Files
zsa_qmk_firmware/quantum/wear_leveling/wear_leveling.c
Drashna Jaelre 98990f50d7
Some checks failed
Build firmware / build-firmware (default) (push) Has been cancelled
Build firmware / build-firmware (oryx) (push) Has been cancelled
Unit Tests / test (push) Has been cancelled
Additional updates for firmware25 (#411)
* add 75_(ansi|iso) Community Layouts to mechlovin/olly/octagon (#22459)

* expand mechlovin/olly/octagon

* Update info.json

* Rename info.json to keyboard.json

* correct matrix position for key

* remove VIA

* [Core] get_keycode_string(): function to format keycodes as strings, for more readable debug logging. (#24787)

* keycode_string(): Format keycodes as strings.

This adds the `keycode_string()` function described in
https://getreuer.info/posts/keyboards/keycode-string/index.html
as a core feature.

* Fix formatting.

* keycode_string review revisions.

* Rename keycode_string() -> get_keycode_string() for consistency with
  existing string utils like get_u8_str().

* Revise custom keycode names with separate _user and _kb tables.

* Correct indent in builddefs/generic_features.mk.

Co-authored-by: Ryan <fauxpark@gmail.com>

* Add KC_NUHS, KC_NUBS, and KC_CAPS.

* Fix linking error with custom names.

* Attempt at simplifying interface.

* Formatting fix.

* Several fixes and revisions.

* Don't use PSTR in KEYCODE_STRING_NAME, since this fails to build on
  AVR. Store custom names in RAM.
* Revise the internal table of common keycode names to use its own
  storage representation, still in PROGMEM, and now more efficiently
  stored flat in 8 bytes per entry.
* Support Swap Hands keycodes and a few other keycodes.

* Revert "Formatting fix."

This reverts commit 2a2771068c7ee545ffac4103aa07e847a9ec3816.

* Revert "Attempt at simplifying interface."

This reverts commit 8eaf67de76e75bc92d106a8b0decc893fbc65fa5.

* Simplify custom names API by sigprof's suggestion.

* Support more keycodes.

* Add QK_LOCK keycode.
* Add Secure keycodes.
* Add Joystick keycodes.
* Add Programmable Button keycodes.
* Add macro MC_ keycodes.
* For remaining keys in known code ranges, stringify them as
  "QK_<feature>+<number>". For instance, "QK_MIDI+7".

* Bug fix and a few improvements.

* Fix missing right-hand bit when displaying 5-bit mods numerically.
* Support KC_HYPR, KC_MEH, HYPR_T(kc), MEH_T(kc).
* Exclude one-shot keycodes when NO_ACTION_ONESHOT is defined.

---------

Co-authored-by: Ryan <fauxpark@gmail.com>

* Align to latest CLI dependencies (#24553)

* Align to latest CLI dependencies

* Update docs

* Add Community Layout support to daskeyboard4 (#23884)

add ansi CL

* Add the plywrks ply8x hotswap variant. (#23558)

* Add hotswap variant

* Update RGB matrix

* Move files around to target develop

* Revert rules.mk for keyboards/jaykeeb/joker/rules.mk

* Update keyboards/plywrks/ply8x/hotswap/keyboard.json

Co-authored-by: Drashna Jaelre <drashna@live.com>

* Apply suggestions from code review

Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>

* Add missing community layouts

* Delete keyboards/plywrks/ply8x/rules.mk

* Update missing keys in RGB matrix

* Add missing key in RGB matrix for hotswap ver

* Remove via keymaps

* Add keyboard alias for plywrks/ply8x to plywrks/ply8x/solder

* Fix typo

* Fix another typo

---------

Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>

* [Core] use `keycode_string` in unit tests (#25042)

* tests: use keycode_string feature

With a proper keycode to string implementation in qmk there is no need
to use the unit tests only implementation anymore.

Signed-off-by: Stefan Kerkmann <karlk90@pm.me>

* tests: remove keycode_util feature

This feature is no longer used as we switched the tests to the
keycode string implementation.

Signed-off-by: Stefan Kerkmann <karlk90@pm.me>

* Non-volatile memory data repository pattern (#24356)

* First batch of eeconfig conversions.

* Offset and length for datablocks.

* `via`, `dynamic_keymap`.

* Fix filename.

* Commentary.

* wilba leds

* satisfaction75

* satisfaction75

* more keyboard whack-a-mole

* satisfaction75

* omnikeyish

* more whack-a-mole

* `generic_features.mk` to automatically pick up nvm repositories

* thievery

* deferred variable resolve

* whitespace

* convert api to structs/unions

* convert api to structs/unions

* convert api to structs/unions

* fixups

* code-side docs

* code size fix

* rollback

* nvm_xxxxx_erase

* Updated location of eeconfig magic numbers so non-EEPROM nvm drivers can use them too.

* Fixup build.

* Fixup compilation error with encoders.

* Build fixes.

* Add `via_ci` keymap to onekey to exercise VIA bindings (and thus dynamic keymap et.al.), fixup compilation errors based on preprocessor+sizeof.

* Build failure rectification.

