Comfort vs Autonomy

>>58769263
the problem is that if you give Google, Microsoft, etc an inch, they'll take a mile. Leave them any loopholes and they'll pry them open and drive trucks through them. So essentially autonomy (and privacy, for that matter) is a binary yes/no choice. If you want it, any of it, you have to put up with the inconveniences of using Linux and avoiding Google services. Its not really a spectrum where you can trade off one for the other a bit at a time. Companies these days make very convenient products but also work very hard to violate your privacy and bind you tightly into their "ecosystem", because that's the revenue-maximizing strategy.

If you have choices, you have more work to do picking among them, inherently. The way that companies make things convenient is by removing choices. You can't have it both ways.

>>58769263
there is a reason that business laptops are endlessly praised here on /g/, especially thinkpads

what other laptop can you install linux on and have it instantly work, including the function keys?

>>58769263
For the GPU passthrough thing, first off that's not something to do on your laptop if that's where you're trying.
Second, you can't pass through certain PCI-e slots. I can't remember what defines that, but usually the top one on a motherboard, the one most likely colored differently, can't be passed through.
Also, you really should be using vfio-pci rather than pci-stub by now. Soon it should enable things like binding the GPU to host and quest systems and using them on either without a reboot or even X restart in between.

The backlight issue is most probably caused by missing ACPI drivers. See if simply

 tee /sys/class/backlight/intel_backlight/brightness <<< 2000

dims the screen.

>>58769417
I've installed linux on exactly one thinkpad. The wireless killswitch didn't work, so the laptop was useless with linux.
That was some time ago already, though. Maybe you could've turned it on in the BIOS or something...

>>58769380
>the problem is that if you give Google, Microsoft, etc an inch, they'll take a mile.
Do you have any examples or news that really show this?

I mean, I can understand why Facebook and MS collect our "personal" data. They're profit-driven corporations, after all, and marketing + ads make tons of profit. Is it *actually* that bad?

Sure, you get the occasional "Yahoo yields data to the FBI" faggotry, but that isn't super common... or is it?
I mean, shit, even Apple (APPLE (R)(TM)) stood up, publicly, against the San Bernardino FBI iPhone backdoor scandal.

>>58769509
>"Yahoo yields data to the FBI" faggotry, but that isn't super common... or is it?
Well Microsoft, Facebook, Yahoo, Google, Apple, and others were all named in the Snowden leaks as being involved in some way in the PRISM program. They all later denied this in some fashion, but they also all have clauses in their user agreements etc that they will disclose pretty much anything in response to a "lawful court order". Which could well mean an NSL that they're gagged from talking about. If you don't want to be visible to mass surveillance (or targeted surveillance, but you probably aren't) then the only way to do that is to not let these companies collect your data to begin with. Even if they don't cooperate, which is doubtful, they're big targets because they have so much information.

More prosaically, why do you think Apple and Google both use the "App Store" model where there's one approved place to get software for their platform? Why do you think Microsoft is moving towards that with the Windows Store? They want to have a veto over what software you can and can't run. That's a loss of autonomy and a gain of convenience. If you want the autonomy back (the ability to run any software you damn well please on a device you own) you have to fight them on it. You not only give up convenience by not having a single source of apps, but they actively try to make it inconvenient to get there - because they make money taking a cut of app purchases and selling in-app ads.

Basically, the reason they offer you convenience is to make it comparatively harder for you to do things that don't make them money.

>>58769486
The backlight works correctly, it's simply the fn + key combo that does not work.

>For the GPU passthrough thing, first off that's not something to do on your laptop if that's where you're trying.
Interesting; why is this?

Also, I didn't know baout the "exclusive" PCI-e slots. Thanks for that.

>>58769688
Either the 'exculsive' PCI-e slots has to do with slot isolation, or it's simply a generation difference from 2.0 to 3.0.
I'm not sure really, I ran into the issue on my motherboard and plugging the GPU to another slot fixed it.

And about laptops, I'm kind of assuming most don't meet the requirements of having proper PCI-e implementations, having two GPUs and using them separately. Some virtualisation extensions might be missing from mobile chips as well.
Though, I'm not sure. What laptop model is this? Has anyone else made it work? Does it fill the requirements on paper?

I'll see if the Fn keys for backlight are registered as keycodes on my laptop real quick, then I'll sleep.

