On Wed, 2023-12-27 at 11:10 +0100, Johan Hovold wrote:

A recent commit restored the original (and still documented) semantics for the HCI_QUIRK_USE_BDADDR_PROPERTY quirk so that the device address is considered invalid unless an address is provided by firmware.

This specifically means that this flag must only be set for devices with invalid addresses, but the Broadcom BCM4377 driver has so far been setting this flag unconditionally.

Fortunately the driver already checks for invalid addresses during setup and sets the HCI_QUIRK_INVALID_BDADDR flag, which can simply be replaced with HCI_QUIRK_USE_BDADDR_PROPERTY to indicate that the default address is invalid but can be overridden by firmware (long term, this should probably just always be allowed).

Fixes: 6945795bc81a ("Bluetooth: fix use-bdaddr-property quirk") Cc: stable@vger.kernel.org      # 6.5 Reported-by: Felix Zhang mrman@mrman314.tech Link: https://lore.kernel.org/r/77419ffacc5b4875e920e038332575a2a5bff29f.camel@mrm... Signed-off-by: Johan Hovold johan+linaro@kernel.org

---

drivers/bluetooth/hci_bcm4377.c | 3 +--  1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/bluetooth/hci_bcm4377.c b/drivers/bluetooth/hci_bcm4377.c index a61757835695..9a7243d5db71 100644 --- a/drivers/bluetooth/hci_bcm4377.c +++ b/drivers/bluetooth/hci_bcm4377.c @@ -1417,7 +1417,7 @@ static int bcm4377_check_bdaddr(struct bcm4377_data *bcm4377)     bda = (struct hci_rp_read_bd_addr *)skb->data;   if (!bcm4377_is_valid_bdaddr(bcm4377, &bda->bdaddr))

• set_bit(HCI_QUIRK_INVALID_BDADDR, &bcm4377->hdev-
  

...

quirks);

• set_bit(HCI_QUIRK_USE_BDADDR_PROPERTY, &bcm4377-
  

...

hdev->quirks);

kfree_skb(skb);   return 0; @@ -2368,7 +2368,6 @@ static int bcm4377_probe(struct pci_dev *pdev, const struct pci_device_id *id)   hdev->set_bdaddr = bcm4377_hci_set_bdaddr;   hdev->setup = bcm4377_hci_setup;

• set_bit(HCI_QUIRK_USE_BDADDR_PROPERTY, &hdev->quirks);

if (bcm4377->hw->broken_mws_transport_config)   set_bit(HCI_QUIRK_BROKEN_MWS_TRANSPORT_CONFIG, &hdev->quirks);   if (bcm4377->hw->broken_ext_scan)

Hi,

Thanks for the patch! I have just tested the patch on my MacBookAir9,1 with some modifications to make it patch with the master branch of the Linux kernel, and it works great! The device is recognized by the system; and I can scan, pair, connect, and play audio just fine. If you want me to send over the modified patch that I have used, please let me know.

Thanks!

On Wed, Dec 27, 2023 at 5:10 AM Johan Hovold johan+linaro@kernel.org wrote:

A recent commit restored the original (and still documented) semantics for the HCI_QUIRK_USE_BDADDR_PROPERTY quirk so that the device address is considered invalid unless an address is provided by firmware.

This specifically means that this flag must only be set for devices with invalid addresses, but the Broadcom BCM4377 driver has so far been setting this flag unconditionally.

Fortunately the driver already checks for invalid addresses during setup and sets the HCI_QUIRK_INVALID_BDADDR flag, which can simply be replaced with HCI_QUIRK_USE_BDADDR_PROPERTY to indicate that the default address is invalid but can be overridden by firmware (long term, this should probably just always be allowed).

Fixes: 6945795bc81a ("Bluetooth: fix use-bdaddr-property quirk") Cc: stable@vger.kernel.org # 6.5 Reported-by: Felix Zhang mrman@mrman314.tech Link: https://lore.kernel.org/r/77419ffacc5b4875e920e038332575a2a5bff29f.camel@mrm... Signed-off-by: Johan Hovold johan+linaro@kernel.org

---

drivers/bluetooth/hci_bcm4377.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/bluetooth/hci_bcm4377.c b/drivers/bluetooth/hci_bcm4377.c index a61757835695..9a7243d5db71 100644 --- a/drivers/bluetooth/hci_bcm4377.c +++ b/drivers/bluetooth/hci_bcm4377.c @@ -1417,7 +1417,7 @@ static int bcm4377_check_bdaddr(struct bcm4377_data *bcm4377)

    bda = (struct hci_rp_read_bd_addr *)skb->data;
    if (!bcm4377_is_valid_bdaddr(bcm4377, &bda->bdaddr))

•           set_bit(HCI_QUIRK_INVALID_BDADDR, &bcm4377->hdev->quirks);
  

•           set_bit(HCI_QUIRK_USE_BDADDR_PROPERTY, &bcm4377->hdev->quirks);
  
    kfree_skb(skb);
    return 0;
  

@@ -2368,7 +2368,6 @@ static int bcm4377_probe(struct pci_dev *pdev, const struct pci_device_id *id) hdev->set_bdaddr = bcm4377_hci_set_bdaddr; hdev->setup = bcm4377_hci_setup;

•   set_bit(HCI_QUIRK_USE_BDADDR_PROPERTY, &hdev->quirks);
    if (bcm4377->hw->broken_mws_transport_config)
            set_bit(HCI_QUIRK_BROKEN_MWS_TRANSPORT_CONFIG, &hdev->quirks);
    if (bcm4377->hw->broken_ext_scan)
  

-- 2.41.0

Looks good to me. This replaces the other patch[1], I take it?

Reviewed-by: Neal Gompa neal@gompa.dev

[1]: https://lore.kernel.org/asahi/aaa107865f4cbd61f8f9006fd3e7ac43b5d1bdad.camel...