* Migrate remaining `split.soft_serial_pin` to `split.serial.pin` (#25046)

* Migrate keyboards/bastardkb

* Migrate keyboards/handwired

* Migrate keyboards/helix

* Fix duplicate serial key

* Fix outdated GPIO control function usage (#25060)

* [Modules] Provide access to current path in `rules.mk`. (#25061)

* Update keymap for keycult 1800 (#25070)

Update keymap

Co-authored-by: yiancar <yiancar@gmail.com>

* Add handwired/erikpeyronson/erkbd (#25030)


Co-authored-by: Erik Peyronson <erik.peyronson@gmail.com>
Co-authored-by: jack <jack@pngu.org>
Co-authored-by: Drashna Jaelre <drashna@live.com>

* Add support for Starry FRL (#24626)


Co-authored-by: jack <jack@pngu.org>
Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: Ryan <fauxpark@gmail.com>

* Add "Large Lad" keyboard (#24727)

Co-authored-by: jack <jack@pngu.org>

* Bump vite from 5.4.12 to 5.4.15 in /builddefs/docsgen (#25065)

Bumps [vite](https://github.com/vitejs/vite/tree/HEAD/packages/vite) from 5.4.12 to 5.4.15.
- [Release notes](https://github.com/vitejs/vite/releases)
- [Changelog](https://github.com/vitejs/vite/blob/v5.4.15/packages/vite/CHANGELOG.md)
- [Commits](https://github.com/vitejs/vite/commits/v5.4.15/packages/vite)

---
updated-dependencies:
- dependency-name: vite
  dependency-type: direct:development
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Allow AnnePro2 to reboot (#24886)

Without this, the QK_REBOOT key did nothing.

* Fix path typo related RP2040 (#25069)

Fix path typo

* Update onekey example for nucleo f446re (#25067)

* use accessible pins for nucleo f446re onekey example

* remove pin collision with matrix in keyboard.json

* use accessible pins for LED

* remove pin collision with matrix

* Update readme.md to reflect pin changes

* Module documentation typo correction (#25073)

* Fix lockups on AVR with `qmk/hello_world` module (#25074)

Fix lockups on AVR.

* At101ish (#25072)

* Dell AT101 replacement pcb support

* Update keyboards/at101ish/readme.md

Co-Authored-By: fauxpark <fauxpark@gmail.com>

* remove empty src clause in makefile

* feature: Update at101ish to qmk v0.28

* feature: Add osdetecting keymap variant.

* refactor: Move at101ish keyboard to handwired folder.

* fix: Adjust at101ish readme-

* fix: review changes.

* chore: Remove unneeded feature.

---------

Co-authored-by: fauxpark <fauxpark@gmail.com>

* Add "license" field to Community Module JSON schema. (#25085)

Add "license" field to community module schema.

* Add kt60HS-T v2 PCB (#25080)

* Add kt60HS-Tv2

* Update keyboards/keyten/kt60hs_t/readme.md

Co-authored-by: Sergey Vlasov <sigprof@gmail.com>

* Update keyboards/keyten/kt60hs_t/v1/readme.md

Co-authored-by: Sergey Vlasov <sigprof@gmail.com>

* Update keyboards/keyten/kt60hs_t/v2/keyboard.json

Co-authored-by: Sergey Vlasov <sigprof@gmail.com>

* Update keyboards/keyten/kt60hs_t/v2/readme.md

Co-authored-by: Sergey Vlasov <sigprof@gmail.com>

* Update keyboards/keyten/kt60hs_t/info.json

Co-authored-by: Sergey Vlasov <sigprof@gmail.com>

* Change of structure

* Moving the keyboard

* Update data/mappings/keyboard_aliases.hjson

Co-authored-by: Sergey Vlasov <sigprof@gmail.com>

* Update keyboards/keyten/kt60hs_t/v1/keyboard.json

Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>

* Update keyboards/keyten/kt60hs_t/v2/keyboard.json

Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>

---------

Co-authored-by: Sergey Vlasov <sigprof@gmail.com>
Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>

* Make sure that unit tests run on all release versions

* [ErgoDox EZ] Fix complication issues due to updates

* [ErgoDox EZ] Fix compilication errors and warnings

We want all green!

* Fix 'qmk lint -kb' argument handling (#25093)

* Refactor Deemen17 Works DE60 (#25088)

* Add Coban Pad 12A (#25039)



Co-authored-by: jack <jack@pngu.org>

* [Keyboard] Add PHDesign PH60/Multi Keyboard PCB (#25086)

* Add PH60/Multi Support

* Add PCB PIcture for README

* Remove MO(_FN2)

* README Typo Fix

* Layout and README Adjustment

* Add README for PHDesign Main Folder

* Keymap Improvement

* Update README.md

* [Keyboard] Add Ortho Slayer (#25099)

* Add Ortho Slayer

* Update keyboards/keyten/ortho_slayer/keymaps/default/keymap.c

Co-authored-by: jack <jack@pngu.org>

* Update keyboards/keyten/ortho_slayer/readme.md

Co-authored-by: jack <jack@pngu.org>

---------

Co-authored-by: jack <jack@pngu.org>

* Fix coban pad9a wrong layout in keyboard.json (#25100)

* Remove `CTPC`/`CONVERT_TO_PROTON_C` options (#25111)

* Remove direct docs.qmk.fm links from docs (#25113)

* Add warning when deprecated 'promicro_rp2040' is used (#25112)

* Add Vida to QMK (#24225)

Co-authored-by: jack <jack@pngu.org>
Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>

* More Windows->Unix style path fixes. (#25119)

* Include `math.h` where necessary. (#25122)

* Cater for use of `__errno_r()` in ChibiOS syscalls.c with newer picolibc revisions (#25121)

* chore: Allow disabling underglow on Work Louder devices (#25123) (#25120)

* Allow disabling Underglow on Work Louder devices

Allows disabling Underglow on Work Louder devices by using `RGBLIGHT_ENABLE = no` on rules.mk

* Update keyboards/work_louder/rgb_functions.c

Suggested by @zvecr on review.

Co-authored-by: Joel Challis <git@zvecr.com>

---------

Co-authored-by: Joel Challis <git@zvecr.com>

* Exclude external userspace from lint checking (#24680)

* fix: Fix startup sound for Preonic (#25132) (#25133)

Add `AUDIO_INIT_DELAY ` to config.h to resolve

* New standard layout for Savage65 (65_ansi_blocker_tsangan_split_bs) (#24690)

* Added a default firmware and layout for the WindStudio Wind X R1
keyboard.

