Qualcomm Bluetooth controllers may not have been provisioned with a valid device address and instead end up using the default address 00:00:00:00:5a:ad.
This address is now used to determine if a controller has a valid address or if one needs to be provided through devicetree or by user space before the controller can be used.
It turns out that the WCN3991 controllers used in Chromium Trogdor machines use a different default address, 39:98:00:00:5a:ad, which also needs to be marked as invalid so that the correct address is fetched from the devicetree.
Qualcomm has unfortunately not yet provided any answers as to whether the 39:98 encodes a hardware id and if there are other variants of the default address that needs to be handled by the driver.
For now, add the Trogdor WCN3991 default address to the device address check to avoid having these controllers start with the default address instead of their assigned addresses.
Fixes: 00567f70051a ("Bluetooth: qca: fix invalid device address check") Cc: stable@vger.kernel.org # 6.5 Cc: Doug Anderson dianders@chromium.org Cc: Janaki Ramaiah Thota quic_janathot@quicinc.com Signed-off-by: Johan Hovold johan+linaro@kernel.org --- drivers/bluetooth/btqca.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
Luiz and Doug,
As the offending commit is now on its way into 6.9, let's just add the default address that the Trogdor machines uses to the address check.
We can always amend this when/if Qualcomm provides some more details, or, in the worst case, when users report that they need to re-pair their Bluetooth gadgets if there are further variations of the default address.
Johan
diff --git a/drivers/bluetooth/btqca.c b/drivers/bluetooth/btqca.c index 216826c31ee3..cfa71708397b 100644 --- a/drivers/bluetooth/btqca.c +++ b/drivers/bluetooth/btqca.c @@ -16,6 +16,7 @@ #define VERSION "0.1"
#define QCA_BDADDR_DEFAULT (&(bdaddr_t) {{ 0xad, 0x5a, 0x00, 0x00, 0x00, 0x00 }}) +#define QCA_BDADDR_WCN3991 (&(bdaddr_t) {{ 0xad, 0x5a, 0x00, 0x00, 0x98, 0x39 }})
int qca_read_soc_version(struct hci_dev *hdev, struct qca_btsoc_version *ver, enum qca_btsoc_type soc_type) @@ -638,8 +639,10 @@ static int qca_check_bdaddr(struct hci_dev *hdev) }
bda = (struct hci_rp_read_bd_addr *)skb->data; - if (!bacmp(&bda->bdaddr, QCA_BDADDR_DEFAULT)) + if (!bacmp(&bda->bdaddr, QCA_BDADDR_DEFAULT) || + !bacmp(&bda->bdaddr, QCA_BDADDR_WCN3991)) { set_bit(HCI_QUIRK_USE_BDADDR_PROPERTY, &hdev->quirks); + }
kfree_skb(skb);
Hi,
On Thu, Apr 25, 2024 at 12:56 AM Johan Hovold johan+linaro@kernel.org wrote:
Qualcomm Bluetooth controllers may not have been provisioned with a valid device address and instead end up using the default address 00:00:00:00:5a:ad.
This address is now used to determine if a controller has a valid address or if one needs to be provided through devicetree or by user space before the controller can be used.
It turns out that the WCN3991 controllers used in Chromium Trogdor machines use a different default address, 39:98:00:00:5a:ad, which also needs to be marked as invalid so that the correct address is fetched from the devicetree.
Qualcomm has unfortunately not yet provided any answers as to whether the 39:98 encodes a hardware id and if there are other variants of the default address that needs to be handled by the driver.
For now, add the Trogdor WCN3991 default address to the device address check to avoid having these controllers start with the default address instead of their assigned addresses.
Fixes: 00567f70051a ("Bluetooth: qca: fix invalid device address check") Cc: stable@vger.kernel.org # 6.5 Cc: Doug Anderson dianders@chromium.org Cc: Janaki Ramaiah Thota quic_janathot@quicinc.com Signed-off-by: Johan Hovold johan+linaro@kernel.org
drivers/bluetooth/btqca.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
Luiz and Doug,
As the offending commit is now on its way into 6.9, let's just add the default address that the Trogdor machines uses to the address check.
We can always amend this when/if Qualcomm provides some more details, or, in the worst case, when users report that they need to re-pair their Bluetooth gadgets if there are further variations of the default address.
I can confirm that this at least gets my boards using their proper BT address. While I still wonder if this is the best strategy to go with, I can agree that this is an expedient fix to land it and works:
Tested-by: Douglas Anderson dianders@chromium.org
We can continue discussion in response to your original patch [1] to figure out if this is going to be our long term strategy or not.
[1] https://lore.kernel.org/r/20240416091509.19995-1-johan+linaro@kernel.org
Hello:
This patch was applied to bluetooth/bluetooth-next.git (master) by Luiz Augusto von Dentz luiz.von.dentz@intel.com:
On Thu, 25 Apr 2024 09:55:03 +0200 you wrote:
Qualcomm Bluetooth controllers may not have been provisioned with a valid device address and instead end up using the default address 00:00:00:00:5a:ad.
This address is now used to determine if a controller has a valid address or if one needs to be provided through devicetree or by user space before the controller can be used.
[...]
Here is the summary with links: - Bluetooth: qca: fix wcn3991 device address check https://git.kernel.org/bluetooth/bluetooth-next/c/0a0f9014b784
You are awesome, thank you!
linux-stable-mirror@lists.linaro.org