Re: [PATCH] PCI: Extend Apple Thunderbolt power quirk to Alpine Ridge

[Darrell Gum]

On Thu, 24 Sep 2026 10:28:15 -0300, Francisco Beltrán Millalén wrote:
> On Macs with an Alpine Ridge Thunderbolt 3 controller, suspending to RAM
[...]

Tested on a second MacBookPro14,3 on 7.2.5 (Omarchy's linux-omarchy
7.2.5-3). It has two Alpine Ridge 4C controllers, with upstream
bridges 8086:1578 at 04:00.0 and 7a:00.0. Your PCI/PM v2 series (with
e18d1abc3bff) and the Apple native PME patch were applied as well.
The full report is in reply to the series:
https://lore.kernel.org/all/20260930141914.6678-1-fbeltranmillalen@xxxxxxxxx/

For this patch:

- The quirk fired on both upstream bridges ("quirk: cutting power to
  Thunderbolt controller...").
- pm_test=platform passes. The stock kernel hard-hung there 3 out of
  3 times, with or without brcmfmac loaded.
- Real S3 resumed on every attempt, about half a dozen cycles
  including lid closes on battery. noirq resume is about 16 s, with
  about 11 s of it in each upstream bridge.
- With a USB 3 stick attached on the 7a:00.0 side (xHCI 7d:00.0),
  the quirk logged for both bridges during a lid-close S3. The stick
  was still there after resume: no disconnect logged, still at
  5000 Mbps, filesystem readable. A replug about 2 min later
  re-enumerated it at SuperSpeed.
- On that cycle the empty side's xHCI logged "xhci_hcd 0000:07:00.0:
  xHC error in resume, USBSTS 0x401, Reinit" and recovered.

The patches were tested together, not bisected. Only one S3 cycle
had a device attached (a USB stick, no real Thunderbolt device). The
no-wake-on-plug trade-off is untested here.

Tested-by: Darrell Gum <d@xxxxxxxx>