* Wind X R1: cleaned-up the folders to make clear that this firmware is
for the release 1 of this keyboard.

* Delete keyboards/windstudio/wind_x/R1 directory

Removing the uppercase R1 folder

* feat(cannonkeys/savage65): Added layout to keyboard.json
- Added the layout LAYOUT_65_ansi_blocker_tsangan_split_bs to the
  community layouts.

* kradoindustries_promenade: add LAYOUT_1x2u (#25090)

* Update shuguet/shu89 (#24780)

* Update keyboard.json

Update mod keys location in RGB layout.

* Update keyboard.json

* Update keyboards/shuguet/shu89/keyboard.json

Co-authored-by: Ryan <fauxpark@gmail.com>

---------

Co-authored-by: Ryan <fauxpark@gmail.com>

* [Keyboard] Add suika83opti (#24991)

* [chore]: move and rename mouse/scroll min/max defines (#25141)

* protocol: move {XY/HV}_REPORT_{MIN,MAX} into report.h

..to allow easier re-use in other code implementations.

* protocol: rename {XY/HV}_REPORT_{MIN/MAX} to MOUSE_REPORT_{XY/HV}_{MIN/MAX}

..to avoid naming collisions.

* [Core] Flow Tap tap-hold option to disable HRMs during fast typing (#25125)

aka Global Quick Tap, Require Prior Idle

* Add Link keyboard (#25058)

Co-authored-by: jack <jack@pngu.org>
Co-authored-by: Drashna Jaelre <drashna@live.com>

* Remove Sofle `rgb_default` keymap & tidy readme's (#25010)

* Added Keyboard LumPy27 (#24967)

Co-authored-by: jack <jack@pngu.org>
Co-authored-by: Drashna Jaelre <drashna@live.com>

* [Keyboard] Kobold r1 (#25161)

* Kobold r1

* Apply suggestions from code review

Co-authored-by: jack <jack@pngu.org>

* `board_init` => `early_hardware_init_post`.

---------

Co-authored-by: jack <jack@pngu.org>

* [Docs] Unify lighting step descriptions (#25167)

unify lighting step descriptions and defaults across docs

* [Keyboard] Add voidhhkb-hotswap (#25007)

* Added files for voidhhkb-hotswap

* Updated keyboard name to resolve build errors

* Implement suggestions from PR. Use 60_hhkb community layout.

* Update keyboards/void/voidhhkb_hotswap/readme.md

Co-authored-by: Ryan <fauxpark@gmail.com>

* Apply suggestions from code review

Co-authored-by: jack <jack@pngu.org>

---------

Co-authored-by: Ryan <fauxpark@gmail.com>
Co-authored-by: jack <jack@pngu.org>

* Remove duplication of RGB Matrix defaults (#25146)

* Remove duplication of RGB Matrix defaults

* Remove more duplication of defaults

* fix

* Fix missing and extra commas in JSON schema (#25057)

* Remove duplication of RGBLight defaults (#25169)

* Ignore the Layer Lock key in Repeat Key and Caps Word. (#25171)

* Allow for disabling EEPROM subsystem entirely. (#25173)

* [keyboard] ymdk/id75/rp2040 (#25157)

Co-authored-by: tao heihei <>

* Remove `bluefruit_le_read_battery_voltage` function (#25129)

* Fix 'Would you like to clone the submodules?' prompt under msys (#24958)

* Fixup eeconfig lighting reset. (#25166)

* DOCS: `qmk-hid` missing in bootloaders list? (#25177)

* Fix for `.clangd`. (#25180)

* Update develop branch to Pico SDK 1.5.1 (#25178)

* Add lint warning for empty url (#25182)

* Implement connection keycode logic (#25176)

* Add handwired/footy (#25151)

Co-authored-by: jack <jack@pngu.org>

* Decrease firmware size for `anavi/macropad8`. (#25185)

Preparation for bootstrapper.

* Align ChibiOS `USB_WAIT_FOR_ENUMERATION` implementation (#25184)

* [Bug][Core] Fix for Flow Tap: fix handling of distinct taps and timer updates. (#25175)

* Flow Tap bug fix.

As reported by @amarz45 and @mwpardue, there is a bug where if two
tap-hold keys are pressed in distinct taps back to back, then Flow Tap
is not applied on the second tap-hold key, but it should be.

In a related bug reported by @NikGovorov, if a tap-hold key is held
followed by a tap of a tap-hold key, then Flow Tap updates its timer on
the release of the held tap-hold key, but it should be ignored.

The problem common to both these bugs is that I incorrectly assumed
`tapping_key` is cleared to noevent once it is released, when actually
`tapping_key` is still maintained for `TAPPING_TERM` ms after release
(for Quick Tap). This commit fixes that. Thanks to @amarz45, @mwpardue,
and @NikGovorov for reporting!

Details:

* Logic for converting the current tap-hold event to a tap is extracted
  to `flow_tap_key_if_within_term()`, which is now invoked also in the
  post-release "interfered with other tap key" case. This fixes the
  distinct taps bug.

* The Flow Tap timer is now updated at the beginning of each call to
  `process_record()`, provided that there is no unsettled tap-hold key
  at that time and that the record is not for a mod or layer switch key.
  By moving this update logic to `process_record()`, it is conceptually
  simpler and more robust.

* Unit tests extended to cover the reported scenarios.

* Fix formatting.

* Revision to fix @NikGovorov's scenario.

The issue is that when another key is pressed while a layer-tap hasn't
been settled yet, the `prev_keycode` remembers the keycode from before
the layer switched. This can then enable Flow Tap for the following key
when it shouldn't, or vice versa.

Thanks to @NikGovorov for reporting!

