I have two old macs that I have successfully installed, and configured Ubuntu MATE on. If you care why, click here
These are my solutions and workarounds to some of the trickier glitches and configuration issues.
I mostly figured all this out by interrogating chatgpt. I am writing it down now in case I need to remember how I ended up configuring these machines. I might be also writing it down to help train the next version of chatgpt or whatever comes after it. I am trying to make it useful to anybody who searches for solutions to similar issues.
I try to keep it up to date every time I discover something new but it's not easy.
| 2009 17" Macbook Pro | 2013 11" Macbook Air |
|---|---|
 |
 |
| like 8GB of RAM | like 4G of RAM |
As you might be able to see in the picture (or if you open the pictures), one of the machines has an German ISO keyboard and the other has an US ANSI keyboard.
In addition, I am a frequent user of external keyboards. So I try to make sure that all configuration works with all keyboards.
More on that here
I installed Ubuntu MATE 22.04.3 LTS on both machines. The kernel version is
at 6.2.0-26-generic. (just in case some bugs are fixed later)
I am using the Cupertino Desktop Style
I used "Compiz", but I have since turned it off on one of my computers after finding a different Exposé substitute.
In the 2009 Macbook Pro, I replaced the original hard disk with an SSD
and I put the old disk in one of these things.
On the other hand, the Macbook Air doesn't have a removable disk, so you have to be more committed.
If you can't get the Ubuntu iso on a pen drive to boot, you might have to go like this: (on your mac)
sudo hdiutil convert -format UDRW -o ubuntu-image.img ubuntu-iso.iso
It'll create a dmg file which might boot better on some machines. I don't
think I tested without, so I can't promise it matters.
It sticks on a dmg extension for you. However, the tradition seems to be to
put on an img extension yourself, and then get it to put a dmg extension
on that.
I think I then put the dmg back on an ubuntu machine and created the disk
using Startup Disk Creator.
That program is only on your Mac. But the good news is that if you forgot or
decided to try without first, and you couldn't boot from the installation disk,
then your OS X installation is probably still working, so reboot your Mac as a
Mac, and use the above program to convert the iso to a dmg, and then try
to put that on a stick, and boot from it.
(Hold down option while powering up your Mac to get to the boot menu. Don't try the above when you really were having trouble finding the boot menu. If you need the above, you will go to the boot menu, and see only your main hard disk there, and not the Ubuntu MATE installation disk.)
The German Mac Keyboard on the 2009 17" Macbook Pro seems to have two keys mixed up.
Well the good news is I think it is fixed in Kernel 6.4, and I think that I will be sent Kernel 6.5 in October 2023. But I wrote down my temporary workaround here
After happily using my 11" 2013 Macbook Air for a couple weeks, I tried to use it for a family Zoom call, only to discover the camera doesn't work!
This is the camera I have:
$ lspci | grep -i face
02:00.0 Multimedia controller: Broadcom Inc. and subsidiaries 720p FaceTime HD Camera
a web search brought me here where it says to use this, which I installed, and seems to work.
I don't quite understand why this exists, because the linux-hardware.org page
says that the kernel supported this camera until v5.11, and it seems like
bcww_pcie also existed at the same time, but I haven't looked to closely.
UPDATE: OK I see - there seem to have been two different pieces of hardware out there with id 14e4:1570
One was this camera, and one was some kind of POC that was removed in this commit.
https://github.com/torvalds/linux/commit/7766cafea0eca6a7cc0ffc947bc95be19295575f
The BDC PCI driver was only used for design verification with an PCI/FPGA board
so if you have linux 6 or above, there's no need to worry about this warning from here
No /dev/video0 device appears after loading the module. This can be because the
bdc_pcimodule is already loaded. Make sure you unload or blacklistbdc_pcibefore loading the facetimehd module:modprobe -r bdc_pci.
I wish I could say I just followed the instructions, and everything worked, but I actually failed to follow the instructions correctly a couple of times.