>>58769787
I've tried once, almost succeeded. I have a laptop with Intel igpu and nvidia dgpu. I used vfio-pci to isolate the GPU. As virt. software I used qemu. The drivers in the guest os installed just fine, modules loaded fine, lspci -k showed my dgpu and the loaded module. However, I was not able to ever start the X. I got too tired of that shit and deleted everything.

I've heard that even if your laptop has two GPUs they're still somehow connected to each other and the display is connected to that system. Sounds fucked up I know, don't know if it's true.

I believe many laptops with dual graphics don't really separate them at a hardware level, like a PCI-E slot is separated from an iGPU in a desktop. Instead, the iGPU is always in charge of the display. When the dGPU is engaged, the iGPU stays in charge, and switches to displaying what the dGPU renders and passes to it.

If I remember right this makes things easier hardwarily because then you only have to worry about powering down and starting up the dGPU, and the iGPU is just always there. It turns a "swap between two things" problem into an "add and remove one thing" problem, which is simpler.

>>58769877
Also, to get GPU pass through working you need:
VT-d support (my CPU did)
motherboard that supports PCI passthrough (mine did)
The GPU has to be in a separate iommu group (mine was)
The GPU also must support PCI lassthrough (I never checked that)

Arch wiki has plenty of information on all of the above

File: apple_unix_ad.jpg (401KB, 1976x1202px) Image search: [Google]

401KB, 1976x1202px

>>58769263
>Both autonomy and comfort are valuable
This is more or less how they used to sell Macs to graybeard *nix-using professionals in the early days of OS X. You would pay Apple to effectively be your sysadmin. Pic related.
>100% autonomy
In practical terms, if you want to run Linux, buy your hardware with this in mind. ThinkPads aren't just a meme. Having your hardware work out of the box increases both your comfort (no frustrating busywork with configuration/maintenance) and autonomy (no need to rely on wikis, forums, etc. to find the right configuration).

On a philosophical note, while you didn't define "autonomy", I think under any reasonable definition 100% software autonomy is simply unachievable. At some point, you will have to delegate, even if you go as far as starting your "software" stack not with Linux, but with a soft CPU core on an FPGA. So the counter balance to autonomy can be called "comfort", but it isn't just about the emotional comfort of avoiding frustrations in your daily computing. It's about being able to do any meaningful work at all within your limited lifespan and not starving while you recreate a large chunk of your field from scratch. Since you have to rely on other people's code, the best option you have is to choose the code you rely on wisely and choose hardware, protocols, etc. that have high-quality software that fits your use case behind them.

>>58769877
So it's a bumblebee laptop?
Wait, you were able to boot the quest os and run it successfully but couldn't start X?
Maybe you simply needed to write a new xorg.conf, if it had previously used a GPU it now couldn't find.

My brightness control keys show up as scancodes and keycodes, but it looks like yours is purely implemented in hardware.
There's this kid though, who seems to have a similar thing working in KDE
https://www.spinics.net/lists/platform-driver-x86/msg09060.html
It's some fujitsu laptop though.

>>58769966
>you didn't define "autonomy"
I think the best practical definition would be "the ability to do something other than things that make the vendor money".

>>58769787
The model I'm using is an Asus g550jk, and it's currently running Ubuntu 16.04.

I checked if my CPU + GPU allowed passthrough, but I did not check that for my mobo. My mistake. Furthermore. as far as my Google-fu has taken me, I have not seen any successful passthrough stories on this model.

File: 9d76bcdc84ec95c043d66e30da68b34b.png (4KB, 300x300px) Image search: [Google]

4KB, 300x300px

>>58770158
I can't find any success stories either, with a quick google of optimus laptop gpu passthrough.
It seems like your issues are mostly
1. The dgpu and igpu share the same framebuffer. The dgpu renders to the framebuffer of the igpu, so you'd either have to virtualise the igpu in software or manage to pass it to the virtual machine as well.
2. Drivers on windows. Especially given that we're dealing with Geforce, it's likely you'll run into error 43 or can't find working drivers at all.

However, if you can't isolate the dgpu with vfio-pci or pci-stub, that'll be the end of it.
If I recall correctly, you don't want to use the nvidia drivers or load any graphics drivers via early modesetting.

My thoughts are becoming incoherent, time for sleep.
Good luck with your laptop, I'm sorry I'm not much use.