AV Linux MX Edition is here!

[fehlix]

It's about the md5sum's found within the antiX folder of the ISO or on the created LiveUSB.
It's not about those checksums of the ISO itself:

Code: Select all

AVL-MXE-2020.11.23-xfce4-amd64.iso.md5
AVL-MXE-2020.11.23-xfce4-amd64.iso.sha512

The ISO-checksum are ok.
Within the antiX folder we have 3 md5sums:

Code: Select all

ls -l 
total 3583776
-rw-r--r-- 1 root root    7662155 Nov 24 07:35 initrd.gz
-rw-r--r-- 1 root root         44 Nov 24 07:35 initrd.gz.md5
-rw-r--r-- 1 root root 3656863744 Nov 24 07:42 linuxfs
-rw-r--r-- 1 root root         42 Nov 24 07:42 linuxfs.md5
-rw-r--r-- 1 root root    5258016 Nov 24 07:35 vmlinuz
-rw-r--r-- 1 root root         42 Nov 24 07:35 vmlinuz.md5

and this one linuxfs.md5 does not match:

Code: Select all

md5sum -c linuxfs.md5
linuxfs: FAILED
md5sum: WARNING: 1 computed checksum did NOT match

md5sum linuxfs
7475ac389c3ead1c7e5a8b228025b0d6  linuxfs
cat  linuxfs.md5
cdaf6aace23e4f2e176001727acb52e3  linuxfs

You can check it also when booted from the LiveMedia either LiveISO or LiveUSB.
by selecting F4 checkmd5 boot option:

AVL-MXE-64bit-F4_option_checkmd5.png

and it shows this error warning:

AVL-MXE-64bit-xfce_ISO-check-md5sum.png

The checkmd5 option is also available on UEFI booting in the text-menu or add as kernel option parameter.
On either GUI LUM or CLI-LUM you get a warning when creating a LiveUSB in full featured mode:
GUI-LUM:

LUM-gui.png

and the warning

LUM-gui-md5sum-warning.png

CLI-LUM:

cli-lum.png

and the warning

cli-lum-md5sum-warning.png

The check made by LUM is mainly to make sure that the USB is created correctly
under the assumption the original md5sum on the ISO match to the files on the ISO.
Of course if you creat a DD/dumpt LiveUSB, the md5sum-check is not made by LUM.
You can manually also check the antiX/*.md5 files,
by just mounting the iso with

Code: Select all

isomount isofile-name

and run md5sum -c on the md5-files within the antiX folder of the mounted ISO.
If one of those other ISO do match and they have been generated on another hardware, suggest to recreate the ISO where md5sum of linuxfs does not match.
If you cannot find a way of recreating ISO's, as suggested, I might prepare a iso-repack-script, which fixes the wrong md5sum's.

[rqdr]

Hello everyone!

I'm really enjoying the new AV Linux, thank you so much for the work!

But, I've found an issue. I'm using the x64 version. When opening a .h2song file with LMMS (installed via stable repo) the program itself closes immediately. I don't understand why.

It'd be really helpful if someone could tell me how to find the problem.

Thank you!

[AVLinux]

Hello everyone!

I'm really enjoying the new AV Linux, thank you so much for the work!

But, I've found an issue. I'm using the x64 version. When opening a .h2song file with LMMS (installed via stable repo) the program itself closes immediately. I don't understand why.

It'd be really helpful if someone could tell me how to find the problem.

Thank you!

Hi!

I am pretty well versed in Ardour and Mixbus but have to admit to not using LMMS very much at all... Is LMMS supposed to import .h2song files? That file extension is for Hydrogen Drum machine not LMMS.. Does the h2song file open in Hydrogen?

[AVLinux]

@fehlix

My apologies if I sound dense.. I am merely running MX Snapshot as any normal user would, I don't really know any other way of creating the ISO's? I'd like to determine what the flaw is in my build environment, I thought the CPU governor thing had fixed the issue because the obvious mismatch problems had gone away.. Is there data or Snapshot diagnostic terminal output I can provide to determine where the break in the chain is?

