You Need to Load the Kernel First

I update my Linux servers regularly. What I do less regularly is restarting them. So I was not really surprised when one of them failed to boot with “you need to load the kernel first” message. What I usually do is: select the old kernel, boot the darn thing up, and then refresh grub. However, this time I overdid it so even the attempt to boot the old kernel produced the same message. It was time for troubleshooting.

Illustration

While I knew which kernels wouldn’t boot, I didn’t actually know which kernels I have available. Fortunately, grub has a command line mode hidden behind ‘c’ key press. There I selected my disk and listed all files:

ls
set root=(hd1,2)
ls /

It helps if you know how your partitions are setup for this to work but, in the worst case scenario, you can also go over each partition to find vmlinuz files. This is also the reason why I leave the boot partition unencrypted. Had my boot partition been encrypted, this step would be impossible and more involved recovery would be needed.

In this case, I was able to find vmlinuz files and I saw that the newest kernel I had installed was 5.4.0.117. Armed with that knowledge I went back to the main grub screen.

Illustration

On the main screen I went to the edit option hidden behind ‘e’ keypress. There it was simple to edit existing commands (both linux and initrd) to update the kernel version to match the one I had on my disk, followed by F10 keypress. Voila! My linux was booting.

Mind you, this wasn’t a permanent solution since the next reboot would leave me with the same issue. What I needed was to update grub (and I might as well update initramfs while I’m at it).

From the prompt, two commands were enough to make my next reboot a boring affair.

sudo update-initramfs -k all -u
sudo update-grub

And that’s it. Now my grub and my kernels are back in sync. At least for a while.

Text for OpenGL

Illustration

I occasionally like to learn something I literally have no use for. It keeps me happy and entertained. This month I decided to deal with OpenGL. And no, I don’t consider learn OpenGL an useless skill. OpenGL is very much useful and, despite newer technologies, still here to stay. It’s just that, since I do no game development, I have no real use for it. But I wanted to dip my toes into that world regardless.

After getting few triangles on the screen, it came time to output text and I was stunned to learn OpenGL has no real text support. And no, OpenGL is not unique here as neither Vulkan or Metal provide much support. Rendering text is simply not an integral part of rendering pipeline. And, once one gives it a thought, it’s clear it doesn’t belong there.

That’s not to say there are no ways to render text. The most common one is treating text as a texture. The less common way is rasterizing font into triangles. Since I really love bitmap fonts and square is easily constructed from two right triangles, I decided to go the rustic route.

The first issue was which font to select. I wanted something old, rather complete, and free. Due to quirks in the copyright law, bitmap fonts are generally not considered copyrightable under US law. Mind you, that holds true only for their final bitmap form. Fonts that come to you as TTF or OTF are definitely copyrightable.

The other issue with selection was completeness. While selecting old ROM font supporting code page 437 (aka US) is easy, the support for various European languages is limited, to say the least. Fortunately, here I came upon Bedstead font family which covered every language I could think off with some extra. While era-faithful setup would include upscaling and even a rudimentary anti-aliasing, I decided to go with a raw 5x9 pixel grid.

For conversion I wrote BedsteadToVertices utility that simply takes all character bitmaps and extracts them into a Vector2 array of triangles. The resulting file is essentially a C# class returning buffer that can be directly drawn. Something like this:

var triangles = BedsteadVerticesFont.GetVertices(text,
                                                 offsetX: -0.95f,
                                                 offsetY: 0.95f,
                                                 scaleX: 0.1f,
                                                 scaleY: 0.2f);
gl.BindBuffer(BufferTargetARB.ArrayBuffer, BufferId);
gl.BufferData<float>(BufferTargetARB.ArrayBuffer, triangles, BufferUsageARB.DynamicDraw);
gl.DrawArrays(GLEnum.Triangles, 0, (uint)triangles.Length / 2);

The very first naïve version of this file ended up generating a 3.8 MB source file. Not a breaking deal but quite a bit larger than I was comfortable with. So I went with a low hanging fruit first. Using float arrays instead of Vector2 instantly dropped the file size to 2.3 MB. Dropping all floats to 4 decimal places dropped it further to 2.0 MB.

And no, I didn’t think about reducing the whitespace. Code generated files don’t need to be ugly and reducing space count to a minimum would do just that. Especially because removing spaces will result in the exactly same compiled code at the expense of readability. Not worth it.

