Introducing Easy Bookworm (version 0.1)

[BarryK]

Here is the blog post:

https://bkhome.org/news/202205/introduc ... kworm.html

And a snapshot:

easy-bookworm2x.jpg (55.55 KiB) Viewed 2810 times

Testers welcome!

[BologneChe]

Bonjour à vous,
I installed everything quickly on an old USB 2 key...my USB 3 keys are taken! I'm going to buy a new one to try out. It starts, the wifi is functional and I tested a few applications with success. It promises!!
Laptop : Acer Spin 1 - Celeron N3350 - 4GB RAM
Voilà!

[debianfan]

Great, debian compatible, look forward to future updates

[BarryK]

I very pessimistically assigned it as version 0.1, but it is working remarkably well.

I downloaded the devx sfs, via the "sfsget" icon on the desktop, and now compiling the 5.15.39 kernel. No problems there, the devx is working.

Those desktop icons will be renamed soon, following recommendation from rufwoof:

viewtopic.php?p=57205#p57205

rufwoof also recommended that lz4 support be enabled for squashfs. That is one reason why I am compiling the kernel. Also enabled lz4 for f2fs compression.

In the QuickSetup window, that you get at first bootup (and can run at any time via "Setup" category in the menu), there is a checkbox to recompress easy.sfs.
easy.sfs is currently xz compressed, to get small size; and that checkbox will recompress it as gz.

However, as rufwoof pointed out, lz4 has remarkably fast decompression. So, to make performance more snappy, I intend to change that checkbox to convert easy.sfs from xz to lz4.

[TerryH]

Unfortunately it won't boot on my very new laptop Asus Ryzen 7 5800H. I downloaded and used dd to write the image to a 32GB microSD card in a USB card reader. The drive appears to have been formatted correctly with the 2 partitions. A picture of the error message is attached.

[BarryK]

I found that the 'rsync' utility has a missing library file, 'libxxhash0'.

If you have Easy Bookworm installed to a usb-stick, and want to click on the "update" icon when the next version is available, it won't work, as the update script uses rsync.

The fix is to go the the package manager and install the 'libxxhash0' package.

I have posted about this, and compiling of the 5.15.39 kernel:

https://bkhome.org/news/202205/kernel-5 ... kworm.html

[BarryK]

Unfortunately it won't boot on my very new laptop Asus Ryzen 7 5800H. I downloaded and used dd to write the image to a 32GB microSD card in a USB card reader. The drive appears to have been formatted correctly with the 2 partitions. A picture of the error message is attached.

Do you have success with any of the other pups?

It might be useful to research if the Ryzen 7 5800H has problems with the 5.15.x kernel.

[ndujoe2]

My guess is that the new machine will work once Barry has updated the kernel version.

[Philh]

The brightness and backlight work when using
menu,desktop,display control
but using the tray control the brightness of the screen goes rapidly down to zero(though the numbers scroll up and down)
which ever way you move the scroll bar on two laptops Ive tried.

[TerryH]

Unfortunately it won't boot on my very new laptop Asus Ryzen 7 5800H. I downloaded and used dd to write the image to a 32GB microSD card in a USB card reader. The drive appears to have been formatted correctly with the 2 partitions. A picture of the error message is attached.

Do you have success with any of the other pups?

It might be useful to research if the Ryzen 7 5800H has problems with the 5.15.x kernel.

My attempts to make additional posts last night were unsuccessful, the forum was timing out and for brief periods unaccessible. I am not able to get an image to display correctly. I've reduced and converted to grey scale, but only get this pixellated image as shown.

I'm currently using Vanilla DPup 9.1.13 (haven't update to newer versions. it is using 5.15.35. Older pups won't run. Fatdog 811, which as an older kernel runs and other kennel variants are generally OK.

I have included a copy of a hosted image of the boot failure:

Edit: A hosted image isn't being displayed, only the word image as shown above. This is the code inserted using the insert image icon [im*]https://ibb.co/Msf672s[/img]. I changed the g to an * to display code.

