Re: [PATCH v4 0/4] Refactor 104-quad-8 to match device operations

[William Breathitt Gray]

On Mon, Apr 17, 2023 at 03:50:46PM -0400, William Breathitt Gray wrote:
> Changes in v4:
>  - Fix endianness errors by utilizing get_unaligned_le24() and
>    put_unaligned_le24()
>  - Mention benefits of using regmap in the commit description
>  - Use "int ret" for regmap_* return values throughout for consistency
>  - Reorganize declaration lists to prioritize longer lines first
> Changes in v3:
>  - Add __always_inline attribute for quad8_control_register_update()
> Changes in v2:
>  - Drop FIELD_MODIFY() macro introduction; u8p_replace_bits() is
>    utilized instead for the same purpose
>  - Replace FIELD_PREP() and FIELD_GET() with u8_encode_bits() and
>    u8_get_bits()
>  - Replace FIELD_MODIFY() with u8p_replace_bits()
>  - Wrap up control register update in quad8_control_register_update()
>  - Utilize ioread8_rep() and iowrite8_rep() to read and write counter
>    data
> 
> The 104-quad-8 driver was initially introduced to the IIO subsystem
> where it didn't quite fit with the existing paradigm [0]; these
> differences eventually led to the creation of the Counter subsystem[1].
> As a result of its awkward beginnings, the design of the 104-quad-8
> driver was structured around maintaining abstract state buffers that
> would eventually be converted to match the actual device registers
> states on-the-fly as needed.
> 
> The original design approach for the 104-quad-8 driver was neither
> efficient nor easy to troubleshoot, but it did allow us to focus on
> implementing and supporting necessary APIs for the nascent Counter
> subsystem. Now that development for the 104-quad-8 driver has shifted
> to maintenance, it is a good time to refactor and clean up the code to
> match closer to what is actually happening on the device. This patchset
> is an attempt to rectify the situation as such.
> 
> The primary change is a transition from maintaining individual
> configuration states independently, to storing buffers of the device
> register configurations. To that end, the bitfield API is leveraged to
> access and retrieve field states. Some helper functions are introduced
> as well to abstract the handling of the PR, FLAG, PSC, and control
> registers. A migration to the regmap API is added as a follow-up patch
> due to its dependence on this patch series.
> 
> [0] https://lore.kernel.org/r/b43e2942b763b87afc85bfa9fe36e5695cba4c44.1475079578.git.vilhelm.gray@xxxxxxxxx/
> [1] https://lore.kernel.org/r/cover.1554184734.git.vilhelm.gray@xxxxxxxxx/
> 
> William Breathitt Gray (4):
>   counter: 104-quad-8: Utilize bitfield access macros
>   counter: 104-quad-8: Refactor to buffer states for CMR, IOR, and IDR
>   counter: 104-quad-8: Utilize helper functions to handle PR, FLAG and
>     PSC
>   counter: 104-quad-8: Migrate to the regmap API
> 
>  drivers/counter/104-quad-8.c | 802 +++++++++++++++++++----------------
>  drivers/counter/Kconfig      |   1 +
>  2 files changed, 440 insertions(+), 363 deletions(-)
> 
> 
> base-commit: 09a9639e56c01c7a00d6c0ca63f4c7c41abe075d
> -- 
> 2.39.2
> 

Queued for counter-next.

William Breathitt Gray

Attachment: signature.asc
Description: PGP signature