However, merging consecutive pixels into one big rectangle was yet another optimization that’s both cheap in implementation and reduces file size significantly. In my case, the end result was 1 MB for 1,500 characters. And yes, this is still a big file but if you exclude all the beautiful Unicode non-ASCII characters, that can bring file size down to 61 KB. Had I wanted to go the binary route, that would be even smaller but I was happy enough with this not to bother.

While the original Bedstead font is monospaced, I decided to throw a wrench into this and remove all extra spacing where I could do so easily. That means that font still feels monospaced but you won’t have excessive spaces really visible. And yes, kerning certain letter pairs (e.g., rl) was too out of scope.

On the OpenGL side, one could also argue that this style of bitmap drawing would be an excellent territory for the use of indices to reduce triangle count. One would be right on technicality but I opted not to complicate my life for dubious gains as modern GPUs (even the integrated ones) are quite capable of handling extra few hundred of triangles.

In any case, I solved my problem and, as always, the source code is available for download.


[2022-06-07: With a bit of optimization, ASCII-only file is at 50 KB.]

Really Prolific?

Illustration

Well, it’s been a while since the FTDI fuckup so I guess it was a time for another IC supplier to go bonkers. Yes, it’s again time for a chip manufacturer to mess with your computer drivers.

Story starts with me searching for 5V USB cable with a 3.3V signal. After finding a suitable device, I did what was needed and forgot about it for a while. A few days ago I needed USB type-A serial device to do a quick loopback test and grabber the same, previously working, device. While the serial port did appear, I couldn’t open it or send any data.

A quick trip to Device Manager has shown a problem: “THIS IS NOT PROLIFIC PL2303. PLEASE CONTACT YOUR SUPPLIER.” Yes, it’s the exact nonsense that FTDI pulled years ago - using Microsoft Windows Update mechanism for their authenticity enforcement.

Illustration

Now, you might thing this is their right. And I can see how they might be annoyed with fake chips using their drivers. However, their beef should be with fake chip suppliers and not with the end customer. For me the concept of bricking device owned by an unsuspected user is a bridge too far.

My case is probably the standard one. I bought device without knowing it has a fake chip in it. I paid the seller, he paid his supplied, his supplied paid the manufacturer and so on. Now my device stopped working. Money is long gone and so is the supplier of the fake chips. I might have lost that money. If I can ask for refund, the seller might be out of money. Manufacturer might be out of money (especially if they didn’t know they’re dealing with fakes). The only person not out of money is probably the guy selling fakes in the first place.

While Prolific might look at me as a potential new customer since I am in the market for a new cable, I believe that’s the wrong assumption. I am never going to knowingly buy a Prolific device again. Why? Because there is no way that I, as a customer, can check if device is indeed original or not.

What I do know is that Prolific is ready to play shenanigans with Microsoft update and brick my devices down the road. Since I cannot verify their authenticity myself, buying any Prolific device is something that might bite me in the ass. Unless something changes, I won’t buy a single Prolific cable ever again. Their product is nothing special and there are many other manufacturers happy to take my money.

I hope that Microsoft will rollback driver since it’s their update that’s causing issues for the customer. I also hope that Prolific will see the error of their ways and stop bricking customer devices. I am hoping, but not holding my breath for either.


PS: And yes, FTDI did say they saw the error of their ways back in 2014. Only to pull the same shit again in 2016. They learned nothing. Chances are neither will Prolific.

PPS: In meantime, you can download the older driver (v3.8.39.0 worked for me) and use it instead.

Installing UEFI ZFS Root on Ubuntu 22.04

Before reading further you should know that Ubuntu has a ZFS setup option since 19.10. You should use it instead of the manual installation procedure unless you need something special. In my case that special something is the native ZFS encryption, UEFI boot, and custom partitioning I find more suitable for a single disk laptop.

After booting into Ubuntu desktop installation (via “Try Ubuntu” option) we want to open a terminal. Since all further commands are going to need root credentials, we can start with that.

sudo -i

The very first step should be setting up a few variables - disk, pool, host name, and user name. This way we can use them going forward and avoid accidental mistakes. Just make sure to replace these values with ones appropriate for your system. I like to use upper-case for ZFS pool as that’s what will appear as password prompt. It just looks nicer and ZFS doesn’t care either way.

