Bullseye build script

[fredx181]

...
I would like to ask: I missed the network_tray.sh in / usr / bin and the icon on the tray. It was just a curiosity, about the peasywifi ...

I'm not sure to understand what you're asking here. EDIT: network_tray.sh is part of frisbee.

Also in the menu, I didn't see the "Repository Configuration" icon anymore ... Is the program no longer needed?

I had my doubts if it's useful for anyone, was it useful for you ?
EDIT: Info here about it (static <> dynamic) viewtopic.php?p=5508#p5508

Fred

[Duprate]

As for the connection icon, it was just a curiosity to change the icon. Everything is working perfectly.

"Repository configuration" was useful. Well, updating the system whenever you make a new "mklive-bullseye" script available, generates a smaller filesystem.squashfs. Updating the previous way, generates a larger file. For me, the way you did it looks great! Currently, I consider DebianDog number one of the puppy species!

Choosing one way or another, I prefer the "mklive-bullseye" script.

[fredx181]

Added "Repository configuration", package "repo-conf" to the custom repositories.
To be able to switch between "standard" and "static" repository, see more info here: viewtopic.php?p=5508#p5508 (Info is focused on Sid, but goes for Bullseye too at this point of time) and here: viewtopic.php?p=5510#p5510
Install with Synaptic or from terminal with apt:

Code: Select all

apt update
apt install repo-conf

Run from Menu > System > Repository configuration
EDIT 2021-01-21, made a small mistake, updated repo-conf to v1.0.1

Fred

[rcrsn51]

Hi Fred: Two things.

1. You might want to add "mtools" to the package lists in mklive-bullseye. Without it, Gparted throws a warning message on FAT32 partitions.

2. When my Bullseye Starter Kit shuts down, there is a longer-than-ususal pause, followed by a red message about some process that it had to kill. Are you seeing this too?

Bill

[Duprate]

Hi! I made a compilation of Kernel 5.11.0 (base 5.11.0-rc4, Kernel.org). All browsers, wine, windows games and other programs are working correctly. All the bugs observed in the 5.10.x series have disappeared (on some PCs, depending on the hardware). I keep testing ...

"Linux 5.11 RC4 is correcting the consequences of the Intel Haswell GT1 graphics card crashing in the last semester. These low-end Intel Haswell graphics pieces should return to work now after several kernel cycles with crashes and startup problems for the affected processors.
However, the rest in relation to Linux 5.11 RC4 is a usual set of bug fixes. Thus, the brief announcement of Linux 5.11-rc4 can be read at lore.kernel.org. In addition, stable Linux 5.11 is due out in February and features many new and improved features. "

[fredx181]

Hi Fred: Two things.

1. You might want to add "mtools" to the package lists in mklive-bullseye. Without it, Gparted throws a warning message on FAT32 partitions.

Yes, will add mtools at next update

2. When my Bullseye Starter Kit shuts down, there is a longer-than-ususal pause, followed by a red message about some process that it had to kill. Are you seeing this too?

I see the red message too:

2021-01-22-003400_666x174_scrot.png (263.16 KiB) Viewed 1495 times

Looks like it has to do with pcmanfm that is still a remaining active process, when I reboot by doing in terminal (as workaround/test);

Code: Select all

killall -9 pcmanfm && reboot

Then the red failed message doesn't appear for me, same for you ?

Fred

[rcrsn51]

Then the red failed message doesn't appear for me, same for you ?

This works for me too. So I added "killall -9 pcmanfm" to the start of the /usr/bin/wmpoweroff script and I get a faster shutdown. Is this OK to do?

[backi]

Hi Fred !
Since your last Fix...regarding "not saving" when booted with "copy2ram" Option ......Problem (as far as i can see).... solved !

[fredx181]

So I added "killall -9 pcmanfm" to the start of the /usr/bin/wmpoweroff script and I get a faster shutdown. Is this OK to do?