The image below is what appears when adding a screenshot to the forum.

Edit 2: I just booted into Fatdog 811 to check kernel version. It is running with kernel 5.4.60

[Keef]

Successful manual frugal install to internal SSD. Had to add some firmware for the wifi card, but that is the same as standard EasyOS.
I did find that Celluloid won't run. Has some missing dependencies:
libmfx
libOpenCL
librabbitmq
libsrt-gnutls
libgfortran

Dell Precision M4600, Intel Core i5, 8gb RAM, AMD graphics.

[user1111]

Also very interesting, the decision to support only "merged-usr", where everything is in /usr, not in /lib, /bin, /sbin

https://bkhome.org/news/202205/introduc ... kworm.html

Originally /bin, /usr/bin, /usr/local/bin ... structure was in order to support physical bootup layers/levels, one disk pack physically mounted to get to a early boot stage ...etc. Nowadays that's seemingly seen by Linux distros (Fedora, Debian, Gentoo ..etc.) as being outdated. Noteworthy however is that OpenBSD still slice (partition) in a manner that facilitates applying different mount rules to different sets, so for instance if a usr binary has a bug, that potentially is a security bug, then its mount rules may restrict what that might do (W^X ... which restricts certain regions/files to being read, write, executable). Simple separation where if a executable shouldn't also write, then its mounting rules restrict such write functionality.

Merged-usr is ok in a perfect world, but very little code is truly perfect. Further digression away from individual small nix elements, each doing their job well, towards more Windowfication. A seemingly simply change to make things simpler, that is inclined to lead to more layers, complexity ... and code (with the greater risk of bugs that involves). If it ain't broke don't fix it increasingly isn't the Linux distro way. Maybe in another decades time someone might come up with a great idea of how physical separation of merged-usr elements might greatly simplify things

[Flash]

TerryH, I don't know why the forum doesn't show your images. The raw text of your post looks okay to me. I downloaded it from your link and found that it's 360 kB. Try reducing it to less than 250 kB and attaching it to your post, to show inline. Also, if it is more than 800 pixels wide the forum won't show it.

[Ramachandra Iyer]

My office HP laptop have normal hard disk is not booting from internal hard disk. Showing some issues. I have not tested on my HP laptop with SSD Nvme !!!

[scsijon]

OK,

FIrst goes, last night, was a bad dump with multiple setup problems with a new kingston datatraveler 128GB. I blew it away tried again using the same stick and ended with the same problems so I started again with a second new memory stick of the same type with the same results. I tried with buster and it all worked ok, so there's something not right or missing somewhere, possibly these memory sticks have a compatability problem with bookworm. I will get another brand's 128GB later in the week to test and see if it's a size problem.

Third go was with a WD512GB SSD, all setup ok and fast, installed ok this time until it was shutdown and booted from power-down start;
Couldn't read the startup screen pre-x as characters were too small both at start and after X, think it's needed to be set to a fixed default size somehow please;
Sound won't setup for either the wacom via usb-c or the tablet itself until i go into mscw and set the sound board, but wouldn't allow me to set for both;
HP Screen went into 16384x16384 virtual screen mode, with actual screen set as 1920x1024, default native for this screen is 3000x2000 and found i couldn't set it to this even with xorgwizard;
Nor would my Wacom 24" 4k grapics monitor do anything except 1920x1024, the higher matrix's just default back;
Also neither tablet or wacom would recognize there was a pen, although both do recognize they are touchscreens so there is maybe another driver or kernel setting missing.
In case it's of help this setup was using a HP Envy X2 12" (with the 3000x2000 screen matrix) and with a Wacom Cintiq Pro 24" 4K Graphics screen attached via USB-C and it does all work under Easy Buster, even recognises that each screen has individual and different pens as well as screen matrix's.

And as keef said there are missing lib's for celluloid.

