[sydney6]

I have a ThinkPad X1 Carbon Gen6, and indeed there is the "Linux" sleep-state option, which has been working for years just fine for me. I'm running a Linux-based OS. Before Lenovo added this option via bios update (via fwupd:), i indeed had to manually manipulate the DSDT to achieve S3 sleep.

[bluedino]

Even with Linux, sleep on my Thinkpads is still a coin flip

[pbnjeh]

sydney6 is referring to a Thinkpad BIOS setting that is labeled "Linux" but really pertains to enabled sleep/power states -- per Linus' video (the OP), this setting works just fine with Windows, too.

This is all described towards the end of the video, for those who don't want to watch the whole ~11 minute thing.

P.S. I should clarify that it works just fine with Windows settings hackery that was being used prior to BIOS implementors disabling or eliminating S3 in their BIOS's.

[sydney6]

It depends, this thinkpad has lpddr memory and if i suspend the system, it can indeed go for weeks (2-3) on a single charge on S3. I used to only update the truly necessary like firefox and the like on do reboots only every 50-60 days with a kernel upgrade and had never issues with multiple daily suspend/resume cycles execpt every 3-4 months, the wifi/bt controller wouldn't correctly load FW and needed reloading, that's about it. _But_, imho, this works only completely "naked", that is, i use a separate usb-c DP-alt-mode cable, a separate usb-c charging cable and keyboard/trackpad via bluetooth. I have had issues with Thunderbolt-Docks _and_ USB-C Docks (so far: Lenovo, Caldigit, Belkin, no-name), USB-C Monitors with integrated Hubs and NICS, USB-NICS, USB-Hubs, pretty much everything external connected, that is not onboard the machine itself. This works well, the rest is a complete disaster and i fear not completely Windows/Mac/Linux Kernel's fault.

[pbnjeh]

I have a CalDigit mini Thunderbolt 3 dock that provides two DisplayPort ports, a 1 Gbps Ethernet port, and a USB (2? I forget) port. I use it to connect two WQHD monitors, Ethernet, and a wired mouse.

It's a real crapshoot whether the dock will properly or fully connect upon cold startup or reboot. Sometimes everything is dead. Sometimes the monitors get signal but Ethernet and the mouse are down. Sometimes only the mouse is down; very occasionally, only Ethernet is down.

About a year ago, I spent multiple hours on the phone in a looong call with a CalDigit support rep. Essentially, we used my system to troubleshoot. We never solved the problem.

With time and observation, I've learned it's best to cold boot (with power connected, my daily work configuration) with this CalDigit Thunderbolt dock physically disconnected. After Windows has booted and gotten to its desktop, I then wait 90 - 100 seconds or a bit longer before I plug in the dock. Doing this, my success at having everything come up working is 50% or even a bit more.

If dock isn't fully functional, repeated warm rebooting is not too likely to correct this. The best bet is to hard shutdown, maybe wait a bit for whatever capacitors to discharge, and then start the above procedure over.

The CalDigit rep thought it was maybe the Thunderbolt support/configuration in my X1C6 that was/is to blame, but I have since formed the impression that Thunderbolt, for all its vaunted power, is actually kind of a shitshow.

The X1C6 did initially have a problem: It rewrote its Thunderbolt configuration upon every boot, into a kind of physically memory not designed to have a lifespan of that many writes. About a year, as I understand it, was what it took for many users to start having problems. Lenovo did eventually ship a firmware update to stop this behavior, and through circumstances I'll skip I ended up de facto avoiding the problem. My point in mentioning this being, maybe the X1C6's Thunderbolt support is still/also borked in some manner that causes my problem with the CalDigit dock.

But, I'm not too convinced of this. And I suspect CalDigit is using the same chipset that many other brands are using for similar docks.

For all the premium they charge for it, Thunderbolt designers/implementors should... be induced to provide a better result. IMHO.

[chenxiaolong]

> The X1C6 did initially have a problem: It rewrote its Thunderbolt configuration upon every boot, into a kind of physically memory not designed to have a lesson of that many writes.

Heh, that's such a Lenovo problem. I hard bricked my old ThinkPad W520 four times (and got warranty replacements) just by hitting F12 to bring up the BIOS boot menu. It did some sort of write that wore out the BIOS flash chip. I don't think they ever fixed that problem.

[pbnjeh]

I actually have a W520 that I'd like to continue to use. (That keyboard, plus a surprisingly decent display panel for Lenovo.)

I hadn't heard of this problem, so thanks for the heads up! I'll be very planful and sparing in my BIOS config access/writes.

I think CoreBoot has made it to some if not all of the -20 series. I wonder whether it would make any difference, although I can't see how if you're still writing to that storage.

[chenxiaolong]

Interesting! I didn't know coreboot was available for the *20 series. I hadn't touched my W520 since the last warranty replacement. Maybe coreboot will be my next weekend project.

[rtpg]

I had always assumed I was doing something wrong, glad to hear it’s not just me. Ubuntu works extremely well except for sleeping just emptying the battery. Got way to used to MacBooks not killing the battery in like 24 hours of sleep…

[theatomheart]

using linux + sleep = ur wrong. turn your computer off. sleep is a lie that never EVER did what it claimed. the only thing it's good for is enabling bad user behavior. period.

[dijit]

I close the lid of my MacBook and come back two days later: my MacBook has lost 3% of its battery.

If I open my laptop within 24hrs then it is already ready by the time I finish opening the lid. This immediacy means I don’t lose my train of thought (ie; the reason I grabbed my laptop in the first place).

I’m a hard Linux lover, and begrudgingly use a MacBook, but sleep has a lot of value, and the Mac stuff seems to work a lot more consistently than what windows is offering. (Linux too, in a number of circumstances: though my old precision 5520 never had an issue with suspend)

[mway]

> the only thing it's good for is enabling bad user behavior.

Okay, I'll bite. Care to expand? What behavior, and why is it bad? Why should users not expect a "suspend" state to work (more or less) as advertised?

edit: Or are you arguing that the only "correct" states are running and complete shutdown? Why would that be preferred?

[deaddodo]

I’ve run Linux on both my Dell XPS 15 (9500) and my Zephyrus G14 (2022). In both cases, the default state is to sleep like Windows 11’s modern standby. In Zephyrus’ case, I can force the old behavior as it’s still supported by the UEFI. However, the Dell can only be powered off or use modern standby. It does not support traditional sleep states. This is, in fact, one of my main reasons for the “downgrade”.

[pbnjeh]

Thank you!