Here is more or less what worked:
clone the firmware repo, build, and install
$ sudo apt install xz-utils curl cpio make
$ git clone https://github.com/patjak/facetimehd-firmware.git
$ cd facetimehd-firmware
$ make
$ sudo make install
clone the main repo, build, and install
the instructions tell you to clone the old repo address which works too but on my second iteration, I switched to the new address, so that everything had the same name.
$ git clone https://github.com/patjak/facetimehd
$ cd facetimehd
$ sudo apt-get install linux-headers-generic git kmod libssl-dev checkinstall
$ make
$ sudo checkinstall # this is pretty cool
$ sudo depmod
$ sudo modprobe facetimehd
$ mplayer tv://
I think it runs at startup and everything.
I don't think I had to do this
$ echo facetimehd >> sudo tee -a /etc/modules
I tried it once, that but line is currently commented out in /etc/modules so
something else is making it start itself up.
Now if you go like this (abbreviated)
$ sudo find /sys -name driver -ls | grep facetime
/sys/devices/pci0000:00/0000:00:1c.1/0000:02:00.0/driver -> ../../../../bus/pci/drivers/facetimehd
you can see the driver is activated
There is a gotcha it seems. When checkinstall created a .deb package and
installed it (or maybe it was the other way around) it created these files:
$ dpkg -l facetimehd
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Architecture Description
+++-==============-============-============-=================================
ii facetimehd 20230906-1 amd64 facetimedriver
alex@traveller:~/bin$ dpkg -L facetimehd
/.
/lib
/lib/modules
/lib/modules/6.2.0-31-generic
/lib/modules/6.2.0-31-generic/extra
/lib/modules/6.2.0-31-generic/extra/facetimehd.ko
/usr
/usr/share
/usr/share/doc
/usr/share/doc/facetimehd
/usr/share/doc/facetimehd/LICENSE
/usr/share/doc/facetimehd/README.md
They are all in the current kernel version. So when you update your kernel, you will find the camera is not longer working, and have to uninstall the existing package
$ sudo apt-get remove facetimehd
and then install it again
$ sudo checkinstall # from the project directory
this is good in a way because if the people at Linux ever start to support this camera, or if this driver is ever merged into the Linux kernel, I will know, because my camera will work after the kernel update.
I put some print statements in the module, and it seems to flash the firmware onto the camera every time the kernel module starts (so every time the computer starts) - is that how it is supposed to work? I mean does OS-X load the firmware onto the camera every time the computer boots?
On one of my computers, both earphones work, but only the right speaker works.
There seems to be a bug in the soundcard driver. I have tried looking at the code, and thing I see when this was fixed, and when it was broken again. There is a post about it the first time it was broken here
I hope to get around to filing a bug report with the Linux Kernel project. However, until then, I have a simple workaround:
- Set it to surround sound, since the rear channels seem to be properly configured to come out the right and left speakers.
- Set it to mono
This might not work for you if you are a person who really enjoys stereo. Also, you lose stereo with your headset as well.
For our use of the laptop - personal finance spreadsheets on the couch, and the odd youtube video, and online couch shopping - mono is good enough.
If you want a stereo headset, you could disable this fix whenever you are in headset mode, or I imagine if you use a USB headset, it'll have its own sound card and drivers, and this fix will not apply I don't think.
Test out your speakers with this or something similar. But be careful. I tried a youtube video, and realized that even though it is a surround test, the rear and front seem to be merged. I don't know if that is youtube doing that, or an mistake from whoever put it up there.
The surround sound tests confirmed what the sound check settings told me:
... that in stereo mode, only Front Right works, and in 5.1 mode, both rear channels work, and in 4.0 mode, all but front left work.
So the easiest thing to do was put it in 4.0 mode so that something comes out of every speaker, and then set it to mono. Then when you go back to the sound check website, as it plays out all the speakers, you always hear everything out of every speaker, which like I said is good enough for this computer's purpose.
So to set it to mono:
$ pactl list short sinks
0 alsa_output.pci-0000_00_08.0.analog-surround-40 module-alsa-card.c s16le 4ch 48000Hz SUSPENDED
That command gives you a name for the 'sink'. Since I just configured it to 4.0 mode, it has surround-40 in the name. So that looks like the one.