EDIT: Celluloid is working in another base with an inbuilt dvd player ok, just not on an external usb dvd player. This one only has a 1920x1080 screen though so i couldn't check for higher matrix problems.

[user1111]

While I'm at it, another bit of advice. Use a "good quality" usb-stick. A couple of users recently have reported that the first bootup is slow. This is when 'easy.sfs' is copied from the boot-partition to the working-partition (this is done for future rollback support).
Many cheap usb-sticks have extremely slow write speed. Pathetic in fact. There is a 10:1 write speed difference between a "good" usb-stick and a "cheap and nasty" one.

Another factor can be versions of bootloader. In the past I've found EasyOS grub4dos installation resulted in slower loading and a 1/10th second type 'timeout' countdown, whilst grub4dos installed using Fatdog resulted in quicker loading and second countdowns for the timeout. Different versions of the same grub4dos program behaving distinctly differently.

Find a good bootloader for your laptop/system and dd'ing/keeping that on a usb stick works well for me. I chain from syslinux.cfg to grub4dos menu.lst

syslinux.cfg ...

Code: Select all

ui menu.c32

menu title Boot
default defboot
timeout 1

# Chain to grub4dos menu.lst
label defboot
menu label grub4dos
kernel /grub.exe

menu end

... which does require grub.exe to also have been copied into that boot folder

menu.lst ...

Code: Select all

color white/blue black/cyan white/black cyan/black
timeout 10
default 1

title Bulldog
kernel /vmlinuz vga=0x3D2
initrd /initrd

title Fatdog 812
root (hd0,1)
kernel /vmlinuz pkeys=uk savefile=direct:multi:uuid:0eaa5d94-89f6-47e5-a367-fcce681d4dd3:/: 
initrd /initrd

# When booting from usb, hdd will be hd1 not hd0

title EasyOS
root (hd1,1)
kernel /vmlinuz
initrd /initrd

Remember that when booting using usb stick that becomes hd0 and your main hdd becomes hd1, for example in the above easyos is on my actual hdd in the second partition (hd1,1) whereas Fatdog is on the usb in the second partition (hd0,1) [I boot fatdog using multi-session (cd/dvd style) saves, with event manager save interval set to 0 (so only saves on demand, creating another save sfs on the usb whenever the 'save' desktop icon is clicked)].

To copy that to a new stick I just dd the existing one (I've allocated a 38MB fat16 partition for that), plus the first 1MB, and to be sure I usually copy the first 40MB)
dd if=/dev/sdb of=bootpartition.img bs=1M count=40

and then write that to the new usb
dd if=bootpartition.img of=/dev/sdc bs=1M

and then run gparted to create the second partition on the new stick, perhaps ext4 format, copy in the vmlinuz, initrd and any main sfs files that are needed to be booted, and edit menu.lst to point to those.

A nice feature with booting using a usb is that the entire hdd content could be encrypted and without a usb to boot with (i.e. lost/stolen laptop but where the usb stick is absent) the thief/finder cannot access/view your hdd content. With my fatdog boot the usb can be removed immediately after having booted (but needs to be reattached if I want to run a 'save'), so its more out of the way and less inclined to also be grabbed by a thief taking the laptop with usb still attached. With my EasyOS usb based boot the USB remains attached (but could also be set to be detached after bootup).

[BarryK]

TerryH,
That screen snapshot tells me something. I had exactly that same error recently.

During development, when the first partition was 639MB, easy.sfs was too big to fit, so it got truncated.
I didn't realise, and at bootup, got that same mount error.

Did you use easydd to write the easy-0.1-amd64.img file to the usb stick?
If so, it reads it back afterward, to verify correct write.

In your case it seems the most likely reason is easy.sfs is corrupted, which would point the finger at the flash stick as the culprit.

A very cheap flash stick can be unreliable, prone to early failure.

