Ciro Santilli OurBigBook.com's questions

I've tried the procedures from:

• Cgroups error: cgroup change of group failed
• How to use cgroup in ubuntu 13.04
• https://unix.stackexchange.com/questions/197718/does-managing-cgroups-require-root-access

but they don't work on Ubuntu 22.04, presumably because of the update to cgroups v2:

sudo cgcreate -a $USER:$USER -g memory:myGroup -t $USER:$USER
sudo cgset -r memory.max=500M myGroup
sudo cgset -r memory.swap.max=0 myGroup
cgexec -g memory:myGroup id

fails with:

cgroup change of group failed

It works if I run with sudo

sudo cgexec -g memory:myGroup id

but then the command runs as root, and I want it to run as the current user instead.

Edit: reported this at: https://bugs.launchpad.net/ubuntu/+source/wayland/+bug/1931387 as well, and it was found to be a duplicate of: https://bugs.launchpad.net/ubuntu/+source/libdrm/+bug/1497593

I was running 20.10 previously with NVIDIA drivers,

Then four days ago, I did a clean 21.04 install from ISO without installing the NVIDIA drivers, and since then the UI has frozen three times on me.

I am unable to determine exactly what action triggered the freezes, I was just basically using Chromium browser normally in most/all cases.

The third time, I was just looking at a webpage, I wasn't even touching my mouse at that exact moment, and then it just froze.

I am then unable to see anything move on the screen, including after Ctrl+Alt+Fn, the last image just stays fixed.

The third time was also a bit different.

A few seconds after freezing, it actually went down to a black terminal screen as during boot messages:

and then I was redirected to the login screen, as if I had been logged out.

I was able to login, my programs were closed, so I was effectively logged out.

Then two Apport windows showed up, and one of them contained several references to Xwayland. Under /var/crash/_usr_bin_Xwayland.1000.crash I see one of them:

• ProcCmdline: /usr/bin/Xwayland :0 -rootless -noreset -accessx -core -auth /run/user/1000/.mutter-Xwaylandauth.48ZQ40 -listen 4 -listen 5 -displayfd 6 -initfd 7

• StackTraceAddressSignature contained mentions of Xwayland and nouveau_dri.so:

  StacktraceAddressSignature: /usr/bin/Xwayland:6:/usr/lib/x86_64-linux-gnu/libc-2.33.so+864:/usr/bin/Xwayland+13cb40:/usr/bin/Xwayland+14866f:/usr/bin/Xwayland+13b1a7:/usr/lib/x86_64-linux-gnu/libc-2.33.so+864:/usr/lib/x86_64-linux-gnu/libc-2.33.so+749:/usr/lib/x86_64-linux-gnu/libc-2.33.so+123d6:/usr/lib/x86_64-linux-gnu/libdrm_nouveau.so.2.0.0+25d7:/usr/lib/x86_64-linux-gnu/libdrm_nouveau.so.2.0.0+2537:/usr/lib/x86_64-linux-gnu/libdrm_nouveau.so.2.0.0+265f:/usr/lib/x86_64-linux-gnu/libdrm_nouveau.so.2.0.0+2a77:/usr/lib/x86_64-linux-gnu/libdrm_nouveau.so.2.0.0+31aa:/usr/lib/x86_64-linux-gnu/dri/nouveau_dri.so+93b943:/usr/lib/x86_64-linux-gnu/dri/nouveau_dri.so+93bc61:/usr/lib/x86_64-linux-gnu/dri/nouveau_dri.so+17cc01
  

I also checked and I am using Wayland (the default since 21.04 apparently): How can I tell if I am running Wayland?

Hardware:

• Lenovo ThinkPad P51
• NVIDIA Quadro M1200 4GB GDDR5 GPU

Just after the crash, dmesg contained the following possibly interesting lines:

