[Regression][SteamDeckOLED][bazzite-deck-gnome:latest] Bluetooth stopped working completely and is root cause for issue of auto-resume from suspend

[Raltay]

Describe the bug

Description

Bluetooth is broken on Steam Deck OLED on bazzite-deck-gnome:latest after version 39.20240328.0 (this is the last version when it worked correctly).

It doesn't work at all - cannot connect to paired devices (fails immediately upon selection), doesn't detect new/unpaired devices. I'm not sure if Bluetooth on/off switch even works at all.

This issue also is most probably (from my superficial log analysis) root cause of device's "'auto-resume' after suspend" issue. When device is suspended (either via power button or via menu option), it 'auto-resumes' after a few seconds pass. It happens every time, regardless of whether Bluetooth switch is ON or OFF. Relevant logs can be found in journalctl_after-suspend.txt file.

Both of those issues occur in both gamescope and desktop mode.

Last version of bazzite-deck-gnome where neither of these issues occured was 39.20240328.0. After updating from it, both of those issues started occurring.

Captured logs

dmesg_after-suspend.txt dmesg_before-suspend.txt journalctl_after-suspend.txt journalctl_before-suspend.txt journalctl-bluetooth_after-suspend.txt journalctl-bluetooth_before-suspend.txt lshw_after-suspend.txt lshw_before-suspend.txt lspci_after-suspend.txt lspci_before-suspend.txt lsusb_after-suspend.txt lsusb_before-suspend.txt

What did you expect to happen?

I expected Bluetooth to continue to work properly (and not cause suspend issue) like in 39.20240328.0.

Output of rpm-ostree status

No response

Hardware

No response

Extra information or context

No response

[dog-in-space]

Can confirm I also have this issue on my SteamDeck OLED. Rebasing to 39.20240328.0 and sleep / bluetooth is functional now

[RodoMa92]

Yeah, that's a Qualcomm driver regression:

[  457.706090] Bluetooth: hci0: SSR or FW download time out
[  457.706106] hci_uart_qca serial0-0: PM: dpm_run_callback(): acpi_subsys_suspend+0x0/0x60 returns -110
[  457.706126] hci_uart_qca serial0-0: PM: failed to suspend: error -110

It's probably worth to open a bluetooth regression issue on the bugzilla kernel repo.

[telorand]

Also having the exact same issue with the same symptoms. Had to rebase to -gnome:39-20240328.0 to fix Bluetooth and sleep.

Hopefully Qualcomm or whoever works on that driver gets this solved quickly.

[KyleGospo]

Fixed on latest testing build, we are working on some unrelated sleep bugs and then will get this kernel to stable.

[Raltay]

I rebased back to bazzite-deck-gnome:stable on my Steam Deck OLED and I can confirm that the issue is fixed on 39-20240414.

Thank you for fast response and have a nice day 🙂

[telorand]

Possibly still occurring. https://github.com/ublue-os/bazzite/issues/960#issuecomment-2063993939

[sstendahl]

Hi,

I just tried with my Steam Deck OLED 1TB model, and the issue described here (non-functioning bluetooth and auto-resume from suspend) is definitely still present in bazzite-deck-gnome:stable in 39-20240419. I tried to rebase to bazzite-deck-gnome:testing, but the issue is there as well, on top of WiFi not working in testing.

I rebased back to 20240328 as @Raltay suggested in the opening post, and it works fine from there. So it's definitely a regression indeed, and it does seem to be in the latest stable. Have not tried 20240414, it could be possible that the regression slipped back in somehow.

[Raltay]

'bazzite-deck-gnome:stable in 39-20240419'

I can confirm that this also started happening again on my OLED with the above version.

Reopening ticket.

[telorand]

Doesn't seem to be occurring on the latest upgrade to 40-20240427. There's still outstanding BT issues with controllers, but Sleep/Suspend and audio devices seem to be working correctly. Additionally, I've noticed that audio doesn't seem to be bugged when switching between Desktop Mode to Game Mode.

[sstendahl]

Yeah this got resolved for me as well a while ago. I think the issue can be closed.