RK3588 EVB1 WLAN support
The EVB1 has an onboard PCIe based WLAN module named AP6275P, which is based on a Broadcom WLAN module. The PCIe device it only detectable when the RTC (hym8563) clock is kept enabled, since that is used by the WLAN/BT module and the necessary GPIOs are toggled correctly. Even then it takes some time for the module to become available. The vendor kernel from Rockchip solved this with a custom rfkill driver combined with some hacks in the PCIe driver.
In the upstream mailinglist there is currently a patchset prepared by Bartosz Golaszewski, who is working on a similar issue with a Qualcomm based design: https://lore.kernel.org/all/20240104130123.37115-1-brgl@bgdev.pl/ . Once this series has landed a driver for the AP6275P power sequence can be added and no hacks in the PCIe driver should be needed.
In addition to the power sequence, upstream's brcmfmac WLAN driver does not yet support the WLAN IP inside of the AP6275P module.
Activity
assigned to @sre
added T42916 label
mentioned in commit f5d03da4
mentioned in commit 369acf97
mentioned in commit 2acc59dd
mentioned in commit 96717617
mentioned in commit c90399fb
PATCHv6 for PCIe power sequencing framework:
https://lore.kernel.org/all/20240325131624.26023-1-brgl@bgdev.pl/
-
PATCHv7 for PCIe power sequencing framework:
https://lore.kernel.org/linux-pm/CAMi1Hd1wc2W02G4SnMhthBgY6GGLW7ZuGzESmoqj5qY_kMkNWA@mail.gmail.com/
-
Discussion adding support for the chip to brcmfmac:
mentioned in commit e03e7f20
mentioned in commit f6944d4a
mentioned in commit a2db328b
mentioned in commit 8cf360b9
mentioned in commit fe6f86f4
-
I started working on a pwrseq driver for the AP6275P. The working branch can be found here:
https://gitlab.collabora.com/hardware-enablement/rockchip-3588/linux/-/tree/rk3588-wlan-pwrseq
mentioned in commit c3a5e3e8
mentioned in commit e6d08d7e
mentioned in commit f7ce5eb2
-
https://lore.kernel.org/all/20240805073425.3492078-1-jacobe.zang@wesion.com/
After some re-testing the PCIe device does probe without the RTC clock. The RTC clock is needed to properly probe the driver, but this is already handled by the series. Once this series landed, a simple DT patch for the EVB1 is enough to get working WLAN.
mentioned in commit 12f9386e
-
PATCHv13: https://lore.kernel.org/all/20240828-wireless-mainline-v13-0-9998b19cfe7e@wesion.com/
There will be yet another version, which drops the
brcm,bcm4329-fmacfallback compatible. -
PATCHv14: https://lore.kernel.org/all/20240910-wireless-mainline-v14-0-9d80fea5326d@wesion.com/
This version works perfectly fine on RK3588 EVB1. The necessary DT changes can be found in the
rk3588-testbranch. I plan to send the patch upstream once the series for the bindings/driver has landed upstream. -
The driver and binding patches got merged to wireless-next.git:
https://lore.kernel.org/linux-wireless/172666764990.3996465.9327030359445823508.kvalo@kernel.org/
mentioned in commit 56440d7e
mentioned in commit 3d041393
mentioned in commit d62b1404
mentioned in commit 7afb8673
mentioned in commit 88fd2b70
mentioned in commit fe4bf8d0
mentioned in commit a66dfaf1
mentioned in commit 3051a77a
mentioned in commit 0ee4736c
mentioned in commit 6aecd91a
mentioned in commit 0e2909c6
mentioned in commit d38e26e3
mentioned in commit 2d2d4f60
mentioned in commit 5641e82c
mentioned in commit f1897f2f
mentioned in commit 1b10a519
mentioned in commit 2c368809
mentioned in commit b0fce54b
mentioned in commit 03511e93
mentioned in commit 1c7b17cf
mentioned in commit b150654f
mentioned in commit 3109d5ff
mentioned in commit 23f00807
mentioned in commit 7be11d34
mentioned in commit b2ffadcc
mentioned in commit a91c4951