How to handle different operating speed for QCA9377 SDIO BT module?

[Christian Hewitt <christianshewitt@xxxxxxxxx>]

I’m working with a QCA9337 SDIO device on an Android (now Linux) set-top box with an Amlogic S905D SoC.

SDIO WiFi (ath10k) is working since 5.7-rc1, but the BT side of the module is still missing.

Most Amlogic devices (95%+) use Broadcom SDIO modules with the following device tree content:

&uart_A {
	bluetooth {
		compatible = "brcm,bcm43438-bt";
		shutdown-gpios = <&gpio GPIOX_17 GPIO_ACTIVE_HIGH>;
		max-speed = <2000000>;
		clocks = <&wifi32k>;
		clock-names = "lpo";
	};
};

I changed the compatible to "qcom,qca9377-bt” and applied the following patch:

diff --git a/drivers/bluetooth/hci_qca.c b/drivers/bluetooth/hci_qca.c
index 439392b1c043..6f0350fbdcd6 100644
--- a/drivers/bluetooth/hci_qca.c
+++ b/drivers/bluetooth/hci_qca.c
@@ -2046,6 +2046,7 @@ static SIMPLE_DEV_PM_OPS(qca_pm_ops, qca_suspend, qca_resume);
 
 static const struct of_device_id qca_bluetooth_of_match[] = {
        { .compatible = "qcom,qca6174-bt" },
+       { .compatible = "qcom,qca9377-bt" },
        { .compatible = "qcom,wcn3990-bt", .data = &qca_soc_data_wcn3990},
        { .compatible = "qcom,wcn3991-bt", .data = &qca_soc_data_wcn3991},
        { .compatible = "qcom,wcn3998-bt", .data = &qca_soc_data_wcn3998},

This results in probing with errors: http://ix.io/2i6P

I noticed "max-speed = <2000000>;” while hci_qca.c sets 3000000. I attempted to set "max-speed = <3000000>;” but this made no difference, so I patched a lower value in hci_qca:

diff --git a/drivers/bluetooth/hci_qca.c b/drivers/bluetooth/hci_qca.c
index 6f0350fbdcd6..b7ea1e9f4904 100644
--- a/drivers/bluetooth/hci_qca.c
+++ b/drivers/bluetooth/hci_qca.c
@@ -1678,7 +1678,7 @@ static const struct hci_uart_proto qca_proto = {
        .name           = "QCA",
        .manufacturer   = 29,
        .init_speed     = 115200,
-       .oper_speed     = 3000000,
+       .oper_speed     = 2000000,
        .open           = qca_open,
        .close          = qca_close,
        .flush          = qca_flush,

This shows a frame reassembly error, but an attempt to load firmware, see: http://ix.io/2i6Q

Using renamed files from https://github.com/boundarydevices/qca-firmware/tree/bd-sdmac-ath10k/qca

There is still a "kernel: Bluetooth: hci0: Frame reassembly failed (-84)” error, but the module is now otherwise up/working - I can scan/see/pair other BT devices.

see: http://ix.io/2i6S and:

SML5442TW:~ # bluetoothctl show
Controller 91:08:00:00:00:00 (public)
	Name: SML5442TW
	Alias: SML5442TW
	Class: 0x000c0000
	Powered: yes
	Discoverable: no
	DiscoverableTimeout: 0x000000b4
	Pairable: yes
	UUID: Audio Source              (0000110a-0000-1000-8000-00805f9b34fb)
	UUID: Generic Attribute Profile (00001801-0000-1000-8000-00805f9b34fb)
	UUID: Generic Access Profile    (00001800-0000-1000-8000-00805f9b34fb)
	UUID: PnP Information           (00001200-0000-1000-8000-00805f9b34fb)
	UUID: A/V Remote Control Target (0000110c-0000-1000-8000-00805f9b34fb)
	UUID: A/V Remote Control        (0000110e-0000-1000-8000-00805f9b34fb)
	UUID: Audio Sink                (0000110b-0000-1000-8000-00805f9b34fb)
	Modalias: usb:v1D6Bp0246d0536
	Discovering: no
Advertising Features:
	ActiveInstances: 0x00
	SupportedInstances: 0x05
	SupportedIncludes: tx-power
	SupportedIncludes: appearance
	SupportedIncludes: local-name

So it looks like hci_qca.c needs modification to handle multiple operating speeds, perhaps using the max-speed description from device-tree if available, or defaulting to the current 3000000 value if not.

I’m not a coding developer so I need to ask for help. If someone can suggest a patch - I can test and confirm it works. Or if someone can explain how this should be implemented, I can see if I can find someone to help with the task.

Christian