Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 669818 - Fans not spinning on some nvidia GPUs on minimal installcd (among LiveCDs)
Summary: Fans not spinning on some nvidia GPUs on minimal installcd (among LiveCDs)
Status: RESOLVED CANTFIX
Alias: None
Product: Gentoo Release Media
Classification: Unclassified
Component: Everything (show other bugs)
Hardware: All Linux
: Normal normal (vote)
Assignee: Gentoo Release Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2018-10-27 20:34 UTC by KostWarCZE
Modified: 2018-10-29 18:45 UTC (History)
2 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description KostWarCZE 2018-10-27 20:34:11 UTC
DISCLAIMER: 
- Chosed critical since this bug has the potential to break end-user hardware (GPU)!

- Possible duplicate of #15419. Decided to make a new bug report since #15419 is marked as resolved and from 2003. 

On minimalCD, LiveDVD and systemrescuecd the GPU on my ASUS ROG G750 (https://www.asus.com/us/ROG-Republic-Of-Gamers/ROG-G750JS/specifications/) is constantly overheating (+- 95C and rising) depending on the spent time in LiveDVD, MinimalCD or systemrescuecd. 

I've noticed that the system prevents the fan from spinning. 

Possible fix based on #15419 is to disable acpi.

Sent me the instructions to help fixing this issue.
Comment 1 Jeroen Roovers (RETIRED) gentoo-dev 2018-10-28 10:45:53 UTC
"Possible" hardware damage, or "actual" hardware damage? "Overheating" or just "heating"?
Comment 2 KostWarCZE 2018-10-28 11:39:58 UTC
(In reply to Jeroen Roovers from comment #1)
> "Possible" hardware damage, or "actual" hardware damage? "Overheating" or
> just "heating"?

Based on my experience i believe that possible hardware damage depending on the hardware used + constant failsaving can cause actual hardware damage.

Overheating since the notebook case had +- 95C and rising.

Based on tests on ubuntu it seems that Gentoo on LiveDVD, MinimalCD and systemrescuecd lacks the /etc/ configuration for sensors to use the fan once i used `sensors-detect` and allow it to update the /etc and rebooted the issue seems to be fixed. 

Related: https://askubuntu.com/questions/703580/fan-not-running-in-ubuntu-14-04

Based on info from IRC this issue CAN also be resolved by installing `fan module?` NeedySeagon on IRC #gentoo seems to have more info about it.

Based on provided info this issue can be solved by making gentoo detect the sensors and apply the configuration to /etc or adding fan modules(?) to default configuration for LiveDVD, minimalCD and systemrescuecd. => reasons to switch status.
Comment 3 KostWarCZE 2018-10-28 11:42:45 UTC
L.W: i'm sure that i can cause actuall hardware damage just by running gentoo compiling for few hours on LiveDVD, minimalCD and systemrescuecd on affected systems, but i'm not willing to make this sacrifice just to prove the bug since its most likely to damage (burn) the GPU core which i can't fix myself.
Comment 4 KostWarCZE 2018-10-28 11:57:10 UTC
EDIT: The overheating is still present even with  https://askubuntu.com/questions/703580/fan-not-running-in-ubuntu-14-04

It's ubuntu issue that is also present on 4.19.0-gentoo that's why it's related.
Comment 5 Jeroen Roovers (RETIRED) gentoo-dev 2018-10-28 12:31:16 UTC
We cannot fix your local hardware issues. If you cannot fix them, ask your hardware manufacturer for support and/or send them the repair bill.
Comment 6 KostWarCZE 2018-10-28 13:49:49 UTC
(In reply to Jeroen Roovers from comment #5)
> We cannot fix your local hardware issues. If you cannot fix them, ask your
> hardware manufacturer for support and/or send them the repair bill.

It's such a PITA to fix anything on gentoo these days srsly..

IT IS NOT HARDWARE ISSUE!!!

Working in BIOS, GRUB, while loading gentoo/ubuntu and on Windows 7,8,10. => software issue => LiveDVD, MinimalCD and systemrescuecd bug.
Comment 7 KostWarCZE 2018-10-28 15:01:50 UTC
New info: the issue with overheating due non-functional GPU fan was resolved by installing nvidia drivers 390 on ubuntu. 

