Re: USB-C DisplayPort display failing to stay active with Intel Barlow Ridge USB4 controller, power-management related issue?

[Aaron Rainbolt <arainbolt@kfocus.org>]

[-- Attachment #1: Type: text/plain, Size: 2454 bytes --]

On Fri, 15 Nov 2024 15:20:22 +0200
Mika Westerberg <mika.westerberg@linux.intel.com> wrote:

> Hi,
> 
> On Thu, Nov 14, 2024 at 10:41:25AM -0600, Aaron Rainbolt wrote:
> > This is production hardware (specifically Clevo's X370SNW1-G and
> > X370SNV1-G laptops), available for purchase from Sager, XOTICPC,
> > Schenker, likely many other resellers, and our own website
> > at https://kfocus.org/spec/spec-m2.html (with a tool that allows users
> > to work around the bug). The firmware is baked into the hardware
> > provided to us by our ODM, and for the sake of stability we do not
> > modify any firmware on the machines with the exception of applying
> > BIOS updates provided to us directly by the ODM. They appear to get
> > their firmware directly from Clevo.  
> 
> Okay thanks.
> 
> > We have requested an updated BIOS from the ODM. If one is
> > available, we will upgrade and run the tests again.  
> 
> Yes, I hope you can get the firmwares. The one you have now is not
> "production quality" firmare so you should not really have that there
> in the first place and Clevo should definitely provide you an
> upgrade. Note this is separate from the BIOS. But your BIOS has issue
> too regarding the USB4 power contract that is required by Microsoft
> so I would expect that you should get that one upgraded too.
> 
> The patch I shared earlier should deal with all the other cases except
> that weird one where we do not seem to get unplugs (and the resource
> is available) which is not how the firmware is expected to work. I was
> planning to submit it upstream after some more validation on our end,
> probably afer v6.13-rc1 is released. I'll CC you.
> 
> If/when you get the new firmare I would definitely appreciate if your
> folks could give it a try.

Hi Mika:

Just to catch you up: 
We find the most recent patch from this thread to be a good improvement.

We tested it on our hardware with USB-C attached displays. In
particular, it allows us to enable displays that do not time out while
using with Thunderbolt. Without the patch, enabling the Thunderbolt
kernel module would disable all attached USB-C displays in 15 seconds.

The patch does not resolve the hot plugging issue on our hardware, but
we have a work-around for that (lspci -k or reloading the thunderbolt
module). We have not been able to acquire new firmware yet, so this is
likely the issue.

The test results are attached if you are interested.

Sincerely, Aaron

[-- Attachment #2: 2024-12-12-tbt-patch-compare.md --]
[-- Type: text/markdown, Size: 1314 bytes --]

## Key
- Display: A USB-C Attached monitor
- A: 6.8.0-45
- B: 6.11.5
- C: 6.11.5-patched
- D: TBT enabled (y/n)
- E: Desktop showed, but timed-out after 15s
- F: Displays do not return on replug
- G: Can be flakey; sometimes a replug or display wake is needed
- N: NO
- Y: YES
- x: Passed

## Test Results
| A | B | C | D | Test           |
| - | - | - | - | -------------- |
| x | x | x | n | Boot with no Displays w/o TBT (modprobe -r thunderbolt)|
| x | x | x | n | Attach         1st  Display   |
| x | x | x | n | Stays Resident 1st  Display   |
| x | x | x | n | Replug > 60s   1st  Display   |
| x | x | x | n | Attach         2nd  Display   |
| x | x | x | n | Replug > 60s   2nd  Display   |
| x | x | x | n | Replug > 60s   BOTH Displays  |
| x | x | x | n | Sleep-Resume   BOTH Displays  |
| x | x | x | n | Reboot with    BOTH Displays  | G

| A | B | C | D | Test           |
| - | - | - | - | -------------- |
| x | x | x | Y | Enable TBT (modprobe thunderbolt)    |
| E | E | x | Y | Stays Resident BOTH Displays  |
| . | . | x | Y | Replug < 15s   1st Display (2nd Remains) |
| . | . | x | Y | Replug > 60s   1st Display (2nd Remains) |
| . | . | F | Y | Replug > 60s Both Displays    |
| . | . | x | Y | Toggle TBT / lspci -k restores Displays |
| . | . | x | Y | Sleep-Resume   BOTH Displays  |