[fehlix]

@fehlix

My apologies if I sound dense.. I am merely running MX Snapshot as any normal user would, I don't really know any other way of creating the ISO's? I'd like to determine what the flaw is in my build environment, I thought the CPU governor thing had fixed the issue because the obvious mismatch problems had gone away.. Is there data or Snapshot diagnostic terminal output I can provide to determine where the break in the chain is?

The linuxfs.md5 checksum is created during mx-snapshot run, just after the linuxfs shquashfile has been generated and before the iso-file is build. There is one "sync" command executed before the md5sum is reading the linuxfs file to create the linuxfs.md5. The code executed looks all ok to me. So we do not know what happened and why the created linuxfs.md5 does not match to the linuxfs file. But as you mentioned you did already had some checksum missmatch of the ISO.md5 sums. I honestly have no clue. One idea was to add some more sync-command and potentially some sleep's between those checksum creations but that would only be a speculation. You can quickly check those md5-sums as mentioned by mounting the iso-file.
BTW, when you had those ISO-md5sum mismatch, was it always only the md5sum which did not match or also the sha256sum. If so it would be an indication, of either adding some more syncs - or something with the md5sum tool itself. If the sha256 always matched, but only the iso-md5sum sometimes not, the indication would be the sequence: the first run is the md5sum and takes some time, the next run is the sha256sum. So if sha256 alwayss was ok, we simply could add another md5sum run after the sha256 again.
But again only weird speculation.

[AVLinux]

Hi fehlix and thank you for your patience,

I really didn't want to come here and be a pain in the neck when you've all been so helpful, I regret that I'm taking up so much time.

OK so something about my system... It's a Threadripper 24 Core/48thread.. When MX-Snapshot ties into building the squashfs it goes full tilt and reports 48 cores and takes about 7 minutes to completely build the ISO, if you are telling me there are sync processes etc running is it possible they are being executed too quickly or some of these processes are overlapping somehow? Is there any way to tell MX Snapshot to use a limited number of cores to slow it down? Are any MX Devs using a machine with similar specs? As to your question about MD5 and SHA512, yes previously when I was running the performance CPU governor I had a few instances where the one checksum would match and the other wouldn't I can't remember now which way the mismatch was though

It is curious that the 32bit ISO spun off an old 32bit Dell Core Duo Inspiron does not have the issues we are seeing here, it takes about 59 minutes to mint an ISO, it certainly points to the fact that speed may be a relevant factor..

[fehlix]

It's probably not the speed itself, but I remember having seen in multiple spots where developer put more than one sync, and I was always wondering why. In theory this should not happen. And to add "our" mx-tools shall always work on such super-hardware regardless of the speed. So in a sense you might help us to make the tool working on such a speedy system. I could tweak mx-snapshot a bit, and if you run again and check whether this would make a difference ...

[Buck Fankers]

But, I've found an issue. I'm using the x64 version. When opening a .h2song file with LMMS (installed via stable repo) the program itself closes immediately. I don't understand why.

Generally in Linux, when some program don't start or closes right away, you want to run in a terminal. That way, you will usually be able to see some error msg, which can then help in troubleshooting.

In your specific case, I don't know that app, I'm not using AV Linux, so someone else will need to tell you, what the command is.
Once you find out which command to run in a terminal, copy/paste output here for further help.

[AVLinux]

fehlix

I will test anything you're willing to throw my way, I want to fix this ASAP! In any case I think it would be nice to be able to put some brakes on the squashfs creation, fast is great but fast and superheated CPU is not so great..

[fehlix]

fehlix

I will test anything you're willing to throw my way, I want to fix this ASAP! In any case I think it would be nice to be able to put some brakes on the squashfs creation, fast is great but fast and superheated CPU is not so great..

Can you try whether the attached makes any difference.