Jump to content


  • Posts

  • Joined

  • Last visited

Profile Information

  • Location
  • Interests

Recent Profile Visitors

5443 profile views
  1. I am just discovering as yourself on this subject (and as I said above, perfectly happy with my armbian config which does not require /dev/spi* nodes) Do you have module spidev loaded (lsmod | grep spi). If not try "modprobe spidev". As far as I know, this is this module that should create device nodes to give userland access to spi devices. Kernel drivers don't need it. It must be enabled by armbian-config for the bus you need for your device. (The dtb overlay is then added to /boot/armbianEnv.txt). Anyway, I found complaints on RPI forum for the lack of /dev/spi interfaces and this may come for dtb for specific devices.
  2. As stated in the tutorial, you should assign an address to the bridge only. The bridge forward everything on data link layer and the interface of the bridge then form/extract IP packet. Otherwise you will have a hard time to manage kernel routing tables...
  3. I've just installed a RPI dvb hat (sony cxd2880 spi device) on a bpim2plus with kernel 5.10.60-sunxi #21.08.2 SMP Tue Sep 14 16:28:44 UTC 2021 armv7l. /dev/spi is not populated but the cxd2880 driver can obviously find spi0 (thanks to a dtb found on armbian forum) because I can stream DVB-T ?!? I suppose the driver don't need spi devices. udev problem on debian bullseye ? (I am afraid to say I don't understand anything on dtb). Kernel seems happy anyway : ls -l /sys/bus/spi/devices/ total 0 lrwxrwxrwx 1 root root 0 Feb 22 12:32 spi0.0 -> ../../../devices/platform/soc/1c68000.spi/spi_master/spi0/spi0.0
  4. After struggling many days with alsa, vlc and usb audio devices, I inevitably was confronted with "RPI usb problem". My feeling so far is that there are some bugs in the linux USB2 kernel driver, but those bugs are making havoc the "RPI firmware", that is to say the underlying OS witch manage the GPU and all devices. The consequence is that any power or driver (video driver, usb driver) problem can turn the RPI in a state where it become unusable (not because of Linux but the firmware itself) until a reboot and electrical reset ! In such case, the fs on a disk plugged in the USB3 bus is also corrupted. (force fsck is greatly encouraged). So, it is quite difficult to reproduce and diagnose linux or devices problems !
  5. Does-it start if you unplug/replug the RJ45 connector (as on RPI...)
  6. find . -type d | awk -F/ '{printf("%"4*NF"s|-- %s\n", " ",$NF)}' (find in current dir all files of type directory and process them threw awk with "/" field separator to print a space with 4 times number of fields width, then a bar and two dash and the last field. awk is your friend
  7. "called by snmpcd (snmp formating)" As you and Werner say : the best solutions are often the simplest. The problem is to find what is really the simplest and to make the difference between simple and basic in a use case. (This is just a bit of general thinking (and nostalgy), as I have no idea about what you are working on). The **simple** network management protocol is surely basic but I never found it was simple and it has been a long time since I saw someone care about it. One more time, this is just general thinking, but I feel we use today a lot of complicated but no so powerful technologies in IT. Linux is a unix derivative, UNix has been designed by people who wanted a simple but powerful system after having worked on MULTix. I try to keep that in mind. But it is not always easy and cannot be achieve if we want our computers to also toast bread and do the laundry.
  8. In shell, pipes and awk are your bests friends : perhaps something like this : $ mkfifo /tmp/timefifo $ awk -W interactive -F: '{printf("cmd=%s s=%f\n", $1, $2*60+$3)}' <>/tmp/timefifo > /tmp/stats & $ /usr/bin/time -o /tmp/timefifo -f "%C:%E" ls -l $ /usr/bin/time -o /tmp/timefifo -f "%C:%E" ls $ /usr/bin/time -o /tmp/timefifo -f "%C:%E" who $ cat /tmp/stats cmd=ls -l s=0.700000 cmd=ls s=0.070000 cmd=who s=0.000000 $ killall awk N.B. : <> open the pipe in read/wite mode to prevent it to be closed when "time" send EOF.
  9. "time" is "built in" into the shell. In order to use the full version described by "man", you should first install the full version in /usr/bin/time : # apt-get install time. And call it with the full path. Also, when man fail to give doc because a package is missing, just google "man cmd".
  10. "Are there other things I can do aside from these I have done?" Lowering DRAM refresh rate was the most efficient powersaving trick. But I don't know if and how it can be done with modern kernel ? https://github.com/armbian/build/blob/master/packages/bsp/h3consumption
  11. Hum ! Is address 6e:e4:fe:d4:82:6d a valid mac-address ? It seems to me that the first bit of first octet shouln't be 1 because it is reserved to multicast addresses. https://communities.vmware.com/t5/VMware-Fusion-Discussions/The-first-byte-of-the-MAC-address-cannot-bit-odd/td-p/1218107 Sorry, my mistake, 6e is odd !!! Sorry, it is even so first bit 0. I go to bed ...
  12. There are a number of issues with the numerous versions of nfs, locking, architecture and export path on server with nfsV4 ... and I wouldn't trust error messages. For example "protocol not supported" can be emitted because of the share path semantic : From : https://wiki.archlinux.org/title/NFS/Troubleshooting#mount.nfs:_Protocol_not_supported mount.nfs: Protocol not supported This error occurs when you include the export root in the path of the NFS source. For example: # mount SERVER:/srv/nfs4/media /mnt mount.nfs4: Protocol not supported Use the relative path instead: # mount SERVER:/media /mnt I have had some troubles in the past with export path and nfsV4 and it works at present although my configuration is not supposed to be correct ?!?
  13. "f you want control and your apps are available, try OpenBSD, very hands on, secure *nix ..." I use a Gentoo system to concentrate and serve all network and security functions. This one would be a good candidate for OpenBSD. But I never used an OpenBSD system. Not really a problem (I have used dozen of *nix systems and BSD was the first one) but I do not know on which SBC, light platform or appliance it could be easily installed (and reliable) ?
  14. systemd-shim is provided by debian. I do not know what may fail without it. (It is said to translate systemd helpers and anyway I have no use for automatic service start and stop on a personally administered system). "dhcp, with the router reserving ip addr for your client mac addr" Well, I am not sure there exists a "good solution". Anyway, I prefer editing a central repository than having (bugged) auto-configuration systems. Auto configuration is mandatory in some cases and a pain in the neck as soon as you have a problem and are not sure where it lies. That is why I removed systemd one more time (I kept it 3 months which is a record-breaking event for me). In this precise case, it is because the network interface needs sometimes a reset at boot (which stall the boot process), that the problem is known to happen in my case with D-link switches - and not handled of course ... I simply cannot patch systemd to resolve all mishappenings by editing meta-configuration files everywhere and praying it will convince systemd to resolve the problems disregarded by mainstream maintainers.
  15. For info, I managed to get rid of systemd on my raspbian desktop. I had to first install dbus-x11 because the dependencies would force half of packages removal. Then installation of sysvinit-core do the trick and trigger install of libnih-dbus1, systemd-shim, libcgmanager0, libnih1, cgmanager and "pinned out" systemd. You probably need to install : ntp and ntpdate and of-course I use a totally static network configuration. (No avahi, no networmanager, no dhcpcd ...) Also, I needed to modify /etc/X11/Xwrapper.config because of DRM driver needing root uid of X server. Of course, I do that sort of things only after reaching a stable configuration that I can maintain without support and with a good system backup !
  • Create New...