It comes from somewhere in that thread. I just tried to attach it to this post but I keep getting a "page not found" error from the attachment manager. I'll try again after posting.

EDIT: No go! I had to do a bloody capcha and then was told that I don't have permission. I bet that's cloudflare sticking its dirty fingers in again. This forum is becoming unusable and it's a real pity, given the quality of its management team.

Thanks hazel.

I feel your pain

I've also been blocked from time to time.

Jan K's post #19 in this thread includes links to a gentoo HowTo plus an update for the kernel patch.

I'll check it out when I have some time.

-- kjh

p.s. don't waste too much time trying to link the code ... I believe I've got it.

I have a request. Can anyone who actually uses this scrolling tty patch please carry out the following test and report back:

At the console, give a command that produces a large number of pages of output very fast. I used a locate command, which just happened to produce a lot of output on my system, but dmesg (without any piping) would probably be more generally effective.

More information at https://www.linuxquestions.org/quest...us-4175711978/

for those who like console paging.....does hold down shift and press page up work for you?

some alternatives (some already discussed here)
https://stackoverflow.com/questions/...-of-a-linux-vm

Yes, it works, though there is a funny glitch in the presentation: some of the scrolled-back display is on the right side of the screen instead of the left side. But it's all there and all legible. This kernel bug only manifests when a lot of data is dumped to the console at once. The boot output, first from the kernel and then from the init scripts, comes more slowly so it doesn't trigger the crash.

Why not use screen?

It allows scrollback, and you don't have to patch your kernel to get it working.

That is an option for regular use, but doesn't help with boot messages which have scrolled away (which is where console scrollback would be most critically needed).

dmesg -H
And you get a colored output at no additional cost.

1 members found this post helpful.

If the system fails to boot, dmesg(1) is unavailable, and this is where access to boot messages is most critical.

I'm not trying to be difficult; this is the use-case previously described in this thread.

2 members found this post helpful.

Do you see this a lot?

I don't, but I'm not using a system with systemd on it.

1 members found this post helpful.

See post #17

You didn't answer my question.

I've seen it a few times, because of fiddling with nconfig and compiling new ones almost twice a week.
Only once or twice it was not my own fault, but the dependency system changed something which I failed to detect in time.
It was generally nothing serious, and I was alway able to revert because backups..

Made me smile.

I would like to get the hability to scroll back back because I do not have screen launched usually, and, well, I've been used to it. And it was simple.

I would not bother to compile my own kernel but would be happy to have the patch merged...

OK. I'll concede the point that if you're regularly compiling your own kernels, the lack of scrollback could be an issue... but if you're doing this often enough for it to be a problem, then surely you've had enough practice to know what to include in your kernel config to ensure that it boots?

I've not compiled a kernel myself in probably 10 years or more. I just don't see the need... life is too short! But thinking back to when I did do it, you'd have to really stuff it up badly for it to fail suddenly and early in the boot process.