#linuxcnc-devel Logs

Apr 13 2018

#linuxcnc-devel Calendar

01:01 AM memfrob: Debian 9.4 (stretch) vanilla straight from debian.org: http://dpaste.com/3PMXJV8
01:03 AM memfrob: xserver-xorg-dev xorg-dev x11proto-xinerama-dev libxinerama-dev all installed
03:44 AM hazzy-dev1 is now known as hazzy-dev
08:57 AM jepler: memfrob: if you manually installed a list of packages, you probably missed one. Here's the guidance our docs give on checking for problems (though it's not 100% the advice I would give, I don't have time to write more about it at the moment) http://linuxcnc.org/docs/master/html/code/building-linuxcnc.html#Satisfying-Build-Dependencies
09:57 AM JT-Shop: when you get time tell me and I'll update the docs
12:52 PM memfrob: I didn't miss one, the file is there but it says "unusable." In a technical sense, what does an unusable header mean?
12:53 PM memfrob: I just read the header itself, and looks like a perfectly normal C header.
12:53 PM memfrob: (/usr/include/X11/extensions/Xinerama.h)
12:57 PM memfrob: cd debian ; ./configure -r (or uspace, doesn't matter) and then running $ dpkg-checkbuilddeps returns: dpkg-checkbuilddeps: error: cannot read debian/control: No such file or directory
12:58 PM memfrob: Ah, ./configure <string> and then $ cd ..
01:12 PM memfrob: This is what's in the include file: https://cgit.freedesktop.org/xorg/lib/libXinerama/tree/include/X11/extensions/Xinerama.h -- It hasn't been touched in 10 years except for a whitespace issue. I think it's usable..
01:38 PM cradek: memfrob: what's your exact error message?
01:49 PM memfrob: http://dpaste.com/3PMXJV8
01:50 PM memfrob: I did a search through the entire debian package database for (xinerama) and all packages were installed.
01:50 PM memfrob: the parenthesis were meant to be quotes.
01:50 PM cradek: dpkg -S Xinerama.h
01:51 PM cradek: libxinerama-dev:amd64: /usr/include/X11/extensions/Xinerama.h
01:51 PM cradek: you have libxinerama-dev installed for your architecture?
01:52 PM memfrob: libxinerama-dev:amd64: /usr/include/X11/extensions/Xinerama.h
01:52 PM memfrob: yes.
01:53 PM cradek: look in config.log and see what's failing?
01:54 PM memfrob: http://dpaste.com/0ZPN19V
01:54 PM cradek: probably shows the actual error in there
01:55 PM cradek: what on earth is all that
01:55 PM memfrob: from config.log :)
01:56 PM cradek: #define RTAPI_KERNEL_VERSION "3.16.52-RTAI
01:56 PM memfrob: Maybe converting RTAI's shell scripts to bash broke xinerama :P
01:56 PM cradek: wherever it got that is wrong
01:56 PM cradek: has nothing to do with xinerama
01:57 PM cradek: check your kernel's include/linux/version.h
01:57 PM memfrob: RTAI's shell scripts were a big mess and really ugly so I re-wrote them. Original: https://github.com/NTULINUX/RTAI/blob/RTAI5_vanilla/base/scripts/rtai-load.in
01:58 PM memfrob: Mine: https://github.com/NTULINUX/RTAI/blob/master/base/scripts/rtai-load.in
01:58 PM cradek: ok this is going to be something peculiar to your rtai or kernel, you're going to have to trace it from here
01:59 PM memfrob: I'm also getting these messages: http://dpaste.com/1SK3PHX
02:02 PM memfrob: where is RTAPI_KERNEL_VERSION coming from, specifically?
02:04 PM memfrob: I guess LinuxCNC can't handle RTAI's scripts being in bash..
02:05 PM memfrob: but if they're left the way they are, they're extremely ugly
02:21 PM jepler: you've got at least two problems, and like cradek says you get to debug them.
02:22 PM jepler: The first, which we can see the consequence of in the definition of RTAPI_KERNEL_VERSION shown in the context of the compile error, is that the value of KERN_VERS is not found correctly. This occurs around line 555 of configure.ac
02:23 PM jepler: first guess, the error just below also has 'UTS' in the identifier, and the grep gets both
02:23 PM jepler: one alternative, make the grep more specific -- e.g., grep -w UTS_RELEASE
02:25 PM jepler: the second, you have a bug in your rewritten rtai-config.in which is probably making it show the usage information when it's invoked with "$RTS --module-ext 2>/dev/null" but at the same time it fails to exit with nonzero status
02:25 PM jepler: that's around line 567
02:26 PM jepler: I will be happy to review any linuxcnc configure.ac patches you submit related to this, please mention @jepler on the PRs
02:28 PM jepler: memfrob: ^^ in case any of that troubleshooting helps you
03:05 PM memfrob: Thank you jepler, I'm going to re-trace all my steps on gentoo and see what's going on. I never had any of these problems on gentoo.
03:05 PM memfrob: I'll be back later, lots of reboots and such. Take care!
04:14 PM memfrob: I had a bug in one of my new bash scripts, that was my bad. Now all I need to fix is that quote in RTAPI_KERNEL_VERSION
04:15 PM memfrob: as opposed to being cute in configure, why not just call $(uname -r) ?
04:16 PM seb_kuzminsky: you might not be running the kernel you're building for?
04:17 PM memfrob: uname -r says 3.16.52-RTAI as well as my grub selection, plus RTAI found it.
04:17 PM memfrob: RTAI calls /lib/modules/$(uname -r)
04:18 PM memfrob: /usr/src/linux-3.16.52/include/generated -> #define UTS_RELEASE "3.16.52-RTAI"
04:18 PM memfrob: ^utsrelease.h
04:20 PM memfrob: plus the RTAI testsuite works and it's the only RTAI kernel I have installed :P
04:24 PM memfrob: I removed all the VERSION_FILE stuff and just changed KERNEL_VERS to $(uname -r) and it works fine
04:24 PM memfrob: Less LOC too :)
04:36 PM memfrob: I forgot what the fix is for this now.. http://dpaste.com/2RY06VY
04:36 PM memfrob: really common problem.
04:40 PM memfrob: security/limits.conf adjustment brb
04:43 PM memfrob: I had to insmod everything by hand, I forgot the proper fix..