This commit revises Flow Tap in the following ways:

* The previous key and timer are both updated from `process_record()`.
  This is slightly later in the sequence of processing than before, and
  by this point, a just-settled layer-tap should have taken effect so
  that the keycode from the correct layer is remembered.

* The Flow Tap previous key and timer are updated now also on key
  release events, except for releases of modifiers and held layer
  switches.

* The Flow Tap previous key and timer are now updated together, for
  simplicity. This makes the logic easier to think about.

* A few additional unit tests, including @NikGovorov's scenario as
  "layer_tap_ignored_with_disabled_key_complex."

* Remove empty `url` fields (#25181)

* Prompt for converter when creating new keymap (#25116)

* High resolution scrolling (without feature report parsing) (#24423)

* hires scrolling without feature report parsing

* fix valid range for exponent

* fix incorrect minimum exponent value documentation

* Avoid duplication in generated community modules `rules.mk` (#25135)

* Bump rlespinasse/github-slug-action from 3 to 5 (#25021)

* Remove `"console":false` from keyboards (#25190)

* Update 'qmk generate-api' to only publish pure DD keymaps (#24782)

* Remove more duplication of defaults (#25189)

* Align `new-keyboard` template to current standards (#25191)

* Bump vite from 5.4.15 to 5.4.18 in /builddefs/docsgen (#25192)

Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Fix `boardsource/beiwagon` RGB Matrix coordinates (#25018)

* Remove `"command":false` from keyboards (#25193)

* Extend lint checks to reject duplication of defaults (#25149)

* modelh: add prerequisites for via support (#24932)

* First TypeK support (#22876)

* Add Lemokey X0 keyboard (#24994)

* keyboards/annepro2/ld: Add per-variant linker scripts (#24999)

C18 has an MCU with 16K SRAM, up from C15's 8K.

Split the linker script into C15 and C18 variants to make use of the
larger RAM capacity of C18.

* Add new keyboard MirageiX (#25054)

Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: jack <0x6a73@protonmail.com>

* [Keymap] Sofle RGB - fixed stuck on numpad layer and layout comments (#24942)

* Add handwired 4x14 LuMaWing keyboard (#24885)

* Add jcpm2 (JC Pro Macro 2) (#24816)

Co-authored-by: jack <jack@pngu.org>
Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: Joel Challis <git@zvecr.com>

* [Keyboard] Add splitkb.com's Halcyon Elora rev2 (#24790)

Co-authored-by: Drashna Jaelre <drashna@live.com>

* [Keyboard] mzmkb/slimdash/rev1 (#24804)

Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: jack <jack@pngu.org>
Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>
Co-authored-by: Ryan <fauxpark@gmail.com>

* Added new keyboard epssp75 (#24756)

Co-authored-by: Drashna Jaelre <drashna@live.com>

* Addition of OK-1 (#24646)

Co-authored-by: Joel Challis <git@zvecr.com>
Co-authored-by: Drashna Jaelre <drashna@live.com>

* Add Umbra keyboard (#24569)

Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>
Co-authored-by: Ryan <fauxpark@gmail.com>
Co-authored-by: Drashna Jaelre <drashna@live.com>

* Amptrics 0420 keyboard addition (#24744)

Co-authored-by: jack <jack@pngu.org>
Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: Ryan <fauxpark@gmail.com>

* [Core] Enhance Flow Tap to work better for rolls over multiple tap-hold keys. (#25200)

* Flow Tap revision for rolling press.

* Remove debugging cruft.

* Formatting fix.

* amptrics/0422 - Prevent OOB in `update_leds_for_layer` (#25209)

* [Keyboard] Add Gravity-45(#25206)

* add gravity-45

* readme.md

* fix readme

* Update keyboards/green_keys/gravity_45/keyboard.json

Co-authored-by: jack <0x6a73@protonmail.com>

* run qmk format-json -i keyboards/green_keys/gravity_45/keyboard.json

* add url

* Update keyboards/green_keys/gravity_45/keyboard.json

Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>

* Update keyboard.json

* Update keyboard.json

* Update keyboards/green_keys/gravity_45/keyboard.json

Co-authored-by: Drashna Jaelre <drashna@live.com>

---------

Co-authored-by: jack <0x6a73@protonmail.com>
Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>
Co-authored-by: Drashna Jaelre <drashna@live.com>

* Remove redundant keyboard headers (#25208)

* Fix Spleeb compile when pointing device is enabled (#25016)

* [Bug] Minimise force-included files (#25194)

* Add additional hooks for Community modules (#25050)

* Workaround for resolving keyboard alias for config file values (#25228)

* Generate versions to keycode headers (#25219)

* Resolve alias for `qmk new-keymap` keyboard prompts (#25210)

* [Keyboard] Add Binepad KN01 (#25224)

* Add Binepad NeoKnob KN01

* post @waffle87 recommendations

* Add battery changed callbacks (#25207)

* Ensure `qmk_userspace_paths` maintains detected order (#25204)

* Bind Bluetooth driver to `host_driver_t` (#25199)

* Deprecate `qmk generate-compilation-database`. (#25237)

* Remove force disable of NKRO when Bluetooth enabled (#25201)

* Keycult 60 (#25213)

Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>
Co-authored-by: jack <jack@pngu.org>

* [Keyboard] Add Binepad KnobX1 (#25222)

Co-authored-by: Drashna Jaelre <drashna@live.com>

* [Keyboard] Update Tractyl Manuform and add F405 (weact) variant (#24764)

* Layout corrections: Zed60 (#25003)

* Fixed print statement after enabling 32-bit layers (#25027)

* Fix Aurora sweep default keymap configuration (#25148)

* Docs update for installing qmk with uv (#24995)

* CXT Studio 12E3: Fix encoder resolutions not applying (#25242)

* add resolution to encoders so they apply

* Tweak default keymap

* replace KC_UNDO with C(KC_Z) as well

* Add debounce to duplicated defaults check (#25246)

* [Docs] Fix typos introduced by PR #25050 (#25250)

It isn't a drashna PR if there aren't some typos in it somewhere.

