#linuxcnc-devel Logs
Jun 05 2020
#linuxcnc-devel Calendar
11:49 AM jepler: hmm still problems upgrading from one of my custom kernels to another
11:53 AM jepler: dpkg: error processing archive ../linux-image-4.19.71-rt24-v7l+_4.19.71-rt24-v7l+-8_armhf.deb (--install):
11:53 AM jepler: unable to make backup link of './boot/System.map-4.19.71-rt24-v7l+' before installing new version: Operation not permitted
11:53 AM jepler: dpkg-deb: error: paste subprocess was killed by signal (Broken pipe)
02:23 PM cerna: skunkworks: That's called 'hoarding'. Not that many *important* things cannot be recreated on the spot and there are actually cloud storage services offering free storage in the vicinity (not exactly 30 GB, but close). So no reason to antagonize webhosting provider.
04:58 PM andypugh: jepler: I guess you found the same inelegant workaround as me for that? Perhaps we just need to document it?
05:12 PM jepler: andypugh: I'm not sure what workaround you mean. My package is intended to be installing and removing the "diverts" just like upstream's package, (but not sharing their exact scripts since the filenames differ)
05:12 PM jepler: This particular file is even supposed to be diverted, but it's not (it says the existing diversion has been left in place)
05:12 PM jepler: I wonder if I had a bad/interrupted update that left diverts AND real files, or something
05:13 PM andypugh: Try sudo apt-get install -o Dpkg::Options::="--force-overwrite" linux-image-4.19.71-rt24-v7l+ ?
05:14 PM andypugh: (or —force-all)
05:34 PM jepler: oh, no, that shouldn't be necessary
05:35 PM andypugh: I know it _shoiuldn’t_. Does it work though?
05:37 PM andypugh: I will try asking on the Debian-live list. It’s off-topic for there, but might find someone who knows how to get round this.
05:41 PM jepler: So here's the thing -- raspbian kernel packages have these lists of "diversions" which are the tricks they use to make their own kernel upgrades work. It should be possible to copy.
05:41 PM jepler: it's not a debian-live issue, it's a "raspbian made the wrong decision to have /boot be a FAT filesystem" issue crossed with a "raspbian kernel packaging isn't like debian kernel packaging OR like linux's in tree debian packaging" crossed with .. I dunno, I'm so tired.
05:42 PM andypugh: I know it isn’t debian live related, but the guys there know a lot about packaging. (Like LinuxCNC know a lot about 3D printers, apparently)
05:44 PM andypugh: Aye, I am with you on the tired. What I want to do this weekend is make the wobble-broach I designed. But I am actually going to spend it trying some more to get to the bottom of the RTAI problem. And deciding at what point we have to give up if 2.8 is ever to be released.
05:47 PM jepler: this time it worked, no force necessary
05:47 PM jepler: shrugging person emoji
05:49 PM jepler: this was wrong in my earliest package, fwiw, though I thought any one that got in the linuxcnc.org package server was "right"
05:49 PM jepler: what changed since my earlier failure attempt was removal of old diverts left because of ??? UNKNOWN ???
05:49 PM jepler: not that users would be capable of that, ugh
05:53 PM jepler: https://nw-digital-radio.groups.io/g/main/topic/udrc_dkms_fails_to_install/31902863?p=,,,20,0,0,0::recentpostdate%2Fsticky,,,20,2,0,31902863 a similar problem
05:53 PM jepler: Leaving 'diversion of /boot/overlays/udrc.dtbo to /usr/share/udrcdkmshack/udrc.dtbo by udrcdkmshack'
05:53 PM jepler: followed by trying to overwrite '/boot/overlays/draws.dtbo', which is also in package raspberrypi-kernel 1.20190517-1
05:53 PM jepler: so it's whatever gets you in the circumstance of 'leaving diversion ...'
06:02 PM jepler: yeah by installing one of my broken work in progress packages I can get back into this state .. let's see if I can clear it again
06:04 PM jepler: yes but I'm not sure these are "right" instructions
06:05 PM jepler: dpkg-divert --list | grep rpikernelhack | while read _ _ a _ _ _ _; do dpkg-divert --package rpikernelhack --no-rename --remove $a; done
06:05 PM andypugh: Are you saying that it is all your fault for making a bad package? :-)
06:05 PM jepler: rm -r /usr/share/rpikernelhack/
06:05 PM jepler: I didn't deliberately disseminate a broken one, but y'know these things happen
06:05 PM jepler: probably
06:06 PM andypugh: Off-topic. Did you see this? https://github.com/carlini/printf-tac-toe
06:06 PM andypugh: Code as art.
06:06 PM jepler: yes
06:07 PM andypugh: Impressive and wrong on so many levels
06:07 PM andypugh: But interesting that printf has so many extra features to misuse.