https://bugzilla.kernel.org/show_bug.cgi?id=222118
Bug ID: 222118
Summary: THUNDEROBOT 911Pro: NVMe in CPU-attached M.2 slot and
NVIDIA dGPU powered off at boot (needs unused power
resource quirk like THUNDEROBOT ZERO)
Product: ACPI
Version: 2.5
Hardware: All
OS: Linux
Status: NEW
Severity: normal
Priority: P3
Component: Power-Other
Assignee: [email protected]
Reporter: [email protected]
Regression: No
Hardware:
- System: THUNDEROBOT 911Pro (sys_vendor "THUNDEROBOT", product_name "911Pro",
product_family "911 series"), board "NLCC"
- BIOS: TP127, 06/30/2021
- CPU: Intel Core i7-11800H (Tiger Lake-H)
- dGPU: NVIDIA GA106M RTX 3060 Laptop [10de:2520] at 0000:01:00.0 (root port
00:01.0, ACPI \_SB.PC00.PEG1)
- NVMe: Longsys/Innogrit [1d97:5216] (Kingmax PCIe SSD 512GB) at 0000:02:00.0
(root port 00:01.1, ACPI \_SB.PC00.PEG2)
Kernel: 7.0.0-31-generic (Ubuntu build of v7.0)
Problem: The NVMe drive in the CPU-attached M.2 slot (0000:02:00.0) and the
NVIDIA dGPU (0000:01:00.0, 0000:01:00.1) are enumerated at boot, but when their
drivers probe them they are already in D3cold and do not respond. Under Windows
10 on the same machine, both devices work normally.
```
[ 0.295685] pci 0000:01:00.0: [10de:2520] type 00 class 0x030000 PCIe Legacy
Endpoint
[ 0.296075] pci 0000:01:00.1: [10de:228e] type 00 class 0x040300 PCIe
Endpoint
[ 0.296249] pci 0000:02:00.0: [1d97:5216] type 00 class 0x010802 PCIe
Endpoint
[ 0.745213] nvme nvme0: pci function 0000:02:00.0
[ 0.745234] nvme 0000:02:00.0: Unable to change power state from D3cold to
D0, device inaccessible
[ 0.749347] nvme 0000:02:00.0: error -ENODEV: probe failed
[ 7.331599] snd_hda_intel 0000:01:00.1: Unable to change power state from D0
to D0, device inaccessible
[ 7.332313] snd_hda_intel 0000:01:00.1: Unable to change power state from
D3cold to D0, device inaccessible
[ 7.433897] snd_hda_intel 0000:01:00.1: GPU sound probed, but not
operational: please add a quirk to driver_denylist
```
The same failure happens with and without
nvme_core.default_ps_max_latency_us=0, pcie_aspm=off and
pcie_aspm.policy=performance.
This looks like the same problem as bug 221087 (THUNDEROBOT ZERO, also Tiger
Lake-H with an RTX 3060). That bug was fixed by commit cd7ef20ba8c6 ("ACPI: PM:
Add unused power resource quirk for THUNDEROBOT ZERO"), which is already in
v7.0 and therefore in the kernel I tested. Its DMI entry only matches product
names containing "ZERO", so it does not apply here (product_name is "911Pro"),
and current mainline has no entry for this machine. On this machine, the SSDT
with OEM Table ID "TglH_Rvp" defines PXP_ power resources under the PEG and RP
scopes.
On the ZERO this behaviour started with a1224f34d72a (v5.16-rc1). I have not
tested any kernel older than 7.0 on this machine.
Request: please add a DMI entry for THUNDEROBOT 911Pro (board NLCC) to
dmi_leave_unused_power_resources_on[] in drivers/acpi/power.c, like
cd7ef20ba8c6 did for the ZERO. I can boot with dyndbg="file
drivers/acpi/power.c +p" and provide that log, the full dmesg and an acpidump,
and I can test a patched kernel.
--
You may reply to this email to add a comment.
You are receiving this mail because:
You are watching the assignee of the bug.
_______________________________________________
acpi-bugzilla mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/acpi-bugzilla
