With lua5.5 and gnome-pkg-tools backported, I could build wireplumber with these patches. The debs are available from Index of /~praveen/mobian
But these patches did not fix the echo in calls. We will have to wait for more changes from Franklin.
With lua5.5 and gnome-pkg-tools backported, I could build wireplumber with these patches. The debs are available from Index of /~praveen/mobian
But these patches did not fix the echo in calls. We will have to wait for more changes from Franklin.
I have also noticed full disk image with jumpdrive is not very reliable - you dd it back, the system does not boot. So taking a tarball of /home/pureos is more reliable. Reflash with librem5-flash-image and then restore the /home/pureos from the tarball.
Both mobian image and pureos dawn full disk image failed to boot. But luckily I had backed up /home/pureos separately.
That doesn’t sound plausible, you should verify whether you haven’t messed anything up. I have juggled many images around this way, sometimes across phones, and it always worked perfectly.
Sounds strange. I know that dd and ddrescue do not do a verification, but you can do that yourself, right??? i.e. Perhaps you should verify the md5sum of the device vs the image???
[Aside: I recently had an issue with an imaged disk (device to device). With both devices plugged in and exact copies of each other (but unmounted) lsblk resulted in a segfault and system freeze requiring a reboot because it apparently could not handle having two devices with exactly the same uuid and it must have had some assumption of uniqueness.]
I guess the starting point is what version of jumpdrive you have.
Also, assuming that you are using LUKS, does “does not boot” mean that you don’t even get to the LUKS unlock passphrase prompt? Or something else?
While I haven’t messed around with using dd to restore on the Librem 5 (thus breaking one of the cardinal rules of backup), I have had Raspberry Pi devices shit themselves many times and had to dd restore a backup (full disk image) and that has always worked for me.
I would have tried to mount the full disk image on the host computer and then you can pull individual trees out of that. That would also shed light on whether the full disk image was any good at all. Do you still have the full disk image around so that you could look more closely at it?
I believe that I have mounted a full disk image backup of the Librem 5 on a host computer but that is of course not a comprehensive test.
![]()
It is a pity that there isn’t a ‘flag’ for that on dd. Or, failing that, some kind of verify option on Gnome Disks.
I usually use sha256sum when moving multi-gigabyte files around.
“Verify” though could be done by a simple re-read and block by block compare, rather than comparing crypto hashes. Comparing the data is perhaps most definitive.
Yes, this is a known issue - although I don’t see segfaults or lockups. I just see that the second copy fails to mount. For this reason, if I want to look at a dd backup of a computer (e.g. selectively restore some files), I would do that on another computer and then copy the files across to the original computer.
Yes. Or have such a thing incorporated into ddrescue. I’ve heard of some other tools to address this, but I’ve never used them (and, so, don’t yet trust them). Specifically bmaptool and dcfldd (enhanced fork of dd for forensics) have built-in integrity checks. [Edit: bmaptool was created by the Tizen project and was mainly wanting to deal with the issue that their images had a lot of zero-blocks … but they also incorporated verification. GitHub - yoctoproject/bmaptool: BMAP Tools · GitHub ].