DISK=/dev/disk/by-id/^^ata_disk^^
POOL=^^Ubuntu^^
HOST=^^desktop^^
USER=^^user^^

General idea of my disk setup is to maximize amount of space available for pool with the minimum of supporting partitions. If you are not planning to have multiple kernels, decreasing boot partition size might be a good idea (512 MB is ok). This time I decided to also add a small swap partition. While hosting swap on top of the pool itself is a perfectly valid scenario, I actually found it sometime causes issues. Separate partition seems to be slightly better.

blkdiscard -f $DISK

sgdisk --zap-all                      $DISK
sgdisk -n1:1M:+63M  -t1:EF00 -c1:EFI  $DISK
sgdisk -n2:0:+960M  -t2:8300 -c2:Boot $DISK
sgdisk -n3:0:+4096M -t3:8200 -c3:Swap $DISK
sgdisk -n4:0:0      -t4:BF00 -c4:Pool $DISK
sgdisk --print                        $DISK

Finally we’re ready to create system ZFS pool. Note that you need to encrypt it at the moment it’s created. And yes, I still like lz4 the best.

zpool create -o ashift=12 -o autotrim=on \
    -O compression=lz4 -O normalization=formD \
    -O acltype=posixacl -O xattr=sa -O dnodesize=auto -O atime=off \
    -O encryption=aes-256-gcm -O keylocation=prompt -O keyformat=passphrase \
    -O canmount=off -O mountpoint=none -R /mnt/install $POOL $DISK-part4

On top of the pool we can create a root dataset.

zfs create -o canmount=noauto -o mountpoint=/ $POOL/root
zfs mount $POOL/root

Over time I went back and forth whether to use a separate home dataset or not. In this iteration, a separate dataset it is. :)

zfs create -o canmount=noauto -o mountpoint=/home $POOL/home
zfs mount $POOL/home
zfs set canmount=on $POOL/home

Assuming we’re done with datasets, we need to do a last minute setting change.

zfs set devices=off $POOL

Assuming UEFI boot, two additional partitions are needed. One for EFI and one for booting. Unlike what you get with the official guide, here I don’t have ZFS pool for boot partition but a plain old ext4. I find potential fixup works better that way and there is a better boot compatibility. If you are thinking about mirroring, making it bigger and ZFS might be a good idea. For a single disk, ext4 will do.

# yes | mkfs.ext4 $DISK-part2
# mkdir /mnt/install/boot
# mount $DISK-part2 /mnt/install/boot/

# mkfs.msdos -F 32 -n EFI -i 4d65646f $DISK-part1
# mkdir /mnt/install/boot/efi
# mount $DISK-part1 /mnt/install/boot/efi

To start the fun we need debootstrap package. Starting this step, you must be connected to the Internet.

apt update && apt install --yes debootstrap

Bootstrapping Ubuntu on the newly created pool comes next. This will take a while.

debootstrap $(basename `ls -d /cdrom/dists/*/ | grep -v stable | head -1`) /mnt/install/

We can use our live system to update a few files on our new installation.

echo $HOST > /mnt/install/etc/hostname
sed "s/ubuntu/$HOST/" /etc/hosts > /mnt/install/etc/hosts
sed '/cdrom/d' /etc/apt/sources.list > /mnt/install/etc/apt/sources.list
cp /etc/netplan/*.yaml /mnt/install/etc/netplan/

If you are installing via WiFi, you might as well copy your wireless credentials. Don’t worry if this returns errors - that just means you are not using wireless.

mkdir -p /mnt/install/etc/NetworkManager/system-connections/
cp /etc/NetworkManager/system-connections/* /mnt/install/etc/NetworkManager/system-connections/

Finally we’re ready to “chroot” into our new system.

mount --rbind /dev  /mnt/install/dev
mount --rbind /proc /mnt/install/proc
mount --rbind /sys  /mnt/install/sys
chroot /mnt/install \
    /usr/bin/env DISK=$DISK USER=$USER \
    bash --login

Let’s not forget to setup locale and time zone.

locale-gen --purge "en\_US.UTF-8" # update-locale LANG=en\_US.UTF-8 LANGUAGE=en\_US # dpkg-reconfigure --frontend noninteractive locales

dpkg-reconfigure tzdata

Now we’re ready to onboard the latest Linux image.

