Re: [PATCH] Bluetooth: HCI: Introduce HCI_QUIRK_BROKEN_LE_CODED

[Bruno Pitrus <brunopitrus@hotmail.com>]

Dnia środa, 23 sierpnia 2023 10:26:31 CEST Paul Menzel pisze:
> Dear Luiz,
> 
> 
> Thank you for your answer.
> 
> Am 22.08.23 um 22:20 schrieb Luiz Augusto von Dentz:
> > Hi Paul,
> > 
> > On Tue, Aug 22, 2023 at 1:01 PM Paul Menzel <pmenzel@molgen.mpg.de> wrote:
> >>
> >> [CC: +Bruno]
> >>
> >> Dear Luiz,
> >>
> >>
> >> Thank you for the patch.
> >>
> >> Am 22.08.23 um 21:14 schrieb Luiz Augusto von Dentz:
> >>> From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
> >>>
> >>> This introduces HCI_QUIRK_BROKEN_LE_CODED is is used to indicate that
> >>
> >> …. It is used …
> >>
> >>> LE Coded PHY shall not be used, it is then set for some Intel models
> >>> that claims to support it but when used causes many problems.
> >>
> >> that claim to …
> >>
> >>> Link: https://github.com/bluez/bluez/issues/577
> >>> Link: https://github.com/bluez/bluez/issues/582
> >>> Link: https://lore.kernel.org/linux-bluetooth/CABBYNZKco-v7wkjHHexxQbgwwSz-S=GZ=dZKbRE1qxT1h4fFbQ@mail.gmail.com/T/#
> >>> Fixes: 288c90224eec ("Bluetooth: Enable all supported LE PHY by default")
> >>> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
> >>> ---
> >>>    drivers/bluetooth/btintel.c      |  2 ++
> >>>    include/net/bluetooth/hci.h      | 10 ++++++++++
> >>>    include/net/bluetooth/hci_core.h |  4 +++-
> >>>    3 files changed, 15 insertions(+), 1 deletion(-)
> >>>
> >>> diff --git a/drivers/bluetooth/btintel.c b/drivers/bluetooth/btintel.c
> >>> index 9b239ce96fa4..3ed60b2b0340 100644
> >>> --- a/drivers/bluetooth/btintel.c
> >>> +++ b/drivers/bluetooth/btintel.c
> >>> @@ -2776,6 +2776,8 @@ static int btintel_setup_combined(struct hci_dev *hdev)
> >>>                case 0x11:      /* JfP */
> >>>                case 0x12:      /* ThP */
> >>>                case 0x13:      /* HrP */
> >>> +                     set_bit(HCI_QUIRK_BROKEN_LE_CODED, &hdev->quirks);
> >>> +                     fallthrough;
> >>>                case 0x14:      /* CcP */
> >>>                        set_bit(HCI_QUIRK_VALID_LE_STATES, &hdev->quirks);
> >>>                        fallthrough;
> >>> diff --git a/include/net/bluetooth/hci.h b/include/net/bluetooth/hci.h
> >>> index c58425d4c592..767921d7f5c1 100644
> >>> --- a/include/net/bluetooth/hci.h
> >>> +++ b/include/net/bluetooth/hci.h
> >>> @@ -319,6 +319,16 @@ enum {
> >>>         * This quirk must be set before hci_register_dev is called.
> >>>         */
> >>>        HCI_QUIRK_USE_MSFT_EXT_ADDRESS_FILTER,
> >>> +
> >>> +     /*
> >>> +      * When this quirk is set, LE Coded PHY is shall not be used. This is
> >>
> >> s/is shall/shall/
> >>
> >>> +      * required for some Intel controllers which erroneously claim to
> >>> +      * support it but it causes problems with extended scanning.
> >>> +      *
> >>> +      * This quirk can be set before hci_register_dev is called or
> >>> +      * during the hdev->setup vendor callback.
> >>> +      */
> >>> +     HCI_QUIRK_BROKEN_LE_CODED,
> >>>    };
> >>>
> >>>    /* HCI device flags */
> >>> diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h
> >>> index 6e2988b11f99..e6359f7346f1 100644
> >>> --- a/include/net/bluetooth/hci_core.h
> >>> +++ b/include/net/bluetooth/hci_core.h
> >>> @@ -1817,7 +1817,9 @@ void hci_conn_del_sysfs(struct hci_conn *conn);
> >>>    #define scan_2m(dev) (((dev)->le_tx_def_phys & HCI_LE_SET_PHY_2M) || \
> >>>                      ((dev)->le_rx_def_phys & HCI_LE_SET_PHY_2M))
> >>>
> >>> -#define le_coded_capable(dev) (((dev)->le_features[1] & HCI_LE_PHY_CODED))
> >>> +#define le_coded_capable(dev) (((dev)->le_features[1] & HCI_LE_PHY_CODED) && \
> >>> +                            !test_bit(HCI_QUIRK_BROKEN_LE_CODED, \
> >>> +                                      &(dev)->quirks))
> >>>
> >>>    #define scan_coded(dev) (((dev)->le_tx_def_phys & HCI_LE_SET_PHY_CODED) || \
> >>>                         ((dev)->le_rx_def_phys & HCI_LE_SET_PHY_CODED))
> >>
> >> Will this be future proof, once firmware for the broken controllers are
> >> fixed?
> > 
> > Yes, it won't cause any regressions if the firmware is fixed in the
> > future, but LE coded PHY will need to be re-enabled by removing the
> > quirks, this is why we say it is broken but we can't depend on runtime
> > detection and should probably print a warning until we have a proper
> > fix for it lands at the firmware side. We could in theory set
> > HCI_QUIRK_BROKEN_LE_CODED based on the firmware version but firmware
> > versioning schemes tend to change so I'm afraid that could also cause
> > regression like this in the future.
> 
> Understood. Maybe you could add the known broken firmware versions to 
> the commit message for posterity though.
> 
> 
> Kind regards,
> 
> Paul
> 
Thanks,
This patch partly works for me. The mouse now takes several dozen seconds to detect where the computer did not find it at all before.
But note that in kernel 6.3.x it was always detected immediately.