site
contact
newsKirkstone-series packages rolled back and forward
Compiling the packages that are intended to be in the new EasyOS Kirkstone-series is a juggling act. I am having to roll some packages back to earlier versions and some forward.
Ghostscript for example, rolled back to 9.20.
OE Kirkstone has ffmpeg 5.0.1; however, vlc in kirkstone won't compile. The OE devs are probably waiting for a later version of vlc that fixes it. But I think that a lot of apps that use ffmpeg are not yet ready for 5.0, due to major api changes. So I rolled back to ffmpeg 4.4.
Kirkstone has rust 1.59; however, I discovered it has some problems, such as unable to create a 'rustc' that will work in the target evironment. OE master has rust 1.65, that I have rolled forward to.
Jon (scsijon in the forum) sent me an email today with a link about mesa dropping classical drivers, starting with 22.0:
https://www.phoronix.com/news/2021-Retire-Mesa-Classic-Main
...they advise the last version to use is 21.3.8. OE kirkstone has mesa 22.0.3. I found that OE has recipes for 21.3.7, so have rolled back to that:
I can't keep playing this game. Yeah, OK, seem to have committed to bring out EasyOS Kirkstone-series, but might get off the merry-go-round after that, just stay with the Kirkstone-series in maintenance-mode and work on some of its features.
But get out of this endless game of upgrading.
This time, I'm aiming for a much bigger package repository. I see that Slackware has 2,292 packages:
https://en.wikipedia.org/wiki/Comparison_of_Linux_distributions
I am aiming for 1,500 - 2,000.
Note, some other distributions such as Debian's 153,621 packages
are misleading, as they split packages into 3-6 smaller packages.
But even if divide by 4, still a huge repo.
Tags: easy
EasyOS advancing to Kirkstone-series
That's the current intention anyway. The Dunfell-series started in early 2020, and I am running into problems with libraries etc. being too old. The OpenEmbedded/Yocto Kirkstone version was released in May 2022, and is an LTS version, now on the fifth iteration (4.0.5):
https://wiki.yoctoproject.org/wiki/Releases
I have been working on the Kirkstone-series very intensely for the last 5-6 days, to the point where my eyes were giving up. So, will slow down for awhile get outdoors.
Anyway, it is looking very good indeed. gcc is version 11.3.0, glibc is 2.35, libreoffice is 7.4.2.3 and chromium is 106.0.5249.119
...yes, even chromium has cross-compiled successfully. And I see on the 'meta-browser' github site that they have already bumped the version. Those guys are keeping chromium up-to-date, and have neglected firefox, which is back at 68.9.0esr. The firefox build failed and I posted a report:
https://github.com/OSSystems/meta-browser/issues/651
Lots of package build recipes still to fix, at least 4-5 more days, then intend to put the new "meta-quirky" onto github as project "oe-qky-kirkstone".
I am just a cog in the large machine that is OE/Yocto, with my "meta-quirky" layer that has a couple of hundred package build recipes created by me. There are so many other layers, that one puts together according to what one wants to compile and for what target:
http://layers.openembedded.org/layerindex/branch/master/layers/
...for example, I am using "meta-office", "meta-browser" and
"meta-clang", as well as the core layers.
Tags: easy
Boot-partition mounts as msdos instead of vfat
If you click on the boot-partition of the usb-stick, it will mount and will show in the file manager, as expected; however, it has mounted as the title says, as an msdos filesystem instead of vfat.
Way back in the early days of the FAT filesystem, all files were "8.3", that is, maximum eight-character name and maximum three-character extension. Microsoft introduced an extension to allow longer filenames, and also to store upper-case and lower-case characters -- though FAT remained case-insensitive.
The old 8.3 is what we call the "msdos" filesystem, and the extension is "vfat". The problem that has mysteriously appeared is that when click on the boot-partition, it mounts as msdos.
To show the difference, first mount as vfat:
# busybox mount -t vfat /dev/sdc1 /mnt/sdc1
...and I have created a file with long name, with mix of upper and lower case characters. Now, unmount and remount as msdos:
# busybox mount -t msdos /dev/sdc1 /mnt/sdc1
...hmmm, it has been a very very long time since I have seen 8.3 files.
I am reporting the problem, don't know why it is happening. I only recall seeing it in about the last month. A change in the kernel FAT driver?
Trying to think what has changed in the last month or so... the kernel yes, but also changed from using busybox 'blkid' to the full 'blkid' utility. OK, comparing them, where sdb1 is the boot-partition on the 4.5.2 usb-stick:
# blkid /dev/sdb1
/dev/sdb1: SEC_TYPE="msdos" UUID="260A-C6D3" BLOCK_SIZE="512" TYPE="vfat" PARTUUID="03fa83c2-01"
# busybox blkid /dev/sdb1
/dev/sdb1: UUID="260A-C6D3" TYPE="vfat"
...interesting, what is that "SEC_TYPE" reported by the full 'blkid'?
