[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250509112032.2980811-1-elder@riscstar.com>
Date: Fri, 9 May 2025 06:20:25 -0500
From: Alex Elder <elder@...cstar.com>
To: robh@...nel.org,
krzk+dt@...nel.org,
conor+dt@...nel.org,
mturquette@...libre.com,
sboyd@...nel.org,
p.zabel@...gutronix.de,
paul.walmsley@...ive.com,
palmer@...belt.com,
aou@...s.berkeley.edu,
alex@...ti.fr,
dlan@...too.org
Cc: heylenay@....org,
inochiama@...look.com,
guodong@...cstar.com,
devicetree@...r.kernel.org,
linux-clk@...r.kernel.org,
spacemit@...ts.linux.dev,
linux-riscv@...ts.infradead.org,
linux-kernel@...r.kernel.org
Subject: [PATCH v8 0/6] reset: spacemit: add K1 reset support
This series adds reset controller support for the SpacemiT K1 SoC.
A SpacemiT reset controller is implemented as an auxiliary device
associated with a clock controller (CCU). A new header file
holds definitions used by both the clock and reset drivers.
This code builds upon the clock controller driver from Haylen Chu.
The only difference from v7 is that the structure containing
the auxiliary device struct is now allocated using kzalloc().
Previously devm_kzalloc() was used, passing the parent device
pointer as its first argument, and as Haylen pointed out, it
was being (doubly) freed in the auxiliary device release
function. Everything else is essentially identical to v7.
This series is based on the "for-next" branch in the SpacemiT
repository:
https://github.com/spacemit-com/linux/tree/for-next
All of these patches are available here:
https://github.com/riscstar/linux/tree/outgoing/reset-v8
-Alex
Between version 7 and version 8:
- The structure containing the auxiliary device is now allocated
using kzalloc(). That means its lifetime is not tied to the
parent device, and auxiliary device's release function is
correct in freeing the structure.
Here is version 7 of this series.
https://lore.kernel.org/lkml/20250508195409.2962633-1-elder@riscstar.com/
Between version 6 and version 7:
- The new shared header file is now named "k1-syscon.h" (suggested
by Haylen Chu)
- The SPACEMIT_CCU_K1 config option has been removed (suggested
by Philipp Zabel)
- The SPACEMIT_CCU config option is now tristate, and selects
AUXILIARY_BUS (suggested by Haylen Chu)
- All code is concentrated into a single file "reset-spacemit.c"
rather than in a directory (suggested by Philipp Zabel)
- A bogus return value has been fixed, and a few irrelevant comments
have been removed (suggested by Philipp Zabel)
- MODULE_AUTHOR(), MODULE_DESCRIPTION(), and MODULE_LICENSE() are
now supplied (suggested by Haylen Chu)
Here is version 6 of this series.
https://lore.kernel.org/lkml/20250506210638.2800228-1-elder@riscstar.com/
Between version 5 and version 6:
- Reworked the code to use the auxiliary device framework.
- Moved the code supporting reset under drivers/reset/spacemit.
- Created a new header file shared by reset and clock.
- Separated generic from SoC-specific code in the reset driver.
- Dropped two Reviewed-by tags.
Here is version 5 of this series.
https://lore.kernel.org/lkml/20250418145401.2603648-1-elder@riscstar.com/
Between version 4 and version 5:
- Added Haylen's Reviewed-by on the second patch.
- Added Philipp's Reviewed-by on the third patch.
- In patch 4, added a const qualifier to some structures, and removed
parentheses surrounding integer constants, as suggested by Philipp
- Now based on the SpacemiT for-next branch
Here is version 4 of this series.
https://lore.kernel.org/lkml/20250414191715.2264758-1-elder@riscstar.com/
Between version 3 and version 4:
- Now based on Haylen Chu's v7 clock code, built on v6.15-rc2.
- Added Krzysztof's Reviewed-by on the first patch.
Here is version 3 of this series.
https://lore.kernel.org/lkml/20250409211741.1171584-1-elder@riscstar.com/
Between version 2 and version 3 there was no feedback, however:
- Haylen posted v6 of the clock series, and it included some changes
that affected the logic in this reset code.
- I was informed that defining CCU nodes without any clocks led to
warnings about "clocks" being a required property when running
"make dtbs_check". For that reason, I made clock properties
optional for reset-only CCU nodes.
- This code is now based on v6.15-rc1, which includes a few commits
that were listed as dependencies previously.
Here is version 2 of this series.
https://lore.kernel.org/lkml/20250328210233.1077035-1-elder@riscstar.com/
Between version 1 and version 2:
- Added Rob's Reviewed-by tag on the first patch
- Renamed the of_match_data data type (and one or two other symbols) to
use "spacemit" rather than "k1".
- Replaced the abbreviated "rst" or "RST" in names of newly-defined
sympols with "reset" or "RESET" respectively.
- Eliminated rcdev_to_controller(), which was only used once.
- Changed a function that unsafely did a read/modify/write of a register
to use regmap_update_bits() instead as suggested by Haylen.
- Eliminated a null check for a pointer known to be non-null.
- Reordered the assignment of reset controller device fields.
- Added a "sentinel" comment as requested by Yixun.
- Updated to be based on Linux v6.14 final.
Here is the first version of this series.
https://lore.kernel.org/lkml/20250321151831.623575-1-elder@riscstar.com/
Alex Elder (6):
dt-bindings: soc: spacemit: define spacemit,k1-ccu resets
soc: spacemit: create a header for clock/reset registers
clk: spacemit: set up reset auxiliary devices
reset: spacemit: add support for SpacemiT CCU resets
reset: spacemit: define three more CCUs
riscv: dts: spacemit: add reset support for the K1 SoC
.../soc/spacemit/spacemit,k1-syscon.yaml | 29 +-
arch/riscv/boot/dts/spacemit/k1.dtsi | 18 ++
drivers/clk/spacemit/Kconfig | 1 +
drivers/clk/spacemit/ccu-k1.c | 225 +++++++------
drivers/reset/Kconfig | 9 +
drivers/reset/Makefile | 1 +
drivers/reset/reset-spacemit.c | 297 ++++++++++++++++++
.../dt-bindings/clock/spacemit,k1-syscon.h | 128 ++++++++
include/soc/spacemit/k1-syscon.h | 160 ++++++++++
9 files changed, 741 insertions(+), 127 deletions(-)
create mode 100644 drivers/reset/reset-spacemit.c
create mode 100644 include/soc/spacemit/k1-syscon.h
base-commit: cb9c3aeae509b36afbdf46942a7a0a0dfc856ce7
--
2.45.2
Powered by blists - more mailing lists