* Allow LVGL onekey keymap to be able compile for other board (#25005)

* [Keyboard] Add Jason Hazel’s Bad Wings v2 (#25252)

Co-authored-by: Florent Allard <florent.allard@savoirfairelinux.com>

* Add raw_hid support to host driver (#25255)

* Fix Wear Leveling compilation (#25254)

* [New Feature/Core] New RGB Matrix Animation "Starlight Smooth" (#25203)

* Enable community modules to define LED matrix and RGB matrix effects. (#25187)

Co-authored-by: Joel Challis <git@zvecr.com>

* Fix OS_DETECTION_KEYBOARD_RESET (#25015)

Co-authored-by: Nick Brassel <nick@tzarc.org>

* Fixes the numlock indicator for Magic Force MF17 numpad (#25260)

* Add Harite v2 keyboard (#24975)

* dlip/haritev2 - Post merge fixes (#25264)

* Remove more USB only branches from NKRO handling (#25263)

* Deprecate `usb.force_nkro`/`FORCE_NKRO` (#25262)

* Add BDN9 Rev. 3 (#25261)

* Remove duplicate of SPI default config from keyboards (#25266)

* Resolve miscellaneous keyboard lint warnings (#25268)

* Add Zeropad (#24737)

Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: Joel Challis <git@zvecr.com>

* Add Waveshare RP2040-Keyboard-3 support (#25269)

* Update PR checklist notes on custom matrix (#25240)

* Chew folders (#24785)

* gcc15 AVR compilation fixes (#25238)

* [Core] STM32G0x1 support (#24301)

* Use relative paths for schemas, instead of $id. Enables VScode validation. (#25251)

* add doio/kb03 (#24815)

Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: Joel Challis <git@zvecr.com>

* Move rookiebwoy to ivndbt (#25142)

* [Chore] use {rgblight,rgb_matrix}_hsv_to_rgb overrides (#25271)

* Remove outdated `nix` support due to bit-rot. (#25280)

* Add `compiler_support.h` (#25274)

* Configure boards to use development_board - 0-9 (#25287)

* [Fix] lib8tion: enable fixed scale8 and blend functions (#25272)

lib8tion: enable fixed scale8 and blend functions

These FastLED derived lib8tion functions have been fixed and enabled by
default in FastLED. QMK just never set these defines, there is no reason
to keep the buggy implementation. It is assumed that nobody relied on
the buggy behavior.

* Configure boards to use development_board - UVWXYZ (#25288)

* [Docs] Fix tap_hold code blocks (#25298)

* salicylic_acid3/getta25 - Fix oled keymap (#25295)

* Configure boards to use development_board - S (#25293)

* Configure boards to use development_board - T (#25294)

* 2025 Q2 changelog (#25297)

Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: Nick Brassel <nick@tzarc.org>

* Update Oryx module for newer code

* Soft reset matrix

* Use improved i2c reset for voyager and moonlander matrix

* Fix formatting

* Remove labeler action (unneeded)

