Summary: | x11-misc/bumblebee-3.2.1_p20210112 not working correctly with x11-drivers/nvidia-drivers-460.39-r1 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Stefano <raistlin> |
Component: | Current packages | Assignee: | Pacho Ramos <pacho> |
Status: | UNCONFIRMED --- | ||
Severity: | normal | CC: | ionen, pacho, proxy-maint, rei4dan |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Stefano
2021-02-20 17:25:10 UTC
If I were to guess it's probably related to 460's reworked power management which is a major change since 450.xx, not that I use bumblebee so I can't help with that. (In reply to Ionen Wolkens from comment #1) > If I were to guess it's probably related to 460's reworked power management > which is a major change since 450.xx, not that I use bumblebee so I can't > help with that. I have the same feeling but no proof. In case it helps, since the new driver comes with systemd stuff, I'll point out I'm not on systemd. Does bumblebee-3.2.1_p20190421-r1 still work for you with that setup? I am also unsure if this change could be related https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=3f48cdfdff62bdeb440900f04d7745dd08365eba So, I tried downgrading bumblebee and upgrading nvidia-drivers, and it seems to work. Then I tried upgrading bumblebee again... and the very same configuration that didn't work at all now works. I'll try to figure out what butterfly flapped their wings somewhere. OK, the relationship of this bug with the power management is getting more and more concrete. The factor that seems to change between everything working and viceversa is booting the machine with, or without, the AC adapter. Now I can dive in the mess of ACPI, bumblebee and laptop-mode for hours. This was fixed somehow with the following versions: Bumblebee 3.2.1_p20210112-r4 Nvidia-drivers 460.56 The only required trick: not starting the vgl service (i.e. commenting out the want vgl in the init script, if using openrc). I would suggest removing that dependency, it really does not seem necessary? (In reply to Stefano from comment #7) > This was fixed somehow with the following versions: > Bumblebee 3.2.1_p20210112-r4 > Nvidia-drivers 460.56 If you could, can you check if all good with =nvidia-drivers-460.67 as well? 460.67 ebuild changed many things but I'm not familiar with bumblebee setups to test, and then it'll be stable in roughly ~3 days if no issues. Knowing if no regression with 465.19.01 would be useful too, it did change a few things wrt acpi handling (not that I'm planning to stable 465 branch anytime soon). Disregard my previous optimistic assumption. This setup literally works or stops working across reboots. I'll work on it some more. (In reply to Ionen Wolkens from comment #8) > If you could, can you check if all good with =nvidia-drivers-460.67 as well? > 460.67 ebuild changed many things but I'm not familiar with bumblebee setups > to test, and then it'll be stable in roughly ~3 days if no issues. It works (with the same random quirks I'm still trying to understand) just as well as .56 |