[Bug 216822] New: Thunderbolt USB Controller (Maple Ridge) and its upstream pcie port fail to runtime resume

[Bug 216822] New: Thunderbolt USB Controller (Maple Ridge) and its upstream pcie port fail to runtime resume

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=216822

            Bug ID: 216822
           Summary: Thunderbolt USB Controller (Maple Ridge) and its
                    upstream pcie port fail to runtime resume
           Product: Drivers
           Version: 2.5
    Kernel Version: kernel 6.0 and later
          Hardware: Intel
                OS: Linux
              Tree: Mainline
            Status: NEW
          Severity: high
          Priority: P1
         Component: USB
          Assignee: drivers_usb@kernel-bugs.kernel.org
          Reporter: chris.chiu@canonical.com
        Regression: No

Created attachment 303426
  --> https://bugzilla.kernel.org/attachment.cgi?id=303426&action=edit
output of lspci -vt

The xHCI host controller(8086:1138) of the TBT4 add-on card fail to detect any
USB-C devices after boot. After forcing the runtime resume by `echo 'on' >
'/sys/bus/usb/devices/usb4/power/control`, the connected usb-c device can be
detected.
After turn on PCI_DEBUG and DEBUG of the usbcode/xhci_hcd, there seems no
problem when the usb port go into runtime suspend
```
[  579.488151] xhci_hcd 0000:13:00.0: set port remote wake mask, actual port
4-1 status  = 0xe0002a0
[  579.488221] xhci_hcd 0000:13:00.0: set port remote wake mask, actual port
4-2 status  = 0xe0002a0
[  579.488239] hub 4-0:1.0: hub_suspend
[  579.488260] usb usb4: bus auto-suspend, wakeup 1
[  579.488283] xhci_hcd 0000:13:00.0: xhci_hub_status_data: stopping usb4 port
polling
[  579.488306] xhci_hcd 0000:13:00.0: config port 4-1 wake bits, portsc:
0xa0002a0, write: 0xa0202a0
[  579.488316] xhci_hcd 0000:13:00.0: config port 4-2 wake bits, portsc:
0xa0002a0, write: 0xa0202a0
[  579.488323] xhci_hcd 0000:13:00.0: config port 3-1 wake bits, portsc:
0xa0002a0, write: 0xa0202a0
[  579.488328] xhci_hcd 0000:13:00.0: config port 3-2 wake bits, portsc:
0xa0002a0, write: 0xa0202a0
[  579.488333] xhci_hcd 0000:13:00.0: xhci_suspend: stopping usb3 port polling.
[  579.488359] xhci_hcd 0000:13:00.0: // Setting command ring address to
0x11aa08001
[  579.488536] xhci_hcd 0000:13:00.0: hcd_pci_runtime_suspend: 0
```

but no messages in kernel after plug in a usb device. I expect to see the
runtime resume first on the upstream port pcieport 0000:00:1d.0, and then the
xhci_hcd 0000:13:00.0 but nothing. Please refer to the attached lspci output
for the device information and suggest what we should do next to find out the
cause. Thanks

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 216822] Thunderbolt USB Controller (Maple Ridge) and its upstream pcie port fail to runtime resume

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=216822

--- Comment #1 from Chris Chiu (chris.chiu@canonical.com) ---
Created attachment 303427
  --> https://bugzilla.kernel.org/attachment.cgi?id=303427&action=edit
output of lspci -vvv

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 216822] Thunderbolt USB Controller (Maple Ridge) and its upstream pcie port fail to runtime resume

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=216822

--- Comment #2 from Chris Chiu (chris.chiu@canonical.com) ---
Created attachment 303428
  --> https://bugzilla.kernel.org/attachment.cgi?id=303428&action=edit
kernel log while booting with a usb storage connected to usb4

The usb-c storage can be detected during boot. But it can't be detected after
re-plug the same device.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 216822] Thunderbolt USB Controller (Maple Ridge) and its upstream pcie port fail to runtime resume

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=216822

Jacob Moroni (mail@jakemoroni.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mail@jakemoroni.com

--- Comment #3 from Jacob Moroni (mail@jakemoroni.com) ---
I see similar behavior with the Gigabyte GC-Maple Ridge add-in card, except I
can’t even get my USB devices to show up after setting power/control to “on”.

There are no kernel log messages printed at all when plugging/removing USBC
devices on this card.

True Thunderbolt devices seem to work though.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 216822] Thunderbolt USB Controller (Maple Ridge) and its upstream pcie port fail to runtime resume

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=216822

--- Comment #4 from Max Lee (max.lee@canonical.com) ---
On this platforms, OEM didn't design RTD3 on Maple Ridge.
Since the MR AIC is to share with other desktop projects, and the NVM was set
to Native + RTD3 enabled.
This cause the driver to get the RTD3 enabled from NVM. and set Linux TBT
controller in D3 state then PME doesn't toggled by device hot plug.
Current BIOS sent _S0W to TBT RP and the NVM native +RTD3 doesn't change to
native only. Is this a correct setting on the NVM to support non-RTD3 platform?
should BIOS sent command to modify NVM offset to disable RTD3?

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 216822] Thunderbolt USB Controller (Maple Ridge) and its upstream pcie port fail to runtime resume

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=216822

--- Comment #5 from Jacob Moroni (mail@jakemoroni.com) ---
I should add that in my case, I’m using this add-in card in a system that has
no BIOS support for Thunderbolt. The motherboard doesn’t have a Thunderbolt
header, so I am pulling pin 1 of the 5 pin connector up to 3.3V to force the
card to turn on, which I’m guessing is the equivalent of writing 1 to
/sys/bus/wmi/devices/86CCFD48-205E-4A77-9C48-2021CBEDE341/force_power on a
system with real support for this card.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.