* Fix module API version for Oryx module

* Use i2cStop instead of trying to work around

---------

Signed-off-by: Stefan Kerkmann <karlk90@pm.me>
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: Duncan Sutherland <dunk2k_2000@hotmail.com>
Co-authored-by: QMK Bot <hello@qmk.fm>
Co-authored-by: Pascal Getreuer <50221757+getreuer@users.noreply.github.com>
Co-authored-by: Ryan <fauxpark@gmail.com>
Co-authored-by: Joel Challis <git@zvecr.com>
Co-authored-by: Ramon Imbao <ramonimbao@gmail.com>
Co-authored-by: Stefan Kerkmann <karlk90@pm.me>
Co-authored-by: Nick Brassel <nick@tzarc.org>
Co-authored-by: jack <jack@pngu.org>
Co-authored-by: yiancar <yiangosyiangou@cytanet.com.cy>
Co-authored-by: yiancar <yiancar@gmail.com>
Co-authored-by: Erik Peyronson <erikpeyronson@gmail.com>
Co-authored-by: Erik Peyronson <erik.peyronson@gmail.com>
Co-authored-by: Sắn <59417802+MaiTheSan@users.noreply.github.com>
Co-authored-by: Hyphen-ated <Hyphen-ated@users.noreply.github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Geoffrey Frogeye <s+github@frogeye.fr>
Co-authored-by: lsh4711 <120231876+lsh4711@users.noreply.github.com>
Co-authored-by: Ben Green <bengreen.uk@gmail.com>
Co-authored-by: フィルターペーパー <76888457+filterpaper@users.noreply.github.com>
Co-authored-by: henrikosorensen <henrik.sorensen@gmail.com>
Co-authored-by: Ivan Gromov <38141348+key10iq@users.noreply.github.com>
Co-authored-by: Sergey Vlasov <sigprof@gmail.com>
Co-authored-by: Pham Duc Minh <95753855+Deemen17@users.noreply.github.com>
Co-authored-by: Dam Vu Duy <RyanDam@users.noreply.github.com>
Co-authored-by: nonameCCC <79012391+nonameCCC@users.noreply.github.com>
Co-authored-by: sudo pacman -Syu <hauvipapro@gmail.com>
Co-authored-by: Andrew Kannan <andrew.kannan@gmail.com>
Co-authored-by: Luis Garcia <luis@bitjester.com>
Co-authored-by: Christian C. Berclaz <christian.berclaz@mac.com>
Co-authored-by: Olivier Mehani <shtrom-github@ssji.net>
Co-authored-by: Sylvain Huguet <sylvain@huguet.me>
Co-authored-by: suikagiken <115451678+suikagiken@users.noreply.github.com>
Co-authored-by: Daniel Reisch <danieljreisch@gmail.com>
Co-authored-by: ClownFish <177758267+clownfish-og@users.noreply.github.com>
Co-authored-by: JamesWilson1996 <47866504+JamesWilson1996@users.noreply.github.com>
Co-authored-by: Less/Rikki <86894501+lesshonor@users.noreply.github.com>
Co-authored-by: Jan Bláha <blaha.j502@gmail.com>
Co-authored-by: Eric Molitor <534583+emolitor@users.noreply.github.com>
Co-authored-by: CJ Pais <cj@cjpais.com>
Co-authored-by: eynsai <47629346+eynsai@users.noreply.github.com>
Co-authored-by: Joel Beckmeyer <joel@beckmeyer.us>
Co-authored-by: Álvaro A. Volpato <alvaro.augusto.volpato@gmail.com>
Co-authored-by: Aidan Gauland <aidalgol@users.noreply.github.com>
Co-authored-by: Michał Kopeć <michal@nozomi.space>
Co-authored-by: takashicompany <t@kashi.company>
Co-authored-by: jack <0x6a73@protonmail.com>
Co-authored-by: Matheus Marques <matheusmbar@gmail.com>
Co-authored-by: LucasMateijsen <l.mateijsen@outlook.com>
Co-authored-by: Jeremy Cook <jscook55@gmail.com>
Co-authored-by: VeyPatch <126267034+VeyPatch@users.noreply.github.com>
Co-authored-by: mizma <omoikane@path-works.net>
Co-authored-by: hen-des <141591967+hen-des@users.noreply.github.com>
Co-authored-by: Cipulot <40441626+Cipulot@users.noreply.github.com>
Co-authored-by: josephawilliamsiv <31166673+josephawilliamsiv@users.noreply.github.com>
Co-authored-by: vchowl <vchowl@outlook.com>
Co-authored-by: Christopher Hoage <iam@chrishoage.com>
Co-authored-by: Silvino R. <366673+silvinor@users.noreply.github.com>
Co-authored-by: dabstractor <dustindschultz@gmail.com>
Co-authored-by: Nathan Cain <13713501+nathanscain@users.noreply.github.com>
Co-authored-by: muge <221161+muge@users.noreply.github.com>
Co-authored-by: HorrorTroll <sonicvipduc@gmail.com>
Co-authored-by: cyxae <cyxae@amphitryon.nrst.fr>
Co-authored-by: Florent Allard <florent.allard@savoirfairelinux.com>
Co-authored-by: art-was-here <mail@buckles.email>
Co-authored-by: Matti Hiljanen <170205+qvr@users.noreply.github.com>
Co-authored-by: Wasteland Fluttershy <ingvardm@gmail.com>
Co-authored-by: Dane Lipscombe <danelipscombe@gmail.com>
Co-authored-by: Danny <nooges@users.noreply.github.com>
Co-authored-by: Infos <136488157+diffrentGuesser@users.noreply.github.com>
Co-authored-by: Florent Linguenheld <f@linguenheld.fr>
Co-authored-by: ivan <81021475+ivndbt@users.noreply.github.com>
Co-authored-by: Pablo Martínez <58857054+elpekenin@users.noreply.github.com>
2025-06-04 15:26:21 +07:00