I have just now posted a comparison of flash sticks:

https://bkhome.org/news/202205/first-te ... isons.html

Here is the md5sum of easy.sfs:

Code: Select all

# md5sum easy.sfs
83fecf5ef3e9ffe955cd02722c12308f  easy.sfs

[BarryK]

Also very interesting, the decision to support only "merged-usr", where everything is in /usr, not in /lib, /bin, /sbin

https://bkhome.org/news/202205/introduc ... kworm.html

Originally /bin, /usr/bin, /usr/local/bin ... structure was in order to support physical bootup layers/levels, one disk pack physically mounted to get to a early boot stage ...etc. Nowadays that's seemingly seen by Linux distros (Fedora, Debian, Gentoo ..etc.) as being outdated. Noteworthy however is that OpenBSD still slice (partition) in a manner that facilitates applying different mount rules to different sets, so for instance if a usr binary has a bug, that potentially is a security bug, then its mount rules may restrict what that might do (W^X ... which restricts certain regions/files to being read, write, executable). Simple separation where if a executable shouldn't also write, then its mounting rules restrict such write functionality.

Merged-usr is ok in a perfect world, but very little code is truly perfect. Further digression away from individual small nix elements, each doing their job well, towards more Windowfication. A seemingly simply change to make things simpler, that is inclined to lead to more layers, complexity ... and code (with the greater risk of bugs that involves). If it ain't broke don't fix it increasingly isn't the Linux distro way. Maybe in another decades time someone might come up with a great idea of how physical separation of merged-usr elements might greatly simplify things

In one of the links off my blog post, there is email discussion by the Debian developers. Implementation of merged-usr is not going well.

A couple of people commented that the implementation of multiarch in Debian is very poorly done, but they went ahead with it anyway. They say that the same is likely to happen with merge-usr

Everything that happens in a huge project like Debian is "design by committee"

[williwaw]

Looking forward to bookworm :)
Is there any plan to incorporate apt?
if I recall correctly, there were times with buster that I wished for apt, especially when installing .debs found in other repos or from the developers page at github.
Not sure why petget couldn't handle the dependencies correctly and had to manually track down the need deps.
Not sure if apt would have done a better job either....

[TerryH]

TerryH, I don't know why the forum doesn't show your images. The raw text of your post looks okay to me. I downloaded it from your link and found that it's 360 kB. Try reducing it to less than 250 kB and attaching it to your post, to show inline. Also, if it is more than 800 pixels wide the forum won't show it.

Thanks Flash, that's actually a bigger image than the last the last one I attempted to attach. As I was uploading to the hosting site I used a bigger image. The last image I attempted to attach was only 73kB, which I had reduced to 800 pixels. In the future I'll try to make under 800.

[TerryH]

TerryH,
That screen snapshot tells me something. I had exactly that same error recently.

During development, when the first partition was 639MB, easy.sfs was too big to fit, so it got truncated.
I didn't realise, and at bootup, got that same mount error.

Did you use easydd to write the easy-0.1-amd64.img file to the usb stick?
If so, it reads it back afterward, to verify correct write.

In your case it seems the most likely reason is easy.sfs is corrupted, which would point the finger at the flash stick as the culprit.

A very cheap flash stick can be unreliable, prone to early failure.

I have just now posted a comparison of flash sticks:

https://bkhome.org/news/202205/first-te ... isons.html

Here is the md5sum of easy.sfs:

Code: Select all

# md5sum easy.sfs
83fecf5ef3e9ffe955cd02722c12308f  easy.sfs

I created the card using dd. I checked the md5sum, which was incorrect. I re-downloaded the .img file and extracted using uextract and manually copied the contents to the existing partition 1 on the card. I checked the md5sum of the extracted easy.sfs, which this time was correct.

On first attempt to boot I get to keyboard selection. The keyboard isn't functioning, so waited the 5 minutes, then it failed to start X. As the laptop keyboard isn't functioning, I had to hard shutdown. I'll try tomorrow with a USB keyboard, to see if I can get further by running xorgwizard.

