#linuxcnc-devel Logs

Jan 14 2020

12:12 AM memfrob: night all!
10:18 AM skunkworks: burklefoo: yes - but I don't know if it works the same in the dhcpdp.conf file in the pi...
10:19 AM skunkworks: what info am I looking for - nothing seemed to change when I re-ran the ethtool -c
10:38 AM burklefoo: you ran: sudo ethtool -C eth0 rx-usecs 0 ?
10:38 AM burklefoo: I would just do it from the command line to make sure
10:50 AM jepler: burklefoo: with the 7c81, can I power the fpga via the pi, or is that not a recommended configuration? It sounds like the 5v buses are just hooked together?
10:51 AM jepler: I have that official usb-c adapter for the pi4
10:52 AM burklefoo: It should work either way (all 5Vs are tied in the plane)
10:53 AM jepler: main thing to check would be if the pi thinks it's undervolted?
10:54 AM burklefoo: Should be OK to power from TB1 as long as the flat cable is short
10:55 AM jepler: yes I'll be using that cable you included
10:55 AM burklefoo: I think I calculated the resistance at about 10 mOhms
10:58 AM jepler: then I just need a straight through "network" cable to the sserial daughtercard 7i73?
10:59 AM jepler: P5 or P6?
11:11 AM jepler: burklefoo: https://gist.github.com/3552d5c3d98112e2083deec2ae053942 sserial worked for me with whatever rev is on the rpi image. I'll confirm what rev that was and then maybe we can figure out what is different.
11:15 AM jepler: hm I can't seem to determine a specific git ref. I'll build tip of master and see if it matters
11:30 AM skunkworks: I have been running smartserial too with the pi image.. (stmbl drive)
11:30 AM burklefoo: The problem might be x86 only
11:34 AM jepler: OK, I don't quite have my x86 system put together to do linuxcnc but I'll do that next. (I do have a second ethernet but no rt kernel at the moment)
11:35 AM jepler: burklefoo: if you install your "newmesa" does it go back to working properly?
11:36 AM jepler: I did introduce a small delta from "newmesa" which was all to do with float vs double as type of C variables..
11:36 AM burklefoo: I was running master with my changes without issue, but I can try adding my changes to 2.8
11:38 AM burklefoo: Since all the errors seem to be timeouts my WAG is that its time or time calculation related
11:38 AM jepler: https://www.bedbathandbeyond.com/store/product/titan-70-inch-x-72-inch-waterproof-fabric-shower-curtain-liner/3312775?strategy=pdp_cav
11:38 AM jepler: haha oops
11:46 AM jepler: OK on pi with tip of master, v2.9.0-pre0-960-ga1ddf4071
11:48 AM skunkworks: fails?
11:55 AM burklefoo: current (1 min ago) master fails for me (note that it does not always fail)
11:56 AM jepler: hhmmmm okay, let me try 10x in a row. It worked for me on x86 as well
11:57 AM burklefoo: errors are always at startup, never later
11:57 AM jepler: OK I got it on the 2nd and following runs
11:59 AM jepler: https://gist.github.com/a6306b2ca8b023c3135343555300c155
12:00 PM burklefoo: that's the type of error I see
12:00 PM burklefoo: quite random
12:00 PM jepler: are you testing with an ethernet device as well?
12:01 PM burklefoo: yes
12:05 PM burklefoo: That's all I have at home now after my 12 YO HP DC7800 died...
12:06 PM jepler: nothing with pci(e) ports in it? that's scary
12:06 PM burklefoo: 2.8 works, Ill try 2.8 with my patches next
12:06 PM burklefoo: all I have a a Elite G800 (no slots)
12:08 PM jepler: I also did 10 starts with 2.8 and it was fine
12:09 PM jepler: I think testing 2.8 + newmesa will be a good test, let me know your result. I'll use "git bisect" over here since I am comfortable with it.
12:12 PM burklefoo: Ack! I need to munge the Makefile
12:12 PM burklefoo: I guess this is why the smart guys use git...