770 lines
30 KiB
C

// Copyright 2022 Nick Brassel (@tzarc)
// SPDX-License-Identifier: GPL-2.0-or-later
#include <stdbool.h>
#include "fnv.h"
#include "wear_leveling.h"
#include "wear_leveling_drivers.h"
#include "wear_leveling_internal.h"
/*
This wear leveling algorithm is adapted from algorithms from previous
implementations in QMK, namely:
- Artur F. (http://engsta.com/stm32-flash-memory-eeprom-emulator/)
- Yiancar -- QMK's base implementation for STM32F303
- Ilya Zhuravlev -- initial wear leveling algorithm
- Don Kjer -- increased flash density algorithm
- Nick Brassel (@tzarc) -- decoupled for use on other peripherals
At this layer, it is assumed that any reads/writes from the backing store
have a "reset state" after erasure of zero.
It is up to the backing store to perform translation of values, such as
taking the complement in order to deal with flash memory's reset value.
Terminology:
- Backing store: this is the storage area used by the wear leveling
algorithm.
- Backing size: this is the amount of storage provided by the backing
store for use by the wear leveling algorithm.
- Backing write size: this is the minimum number of bytes the backing
store can write in a single operation.
- Logical data: this is the externally-visible "emulated EEPROM" that
external subsystems "see" when performing reads/writes.
- Logical size: this is the amount of storage available for use
externally. Effectively, the "size of the EEPROM".
- Write log: this is a section of the backing store used to keep track
of modifications without overwriting existing data. This log is
"played back" on startup such that any subsequent reads are capable
of returning the latest data.
- Consolidated data: this is a section of the backing store reserved for
use for the latest copy of logical data. This is only ever written
when the write log is full -- the latest values for the logical data
are written here and the write log is cleared.
Configurables:
- BACKING_STORE_WRITE_SIZE: The number of bytes requires for a write
operation. This is defined by the capabilities of the backing store.
- WEAR_LEVELING_BACKING_SIZE: The number of bytes provided by the
backing store for use by the wear leveling algorithm. This is
defined by the capabilities of the backing store. This value must
also be at least twice the size of the logical size, as well as a
multiple of the logical size.
- WEAR_LEVELING_LOGICAL_SIZE: The number of bytes externally visible
to other subsystems performing reads/writes. This must be a multiple
of the write size.
General algorithm:
During initialization:
* The contents of the consolidated data section are read into cache.
* The contents of the write log are "played back" and update the
cache accordingly.
During reads:
* Logical data is served from the cache.
During writes:
* The cache is updated with the new data.
* A new write log entry is appended to the log.
* If the log's full, data is consolidated and the write log cleared.
Write log structure:
The first 8 bytes of the write log are a FNV1a_64 hash of the contents
of the consolidated data area, in an attempt to detect and guard against
any data corruption.
The write log follows the hash:
Given that the algorithm needs to cater for 2-, 4-, and 8-byte writes,
a variable-length write log entry is used such that the minimal amount
of storage is used based off the backing store write size.
Firstly, an empty log entry is expected to be all zeros. If the backing
store uses 0xFF for cleared bytes, it should return the complement, such
that this wear-leveling algorithm "receives" zeros.
For multi-byte writes, up to 8 bytes will be used for each log entry,
depending on the size of backing store writes:
╔ Multi-byte Log Entry (2, 4-byte) ═╗
║00XXXYYY║YYYYYYYY║YYYYYYYY║AAAAAAAA║
║ └┬┘└┬┘║└──┬───┘║└──┬───┘║└──┬───┘║
║ LenAdd║ Address║ Address║Value[0]║
╚════════╩════════╩════════╩════════╝
╔ Multi-byte Log Entry (2-byte) ══════════════════════╗
║00XXXYYY║YYYYYYYY║YYYYYYYY║AAAAAAAA║BBBBBBBB║CCCCCCCC║
║ └┬┘└┬┘║└──┬───┘║└──┬───┘║└──┬───┘║└──┬───┘║└──┬───┘║
║ LenAdd║ Address║ Address║Value[0]║Value[1]║Value[2]║
╚════════╩════════╩════════╩════════╩════════╩════════╝
╔ Multi-byte Log Entry (2, 4, 8-byte) ══════════════════════════════════╗
║00XXXYYY║YYYYYYYY║YYYYYYYY║AAAAAAAA║BBBBBBBB║CCCCCCCC║DDDDDDDD║EEEEEEEE║
║ └┬┘└┬┘║└──┬───┘║└──┬───┘║└──┬───┘║└──┬───┘║└──┬───┘║└──┬───┘║└──┬───┘║
║ LenAdd║ Address║ Address║Value[0]║Value[1]║Value[2]║Value[3]║Value[4]║
╚════════╩════════╩════════╩════════╩════════╩════════╩════════╩════════╝
19 bits are used for the address, which allows for a max logical size of
512kB. Up to 5 bytes can be included in a single log entry.
For 2-byte backing store writes, the last two bytes are optional
depending on the length of data to be written. Accordingly, either 3
or 4 backing store write operations will occur.
For 4-byte backing store writes, either one or two write operations
occur, depending on the length.
For 8-byte backing store writes, one write operation occur.
2-byte backing store optimizations:
For single byte writes, addresses between 0...63 are encoded in a single
backing store write operation. 4- and 8-byte backing stores do not have
this optimization as it does not minimize the number of bytes written.
╔ Byte-Entry ════╗
║01XXXXXXYYYYYYYY║
║ └─┬──┘└──┬───┘║
║ Address Value ║
╚════════════════╝
0 <= Address < 0x40 (64)
A second optimization takes into account uint16_t writes of 0 or 1,
specifically catering for KC_NO and KC_TRANSPARENT in the dynamic keymap
subsystem. This is valid only for the first 16kB of logical data --
addresses outside this range will use the multi-byte encoding above.
╔ U16-Encoded 0 ═╗
║100XXXXXXXXXXXXX║
║ │└─────┬─────┘║
║ │Address >> 1 ║
║ └── Value: 0 ║
╚════════════════╝
0 <= Address <= 0x3FFE (16382)
╔ U16-Encoded 1 ═╗
║101XXXXXXXXXXXXX║
║ │└─────┬─────┘║
║ │Address >> 1 ║
║ └── Value: 1 ║
╚════════════════╝
0 <= Address <= 0x3FFE (16382) */
/**
* Storage area for the wear-leveling cache.
*/
static struct __attribute__((__aligned__(BACKING_STORE_WRITE_SIZE))) {
__attribute__((__aligned__(BACKING_STORE_WRITE_SIZE))) uint8_t cache[(WEAR_LEVELING_LOGICAL_SIZE)];
uint32_t write_address;