[ 9556.450894] nouveau 0000:01:00.0: gr: TRAP ch 3 [00ff8cf000 Xwayland[3831]]
[ 9556.450907] nouveau 0000:01:00.0: gr: GPC0/TPC2/MP trap: global 00000004 [MULTIPLE_WARP_ERRORS] warp 3c000d [OOR_REG]
[ 9556.454264] nouveau 0000:01:00.0: gr: TRAP ch 3 [00ff8cf000 Xwayland[3831]]
[ 9556.454283] nouveau 0000:01:00.0: gr: GPC0/TPC0/MP trap: global 00000004 [MULTIPLE_WARP_ERRORS] warp 3e000d [OOR_REG]
[ 9556.466339] nouveau 0000:01:00.0: gr: TRAP ch 3 [00ff8cf000 Xwayland[3831]]
[ 9556.466351] nouveau 0000:01:00.0: gr: GPC0/TPC3/MP trap: global 00000004 [MULTIPLE_WARP_ERRORS] warp 3e000d [OOR_REG]
[ 9698.230726] nouveau 0000:01:00.0: Xwayland[3831]: failed to idle channel 5 [Xwayland[3831]]
[ 9713.230743] nouveau 0000:01:00.0: Xwayland[3831]: failed to idle channel 5 [Xwayland[3831]]
[ 9713.231918] nouveau 0000:01:00.0: fifo: fault 00 [READ] at 0000000000056000 engine 07 [HOST0] client 06 [HUB/HOST] reason 02 [PTE] on channel 5 [00ff0d9000 Xwayland[3831]]
[ 9713.231930] nouveau 0000:01:00.0: fifo: channel 5: killed
[ 9713.231932] nouveau 0000:01:00.0: fifo: runlist 0: scheduled for recovery
[ 9713.231939] nouveau 0000:01:00.0: fifo: engine 5: scheduled for recovery
[ 9713.231943] nouveau 0000:01:00.0: Xwayland[3831]: channel 5 killed!
[ 9717.642691] nouveau 0000:01:00.0: fifo: SCHED_ERROR 0a [CTXSW_TIMEOUT]
[ 9717.642702] nouveau 0000:01:00.0: fifo: runlist 0: scheduled for recovery
[ 9717.642709] nouveau 0000:01:00.0: fifo: channel 3: killed
[ 9717.642714] nouveau 0000:01:00.0: fifo: engine 0: scheduled for recovery
[ 9717.643113] nouveau 0000:01:00.0: Xwayland[3831]: channel 3 killed!
[ 9719.351714] apport-gtk[119654]: segfault at 18 ip 00007f337637d194 sp 00007ffd95094dc0 error 4 in libgtk-3.so.0.2404.21[7f3376274000+385000]
[ 9719.351721] Code: c4 08 5b 5d c3 90 f3 0f 1e fa 48 8b 7f 10 48 85 ff 74 0b e9 ce c6 ff ff 66 0f 1f 44 00 00 48 83 ec 08 48 89 d7 e8 0c 36 17 00 <48> 8b 40 18 48 8b 78 10 e8 ef 03 09 00 48 83 c4 08 48 89 c7 e9 a3
[ 9744.187533] rfkill: input handler enabled
[ 9745.940823] rfkill: input handler disabled
[ 9775.737572] rfkill: input handler enabled
[ 9777.953055] rfkill: input handler disabled

Related:

• Ubuntu 21.04 Wayland, GIMP black window GIMP works fine
• Blackscreen - failed to idle channel contains one of the key dmesg messages I saw similar to "Xwayland[3831]: channel 5 killed!"

Fuller logs at: https://gist.github.com/cirosantilli/f679b17f5e18253036bf58fc3ff8690d

Okular 20.8 and earlier has a feature that I miss badly: the ability to search for results only in the header titles, AKA "Outline" tab in the Evince UI.

But in Evince 3.38, I don't see any search box specific for the Outline, and if I do the using Ctrl + F, it just opens a regular full text search that finds hits anywhere in the document, not only in headers.

In Okular, as I start typing the search box of their outline, it filters only the headers which match the search string, which is very handy.

Created a feature request for it at: https://gitlab.gnome.org/GNOME/evince/-/issues/1555

Tested on Ubuntu 20.10.

This is a 20.10 regression that did not happen in 20.04. It is a graphics problem, not about setting the actual time.

Sometimes if I leave my computer idle, when I come back to it the clock is out of date.

Or sometimes it shows two digits superimposed one on top of the other. This strongly indicates that it is a graphics issue.

Then if I start messing with windows/clicking around it soon updates to the correct current result.

This can be very misleading, and lead to being late for an appointment.

Is there a bug report/fix on the way, and if not what's the best place to report it?

