On Fri, 25 Feb 2022 00:54:41 -0800 (PST)
antoine besnier <
antoine.bes...@gmail.com> wrote:
> Follow up: it's been a long and hard process to get 5.10.2 working.
> I could not manage to get 5.15 working. At one point, I could connect to my
> pc1 at 192.168.10.1, but it could never get an IP frommy router/dhcp.
> I downgraded to 5.10.1, and it took maybe two full tries (full wipe of the
> drive, use the
> latest librecmc-ramips-mt7621-gb-pc1-squashfs-sysupgrade_*.bi
> <
https://github.com/gnubee-git/gnubee-git.github.io/tree/master/debian>n,
> install jessie, the modules, update to stretch, burn the new firmware,
> etc...
Those are libreCMC images with old kernels.
>
> But it finally works and I will leave it at 5.10.1
>
> By the way, I have an issue with the network interfaces.
>
> */etc/network/interfaces:*
> auto lo
> iface lo inet loopback
>
> auto ethblack
> iface ethblack inet static
> address 192.168.10.1
> netmask 255.255.255.0
>
> auto ethblue
> iface ethblue inet dhcp
>
>
> BUT, *ip a* gives me info about the following interfaces:
> lo (expected)
> eth0 (expeceted)
> *rename3*@eth0 (surprising)
> ethblue@eth0 (expected)
>
> This info is confirmed by *lshw*:
> *-network:0
> description: Ethernet interface
> physical id: 3
> logical name: eth0
> *-network:1 DISABLED
> description: Ethernet interface
> physical id: 4
> logical name: *rename3 *
> *-network:2
> description: Ethernet interface
> physical id: 5
> logical name: ethblue
>
> I noticed this with both 5.10.1 and 5.15.
>
> The mismatch between ethblack and rename3 prevents me from restarting the
> network service, or using ifup/ifdown on ethblack, and is probably the
> cause of my connectivity problems during the previous attempts.
>
> I tried to dig through the usual suspects udev rules, but could not find
> any thing, and therefore, I don't understand what is happening.
>
>
> OK, strike all that... I upgraded to bullseye, and after a reboot, rename3
> is now back to ethblack... However, I maintain that this issue exists!
> Maybe a few reboots make it go away? And maybe that's also what happened 3
> years ago when I first tried to install Neil's firmware.
> (
https://groups.google.com/g/gnubee/c/YVM08lfWUUc/m/rr5UpEFIAAAJ)
I'm still using 5.10, I tried building my own 5.15 kernel config with
some different filesystem drivers, took a while to get it to fit into
the flash partition size, but then it wouldn't boot, I didn't have time
yet to troubleshoot the issue (I'm sure it's something with my Kconfig,
I just need to experiment with different options and diff it against
Neil's).
Did you try the cross-compile (chroot) method? You can compile on
whichever machine you have is the fastest, you don't need to do it on
the Gnubee, doing so would indeed be very slow. I did that for the old
kernel versions, including the currently working 5.10, so it definitely
works.
I did do some modifications to Neil's scripts for my own needs,
which are currently in my fork (Neil didn't want to merge them, which I
fully understand). Mostly to conform to shell coding guidelines, with
at least one bug fix I could find.
All I have in my /etc/network/interfaces(.d) is this:
root@gnubee:~# cat /etc/network/interfaces.d/01-lo
auto lo
iface lo inet loopback
root@gnubee:~# cat /etc/network/interfaces.d/02-ethblack
auto ethblack
iface ethblack inet dhcp
pre-up /etc/network/scripts.d/no_switch
> Le mercredi 23 février 2022 à 09:50:22 UTC+1, antoine besnier a écrit :
>
> > Thanks!
> >
> > Le mercredi 23 février 2022 à 04:37:16 UTC+1,
ne...@brown.name a écrit :
> >
> >> I've fixed the README. It is "GNUBEE-ROOT" with a hyphen, not an
> >> underscore.
> >> Thanks.
> >>
> >> On Wednesday, February 23, 2022 at 1:18:11 AM UTC+11 antoine besnier
> >> wrote:
> >>
> >>> Additional useless info: on my gnubee PC1, it took 4h30 to run gbmake
> >>> firmware gbpc1-5.15 (with 2GB of swap enabled)
> >>>
> >>> Le mardi 22 février 2022 à 10:43:46 UTC+1, antoine besnier a écrit :
> >>>
> >>>> Hi Neil,
> >>>>
> >>>> Stupid question on your kernel build:
> >>>>
> >>>> The Readme on your site (
http://neil.brown.name/gnubee/README) states
> >>>> that your image looks for a file system labeled GNUBEE*_*ROOT (with an
> >>>> underscore), while the readme on github (
> >>>>
https://github.com/neilbrown/gnubee-tools) mentions GNUBEE*-*ROOT