Thanks for the disconnect-trigger test, it helps. s2idle: the card survives resume and the firmware *does* detect the event, but its wakeup never reaches the host. So WoW is effectively dead in s2idle because the card's PME doesn't propagate up the chain.
S3: This is hard to proceed without a PCIe protocol analyzer. Generally we can rely on firmware dump, however it is not possible here since link is down. If you happen to have access to a PCIe analyzer that can sit on that slot, a capture across the S3 wake would be gold; otherwise that side is largely a platform/BIOS question for the bridge chain. The s2idle side is more tractable because the card is alive, so let's chase "the card's PME never reaches the host" there. Could you grab, on the s2idle + WoWLAN-armed runs: 1. The root port above the card, 00:02.1, during the WoW window — is its PME interrupt actually enabled and is it a wake source? With your no- usb-pme hook disabled: setpci -s 00:02.1 CAP_EXP+0x1c.w (bit3 = PME Interrupt Enable) cat /sys/bus/pci/devices/0000:00:02.1/power/wakeup 2. Whether the card even asserted PME. Read its wakeup_count before suspend and after the (power-button) resume: cat /sys/bus/pci/devices/0000:09:00.0/power/wakeup_count If it goes up, the card did drive PME_Status but it was lost on the way up the chain; if it doesn't, the card never asserted PME. 3. One retest: s2idle + WoWLAN armed, but with ASPM L1 and L1 PM Substates disabled on every link in the chain (not just the endpoint), then try the WoW wake. In-band PME needs the link to exit L1; if a switch on this chain can't bring the link out of L1 it would swallow the PME. If the wake starts working with L1 off, that localizes it. That should tell us whether the card is signaling and where the signal dies. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2170503 Title: ath12k WCN7850: system hangs on resume from suspend when WoWLAN magic- packet is enabled To manage notifications about this bug go to: https://bugs.launchpad.net/linux/+bug/2170503/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
