Some UART / serdev based Bluetooth drivers need to be able to reset the
hardware during initialization or recovery. This is not possible with
the current power sequencing API if a unit is shared between USB and
UART interfaces.

For now, split the Bluetooth unit into two, one for each interface.
This allows either interface to independently toggle the power, with
the last action prevailing, thereby unbreaking the Qualcomm Bluetooth
serdev driver. This is also needed for the Realtek Bluetooth serdev
driver (hci_h5).

Having independent control from either interface unfortunately means
that userspace is able to shut down the controller from sysfs using
the USB port's "disable" setting without the serdev driver knowing
about it. On the USB side, independent control is also desired, as it
allows the USB core to power cycle the port/device during faults, and
for userspace to initiate reset and recovery using the aforementioned
"disable" setting. However when USB is used, a serdev device is not
created, and there is no conflicting usage, which allows the power
sequencing to work even without the split unit.

Suggested-by: Wei Deng <redacted>
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
---
Changes since v4:
- New patch

This patch is _not_ squashed into the previous to provide clear context
on the change, and the less than perfect nature of it.

I think what we need in this case is some sort of lockout. The serdev
consumer would want "exclusive" access, locking out the USB consumer,
which can tolerate "shared" access.

Something similar would be needed for the WiFi if we ever add support
for SDIO for toggling the state of W_DISABLE#1. At least the SDIO and
PCIe reset signals are separate.
---
 drivers/power/sequencing/pwrseq-pcie-m2.c | 22 ++++++++++++++++++----
 1 file changed, 18 insertions(+), 4 deletions(-)

diff --git a/drivers/power/sequencing/pwrseq-pcie-m2.c b/drivers/power/sequencing/pwrseq-pcie-m2.c
index 693b0d33f8cf..56e4f40e0480 100644
--- a/drivers/power/sequencing/pwrseq-pcie-m2.c
+++ b/drivers/power/sequencing/pwrseq-pcie-m2.c

@@ -83,8 +83,22 @@ static int pwrseq_pci_m2_e_bt_disable(struct pwrseq_device *pwrseq)
 	return gpiod_set_value_cansleep(ctx->w_disable2_gpio, 1);
 }
 
-static const struct pwrseq_unit_data pwrseq_pcie_m2_e_bt_unit_data = {
-	.name = "bt-enable",
+/*
+ * XXX There are two Bluetooth units to allow either one to be able to power
+ * off and thus reset the controller. In practice only one of the interfaces
+ * is used, so there is no conflict. However userspace could power off the
+ * USB unit by disabling the associated USB port, without the UART unit or
+ * its consumer ever knowing.
+ */
+static const struct pwrseq_unit_data pwrseq_pcie_m2_e_bt_uart_unit_data = {
+	.name = "bt-uart-enable",
+	.deps = pwrseq_pcie_m2_unit_deps,
+	.enable = pwrseq_pci_m2_e_bt_enable,
+	.disable = pwrseq_pci_m2_e_bt_disable,
+};
+
+static const struct pwrseq_unit_data pwrseq_pcie_m2_e_bt_usb_unit_data = {
+	.name = "bt-usb-enable",
 	.deps = pwrseq_pcie_m2_unit_deps,
 	.enable = pwrseq_pci_m2_e_bt_enable,
 	.disable = pwrseq_pci_m2_e_bt_disable,

@@ -130,13 +144,13 @@ static int pwrseq_pcie_m2_e_pwup_delay(struct pwrseq_device *pwrseq)
 
 static const struct pwrseq_target_data pwrseq_pcie_m2_e_uart_target_data = {
 	.name = "uart",
-	.unit = &pwrseq_pcie_m2_e_bt_unit_data,
+	.unit = &pwrseq_pcie_m2_e_bt_uart_unit_data,
 	.post_enable = pwrseq_pcie_m2_e_pwup_delay,
 };
 
 static const struct pwrseq_target_data pwrseq_pcie_m2_e_usb_target_data = {
 	.name = "usb",
-	.unit = &pwrseq_pcie_m2_e_bt_unit_data,
+	.unit = &pwrseq_pcie_m2_e_bt_usb_unit_data,
 };
 
 static const struct pwrseq_target_data pwrseq_pcie_m2_e_pcie_target_data = {

-- 
2.55.0.795.g602f6c329a-goog