Suspect that the nvidia drivers have the driver for fan to work.
Comment 8 Ben Kohler gentoo-dev 2018-10-28 15:23:57 UTC
What do you suggest we do?
Comment 9 Andreas Sturmlechner gentoo-dev 2018-10-28 15:54:47 UTC
What kind of self-respecting hardware would need a userspace driver for the fan to work?
Comment 10 KostWarCZE 2018-10-28 21:14:35 UTC
New info: Based on tests on ubuntu: Nouveau does NOT have the drivers to make the fan working.

(In reply to Ben Kohler from comment #8)
> What do you suggest we do?

Hotfix? If (mobile) Nvidia GPU is detected on LiveDVD, MinimalCD and systemrescuecd then install nvidia closed-source drivers to prevent the issue and potential hardware damage..

For the permanent fix we need more info about why is closed-source nvidia drivers required to make the fan working on nvidia notebooks. As i said on IRC i was told there is a module for this issue investigating it now.. In theory just include the module for fan functionality?

Relevant to ACPI in theory? those notebooks have the driver in it's bios so if linux does NOT use ACPI or software that takes over the fan control (sensors in theory) it might be an effective solution?

Can we in theory reverse-engineer the closed-source drivers to figure out what is required?

(In reply to Andreas Sturmlechner from comment #9)
> What kind of self-respecting hardware would need a userspace driver for the
> fan to work?

mine and most likely any Nvidia notebook.. I think those systems are a victim of torvald's `Nvidia.. fuck you`.. 

...

Do we have enough info to confirm the bug? This bug is (i believe) critical since it practically turn gentoo into a destructive malware.. The end-user should be at minimal warned about this when Mobile Nvidia GPU is detected..
Comment 11 Ben Kohler gentoo-dev 2018-10-28 21:32:38 UTC
I think you need to slow down and stop panicking, and think about what we can *realistically* do.

I'm not sure we're able to ship nvidia-drivers preinstalled on the minimal installcd, for licensing reasons.  We certainly cannot install it on demand, as that would require full working portage & toolchain.  We don't have those.

I can mark this bug "confirmed" but I still have no idea if we can or should actually do anything about it.

This isn't affecting all notebooks with nvidia GPUs, it's an isolated case.
Comment 12 Jorge Manuel B. S. Vicetto (RETIRED) Gentoo Infrastructure gentoo-dev 2018-10-29 10:05:50 UTC
Gentoo has stopped providing any X based install media a long time ago.
The "issue" you report is related to kernel, not to user land or any packaging done by Gentoo, so this isn't a "Gentoo issue" but, if anything, a "Linux issue".
Comment 13 KostWarCZE 2018-10-29 18:42:47 UTC
(In reply to Ben Kohler from comment #11)
> I think you need to slow down and stop panicking, and think about what we
> can *realistically* do.
> 
> I'm not sure we're able to ship nvidia-drivers preinstalled on the minimal
> installcd, for licensing reasons.  We certainly cannot install it on demand,
> as that would require full working portage & toolchain.  We don't have those.

So what can we do? There is a utility that allows the end-user to install Nvidia-drivers via GUI on one-click.. can't we do something simmilar?

> I can mark this bug "confirmed" but I still have no idea if we can or should
> actually do anything about it.
> 
> This isn't affecting all notebooks with nvidia GPUs, it's an isolated case.

Good to know.. i try to provide more info and solution once i get to it.
Comment 14 KostWarCZE 2018-10-29 18:45:19 UTC
(In reply to Jorge Manuel B. S. Vicetto from comment #12)
> Gentoo has stopped providing any X based install media a long time ago.
> The "issue" you report is related to kernel, not to user land or any
> packaging done by Gentoo, so this isn't a "Gentoo issue" but, if anything, a
> "Linux issue".

You still provide LiveDVD and since this issue can't be resolved by installing gentoo package like Nouveau then anyone with affected machine using gentoo is affected.. => Gentoo issue