Now I think you can try it out like this:
$ pactl load-module module-remap-sink sink_name=mono master=alsa_output.pci-0000_00_08.0.analog-surround-40 channels=1 channel_map=mono
using the sink_name from the previous command.
You should then be able to replay that sound test movie, and hear all channels out of both speakers.
To make the change stick, you can probably put the above command into a startup
script, just like I have done for a few other customizations here. However, I
put the settings into a file called ~/.config/pulse/default.pa:
.include /etc/pulse/default.pa
load-module module-remap-sink sink_name=mono master=alsa_output.pci-0000_00_08.0.analog-surround-40 channels=1 channel_map=mono
and then restarted pulse like this:
systemctl --user restart pulseaudio
If you want something that turns on and off depending if you are using
earphones, you might want to just put the above pactl command in a script
with a fancy button or something.
If you sometimes use this computer at a desk, you could also get external speakers with their own USB sound card, and then plug into them when you are at the desk, and use headphones when you are on the go.
When touchpads have the buttons next to them, some people like to enable a tap feature where they can just tap on the touch pad, and it acts like a click. However, with these mac touchpads, is totally integrated. It's like tapping, except you need to tap a bit harder, until it clicks. So the tap feature is redundant:
So just make sure "Enable mouse clicks with touchpad" is unchecked.
In that configuration window, underneath the red circle, there are some two finger and three finger right and middle click emulation checkboxes, covered here
I haven't worked out what I am going to do without SysRq/PrtScn yet.
Go here for my brainstorm on the subject.
Sometimes when my computer goes to sleep or locks, and then I wake it, weird things happen on the external screen. I just unplug it and plug it back in.
However, sometimes when I plug in an external monitor, either because I am solving the problem I just mentioned, or for a normal reason, weird things happen like the dock starts floating in the middle of the screen.
These things sometimes work:
mate-panel --replace &
killall plank
plank &
disown
or
nohup plank &>/dev/null &
I have since added a keyboard shortcut to handle this by opening "Keyboard
Shortcuts" in the "start menu", and adding a custom one called plank. I have
set Shift+Ctrl+Alt+P to a script called ~/bin/plank.sh (except don't try to
put a ~ in the shortcuts settings - put you full home directory path)
The script looks like this
$ cat ~/bin/plank.sh
#! /usr/bin/env bash
killall plank
plank &
then when I wake up my computer, if the dock is floating in the middle of the screen, I just Shift+Ctrl+Alt+P and get to back to work.
My partner had a few deal breaker demands for Mac like UI. They might not be the same as your deal breaker demands for Mac like UI. It took some trial and error to find out what "being used to a Mac" and not wanting to change meant.
I made all the configuration changes to both computers so that I can fix the configuration issues on mine when possible, and just share the solution.
I switched to the Cupertino desktop, which has a dock, and puts a lot of things where they are on a Mac. You can find it in MATE Tweak.
I am not doing anything to reconcile the differences between ⌘ and Ctrl placement on mac and normal keyboard, and the differences between their function in OS X and Linux. I just use the key for what it does in that OS, and accept where it is placed on the keyboard that I am using.
I have shared a few more thoughts on this here
The way this work by default, is that you click in the lower right hand corner of the trackpad to right click, and in the lower middle to middle click. There are like.. zones... clicking areas.. (which I guess work as tapping areas too if you have tapping on, but I turned off tapping here)
I guess it is meant to work like external buttons along the bottom of a trackpad do. It's a good idea, except if you don't remember where the virtual bottons are you do stuff like middle click on browser tabs you are trying to activate and close them.
The alternative scheme is click with two fingers for right click, and three fingers for middle click. I think this is a little more like Mac OS.
You can turn it on in mouse settings like this:
Somehow that setting didn't work as expected the first time I tried it, or I thought by the menu structure that it would turn on multi-finger mode without turning off the clicking areas, so I found a much harder way to accomplish the same thing which is now documented here
Another potential deal breaker was this button:
The feature in OS X used to be called Exposé but now I guess it has a new name.