Yes, should be OK, however, better IMO is in '/etc/init.d/snapexit' (should be activated by default and run at shutdown) because when running poweroff or reboot from terminal or console then the pcmanfm process will be killed also:

Code: Select all

  stop)
     [ $(pidof pcmanfm) ] && killall -9 pcmanfm
	if [ -f /mnt/live/tmp/modules ]; then
	#clear
	exec /usr/local/bin/snap-ex </dev/console >/dev/console 2>&1
	fi
        ;;

@backi thanks, good that it works OK now !

Fred

[rcrsn51]

however, better IMO is in '/etc/init.d/snapexit'

Thanks. That works for me.

[fredx181]

New! Appimage with dependencies included, running a build from a Puppy OS as host should work now this way.
Info and download: https://debiandog.github.io/MakeLive/Re ... yedog.html

- Some small fixes, re-attached mklive-bullseye script at first post, see also Changes and Fixes 2021-01-23
EDIT: To apply these fixes on an existing install:

Code: Select all

apt update
apt install mtools porteusbootscripts

- Added package "deadbeef' audio player (v1.8.2) to the custom repository (taken from Antix repo).
Install with synaptic or from terminal with apt:

Code: Select all

apt update
apt install deadbeef

Fred

[Duprate]

Good afternoon Fred! I made a new filesystem with its script "mklive-bullseye-2021-01-23". Very good! On my PC, I noticed a faster boot.

I tried the "Activate module" and "deactivate module" options, with a "Clamav.sfs" module and another "Seamonkey.squashfs".
Worked perfectly. As well as in AUFS.

As you already know, I am trying the 5.11.0-rc4 kernel that came with fixes and no longer causes problems for some Intel processors (as in the 5.10.x series).

I like new things, kernel and system updated regularly. DebianDog with Overlayfs, allows me to stay that way ...

Good job, Fred!

[fredx181]

I tried the "Activate module" and "deactivate module" options, with a "Clamav.sfs" module and another "Seamonkey.squashfs".
Worked perfectly. As well as in AUFS.

Sfs-load method with overlay is not much tested yet, so good to hear !

Fred

[rcrsn51]

Hi Fred: Something to consider for the future.

Now that Porteus-boot works with a wide variety of hardware, is there any reason to keep supporting Live-boot? Setting up persistence with Live-boot was always awkward and it takes up a lot of space in the ISO.

Bill

[fredx181]

Hi Fred: Something to consider for the future.

Now that Porteus-boot works with a wide variety of hardware, is there any reason to keep supporting Live-boot? Setting up persistence with Live-boot was always awkward and it takes up a lot of space in the ISO.

Bill

Yes, you got a point, I tend to keeping it, but probably I'll add a yes/no choice in next mklive-bullseye update:

Include initrd.img for live-boot ?

2021-01-25-131822_871x95_scrot.png (18.02 KiB) Viewed 1769 times

Fred

[dancytron]

I created the ChromeDog in the latest version of the AppImage on my HP Laptop.

All worked as expected. Install bt4stretch and it worked the first time.

Great job as usual.

[dancytron]

I've been trying to set up my Bullseye build to be my TV watching laptop distro.

Is there a trick to getting Bill's bluetooth (bt4stretch) to work with vlc (running as root with the hack)?

It works perfectly the 1st time with mpv, but with vlc I don't see the "bluetooth" choice under "audio-audio device" unless I start mpv 1st. Then when I choose it, it stops working after a maybe 10 seconds.

I started vlc from terminal, and this is what I got when I picked "bluetooth" and then played an .mp4.

Code: Select all

root@live:~# vlc
VLC media player 3.0.12 Vetinari (revision 3.0.12-1-0-gd147bb5e7e)
[000055db1a9e1c90] vlcpulse audio output error: PulseAudio server connection failure: Connection refused
[000055db1a920580] main libvlc: Running vlc with the default interface. Use 'cvlc' to use vlc without interface.
[000055db1a9b1340] main playlist: playlist is empty
[00007f2ee8004c00] gl gl: Initialized libplacebo v2.72.0 (API v72)
libva info: VA-API version 1.10.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: va_openDriver() returns -1
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/i965_drv_video.so
libva info: va_openDriver() returns -1
[00007f2ee8004c00] glconv_vaapi_x11 gl error: vaInitialize: unknown libva error
libva info: VA-API version 1.10.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: va_openDriver() returns -1
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/i965_drv_video.so
libva info: va_openDriver() returns -1
[00007f2ee8004c00] glconv_vaapi_drm gl error: vaInitialize: unknown libva error
libva info: VA-API version 1.10.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
libva info: va_openDriver() returns -1
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/i965_drv_video.so
libva info: va_openDriver() returns -1
[00007f2ee8004c00] glconv_vaapi_drm gl error: vaInitialize: unknown libva error
[00007f2ee8004c00] gl gl: Initialized libplacebo v2.72.0 (API v72)
Failed to open VDPAU backend libvdpau_va_gl.so: cannot open shared object file: No such file or directory
[00007f2ee8004c00] gl gl: Initialized libplacebo v2.72.0 (API v72)

