Sam Van den Eynde's questions

I rebuild the kernel based on the git sources. These are the steps that I take:

1. git clone git://kernel.ubuntu.com/ubuntu/ubuntu-focal.git
2. cd ubuntu-focal
3. vi debian.master/changelog
4. apply patches
5. fakeroot debian/rules clean binary-headers binary-generic

If I then apply another patch and want to compile the same version again, I get an error because certain folders are not empty. I can issue 'fakeroot debian/rules clean' to solve that, but then I need to compile the whole kernel.

Is there a way to recompiling only the changed files and build the debs based on that? It seems a trivial thing I just haven't figured out yet.

If I take a snapshot from my root filesystem (so an identical copy), mount it, chroot into it, and try to upgrade it using do-release-upgrade, the process fails. I have seen for example it's complaining about /boot/efi not being mounted, but of course I don't want it to mess with that (yet).

Any ideas? I am thinking about this to stage upgrades at home for remote family members, could save me a lot of time when a new LTS comes along.

I have followed multiple posts here as well as https://help.ubuntu.com/community/HdaIntelSoundHowto, but it seems issues with audio chipsets are often quite particular.

My system has an ASUS P10S-WS motherboard with 7.1 audio. I have stereo speakers connected to the line out, which is one of the output pins on the back (the lime one, or Front Speaker out when connected in a surround setup). The problem is that this gives mono sound only, so output on only one speaker. If I connect the headphone jack on the front of the computer, stereo output is ok, so the speakers work.

The system is a regular Ubuntu 18.04, latest available updates and kernel installed (4.15.0-23-generic).

My current actions and results:

• I checked and double checked the correct profile is loaded, using sound settings as well as pavucontrol. I have tried both stereo out and stereo duplex.
• I have tried loading the module specifying different models as described in the guide above, but unfortunately the ALC1150 is not available. I have tried several ASUS profiles and some generic ones. Some models disabled sound, so I know at least these actions must have changed the driver configuration.
• I have used hdajackretask to remap the output pin. This was the only action so far that gave a result: I was able to get output on the second speaker, but is was heavily distorted. I must admit I'm not really sure what I'm doing here, I have tried to remap it to various line out and headphones entries in the drop down list.

Some details from alsa:

pactl list cards

Card #3
Name: alsa_card.pci-0000_00_1f.3
Driver: module-alsa-card.c
Owner Module: 10
Properties:
    alsa.card = "0"
    alsa.card_name = "HDA Intel PCH"
    alsa.long_card_name = "HDA Intel PCH at 0xdf620000 irq 160"
    alsa.driver_name = "snd_hda_intel"
    device.bus_path = "pci-0000:00:1f.3"
    sysfs.path = "/devices/pci0000:00/0000:00:1f.3/sound/card0"
    device.bus = "pci"
    device.vendor.id = "8086"
    device.vendor.name = "Intel Corporation"
    device.product.id = "a170"
    device.product.name = "Sunrise Point-H HD Audio"
    device.form_factor = "internal"
    device.string = "0"
    device.description = "Built-in Audio"
    module-udev-detect.discovered = "1"
    device.icon_name = "audio-card-pci">

cat /proc/asound/card0/codec* | head

Codec: Realtek ALC1150
Address: 0
AFG Function Id: 0x1 (unsol 1)
Vendor Id: 0x10ec0900
Subsystem Id: 0x1043868f
Revision Id: 0x100001
No Modem Function Group found
Default PCM:
    rates [0x5f0]: 32000 44100 48000 88200 96000 192000
    bits [0xe]: 16 20 24
    formats [0x1]: PCM

I'd just like to know how this works and what I can do about it. Any takers?

If I have two folders E en D and a passphrase P, in a typical ecryptfs setup I would something like this:

• I setup folder E to be the encrypted version of folder D, using passphrase P to do the encryption/decryption.
• Folder E stores the actual data to disk, encrypted.
• Folder D is my live interface as a user, representing the data to me, decrypted on the fly.

But, can I also use this the other way around?

• I still set up folder E as the encrypted version of folder D, using passphrase P as the key for encryption.
• However, I interact directly using folder E. Folder E contains from my perspective as a user, my regular "decrypted" data.
• Folder D considers E as being encrypted, and represents a live view of that data "decrypted" using passphrase P, actually encrypting it in the process.
• I would be able to use folder D as the primary source for a backup of my data in folder E to a public cloud provider.

Any thoughts?

I would like to understand what the difference is between getting a new base image (e.g. Ubuntu 16.04) and apt updating running containers.

Do I need to update the containers myself, or can I rely on the image being updated? Or is that decoupled as soon as the container gets created?

I've been updating them myself so far.

I'm planning on running a desktop system with 2 GPU's on one monitor: am I correct that GPU switching is not available in this scenario? So I'll have to boot either using the Intel GPU (but I'll be able to use the NVidia for CUDA stuff/rendering or VGA pass-through to VM) or using the NVidia GPU (for Steam)?

Or is Optimus/Bumblebee also available on the desktop, driving the monitor using one cable with the cards "automagically" aware of one another?

What are your set-ups/experiences?

I'm looking into possibly getting a Lenovo Yoga Tablet 2. This is a pure tablet device, that comes with either Windows or Android. It has an Intel cpu.

Was anybody able to install Ubuntu on either one of those 2 flavors? What was the experience like? I don't mind using Ubuntu Desktop or Touch (or Ubuntu Next).

I am looking for a consistent way to configure cgroups on a Ubuntu 14.04 Desktop system with limited resources.

What I would like to do:

• create a few groups with limited memory and cpu assigned for specific tasks
• auto-enable these groups on reboot
• assign certain commands/programs to each group

I understand cgroup configuration, and how to start processes for a specific group using cgexec. The problem is the next step.

I have build a working /etc/cgconfig.conf file, that I am able to parse and activate successfully using cgconfigparser to get configured cgroups. However this is as far as I get.

• I don't find a service to read and parse /etc/cgconfig.conf correctly on system boot. The current cgroup packages don't seem to include such a service.
• As I understand /etc/cgconfig.conf is accompanied by a /etc/cgrules.conf file, that described how processes are started by default in a certain group. I can't identify the service or mechanism that parses it. Without this functionality I need to start all processes using cgexec to get them in a cgroup.

I have tried to use cgmanager, but that service does not seem to configure from /etc/cgconfig.conf either. In general I find the information how to configure and use cgroups on Ubuntu Desktop systems hard to locate.

So, is anybody able to point me in the right direction? What service do I need to install, and what config files does it obey?

Just to be clear: I am not looking for containers. It is to structure resource handling on a single dedicated system.

Many thanks :)