Lenovo ThinkPad P51, NVIDIA Quadro M1200 GPU, NVIDIA driver version 460.32.03.

At How to run Ubuntu 16.04 Desktop on QEMU? I have described in detail how to boot and install the desktop Ubuntu image on QEMU.

Now I wanted to do a similar setup for the server image to try keep things more minimal:

wget http://releases.ubuntu.com/18.04/ubuntu-18.04.1.0-live-server-amd64.iso

I have installed the server image with the GUI previously, and the setup seems to be text based, so I thought it might work.

But then, if I just add the usual:

-nographic -serial mon:stdio

as in:

qemu-system-x86_64 -cdrom ubuntu-18.04.1.0-live-server-amd64.iso \
  -drive file=ubuntu-18.04.1.0-live-server-amd64.img.qcow2,format=qcow2 \
  -enable-kvm   -m 2G   -smp 2   -vga virtio -nographic -serial mon:stdio

I just never get anything on the terminal serial.

I also tried to mount the image to see what was inside:

mkdir mydir
sudo mount ubuntu-18.04.1.0-live-server-amd64.iso mydir

to see what was inside, and:

cat mydir/boot/grub/grub.cfg

indicates that there is no console=ttyS0 option on the command line, which might explain why the terminal is empty.

I can't edit the ISO file however without first going into a GUI session, as it is mounted readonly.

I have already used the Cloud image previously, and it is great: Is there any prebuilt QEMU Ubuntu image(32bit) online? but I can't use it this time because I need GRUB for something (edit: nevermind, the cloud image does have GRUB, just it is hidden by default, see: How to get to the GRUB menu at boot-time using serial console? )

Related: How to get to the GRUB menu at boot-time using serial console? I managed to get GRUB to show on the terminal after changing the GRUB configuration with the GUI as shown in that thread.

Tested on an Ubuntu 16.04 host.

The project page http://libguestfs.org/ claims that we can do operations without sudo, which is awesome.

However, if I try it out:

sudo apt-get install libguestfs-tools
mkdir sysroot
# Just a test file.
dd if=/dev/urandom of=sysroot/myfile bs=1024 count=1024
virt-make-fs --format=ext2 --type=qcow2 sysroot sysroot.qcow2 

I get the error:

libguestfs: error: /usr/bin/supermin exited with error status 1.
To see full error messages you may need to enable debugging.
Do:
  export LIBGUESTFS_DEBUG=1 LIBGUESTFS_TRACE=1
and run the command again.  For further information, read:
  http://libguestfs.org/guestfs-faq.1.html#debugging-libguestfs
You can also run 'libguestfs-test-tool' and post the *complete* output
into a bug report or message to the libguestfs mailing list.

How to solve that?

Tested in Ubuntu 16.04.4, libguestfs-tools 1:1.32.2-4ubuntu2.

I'm interested in the official Ubuntu server ARM image: https://www.ubuntu.com/download/server/arm

How to properly cross compile for it from my x86_64 host?

While a gcc-5-aarch64-linux-gnu C hello world compiled from my 16.04 x86_64 host did work on the 16.04 ARM guest, I don't think it is simple to guarantee that the libc will be binary compatible across Ubuntu versions.

Is it documented anywhere what is the correct cross compiler to use?

I could only find server ARM images at: https://www.ubuntu.com/download/server/arm

But are there any ARM images that come with a graphical desktop installed by default?

I don't have specific hardware requirements, even just running it on QEMU would be fine for now.

On Ubuntu 16.04, I've done:

# Go to TTY with Ctrl + Alt + F1
sudo apt-get install netsurf-fb
sudo usermod -aG 'audio,input,video' "$USER"
# Logout, login.
netsurf-fb

And it fails with:

Unable to set video: Could not set console screen info fbdev

sudo didn't help either.

netsurf version 3.2.

fbi and vlc did work on the other hand.

When I reopen Okular, I want to have the same windows / tabs as when I last closed it, usually at system shutdown / startup.

Is that possible in Ubuntu (not Kubuntu, running the standard desktop). 15.10 here, but all versions appreciated.

There seems to be a way to save the global KDE session, so that may be the way to go: https://bugs.kde.org/show_bug.cgi?id=300985

To reproduce:

seq 10000 | /usr/bin/xsel -b
/usr/bin/xsel -b

Observed output: either of:

• nothing
• the correct output
• the correct output + a few lines of trash at the end

The output changes randomly between those three even if I run xsel -b multiple times without resetting it with the pipe to stdin!