[rcrsn51]

@dancytron: Those messages appear to be associated with video - do you also get them when playing VLC to your regular audio output?

Are you using bt4stretch v2.1? The intent is to make BT your default ALSA output device, so you shouldn't need to pick a device in VLC. You can check the device by opening alsamixer.

I tested this in one of my Bullseye setups and VLC played correctly to BT. I will test some more.

Have you tried playing an ordinary mp3? Is this situation new to Bullseye? Did it work OK in Buster?

BTW, VLC is the first app I have seen that did not migrate from Buster to Bullseye and needed a new squashfs module.

[dancytron]

Thanks for the reply.

It might be a day or 2 before I get to looking at it.

It didn't work in buster, but I just tried once. I was satisfied with mpv because I didn't want to watch OTA TV.

Thanks,

Dan

[fredx181]

How to install from deb-multimedia Bullseye repository:
(deb-multimedia has more multimedia packages (and often newer) than the standard Debian repositories)

1) Install deb-multimedia-keyring:

Code: Select all

wget http://www.deb-multimedia.org/pool/main/d/deb-multimedia-keyring/deb-multimedia-keyring_2016.8.1_all.deb
dpkg -i deb-multimedia-keyring_2016.8.1_all.deb

2) Add to /etc/apt/sources.list (edit with text-editor) this line and save the file:

Code: Select all

deb https://www.deb-multimedia.org bullseye main non-free

3) Update the package lists:

Code: Select all

apt update

That's it.

Fred

[fredx181]

*** Updated mklive-bullseye ***
(probably last update for some time, unless bugs found)

Some (not major) changes:
- Added choice to include initrd.img or not
(initrd.img is for the "live-boot" option, the porteus-boot option (initrd1.xz) will be included anyway)
See also here: viewtopic.php?p=15978#p15978
- The choice for mksquashfs "gzip' or "xz" goes now for the kernel .squashfs too.
(so, for example, when chosen xz, the e.g. k-5.10.0-1-amd64.squashfs and 01-filesystem.squashfs will both be compressed with xz)

Did a test build using first Desktop choice (Openbox minimal, included firefox-esr), skipped creating initrd.img and chose xz compression, and the resulting ISO size became only 255MB

Script re-attached at post #1, download appimages Here

Fred

[rcrsn51]

skipped creating initrd.img and chose xz compression,

I have never used the xz compression option in my builds. Is there any performance penalty when the modules get loaded and used?

[fredx181]

skipped creating initrd.img and chose xz compression,

I have never used the xz compression option in my builds. Is there any performance penalty when the modules get loaded and used?

Yes, probably booting goes a bit slower and initial RAM usage a bit higher (but may depend on your hardware, so not sure).
FYI, using lz4 compression gives the best performance (e.g. when converting 01-filesystem.squashfs to lz4) but creates much larger file.

Fred

[rcrsn51]

(probably last update for some time, unless bugs found)

When Bullseye is officially released, will the sources.lst definitions in the script need to change?

[fredx181]

(probably last update for some time, unless bugs found)

When Bullseye is officially released, will the sources.lst definitions in the script need to change?

No, can stay the same, because next stable release will be called bullseye (as already defined in sources.list).

For who wants to keep running the "testing" branch (which bullseye currently still is), change all instances of bullseye to "testing" in sources.list (I've read online that the codename for the new "testing" will be "bookworm" in the future).

