that seems to work)
03:14 PM rmu: trying to trace it currently
03:17 PM rmu: https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/drivers/mesa-hostmot2/sserial.c#L2052 https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/drivers/mesa-hostmot2/sserial.c#L1825 this seem to be the places that convert different kind of floats for process data
03:17 PM rmu: not sure if this is used for parameters, i suspect it is not
03:20 PM rmu: ok, pretty sure the bug is in driver and not another messup in serial comms ;-)
03:21 PM rmu: https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/drivers/mesa-hostmot2/sserial.c#L388 i think this and probably other places also need conversion of float types.
03:22 PM rmu: thats just copying datalength/8 bytes into a double
03:34 PM linuxcnc-build2: Build [#452](http://buildbot2.highlab.com/buildbot/#builders/11/builds/452) of `10-rip.debian-12-bookworm-rtpreempt-amd64` 4failed.
04:31 PM linuxcnc-build2: Build [#449](http://buildbot2.highlab.com/buildbot/#builders/7/builds/449) of `10-rip.debian-11-bullseye-rtpreempt-amd64` 4failed.
04:35 PM linuxcnc-build2: Build [#449](http://buildbot2.highlab.com/buildbot/#builders/14/builds/449) of `10-rip.debian-10-buster-rtpreempt-amd64` 4failed.
04:47 PM linuxcnc-build2: Build [#450](http://buildbot2.highlab.com/buildbot/#builders/14/builds/450) of `10-rip.debian-10-buster-rtpreempt-amd64` 8completed with warnings.
04:47 PM linuxcnc-build2: Build [#450](http://buildbot2.highlab.com/buildbot/#builders/7/builds/450) of `10-rip.debian-11-bullseye-rtpreempt-amd64` 8completed with warnings.
04:49 PM linuxcnc-build2: Build [#453](http://buildbot2.highlab.com/buildbot/#builders/11/builds/453) of `10-rip.debian-12-bookworm-rtpreempt-amd64` 8completed with warnings.
06:29 PM linuxcnc-build2: Build [#455](http://buildbot2.highlab.com/buildbot/#builders/11/builds/455) of `10-rip.debian-12-bookworm-rtpreempt-amd64` 4failed.
07:27 PM linuxcnc-build2: Build [#452](http://buildbot2.highlab.com/buildbot/#builders/7/builds/452) of `10-rip.debian-11-bullseye-rtpreempt-amd64` 4failed.
07:29 PM linuxcnc-build2: Build [#452](http://buildbot2.highlab.com/buildbot/#builders/14/builds/452) of `10-rip.debian-10-buster-rtpreempt-amd64` 4failed.