You press it, and it shows all your windows. In normal Ubuntu (Gnome 3) when
you press ⌘ (or the windows button) you get the application launcher with
something similar, and in KDE there is something similar too.
To get this in MATE, I had to install "compiz" and then use a feature called Scale.
I think these are all the compiz packages I have installed:
compiz-core
compiz-gnome
compiz-plugins
libcompizconfig0
python3-compizconfig
In the MATE Tweak control panel, under Windows there is a button that says "OPEN CCSM". Then under Window Management, there is something called Scale. I activated that and whatever it makes me activate at the same time.
Once it is activated, you should already be able to use it with the default keys. ⌘-w and ⌘-SHIFT-w - one of them shows all the windows in all workspaces, and one of them just in this workspace. I only usually use one workspace, so if I can't find a window, it could have slipped into another workspace, so I mapped the Exposé key to "Initiate Window Picker For All Windows"
However, that was a bit tricky too...
I tried to set the keyboard shortcut for the Exposé key. It puts the code
LaunchA into the settings interface when I try to detect it. However, it
doesn't save properly.
As a workarouind, I used xmodmap to switch the code for the Exposé key to
F20 (from I128) which I am not using, since even my extended keyboard only
goes up to f19.
Then I used the MATE control panel to detect and configure that key again. There
is more about how this all seems to work here. But
basically, I found out with xev that the Exposé key is 128.
I don't know where the code LaunchA that the interface displays comes from.
because the code that points at it seems to be I128:
$ grep 128 /usr/share/X11/xkb/keycodes/evdev
<I128> = 128; // #define KEY_SCALE 120
alias <I136> = <STOP>; // #define KEY_STOP 128
but you can see from the comments that they ported it from somewhere that called it "SCALE". I wonder if that is a coincidence.
Anyway, I created a file called ~/.Xmodmap that looks like this:
keycode 128 = F20
and then activated it like this:
$ xmodmap ~/.Xmodmap
Then you should be able to try to assign that key to the window picker again in the Scale settings, and it should show F20 as the key that you pressed, and save it properly.
To make this change stick, I added a MATE startup application with the
xmodmap command. However, I learned the hard way that you can't use ~ in
the command. You have to put the full path to your home directory.
Do not do this:
Use the full path to .Xmodmap
Also, the file can really be called anything you want I think. You are just
calling it with xmodmap. I think .Xmodmap is a tradition. You could have
different startup scripts for different xmodmap operations on different files
that served different purposes.
One issue with this change is that if I am using a non mac keyboard, which I often do, I can't use this feature. However, I never used this feature before anyway, so if I get hooked on it now that I have had to configure it for somebody else, I'll rethink my key combinations for when I am using a PC keyboard, or remap another key!
I was having some strange issues when plugging in and unplugging my second screen, and wondered if the problem was Compiz. The only reason I had it on was for "Scale", and as I am only an occasional exposé user, I searched for a simpler solution.
Also, while investigating the "tilde" bug I have learned a bit more about key mappings, and don't feel as comfortable using xmodmap if I don't have to - more on that some other time.
I found that years ago, somebody created something called skippy-xd. I think it was "famous" in its day, but is not longer maintained, and if it was packaged with distros before, it isn't anymore.
Somebody named dreamcat4 has forked it, and is maintaining it
here.
I don't know how I am supposed to make sure that it isn't spyware. So I just
took a leap.
It's a bit clunkier than Compiz's "scale", but it has other advantages like not having to remap keys.
I think this is what I did to install it. This is the version I used:
~/skippy-xd$ git rev-parse HEAD
181b143e225a993b60b255d2a0548ca588c14f88
I might not have "installed" it. looking at it, my ~/bin directory has
symlinks straight to the directory cloned repo:
$ ls -l skippy*
skippy-xd -> ~/skippy-xd/skippy-xd
skippy-xd-runner -> ~/skippy-xd/skippy-xd-runner
(the above output has been abbreviated)
So you might be able to just compile it, and then make your own symlinks. I wish I had written this down immediately.
