Thread (23 messages) flat view 23 messages, 3 authors, 2021-07-30

Re: [PATCH v5 05/11] Bluetooth: btintel: Fix the first HCI command not work with ROM device

From: Tedd Ho-Jeong An <hidden>
Date: 2021-07-29 21:59:51

Hi Marcel,

On Thu, 2021-07-29 at 21:35 +0200, Marcel Holtmann wrote:
Hi Tedd,
quoted
The some legacy ROM controllers have a bug with the first HCI command
sent to it returning number of completed commands as zero, which would
stall the command processing in the Bluetooth core.

As a workaround, send HCI Rest command first which will reset the
controller to fix the issue.

Signed-off-by: Tedd Ho-Jeong An <redacted>
---
drivers/bluetooth/btintel.c | 21 +++++++++++++++++++++
drivers/bluetooth/btintel.h |  1 +
drivers/bluetooth/btusb.c   | 16 ++++++++++++++--
3 files changed, 36 insertions(+), 2 deletions(-)
diff --git a/drivers/bluetooth/btintel.c b/drivers/bluetooth/btintel.c
index bf0ad05b80fe..65ecf2ae9a10 100644
--- a/drivers/bluetooth/btintel.c
+++ b/drivers/bluetooth/btintel.c
@@ -1659,6 +1659,7 @@ static int btintel_legacy_rom_setup(struct hci_dev *hdev,
int btintel_setup_combined(struct hci_dev *hdev)
{
+	struct btintel_data *intel = hci_get_priv(hdev);
	const u8 param[1] = { 0xFF };
	struct intel_version ver;
	struct intel_version_tlv ver_tlv;
@@ -1667,6 +1668,26 @@ int btintel_setup_combined(struct hci_dev *hdev)
	BT_DBG("%s", hdev->name);

+	/* The some controllers have a bug with the first HCI command sent to it
+	 * returning number of completed commands as zero. This would stall the
+	 * command processing in the Bluetooth core.
+	 *
+	 * As a workaround, send HCI Reset command first which will reset the
+	 * number of completed commands and allow normal command processing
+	 * from now on.
+	 */
+	if (test_bit(INTEL_BROKEN_READ_VERSION, &intel->flags)) {
+		skb = __hci_cmd_sync(hdev, HCI_OP_RESET, 0, NULL,
+				     HCI_INIT_TIMEOUT);
+		if (IS_ERR(skb)) {
+			bt_dev_err(hdev,
+				   "sending initial HCI reset failed (%ld)",
+				   PTR_ERR(skb));
+			return PTR_ERR(skb);
+		}
+		kfree_skb(skb);
+	}
+
	/* Starting from TyP device, the command parameter and response are
	 * changed even though the OCF for HCI_Intel_Read_Version command
	 * remains same. The legacy devices can handle even if the
diff --git a/drivers/bluetooth/btintel.h b/drivers/bluetooth/btintel.h
index df7aa30142b4..29b678364a79 100644
--- a/drivers/bluetooth/btintel.h
+++ b/drivers/bluetooth/btintel.h
@@ -143,6 +143,7 @@ struct intel_debug_features {
#define INTEL_FIRMWARE_LOADED		2
#define INTEL_FIRMWARE_FAILED		3
#define INTEL_BOOTING			4
+#define INTEL_BROKEN_READ_VERSION	5

struct btintel_data {
	unsigned long flags;
diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
index 8c54ab03ee63..a64473c525eb 100644
--- a/drivers/bluetooth/btusb.c
+++ b/drivers/bluetooth/btusb.c
@@ -62,6 +62,7 @@ static struct usb_driver btusb_driver;
#define BTUSB_QCA_WCN6855	0x1000000
#define BTUSB_INTEL_NEWGEN	0x2000000
#define BTUSB_INTEL_COMBINED	0x4000000
+#define BTUSB_INTEL_BROKEN_READ_VERSION 0x8000000

