On Wed, Apr 24, 2024 at 3:31 PM Wren Turkal <wt@xxxxxxxxxxxxxxxx> wrote:
>
> On 4/24/24 6:22 AM, quic_zijuhu wrote:
> > On 4/24/2024 9:18 PM, Bartosz Golaszewski wrote:
> >> On Wed, 24 Apr 2024 at 15:10, Wren Turkal <wt@xxxxxxxxxxxxxxxx> wrote:
> >>>
> >>> On 4/24/24 5:29 AM, Bartosz Golaszewski wrote:
> >>>> From: Bartosz Golaszewski<bartosz.golaszewski@xxxxxxxxxx>
> >>>>
> >>>> Any return value from gpiod_get_optional() other than a pointer to a
> >>>> GPIO descriptor or a NULL-pointer is an error and the driver should
> >>>> abort probing. That being said: commit 56d074d26c58 ("Bluetooth: hci_qca:
> >>>> don't use IS_ERR_OR_NULL() with gpiod_get_optional()") no longer sets
> >>>> power_ctrl_enabled on NULL-pointer returned by
> >>>> devm_gpiod_get_optional(). Restore this behavior but bail-out on errors.
> >>>> While at it: also bail-out on error returned when trying to get the
> >>>> "swctrl" GPIO.
> >>>>
> >>>> Reported-by: Wren Turkal<wt@xxxxxxxxxxxxxxxx>
> >>>> Reported-by: Zijun Hu<quic_zijuhu@xxxxxxxxxxx>
> >>>> Closes:https://lore.kernel.org/linux-bluetooth/1713449192-25926-2-git-send-email-quic_zijuhu@xxxxxxxxxxx/
> >>>> Fixes: 56d074d26c58 ("Bluetooth: hci_qca: don't use IS_ERR_OR_NULL() with gpiod_get_optional()")
> >>>> Reviewed-by: Krzysztof Kozlowski<krzysztof.kozlowski@xxxxxxxxxx>
> >>>> Signed-off-by: Bartosz Golaszewski<bartosz.golaszewski@xxxxxxxxxx>
> >>>
> >>> Tested-by: "Wren Turkal" <wt@xxxxxxxxxxxxxxxx>
> >>>
> >>>
> >>> Like this?
> >>
> >> Yes, awesome, thanks.
> >>
> >> This is how reviewing works too in the kernel, look at what Krzysztof
> >> did under v1, he just wrote:
> >>
> >> Reviewed-by: Krzysztof Kozlowski<krzysztof.kozlowski@xxxxxxxxxx>
> >>
> > v1 have obvious something wrong as i pointed and verified.
> > so i think it is not suitable to attach v1's review-by tag to v2 anyway.
>
> @Zijun, your concern is that current DTs may not define the gpio and
> this will cause the bluetooth not to work?
>
This is simply not true. If the GPIO is not specified,
gpiod_get_optional() will return NULL and GPIO APIs will work just
fine.
That being said: the contract for whether a GPIO is needed or not is
not in the driver C code or released DT sources. It's in the bindings
documents under Documentation/devicetree/bindings/. This is the source
of truth. So if the binding for a given model has always said
"required: enable-gpios" then we're absolutely in our rights to fix
the driver to conform to that even if previously omitted. If you think
otherwise - relax the bindings first.
Bart
> Would that not more appropriately be fixed by machine-specific fixups
> for the DT?
>
> >
> >> And mailing list tools will pick it up.
> >>
> >> Bartosz
> >
>
> --
> You're more amazing than you think!