Fred

[dancytron]

@dancytron: Those messages appear to be associated with video - do you also get them when playing VLC to your regular audio output?

Are you using bt4stretch v2.1? The intent is to make BT your default ALSA output device, so you shouldn't need to pick a device in VLC. You can check the device by opening alsamixer.

I tested this in one of my Bullseye setups and VLC played correctly to BT. I will test some more.

Have you tried playing an ordinary mp3? Is this situation new to Bullseye? Did it work OK in Buster?

BTW, VLC is the first app I have seen that did not migrate from Buster to Bullseye and needed a new squashfs module.

Sorry it took so long to get back to this.

It works correctly when I open vlc first and then choose something to play, but not when I try to open a file directly from pcmanfm. It show bluealsa as a choice when I open vlc 1st, but not when I try to open the file via pcmanfm.

Unless there is an easy fix, I'm okay with this.

Yes I am using bt4stretch v. 2.1.

Alsamixer shows BlueALSA as card.

[rcrsn51]

but not when I try to open a file directly from pcmanfm. It show bluealsa as a choice when I open vlc 1st, but not when I try to open the file via pcmanfm.

This works OK for me. In VLC, I open Audio > Audio Device. It has "Default" selected at the end of the list. So when I run a file from pcmanfm, it automatically plays to BT.

[Duprate]

Hello, Fred! A question: Since the DebianDog initrd is based on the Porteus initrd, why is there a need to have /lib/modules/5.xx.x the kernel modules if this folder is already inside the 02-kernel-5 module. xx.x.squashfs? I noticed that in the Porteus initrd there isn't, being much smaller. Does Debian require this?

[Clarity]

I am not answering anyone here, this post is a mere footnote.

[I have never used the xz compression option in my builds. Is there any performance penalty when the modules get loaded and used?

Yes, probably booting goes a bit slower and initial RAM usage a bit higher (but may depend on your hardware, so not sure).
FYI, using lz4 compression gives the best performance (e.g. when converting 01-filesystem.squashfs to lz4) but creates much larger file.

Fred

Agree and timed in each's booting, tested to verify. Further, there is NO difference in desktop performance as far as the performance reports show. Thus, no matter the compression in build or the actual size of the outcome, the desktop yields are always the same.

The ONLY difference, aside from a slight boot timing is:

• XZ gives a smaller compression of the system file for ISO/IMG transport over the internet and/or across your LAN or your I/O subsystem.

  1. Internet transits over the cell network constitutes metering and charges for users.

  2. LAN transit cost users nothing.

Again, this is a mere FYI

[fredx181]

Hello, Fred! A question: Since the DebianDog initrd is based on the Porteus initrd, why is there a need to have /lib/modules/5.xx.x the kernel modules if this folder is already inside the 02-kernel-5 module. xx.x.squashfs? I noticed that in the Porteus initrd there isn't, being much smaller. Does Debian require this?

Hi Duprate, not sure, but I think your question is why the porteus-boot initrd1.xz from DebianDog is so much bigger than the initrd.xz from official Porteus distro (tell me if I understood wrong).
Well, my knowledge is limited about these things, but I think the initrd.xz from official Porteus has much more builtin modules, just enough to start booting (e.g. find devices, load squashfs, aufs or overlay etc...)

For info, what I do mostly:
Basically I use nowadays "mkinitramfs" to built initrd1.xz from Debian stock kernel, e.g.

Code: Select all

mkinitramfs -k -o /dev/null 5.10.0-1-amd64

("-o /dev/null" makes it only create the directory in /var/tmp, no initrd)
Then from /var/tmp/mkinitramfs_* I copy the contents of lib/modules to the porteus-boot skeleton and compress it.
Can be configured in '/etc/initramfs-tools/initramfs.conf', default is MODULES=most and I kept that.
In the past I've experimented with other methods than mkinitramfs and got a much smaller initrd1.xz (5M), it booted fine with several hardware but not all, (Bill did some tests back then, I have little hardware to test on) So it seemed best to me to use mkinitramfs (with MODULES=most) to be on the safe side (at the cost of a bigger ISO).
Hopefully this answers your question.

Fred