static const struct usb_device_id btusb_table[] = {
	/* Generic Bluetooth USB device */
@@ -376,11 +377,14 @@ static const struct usb_device_id blacklist_table[] = {
						     BTUSB_WIDEBAND_SPEECH |
						     BTUSB_VALID_LE_STATES },
	{ USB_DEVICE(0x8087, 0x07da), .driver_info = BTUSB_CSR },
-	{ USB_DEVICE(0x8087, 0x07dc), .driver_info = BTUSB_INTEL_COMBINED },
-	{ USB_DEVICE(0x8087, 0x0a2a), .driver_info = BTUSB_INTEL_COMBINED },
+	{ USB_DEVICE(0x8087, 0x07dc), .driver_info = BTUSB_INTEL_COMBINED |
+						     BTUSB_INTEL_BROKEN_READ_VERSION },
+	{ USB_DEVICE(0x8087, 0x0a2a), .driver_info = BTUSB_INTEL_COMBINED |
+						     BTUSB_INTEL_BROKEN_READ_VERSION },
	{ USB_DEVICE(0x8087, 0x0a2b), .driver_info = BTUSB_INTEL_NEW |
						     BTUSB_WIDEBAND_SPEECH },
	{ USB_DEVICE(0x8087, 0x0aa7), .driver_info = BTUSB_INTEL_COMBINED |
+						     BTUSB_INTEL_BROKEN_READ_VERSION |
						     BTUSB_WIDEBAND_SPEECH },
	{ USB_DEVICE(0x8087, 0x0aaa), .driver_info = BTUSB_INTEL_NEW |
						     BTUSB_WIDEBAND_SPEECH |
can you check that all 3 have this problem? Don’t we ever produced a ROM where this is fixed?
It looks like the early version of ROM (WP2) causes the problem. StP and SdP don't have the problem.
I will update accordingly.

WP2 - I only had mini-PCIe form factor card and it is broken.
< HCI Command: Intel Read Version (0x3f|0x0005) plen 1                        #1 [hci0] 9.212217
        Requested Type:
          All Supported Types(0xff)
HCI Event: Command Complete (0x0e) plen 13                                      #2 [hci0] 9.213338
      Intel Read Version (0x3f|0x0005) ncmd 0
        Status: Success (0x00)
        Hardware platform: 0x37
        Hardware variant: 0x07
        Hardware revision: 1.0
        Firmware variant: 0x01
        Firmware revision: 8.0
        Firmware build: 2-3.2013
        Firmware patch: 0

StP - OK. 
< HCI Command: Intel Read Version (0x3f|0x0005) plen 1                      #3 [hci0] 108.053455
        Requested Type:
          All Supported Types(0xff)
HCI Event: Command Complete (0x0e) plen 13                                    #4 [hci0] 108.054034
      Intel Read Version (0x3f|0x0005) ncmd 1
        Status: Success (0x00)
        Hardware platform: 0x37
        Hardware variant: 0x08
        Hardware revision: 1.0
        Firmware variant: 0x01
        Firmware revision: 1.0
        Firmware build: 3-17.2014
        Firmware patch: 0

SdP - OK.
< HCI Command: Intel Read Version (0x3f|0x0005) plen 1                    #400 [hci0] 173.911992
        Requested Type:
          All Supported Types(0xff)
HCI Event: Command Complete (0x0e) plen 13                                  #401 [hci0] 173.912576
      Intel Read Version (0x3f|0x0005) ncmd 1
        Status: Success (0x00)
        Hardware platform: 0x37
        Hardware variant: 0x08
        Hardware revision: 1.0
        Firmware variant: 0x22
        Firmware revision: 5.0
        Firmware build: 25-20.2015
        Firmware patch: 0

quoted
@@ -4221,6 +4225,11 @@ static int btusb_probe(struct usb_interface *intf,
	priv_size = 0;

+	if (id->driver_info & BTUSB_INTEL_COMBINED) {
+		/* Allocate extra space for Intel device */
+		priv_size += sizeof(struct btintel_data);
+	}
+
This needs to be in the previous patch.

Regards

Marcel
  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help