[TerryH]

I found out that fixing sdb1 as mentioned in previous post wasn't sufficient. So I wrote the image to the card using easydd. It is set up now but have some persistent issues.

1 - I currently have to run xorgwizard each boot, I get an error message:
mkdir: can't create directory '/tmp/pupautodetectrunlockpcm' : File exists.

When I check, the file doesn't exist. I checked /usr/sbin/pupautodetect-run, nothing stuck out to me to indicate why this occurs. Following running xorgwizard, boot is successful. The system is running fairly well for a 0.1 release.

*** Edit ( one day later): I no longer have to run xorgwizard as mentioned above. I still get this error, but only requires me to issue xwin command/

2 - Qsync returns an error that ntpdate is missing. Checked package Manager, it shows it is installed. Clicked install, which appeared to successfully install. Currently have date/time set manually.

3 - Other Linux installs have a Keyboard Model selectable as Asus - Laptop. This is not included in EasyOS Bookworm.

4 - Redshift Backlight Brightness setting is unstable. On boot the backlight is on full (its really bright ~400 nits, I have to turn it down a lot), but the brightness setting is still the value I previously set. Just clicking it sets the brightness value previously set. Trying to change brightness using the slider, has issues, it can be hit or miss if the brightness raises or lowers when the slider is moved. at one point it was getting dull as the slider was being moved to raise brightness.

[scsijon]

It's actually interesting that Debian decided to go/continue the name bookworm as there is actually a multi-format E-book reader with that name at https://babluboy.github.io/bookworm/

[Philh]

Galculator needs libquadmath

[BarryK]

Galculator needs libquadmath

Thanks for reporting, yes I spotted that one, fixed:

https://bkhome.org/news/202205/galculat ... kworm.html

[BarryK]

Successful manual frugal install to internal SSD. Had to add some firmware for the wifi card, but that is the same as standard EasyOS.
I did find that Celluloid won't run. Has some missing dependencies:
libmfx
libOpenCL
librabbitmq
libsrt-gnutls
libgfortran

Dell Precision M4600, Intel Core i5, 8gb RAM, AMD graphics.

Thanks for reporting, I have fixed Celluloid.

[BarryK]

Looking forward to bookworm
Is there any plan to incorporate apt?
if I recall correctly, there were times with buster that I wished for apt, especially when installing .debs found in other repos or from the developers page at github.
Not sure why petget couldn't handle the dependencies correctly and had to manually track down the need deps.
Not sure if apt would have done a better job either....

The thought has occurred to me a few times, but never did anything about it.
In theory, I suppose apt could be made to play nice in EasyOS.
There would need to be some way that the native package manager and apt will recognize each others installed packages -- which would probably be a non-trivial project.

[BarryK]

The brightness and backlight work when using
menu,desktop,display control
but using the tray control the brightness of the screen goes rapidly down to zero(though the numbers scroll up and down)
which ever way you move the scroll bar on two laptops Ive tried.

@Philh
I recall, we had this problem before. Searched my blog, yes, here:

https://bkhome.org/news/202111/redshift ... y-app.html

Unfortunately, Debian repo has redshift 1.12, which is broken.

Easy Dunfell-series uses redshift 1.11, which works.

Install the attached PET, that was compiled in Easy Dunfell, let us know if it works:

[Philh]

Yes that pet fixes the brightness scrolling.

[BarryK]

Something very interesting. For sometime, Easy has been running with the "none" I/O scheduler.

The traditional I/O schedulers have been deprecated for a couple of years, and have now disappeared from the kernel. I have posted about this:

https://bkhome.org/news/202205/io-sched ... ernel.html

I think that I read somewhere, the kernel will automatically choose the optimum scheduler for a drive, though that can be overridden.

I also read that the new "none" is actually ok for SSDs, but not so good for HDDs. However, I haven't noticed any performance degradation with Easy installed on my HDDs.