One of the internal drives has a fat32 partition, sda2:
# blkid /dev/sda2
/dev/sda2: LABEL_FATBOOT="HDDESP" LABEL="HDDESP" UUID="9473-48E1" BLOCK_SIZE="512" TYPE="vfat" PARTUUID="90182abe-5fe0-4cf5-b2b5-ddb6b732f09c"
# busybox blkid /dev/sda2
/dev/sda2: LABEL="HDDESP" UUID="9473-48E1" TYPE="vfat"
...no problem with that, mounts as vfat.
Just reporting the problem, as working on something else right
now and don't want to break the train of thought.
A thought: fat12 is so ancient, might have to go up to fat16, but
also will need to increase the size of the
boot-partition.
Tags: easy
iftop compiled in OE
This is what iftop does:
https://linux.die.net/man/8/iftop
Have compiled in OpenEmbedded:
https://github.com/bkauler/oe-qky-dunfell/commit/b2694f78bad9755720163a7a77a102423725fc3a
Added to the repo in woofQ:
https://github.com/bkauler/woofq/commit/0639bbba080cf2286a2ba3c5af1ea249f8c49f31
Click on "pkg" icon and choose to update the database, and it can
be downloaded and installed.
Tags: easy
Samba fixed in EasyShare again
Yeah, again, because this bug was supposed to be fixed. Thanks to don570 who reported it:
https://forum.puppylinux.com/viewtopic.php?p=73981#p73981
Have made a commit to github:
https://github.com/bkauler/woofq/commit/434c3041656e11e18e26284aa7691f311600346e
For now, if you are running Easy 4.5.1, you could manually edit /usr/local/EasyShare/samba-setup, change line 126 to this:
pid directory = /var/run/samba
And also do the same for /etc/samba/smb.conf
You will have to turn off samba then back on, via the radiobuttons in EasyShare, or from a terminal:
# /etc/init.d/rc.samba restart
Sorry about that, the bug slipped through.
One very good thing for 4.5.2 is the kernel module vmd.ko now builtin. This has fixed the kernel unable to detect the internal drive in some modern computers.
The result of that problem was that when you do a frugal install and create an entry for the boot-loader, GRUB, rEFInd, Limine or whatever, at bootup in the initrd there were messages like "/etc/fstab not found" and "easy.sfs not found".
The first message was output by the 'mount' utility when it
failed to mount the partition, due to the kernel unable to see the
drive.
Tags: easy
EasyOS Dunfell-series version 4.5.2 released
This is another bug-fix release for version 4.5. Please read the announcements for 4.5 and 4.5.1:
- EasyOS Dunfell-series 64-bit version 4.5 released — November 12, 2022
- EasyOS Dunfell-series 64-bit version 4.5.1 released — November 19, 2022
Release notes for 4.5.2 are here:
https://distro.ibiblio.org/easyos/amd64/releases/dunfell/2022/4.5.2/release-notes.htm
If you want to see more details of the changes since 4.5.1, github commits are here:
https://github.com/bkauler/woofq/commits/8f8d6b2e8ab449cd0b3e0e52d32bd3f328fafafa
Download English, French and German builds here:
https://distro.ibiblio.org/easyos/amd64/releases/dunfell/2022/4.5.2/
Or mirrored courtesy of NLUUG here:
https://ftp.nluug.nl/os/Linux/distr/easyos/amd64/releases/dunfell/2022/4.5.2/
Feedback welcome at the forum:
https://forum.puppylinux.com/viewtopic.php?p=73787#p73787
Have fun!
Tags: easy
Kernel 5.15.79 compiled with vmd.ko builtin
Sometime ago, Ramachandra Iyer purchased a new HP laptop, but found that EasyOS and some of the pups would not recognize the NVME SSD at early bootup. This meant that unable to install any of these distros to the internal drive.
He is able to boot from a USB-stick, and the NVME SSD is recognized after bootup, which does seem to indicate that a kernel module is required to be loaded. I searched online and found some posts that indicate the culprit is the 'vmd.ko' module, for example:
https://www.reddit.com/r/archlinux/comments/lxx54z/arch_linux_not_detecting_nvme_storage_on_boot/
I posted to the forum, asking him to test if vmd.ko gets loaded:
https://forum.puppylinux.com/viewtopic.php?p=73525#p73525
Other linux distributions are able to copy a kernel module into the initrd, and it will be loaded at early bootup. EasyOS does not load modules in the initrd, and relies on a big kernel with all required drivers to detect the drives being builtin to the kernel.
I have compiled the 5.15.79 kernel with vmd.ko builtin. This does not guarantee to fix Ramachandra Iyer's problem, but from online reading it does seem likely. Keeping the fingers crossed that it does.
Device Drivers > PCI support > PCI controller drivers
<*> Intel Volume Management Device Driver CONFIG_VMD
Easy 4.5.2 will be released very soon.
Tags: easy