12:18 PM burklefoo: OK 2.8 with newmesa patches is OK (10 starts ok)
12:18 PM jepler: thank you
12:19 PM jepler: that seems to agree with what my bisect testing is converging on, that it's something I added on top of newmesa that broke it
12:20 PM jepler: 67ea53e21be502991d91035f9a5473e4e6549ead is the first bad commit
12:21 PM jepler: > Fix places where the "float" type is used
12:38 PM jepler: burklefoo: is this structure required to mirror the memory layout of something in the device? I think I've determined that my change of two of the structure members to "double" is what broke it, but I don't know why. https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/drivers/mesa-hostmot2/sserial.h#L87
12:42 PM jepler: > hm2_sserial_data_t data;
12:42 PM jepler: > if ((addr = hm2_sserial_get_bytes(hm2, chan, &data, addr, 14)) < 0) {
12:42 PM jepler: yes, yes it is
12:51 PM jepler: OK I pushed what should be a fix for that problem.
12:52 PM jepler: right to master, bypassed the pull request process :slaps own hand:
12:53 PM burklefoo: Ill try in a bit
12:54 PM burklefoo: Thanks!
12:58 PM jepler: sure thing, I hope I successfully tracked it down. also I now have an ethernet board set up for future testing so that's good too.
12:59 PM jepler: afk for some lunch, then probably on to circuitpython stuff. sometime I need to talk to seb_kuzminsky about how we want to expose these pi packages on linuxcnc.org...
01:13 PM linuxcnc-build: build #6344 of 1300.rip-precise-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1300.rip-precise-i386/builds/6344 blamelist: Jeff Epler <jepler@gmail.com>
01:13 PM jepler: argh it me
01:15 PM linuxcnc-build: build #6347 of 1306.rip-precise-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1306.rip-precise-amd64/builds/6347 blamelist: Jeff Epler <jepler@gmail.com>
01:21 PM linuxcnc-build: build #4505 of 1400.rip-wheezy-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1400.rip-wheezy-i386/builds/4505 blamelist: Jeff Epler <jepler@gmail.com>
01:21 PM linuxcnc-build: build #4506 of 1403.rip-wheezy-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1403.rip-wheezy-amd64/builds/4506 blamelist: Jeff Epler <jepler@gmail.com>
01:23 PM linuxcnc-build: build #1362 of 1630.rip-stretch-rtpreempt-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1630.rip-stretch-rtpreempt-amd64/builds/1362 blamelist: Jeff Epler <jepler@gmail.com>
01:23 PM linuxcnc-build: build #2972 of 1510.rip-jessie-rtpreempt-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1510.rip-jessie-rtpreempt-i386/builds/2972 blamelist: Jeff Epler <jepler@gmail.com>
01:23 PM linuxcnc-build: build #2972 of 1500.rip-jessie-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1500.rip-jessie-i386/builds/2972 blamelist: Jeff Epler <jepler@gmail.com>
01:23 PM linuxcnc-build: build #2974 of 1520.rip-jessie-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1520.rip-jessie-amd64/builds/2974 blamelist: Jeff Epler <jepler@gmail.com>
01:23 PM linuxcnc-build: build #2974 of 1530.rip-jessie-rtpreempt-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1530.rip-jessie-rtpreempt-amd64/builds/2974 blamelist: Jeff Epler <jepler@gmail.com>
01:23 PM linuxcnc-build: build #4024 of 1402.rip-wheezy-rtpreempt-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1402.rip-wheezy-rtpreempt-i386/builds/4024 blamelist: Jeff Epler <jepler@gmail.com>
01:23 PM linuxcnc-build: build #4709 of 1404.rip-wheezy-rtpreempt-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1404.rip-wheezy-rtpreempt-amd64/builds/4709 blamelist: Jeff Epler <jepler@gmail.com>
01:24 PM linuxcnc-build: build #1362 of 1610.rip-stretch-rtpreempt-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1610.rip-stretch-rtpreempt-i386/builds/1362 blamelist: Jeff Epler <jepler@gmail.com>
01:26 PM linuxcnc-build: build #6361 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/6361 blamelist: Jeff Epler <jepler@gmail.com>
05:23 PM cerna: jepler: Have you tried to change the CLOCK_MONOTONIC to CLOCK_MONOTONIC_RAW to circumvent the clock update issue?
08:42 PM jepler: cerna: You cannot "sleep" until a CLOCK_MONOTONIC_RAW timestamp with clock_nanosleep, so I don't see a feasible way to try it
08:43 PM jepler: https://stackoverflow.com/questions/39061851/is-clock-nanosleep-affected-by-adjtime-and-ntp