apt update
apt install --yes --no-install-recommends linux-image-generic linux-headers-generic

Followed by boot environment packages.

apt install --yes zfs-initramfs cryptsetup keyutils grub-efi-amd64-signed shim-signed

Since data is encrypted, we might as well use random key to encrypt our swap.

echo "swap $DISK-part3 /dev/urandom \
    swap,cipher=aes-xts-plain64,size=256,plain" >> /etc/crypttab
cat /etc/crypttab

To mount EFI, boot, and swap partitions, we need to do some fstab setup.

echo "PARTUUID=$(blkid -s PARTUUID -o value $DISK-part2) \
    /boot ext4 noatime,nofail,x-systemd.device-timeout=5s 0 1" >> /etc/fstab
echo "PARTUUID=$(blkid -s PARTUUID -o value $DISK-part1) \
    /boot/efi vfat noatime,nofail,x-systemd.device-timeout=5s 0 1" >> /etc/fstab
echo "/dev/mapper/swap none swap defaults 0 0" >> /etc/fstab
cat /etc/fstab

We might as well activate the swap now.

/etc/init.d/cryptdisks restart && sleep 5
swapon -a

Now we get grub started and update our boot environment.

KERNEL=`ls /usr/lib/modules/ | cut -d/ -f1 | sed 's/linux-image-//'`
update-initramfs -u -k $KERNEL

Grub update is what makes EFI tick.

update-grub
grub-install --target=x86_64-efi --efi-directory=/boot/efi --bootloader-id=Ubuntu \
    --recheck --no-floppy

Finally we install out GUI environment. I personally like ubuntu-desktop-minimal but you can opt for ubuntu-desktop. In any case, it’ll take a considerable amount of time.

apt install --yes ubuntu-desktop-minimal

A short package upgrade will not hurt.

add-apt-repository universe
apt update && apt dist-upgrade --yes

If one is so inclined, /home directory can get a separate dataset too but I usually skip it these days. I just proceed to create the user, assign a few extra groups to it, and make sure its home has correct owner.

adduser --disabled-password --gecos '' $USER
usermod -a -G adm,cdrom,dip,lpadmin,plugdev,sudo $USER
echo "$USER ALL=NOPASSWD:ALL" > /etc/sudoers.d/$USER
passwd $USER

As install is ready, we can exit our chroot environment.

exit

And cleanup our mount points.

umount /mnt/install/boot/efi
umount /mnt/install/boot
mount | grep -v zfs | tac | awk '/\/mnt/ {print $3}' | xargs -i{} umount -lf {}
umount /mnt/install/home
umount /mnt/install
zpool export -a

After the reboot you should be able to enjoy your installation.

reboot

PS: There are versions of this guide using the native ZFS encryption for other Ubuntu versions: 21.10 and 20.04

PPS: For LUKS-based ZFS setup, check the following posts: 20.04, 19.10, 19.04, and 18.10.

RS-232 Framework Expansion Card

Boards for this project were sponsored by PCBWay (affiliate link).


A while ago I created a CAN-bus expansion card for Framework laptop. However, I consider this an utter failure. While the card did work, the lack of CAN-bus capable parts meant you cannot actually create one yourself. At least until this IC shortage comes to an end. However, experience was fun enough that I decided to make something else I do need and that one could actually manufacture in today’s world - a RS-232 expansion card.

For those uninitiated, RS-232 is an overall standard that defined connection between different devices for a long time. As many standards go, RS-232 is quite an expansive beast with many different signals. However, over the time, it became really common to have it boiled down to a three-wire setup: RXD (receive), TXD (transmit), and GND (ground). While RS-232 did disappear from desktops and laptops alike, it’s far from dead. Be it industrial equipment or a network switch, RS-232 still has its place.

Illustration

Design was easy enough. To get from USB to TTL UART, I used MCP2221, but any similar chip will work equally well. Resulting TTL signal is then fed into MAX232 where charge pump brings it to RS-232 voltage levels, about ±8 V. This is quite sufficient to be fully RS-232 compliant. Even better, MAX232 will allow for receiving a full RS-232 ±15V signal with some margin. The end result should be that you can interconnect with pretty much any RS-232 compliant device out there.