There is a chance that I never saw the above installation instructions from the wike, and just compiled it, and improvised.
So either way, once you have skippy-xd and skippy-xd-runner in your path,
use Ubuntu MATE's Startup Applications to start skippy-xd-runner --daemon
at startup. Then use Ubuntu MATE's Keyboard Shortcuts to point the Exposé
key at skippy-xd-runner --expose
Unlike the Compiz shorcuts, LaunchA works just fine, so there is no need to remap the Exposé key to F20. If you still have it mapped, it will work, but it is better to remove the mapping first.
I tried running skippy-xd in a few other ways - like not starting the daemon,
and just running the application (just skippy-xd I think) and it seemed to
work but got confused when I pressed the key more than once.
The final deal-breaker was swiping the keypad to show the desktop.
I accomplished something close enough using touchegg
I don't know if compiz is required.
Don't use the version of touchegg that comes with Ubuntu; it's totally
different. Or do use it, but don't try to follow these instructions, unless you
are reading this in the future, when the one I am using starts to come with
Ubuntu.
I am using this version:
$ touchegg -v
Touchégg v2.0.16.
As it says in the instructions, add this PPA, and install:
$ sudo add-apt-repository ppa:touchegg/stable
$ sudo apt update
$ sudo apt install touchegg
(Actually that's not how I actually installed it on my second computer. Read about what happened here)
He says to configure it using an interface called Touché, except I couldn't figure out how to get Touché configured, so I had twice the problems.
So following the instructions, I created a file called
~/.config/touchegg/touchegg.conf with this in it:
<touchégg>
<settings>
<property name="action_execute_threshold">10</property>
</settings>
<application name="All">
<gesture type="SWIPE" fingers="3" direction="RIGHT">
<action type="SHOW_DESKTOP">
<animate>true</animate>
</action>
</gesture>
<gesture type="SWIPE" fingers="3" direction="UP">
<action type="SHOW_DESKTOP">
</action>
</gesture>
</application>
</touchégg>
Apparently the Mac OS swipe to show desktop is up and to the right, so this
just makes both work - UP and RIGHT. Also, I am not sure about exactly how
many fingers you need on a mac, but this is close enough and easy enough to
learn for somebody who has already made a habit out of the other gesture I hope.
I lowered the execution threshold this machine since it was not always working.
The other machine works fine with the default of 20. But actually, I think it
just started getting faster when I restarting my computer, and had the client
running in the background.
I made one direction animate and one direction not, to see if I could tell the
difference, but I can't.
You can now test it by running the touchegg client in the foreground:
$ touchegg
Touchégg v2.0.16.
Starting Touchégg in client mode
Parsing your configuration file...
Using configuration file "~/.config/touchegg/touchegg.conf"
Configuration parsed successfully
Connecting to Touchégg daemon...
Connection with Touchégg established
And while it is running, I can swipe three fingers across the trackpad either up or to the right, or both, and the desktop becomes visible, and then swipe the same direction again, and see the windows again.
Now, to turn it on permanently, there is a .desktop file that comes with the
package. The client should just turn on the next time you start.
$ dpkg -L touchegg | grep desktop
/etc/xdg/autostart/touchegg.desktop
and that is already placed in the hidden section of Startup Application. In
fact the startup application I added for xmodmap before actually just creates
one of these desktop files.
I got this awesome Penguin sticker off Etsy from a company called Denver Decals It looks even more awesome when the light is lit, but my camera probably wouldn't do it justice.
Well I am using the Macbook Air to write this document in emacs and it, use firefox for research, and previewing, and it seems to work pretty well.
UPDATE: I have been using the 2013 11" Macbook Air for some time now as my main home computer at a desk, and as my tiny travel computer that fits in any bag and am totally happy with it.
My partner has been using the 2009 Macbook Pro a browsing on the couch computer as well, and while not as excited about the whole thing as me, has no problem with it (that I can't solve). It's great that this nice piece of hardware with a fantastic giant screen has been given another life.