mbox series

[0/3] gpio: bcm-kona: Various GPIO fixups

Message ID 20250130-kona-gpio-fixes-v1-0-dd61e0c0a9e7@gmail.com
Headers show
Series gpio: bcm-kona: Various GPIO fixups | expand

Message

Artur Weber Jan. 30, 2025, 5:10 p.m. UTC
Fixes two issues that were preventing GPIO from working correctly:

- Lock/unlock functions tried to write the wrong bit to the unlock
  registers for GPIOs with numbers larger than 32

- GPIOs only initialized as IRQs did not unlock the configuration
  registers, causing IRQ-related configuration (e.g. setting the IRQ
  type) to fail.

Also includes a minor fix to add a missing newline to an error message.

Tested on a Samsung Galaxy Grand Neo (baffinlite rev02) with a BCM23550
(DTS not yet in mainline).

Signed-off-by: Artur Weber <aweber.kernel@gmail.com>
---
Artur Weber (3):
      gpio: bcm-kona: Fix GPIO lock/unlock for banks above bank 0
      gpio: bcm-kona: Make sure GPIO bits are unlocked when requesting IRQ
      gpio: bcm-kona: Add missing newline to dev_err format string

 drivers/gpio/gpio-bcm-kona.c | 73 ++++++++++++++++++++++++++++++++++++--------
 1 file changed, 60 insertions(+), 13 deletions(-)
---
base-commit: ffd294d346d185b70e28b1a28abe367bbfe53c04
change-id: 20250130-kona-gpio-fixes-6e74595e69f2

Best regards,

Comments

Linus Walleij Feb. 6, 2025, 9:34 a.m. UTC | #1
Hi Artur,

On Thu, Jan 30, 2025 at 6:10 PM Artur Weber <aweber.kernel@gmail.com> wrote:

> Fixes two issues that were preventing GPIO from working correctly:
>
> - Lock/unlock functions tried to write the wrong bit to the unlock
>   registers for GPIOs with numbers larger than 32
>
> - GPIOs only initialized as IRQs did not unlock the configuration
>   registers, causing IRQ-related configuration (e.g. setting the IRQ
>   type) to fail.
>
> Also includes a minor fix to add a missing newline to an error message.
>
> Tested on a Samsung Galaxy Grand Neo (baffinlite rev02) with a BCM23550
> (DTS not yet in mainline).
>
> Signed-off-by: Artur Weber <aweber.kernel@gmail.com>

Looks good to me, do you want to resend the patches picking up the
Acks or do you want me to apply as-is?

Should this go in as urgent fixes (-rcN) or as nonurgent?

The DTS not being mainline suggests nonurgent but are there
other systems suffering?

Yours,
Linus Walleij
Linus Walleij Feb. 6, 2025, 9:36 a.m. UTC | #2
On Thu, Feb 6, 2025 at 10:34 AM Linus Walleij <linus.walleij@linaro.org> wrote:
> On Thu, Jan 30, 2025 at 6:10 PM Artur Weber <aweber.kernel@gmail.com> wrote:
>
> > Fixes two issues that were preventing GPIO from working correctly:
> >
> > - Lock/unlock functions tried to write the wrong bit to the unlock
> >   registers for GPIOs with numbers larger than 32
> >
> > - GPIOs only initialized as IRQs did not unlock the configuration
> >   registers, causing IRQ-related configuration (e.g. setting the IRQ
> >   type) to fail.
> >
> > Also includes a minor fix to add a missing newline to an error message.
> >
> > Tested on a Samsung Galaxy Grand Neo (baffinlite rev02) with a BCM23550
> > (DTS not yet in mainline).
> >
> > Signed-off-by: Artur Weber <aweber.kernel@gmail.com>
>
> Looks good to me, do you want to resend the patches picking up the
> Acks or do you want me to apply as-is?

Actually it's Bartosz applying, heh confused this for pinctrl.

But most comments go anyway I think.