Unlike for my CAN-bus expansion card, here I went for a slightly bigger connector of JST XH variety. While this one actually required a bit of modification to the expansion card casing, it’s actually much more convenient than its 2.0 mm brethren. First of all, spacing is 2.5 mm which is close enough to 0.1" spacing that you can easily use a standard jumper wires to connect. This means you don’t need any specialized tools or bother with creating a cable yourself. Further more, if you plan to have something more permanent connected and you want the polarity protection you get from JST XH, it’s trivial to find many cables with 3-pin JST XH already present.

Illustration

Pinout was another difficult choice and in the end I decided onto RXD GND TXD. Benefits of this approach are mostly in the area of safety. If GND wire was at pin 1, there is some place for doubt whether you start counting on left or right. With it being in the middle, that doubt is removed. Further more, to do the self-test, one just needs to connect two outer pins together. Again, it describes connection uniquely without even mentioning pin numbers. Yes, you need pin numbers for RXD and TXD regardless but swapping those two will never cause any damage so the guessing game is less dangerous.

Illustration

As noted above, the boards were manufactured by PCBWay and there were a few things to watch out for. Due to the connector, these boards had to be 0.8mm. PCBWay does support this thickness without any extra cost and it’s only a click away.

Another thing to watch for, was that this board really does require 6/6 manufacturing process. Interestingly, I accidentally had one via that was too close but my board passed all the checks PCBWay had. When I got it back, I saw that via’s ring was adjusted to be slightly thinner on the offending side so everything ended up working just fine. While I appreciate such help, I would actually prefer the check to fail so I can correct the boards myself instead of silently fixing the issue.

One important thing that had me worried was if small PCB features needed for USB type-C edge connector would be routed correctly. What I discovered by accident was that some board houses do their routing with a 0.068" mill end which is a bit too large for these cutouts. If that happens, you simply cannot fit this connector onto it. What you want is 0.04" mill end to correctly handle this board outline. Either someone was paying attention or PCBWay does this by default. Either way, I got my boards perfectly milled.

If you look at pictures carefully, you will also notice something is off with the silk screen and here I must take a full fault for this. I do love Bahnschrift font and how well it renders at small sizes so I decided to use it for silk-screen markings. But I got greedy and went with its condensed variant that, in retrospective, had no chance of being readable. This one was purely on me.

In any case, boards were in my hands within 10 days. This also meant they arrived faster than I was ready for them but I cannot really hold this against PCBWay. :)

Illustration

As far as case goes, I toyed with an idea to fit it into the existing USB A case. While this might have been just possible, it also meant for each device one would need to dismantle a perfectly good official expansion card. Thus, I went with 3D printed case with slight modifications. When combined with friction-fit top, card actually doesn’t look half bad.

With everything else out of the way, now comes the part where I tell you about potential issues. The major one is that this is non-insulated interface. RS-322 driver can easily handle ±30 V so this would not be an issue normally. However, for signals to be measured, one needs to have both communicating devices on the same ground potential. And therein lies the major trouble. If you accidentally connect something that is not ground into this, you will pass a significant current and maybe burn down the transceiver. In the worst case you could also damage the laptop as all grounds are connected together.

All that said, as long as you connect ground to ground you should be fine. If you are not sure about the ground potential on the other side, a neat trick is to simply disconnect your laptop from the AC adapter. If you run your laptop of the battery, its ground will “float” and thus you again can rely on ±30 V range of MAX232 chip. This simple trick will give you enough leeway to make many mistakes.

Lastly, there is a slightly unconventional fast (20 ms) fuse on the ground path that might save your butt if things go really bad. Notice that I said “might” and not “will” here. Fuses are great but even a fast fuse will take a decent amount of time to break the connection. Fortunately, type-C ports are quite rugged so odds are decent the motherboard won’t be damaged. But I wouldn’t bet on it.

Illustration

My personal approach when connecting anything to this expansion card would be to make sure I’m connecting ground on both sides and, as an extra precaution, to test connection with laptop disconnected for the mains. If communication works fine, you can connect laptop back to the mains.

Also note this is nothing specific to this adapter - all non-insulated USB adapters (and most of them are) suffer from the same potential problems. However, due to the size of this expansion card, it’s much easier to misconnect wires then when you’re connecting to the DB-9.

In any case, as always, the design is freely available. If you don’t want to bother soldering one yourself, they are also available for purchase (USA only, at this time).