The Linux 6.7 Merge Window Is Massive With Many New Features

[oleid]

I find it pretty sad that the in-kernel console is being removed. There should be a way to interact with the kernel even if userspace blows up.

Userspace can get restarted.

[DarkFoss]

Hopefully CoreCtrl will work correctly with the 7900 cards.

https://gitlab.com/corectrl/corectrl/-/issues/344

[rogerx]

I find it pretty sad that the in-kernel console is being removed. There should be a way to interact with the kernel even if userspace blows up.

The only reason in-kernel or virtual terminals will be removed is if Wayland/systemd developers and/or RedHat/Ubuntu demand it's removal.

Now with that stated, I'm pretty sure there will be more serious efforts for virtual terminals complete removal, since it's a successful competing function with Microsoft Windows. ( ... and we cannot have Linux looking and working better than Windows!)

As for I, I'm always resorting to virtual terminals on a daily bases, for working around HDMI bugs and all things recently created within the past decade or so. Else, I would reboot like Windows.

If I could, I'd live entire days within a virtual terminal. I used to do so, a decade ago! Currently, I run graphical desktops atop of the virtual terminals, granted most other Linux users do so too, but with the terminal interface hidden from typical users.

[darkbasic]

Did the AMD Preferred Core patches made their way into 6.7?

[Quackdoc]

The only reason in-kernel or virtual terminals will be removed is if Wayland/systemd developers and/or RedHat/Ubuntu demand it's removal.

Now with that stated, I'm pretty sure there will be more serious efforts for virtual terminals complete removal, since it's a successful competing function with Microsoft Windows. ( ... and we cannot have Linux looking and working better than Windows!)

As for I, I'm always resorting to virtual terminals on a daily bases, for working around HDMI bugs and all things recently created within the past decade or so. Else, I would reboot like Windows.

If I could, I'd live entire days within a virtual terminal. I used to do so, a decade ago! Currently, I run graphical desktops atop of the virtual terminals, granted most other Linux users do so too, but with the terminal interface hidden from typical users.

I doubt that would be enough grounds to remove it.

I find it pretty sad that the in-kernel console is being removed. There should be a way to interact with the kernel even if userspace blows up.

I not sure anyone wants to remove them, I havent seen anyone advocate for that, disabling it by default however is for sure something people want

[varikonniemi]

Userspace can get restarted.

How is that done if init is dead and you cannot launch another one using kernel console?

[oleid]

How is that done if init is dead and you cannot launch another one using kernel console?

Technically, the kernel could launch a userspace process.

[ehansin]

Thanks everyone for the VT discussion, I'm learning stuff here. I read the Reddit post and comments. Sounds like differences in opinion, different wants and needs. Decisions here are above my pay grade, so nothing in my control. But at least I now understand things better whatever happens.

[varikonniemi]

Technically, the kernel could launch a userspace process.

Yes, and have the machinery to specify that process. And before that, detect that userspace is fucked/stalled. And mop it up before spawning new.

When alternative is to have a simple (or extensive) console in kernel that can always be accessed unless the kernel has experienced memory corruption.

I wonder what kind of Linux using person has never had to switch to kernel console to rectify things going wrong in userspace.

[oleid]

Yes, and have the machinery to specify that process. And before that, detect that userspace is fucked/stalled. And mop it up before spawning new.

Well, it could be a statically linked binary that the kernel manages. I mean, how do micro kernel operating systems handle it.

When alternative is to have a simple (or extensive) console in kernel that can always be accessed unless the kernel has experienced memory corruption.

Consoles are less simple than one thinks. And they provide an interface to the outside world. That's better hardened.

I wonder what kind of Linux using person has never had to switch to kernel console to rectify things going wrong in userspace.

Oh, I regularly use the console. But if things really go wrong, I'm using a serial interface.
Userspace is typically not the problem. And I've never seen PID1 crash.