Bug 220111

Summary:	AHCI: Regression Bug Affects ASUSPRO-D840SA system causes integrated GPU buffer underrun.
Product:	IO/Storage	Reporter:	kernel-dev
Component:	Serial ATA	Assignee:	Virtual Assignee for IO Storage/Serial ATA (io_serialata)
Status:	NEW ---
Severity:	normal	CC:	cassel, dlemoal, kernel-dev
Priority:	P3
Hardware:	Intel
OS:	Linux
See Also:	https://github.com/qubesos/qubes-issues/issues/9887
Kernel Version:	6.9	Subsystem:
Regression:	Yes	Bisected commit-id:	7627a0edef54

Description kernel-dev 2025-05-12 22:10:03 UTC

This regression was introduced in commit "7627a0edef54." It appears that ahci.c is setting non-host controller devices (in this case, Intel VGA) to ATA_LPM_MIN_POWER, which causes buffer underruns in the Intel iGPU on i7-9700 systems, rendering the display unusable in most cases (129 out of 132 boots). Currently, this can be fixed with the kernel parameter `ahci.mobile_lpm_policy=0` as a workaround.

Related Issues:
- https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/14213
- https://github.com/qubesos/qubes-issues/issues/9887

### Bisect History on stable/linux.
```
480e035fc4c7 -> BAD
e5e038b7ae9d -> BAD
...
07abb19a9b20 -> BAD
...
6cdebf62a159 -> BAD
...
57aaf9134c03 -> BAD <<<
    a9dde2823c33^..7edbb6059274 -> BAD
    ...         ^..7627a0edef54 -> BAD <<< 
    ...         ^..ae1f3db006b7 -> GOOD    
7d62cb2a59f4 -> GOOD
8c9c2f851b5a -> GOOD
...
f88c3fb81c4b -> GOOD
...
216532e147b2 -> GOOD
...
e66c58f74351 -> GOOD
```

Comment 1 kernel-dev 2025-05-15 21:36:55 UTC

Based on the kernel log, it appears that the AHCI is correctly recognizing the device; it _doesn't_ change anything about the iGPU itself. But somehow, it seems that setting the AHCI device to ATA_LPM_MIN_POWER is causing the integrated GPU not to function correctly. I'm not sure if this issue is specific to the ASUSPRO-D840SA or if it affects all i7-9700 systems. I'm quite confused about this.