#asahi-dev · 2026-05-31 19:22
pjvhttps://github.com/AsahiLinux/linux/issues/507 -> https://github.com/AsahiLinux/linux/pull/508#asahi-dev · 2026-05-31 18:57
pjvThis is on a Macbook Pro M1#asahi-dev · 2026-05-31 18:56
pjvuhuh#asahi-dev · 2026-05-31 18:54
pjvVerified on Fedora Asahi Remix with the custom kernel `6.19.14-suspendfix+`. Multiple consecutive manual suspends and resumes successfully retained WiFi and Bluetooth connectivity without timeouts.#asahi-dev · 2026-05-31 18:54
pjv### Testing#asahi-dev · 2026-05-31 18:54
pjv3. **Coordinated Quiesce:** Modified `bcm4377_suspend` to locate the sibling WiFi PCI function (function 0) and wait until it transitions to `PCI_D3hot` or `PCI_D3cold` before writing `BCM4377_BAR0_SLEEP_CONTROL_QUIESCE`.#asahi-dev · 2026-05-31 18:54
pjv2. **Bluetooth Driver:** Added retry loops on `bootstage` and `rti_status` checks in `bcm4377_boot` to allow slower wakeups.#asahi-dev · 2026-05-31 18:54
pjv1. **WiFi Driver:** Increased `BRCMF_PCIE_MBDATA_TIMEOUT` to 5 seconds and added return code validation to `brcmf_pcie_pm_enter_D3`.#asahi-dev · 2026-05-31 18:54
pjv### Proposed/Verified Fixes#asahi-dev · 2026-05-31 18:54
pjv3. **Error Propagation:** The WiFi driver ignores return values from mailbox data writes during suspend.#asahi-dev · 2026-05-31 18:54
pjv2. **Cross-Function Race:** Sibling WiFi (01:00.0) and Bluetooth (01:00.1) drivers independently access the shared sleep control register (`BAR0_SLEEP_CONTROL`). If Bluetooth quiesces the chip while WiFi is waiting for a response, the firmware hangs.#asahi-dev · 2026-05-31 18:54
pjv1. **Handshake Timeout:** The 2-second timeout (`BRCMF_PCIE_MBDATA_TIMEOUT`) is insufficient during connection teardown.#asahi-dev · 2026-05-31 18:54
pjv### Root Causes#asahi-dev · 2026-05-31 18:54
pjvThe Broadcom BCM4387 PCIe device consistently fails to enter the D3 power state during system suspend. This results in the hardware entering an unrecoverable state or timing out upon resume unless modules are unloaded before sleep.#asahi-dev · 2026-05-31 18:54
pjv### Description#asahi-dev · 2026-05-31 18:52
pjvit's more than just increasing a timeout#asahi-dev · 2026-05-31 18:52
pjvWithout deep understanding: both bluetooth and wifi are using a same bus, and when sending the bluetooth driver to D3 sleep mode, this message never gets sent to the device because wifi keeps the bus full#asahi-dev · 2026-05-31 18:51
pjvIt increases a handshake timeout and solves a race condition#asahi-dev · 2026-05-31 18:46
pjvBluetooth connections no longer come back after suspend. I wrote a post here: https://discussion.fedoraproject.org/t/bluetooth-connections-not-possible-after-sleep/192814 . I will create a github issue and submit a PR that follows asahi contributor policy, as I had AI create a bugfix. I don'have any experience with kernel development, but I progression-tested the fix on my laptop. Does that work for you?