Interestingly, I am running a 10th Gen Intel Iris Pro GPU and had no problems on these kernels, (Arch)/

There were minor issues on 5.5 at the start of the cycle when some patches were overlooked, but that's all fixed now.

I honestly don't experience more problems than with my MacBook, (especially after major upgrades).

Maybe the issue is people just keep defaulting to Ubuntu and equating their experiences there with Linux as a whole. Maybe it just isn't the best distro around.

I'd honestly go with EndeavourOS, Manjaro or if you want to stay in the Ubuntu ecosystem, maybe Mint. I know everybody always says just switch distros, but honestly Ubuntu hasn't been the best distro around for anything for over a decade.

I would be really surprised if Ubuntu had issues that Manjaro or Arch don't have. Debian and Ubuntu is usually pretty ahead on testing different hardware combinations. Most likely this is some hardware-specific issue and you don't own the same hardware as GP.

No doubt Canonical tests more, but there's something about the way they configure the kernel and packages that gave me more problems over the years than most other distros. In fact, the first Ubuntu install ever, back in '06 failed for me completely, while Mandrake worked.

I do appreciate that this is subjective, I just wished Ubuntu isn't the only choice people evaluated when trying Linux.

Ubuntu has been fairly solid for me, over several laptops and almost a decade of releases.

Freezes due to graphics drivers (on Intel) is something from ~3-5 years ago, I'd say.

And let's be realistic, when people are just trying GNU/Linux (for the first time), they will go for the most popular and user-friendly distro available. Or, at least, they should.

I've been using ubuntu since the first release. I have issues with the switch unity gnome plus the odd kernel issue. Battery performance also at times has been an issue.

I stick to several year old thinkpads for hardware and LTS releases though. Mostly I just want the OS to get out of the way.

Normally it just works though.

I am not using Ubuntu or an Ubuntu derivative. Thank god I am using NixOS, which makes it very trivial to switch between kernel versions including external modules such as ZFS.

When it comes to Arch, there is a long thread about this issue in the forums:

https://bbs.archlinux.org/viewtopic.php?id=250765

As I said, they forgot to include Intel GPU patches at the start of the 5.5 kernel series[1], that's been resolved, I personally haven't seen it elsewhere, but of course incompatibilities happen.

1 - https://www.phoronix.com/scan.php?page=news_item&px=Linux-5....

That’s strange, usually if you buy hardware with Linux in mind and use stable software packages then you have little to worry about.

Typically Intel GPUs were a safe bet across the board and you had to avoid some of the other ones.

I’m curious to read more into what went wrong, it seems to be across a few bug reports.

> Linux in mind and use stable software packages then you have little to worry about

Linux, and by that I mean all distros, have always had difficulties with basic things such as wifi or hibernation for the better part of the past twenty years.

It's baffling to me this is still an issue in 2020.

I don't understand why Linux still does not have reliable hibernation. Sleep requires working with BIOS and other hardware components, so I could understand that without proper drivers things might be shaky. But hibernate is hardware-agnostic feature so it should be rock-solid by now.

No, it's not really. There's ACPI, but a combination of wiggle room in the standard and vendors saying fuck you to any attempt at rigorous compliance testing means that there's no way to assure that any ACPI implementation will work with drivers the vendor didn't write. The whole state of the PC industry is "does Windows boot/run with this? It does? Ship it."

Even in Windows they have to workaround some stupid bugs from the OEM or ODM side.

Sounds kinda like most web development: if it runs on Chrome, it's good enough. (Guilty myself)

Perhaps System76 should provide Linus with a free laptop to see what the issues are from a Linux standpoint and get them fixed. I think he was using desktops mostly. Probably because of compile times.

Maybe economics can help.

When anyone else can resell your code, your work becomes a public good.

Public goods are underprovided in equilibrium.

I was reading about who supports i915 and it looks like it was started by a company Tungsten Graphics which got sold to VMWare in 2008. I'm curious if over time the people at VMWare working on it left or put less effort into the codebase (they did a big shakeup of the company in 2016).

If that's the case hopefully it gets more attention eventually. Often with OSS it's some random developer working at some big company supporting these various drivers, not just people in their spare time.

This will always be problem with all software that changes, especially ones supporting hardware that also constantly evolves.

Linux' hardware support for laptops is basically an inverted parabola. It starts off bad, then gets good, then goes bad again, as new bugs seep in that never get fixed (Wifi becoming unreliable, lockups during suspend/resume, audio issues, ...).

Fill in Windows and you get the same shit. I think it is often x86 and shitty drivers all around...

My thinkpads haven't really had issues. Had linux exclusively since 1999.

There's likely more than one bug, but this is the one I've followed as the most egregious experience I've had in quite some time, if ever, on Linux.

GPU hang on transition to idle https://gitlab.freedesktop.org/drm/intel/issues/673

I don't know whether this qualifies as a break with no regressions in user space rule. I didn't run into it on either 5.3 or 5.5. My understanding is the backport to fix on 5.4 would be too complicated. Therefore I wonder why not just revert the whole thing from 5.4? I don't know the answer to that. I gave up on it totally and moved to 5.5rc kernels, rather than deal with 5.4.

Given this fact of 5.4's behavior, I think it's a bad idea for a distribution to pick that kernel to use, if they can't do the backport themselves, and yet expect to support i915 hardware. Yep, it's an LTS kernel. Ordinarily it's reasonable to pick an LTS for a distribution, but not this time around. Maybe Ubuntu should have shipped a late 5.3.x and then updated to 5.4 once fixed? I don't know if it even has been fixed now in 5.4, although I haven't gotten any emails on that bug since March 10.