The input has to be large enough: seq 1000 does not work. The actual bytes don't seem to matter, only the input size.

Once I do seq 10000 | /usr/bin/xsel -b:

• xclip -selection clipboard -o is also corrupted, but in a different weird way
• pasting with Ctrl + V on a textarea seems to work normally

I do not reproduce this if I try to set the textarea with:

• Ctrl + C
• seq 10000 | xclip -selection clipboard

so it seems to be a problem in the way xsel handles the stdin input.

Using -i and -o did not change anything.

The problem started today "by itself".

If you reproduce, please add a comment below.

Ubuntu 14.04 AMD64, xsel 1.2.0.

The selected text does not have be Chinese characters, any character just disappears.

The problem happens both on:

• mouse drag
• ctrl + A

selection methods.

I installed the Chinese input through the simplest GUI method possible on the control panel.

This does not happen on Chromium, only Firefox.

Happens with Ubuntu 14.04, and Firefox on --safe-mode to turn off all addons.

Similar questions:

• Selected Text missing when press ctrl or alt , but for me it only happens on Firefox, and for any selection method (mouse drag, ctrl + a)

When I started Firefox today all background colors were gone as in the following screenshot:

This probably happened because I have set up automatic software updates, so something must have gotten broken.

What I've tried to no avail:

• start Firefox with --safe-mode to disable all plugins
• rested GNOME color themes to Ubuntu's default
• sudo apt-get update && sudo apt-get upgrade. This upgraded from Firefox 36 to 37.

I have obtained the sqlite source code with:

sudo apt-get install fossil
mkdir sqlite
cd sqlite
fossil clone http://www.sqlite.org/cgi/src/doc/trunk a
fossil open a
rm a

and compiled it with the instructions on the README. Now I want to run make test.

The README says that that requires Tcl development files. On Ubuntu 14.04 I run:

sudo apt-get install tcl8.6-dev

Then if I make test it gave tcl.h not found

So I located tcl.h and run:

CPATH="$CPATH:/usr/include/tcl8.6" make test

But now it gives:

/tmp/cc4jwHgX.o: In function `dbFreeStmt':
/home/ciro/git/sqlite/./src/tclsqlite.c:451: undefined reference to `Tcl_Free'
/tmp/cc4jwHgX.o: In function `closeIncrblobChannels':
/home/ciro/git/sqlite/./src/tclsqlite.c:189: undefined reference to `Tcl_UnregisterChann

 ...

I have tried:

cd /usr/lib/x86_64-linux-gnu
sudo ln -s libtcl8.6.so libtcl.so

with not success.

If I do:

vlc dir

it also plays files like images for 10 seconds. Is there a way to prevent that and play only the music?

I also have music files on arbitrary subfolders: dir/jazz, dir/classic, dir/jazz/miles-davies, etc.

I can do it with find and some bash-fu, but I'd prefer a simple vlc solution, or some other very simple solution that does not involve coding a complex script, in particular deciding what is a music file or not.

I have seen: How to play music files directly from folders? , but it is not VLC specific.

Also opened on the bug tracker: https://trac.videolan.org/vlc/ticket/12819

If I understand https://help.ubuntu.com/community/Repositories/Ubuntu correctly, both are not free and not supported by Canonical. So what is the difference between them?

Where can I get a list of the partners and their repositories? Is there an official (non wiki) documentation from Canonical on this subject?

Specifically:

1. Can multiverse software be closed source? If not, how and why can non free software be open source?

2. What are "licensing requirements"? An example would be great.

3. Closed source does not imply proprietary?

4. How is proprietary different from non free?

5. Partner software gets support from whom? Enterprises which have contracts with Canonical stating they will provide support? Is this what partners are?

How can I determine if a package is a meta-package from the command line, possibly via apt-get, aptitude or apt-cache?

I have tried:

apt-cache show texlive-full
apt-cache showpkg texlive-full

but the only way I can tell this package is meta is by reading the "en-description" field.

Is there a more automatic way of doing this, that will give me a yes/no response, or at least have a field such as then "en-description" dedicated to this?