Yours,
Linus Walleij
Artur Weber Feb. 6, 2025, 3:37 p.m. UTC | #3
On 6.02.2025 10:34, Linus Walleij wrote:
> Hi Artur,
> 
> On Thu, Jan 30, 2025 at 6:10 PM Artur Weber <aweber.kernel@gmail.com> wrote:
> 
>> Fixes two issues that were preventing GPIO from working correctly:
>>
>> - Lock/unlock functions tried to write the wrong bit to the unlock
>>    registers for GPIOs with numbers larger than 32
>>
>> - GPIOs only initialized as IRQs did not unlock the configuration
>>    registers, causing IRQ-related configuration (e.g. setting the IRQ
>>    type) to fail.
>>
>> Also includes a minor fix to add a missing newline to an error message.
>>
>> Tested on a Samsung Galaxy Grand Neo (baffinlite rev02) with a BCM23550
>> (DTS not yet in mainline).
>>
>> Signed-off-by: Artur Weber <aweber.kernel@gmail.com>
> 
> Looks good to me, do you want to resend the patches picking up the
> Acks or do you want me to apply as-is?

I'll send a v2 in a second with the style fixes suggested by Markus[1].
After that, it should be good to go.

> Should this go in as urgent fixes (-rcN) or as nonurgent?
> 
> The DTS not being mainline suggests nonurgent but are there
> other systems suffering?

Non-urgent is fine. AFAICT the only upstream-supported Broadcom Kona
devices at the moment are old Broadcom dev boards; while they
technically are also affected by the bug, I doubt anyone's actively
using them. As for the devices I'm testing with, there's still quite
a few fixups I plan to make before I consider them ready to upstream.

Best regards
Artur

[1] https://lore.kernel.org/lkml/CAGt4E5sqd_Aojk+boD5K5EiRfOsiU+jYY5EV0DP6TFut291HnQ@mail.gmail.com/
Artur Weber Feb. 6, 2025, 5:48 p.m. UTC | #4
On 6.02.2025 16:37, Artur Weber wrote:
> On 6.02.2025 10:34, Linus Walleij wrote:
>> Hi Artur,
>>
>> On Thu, Jan 30, 2025 at 6:10 PM Artur Weber <aweber.kernel@gmail.com> 
>> wrote:
>>
>>> Fixes two issues that were preventing GPIO from working correctly:
>>>
>>> - Lock/unlock functions tried to write the wrong bit to the unlock
>>>    registers for GPIOs with numbers larger than 32
>>>
>>> - GPIOs only initialized as IRQs did not unlock the configuration
>>>    registers, causing IRQ-related configuration (e.g. setting the IRQ
>>>    type) to fail.
>>>
>>> Also includes a minor fix to add a missing newline to an error message.
>>>
>>> Tested on a Samsung Galaxy Grand Neo (baffinlite rev02) with a BCM23550
>>> (DTS not yet in mainline).
>>>
>>> Signed-off-by: Artur Weber <aweber.kernel@gmail.com>
>>
>> Looks good to me, do you want to resend the patches picking up the
>> Acks or do you want me to apply as-is?
> 
> I'll send a v2 in a second with the style fixes suggested by Markus[1].
> After that, it should be good to go.
> 
>> Should this go in as urgent fixes (-rcN) or as nonurgent?
>>
>> The DTS not being mainline suggests nonurgent but are there
>> other systems suffering?
> 
> Non-urgent is fine. AFAICT the only upstream-supported Broadcom Kona
> devices at the moment are old Broadcom dev boards; while they
> technically are also affected by the bug, I doubt anyone's actively
> using them. As for the devices I'm testing with, there's still quite
> a few fixups I plan to make before I consider them ready to upstream.
> 
> Best regards
> Artur
> 
> [1] https://lore.kernel.org/lkml/ 
> CAGt4E5sqd_Aojk+boD5K5EiRfOsiU+jYY5EV0DP6TFut291HnQ@mail.gmail.com/

v2 sent: https://lore.kernel.org/lkml/20250206-kona-gpio-fixes-v2-0-409135eab780@gmail.com/T/#t

Best regards
Artur