bool unlocked;
} wear_leveling;
/**
* Locking helper: status
*/
typedef enum backing_store_lock_status_t { STATUS_FAILURE = 0, STATUS_SUCCESS, STATUS_UNCHANGED } backing_store_lock_status_t;
/**
* Locking helper: unlock
*/
static inline backing_store_lock_status_t wear_leveling_unlock(void) {
if (wear_leveling.unlocked) {
return STATUS_UNCHANGED;
}
if (!backing_store_unlock()) {
return STATUS_FAILURE;
}
wear_leveling.unlocked = true;
return STATUS_SUCCESS;
}
/**
* Locking helper: lock
*/
static inline backing_store_lock_status_t wear_leveling_lock(void) {
if (!wear_leveling.unlocked) {
return STATUS_UNCHANGED;
}
if (!backing_store_lock()) {
return STATUS_FAILURE;
}
wear_leveling.unlocked = false;
return STATUS_SUCCESS;
}
/**
* Resets the cache, ensuring the write address is correctly initialised.
*/
static void wear_leveling_clear_cache(void) {
memset(wear_leveling.cache, 0, (WEAR_LEVELING_LOGICAL_SIZE));
wear_leveling.write_address = (WEAR_LEVELING_LOGICAL_SIZE) + 8; // +8 is due to the FNV1a_64 of the consolidated buffer
}
/**
* Reads the consolidated data from the backing store into the cache.
* Does not consider the write log.
*/
static wear_leveling_status_t wear_leveling_read_consolidated(void) {
wl_dprintf("Reading consolidated data\n");
wear_leveling_status_t status = WEAR_LEVELING_SUCCESS;
if (!backing_store_read_bulk(0, (backing_store_int_t *)wear_leveling.cache, sizeof(wear_leveling.cache) / sizeof(backing_store_int_t))) {
wl_dprintf("Failed to read from backing store\n");
status = WEAR_LEVELING_FAILED;
}
// Verify the FNV1a_64 result
if (status != WEAR_LEVELING_FAILED) {
uint64_t expected = fnv_64a_buf(wear_leveling.cache, (WEAR_LEVELING_LOGICAL_SIZE), FNV1A_64_INIT);
write_log_entry_t entry;
wl_dprintf("Reading checksum\n");
#if BACKING_STORE_WRITE_SIZE == 2
backing_store_read_bulk((WEAR_LEVELING_LOGICAL_SIZE), entry.raw16, 4);
#elif BACKING_STORE_WRITE_SIZE == 4
backing_store_read_bulk((WEAR_LEVELING_LOGICAL_SIZE), entry.raw32, 2);
#elif BACKING_STORE_WRITE_SIZE == 8
backing_store_read((WEAR_LEVELING_LOGICAL_SIZE) + 0, &entry.raw64);
#endif
// If we have a mismatch, clear the cache but do not flag a failure,
// which will cater for the completely clean MCU case.
if (entry.raw64 == expected) {
wl_dprintf("Checksum matches, consolidated data is correct\n");
} else {
wl_dprintf("Checksum mismatch, clearing cache\n");
wear_leveling_clear_cache();
}
}
// If we failed for any reason, then clear the cache
if (status == WEAR_LEVELING_FAILED) {
wear_leveling_clear_cache();
}
return status;
}
/**
* Writes the current cache to consolidated data at the beginning of the backing store.
* Does not clear the write log.
* Pre-condition: this is just after an erase, so we can write directly without reading.
*/
static wear_leveling_status_t wear_leveling_write_consolidated(void) {
wl_dprintf("Writing consolidated data\n");
backing_store_lock_status_t lock_status = wear_leveling_unlock();
wear_leveling_status_t status = WEAR_LEVELING_CONSOLIDATED;
if (!backing_store_write_bulk(0, (backing_store_int_t *)wear_leveling.cache, sizeof(wear_leveling.cache) / sizeof(backing_store_int_t))) {
wl_dprintf("Failed to write to backing store\n");
status = WEAR_LEVELING_FAILED;
}
if (status != WEAR_LEVELING_FAILED) {
// Write out the FNV1a_64 result of the consolidated data
write_log_entry_t entry;
entry.raw64 = fnv_64a_buf(wear_leveling.cache, (WEAR_LEVELING_LOGICAL_SIZE), FNV1A_64_INIT);
wl_dprintf("Writing checksum\n");
do {
#if BACKING_STORE_WRITE_SIZE == 2
if (!backing_store_write_bulk((WEAR_LEVELING_LOGICAL_SIZE), entry.raw16, 4)) {
status = WEAR_LEVELING_FAILED;
break;
}
#elif BACKING_STORE_WRITE_SIZE == 4
if (!backing_store_write_bulk((WEAR_LEVELING_LOGICAL_SIZE), entry.raw32, 2)) {
status = WEAR_LEVELING_FAILED;
break;
}
#elif BACKING_STORE_WRITE_SIZE == 8
if (!backing_store_write((WEAR_LEVELING_LOGICAL_SIZE), entry.raw64)) {
status = WEAR_LEVELING_FAILED;
break;
}
#endif
} while (0);
}
if (lock_status == STATUS_SUCCESS) {
wear_leveling_lock();
}
return status;
}
/**
* Forces a write of the current cache.
* Erases the backing store, including the write log.
* During this operation, there is the potential for data loss if a power loss occurs.
*/
static wear_leveling_status_t wear_leveling_consolidate_force(void) {
wl_dprintf("Erasing backing store\n");
// Erase the backing store. Expectation is that any un-written values that are read back after this call come back as zero.
bool ok = backing_store_erase();
if (!ok) {
wl_dprintf("Failed to erase backing store\n");
return WEAR_LEVELING_FAILED;
}
// Write the cache to the first section of the backing store.
wear_leveling_status_t status = wear_leveling_write_consolidated();
if (status == WEAR_LEVELING_FAILED) {
wl_dprintf("Failed to write consolidated data\n");
}
// Next write of the log occurs after the consolidated values at the start of the backing store.
wear_leveling.write_address = (WEAR_LEVELING_LOGICAL_SIZE) + 8; // +8 due to the FNV1a_64 of the consolidated area
return status;
}
/**
* Potential write of the current cache to the backing store.
* Skipped if the current write log position is not at the end of the backing store.
* During this operation, there is the potential for data loss if a power loss occurs.
*
* @return true if consolidation occurred
*/
static wear_leveling_status_t wear_leveling_consolidate_if_needed(void) {
if (wear_leveling.write_address >= (WEAR_LEVELING_BACKING_SIZE)) {
return wear_leveling_consolidate_force();
}
return WEAR_LEVELING_SUCCESS;
}
/**
* Appends the supplied fixed-width entry to the write log, optionally consolidating if the log is full.
*
* @return true if consolidation occurred
*/
static wear_leveling_status_t wear_leveling_append_raw(backing_store_int_t value) {
bool ok = backing_store_write(wear_leveling.write_address, value);
if (!ok) {
wl_dprintf("Failed to write to backing store\n");
return WEAR_LEVELING_FAILED;
}
wear_leveling.write_address += (BACKING_STORE_WRITE_SIZE);
return wear_leveling_consolidate_if_needed();
}
/**
* Handles writing multi_byte-encoded data to the backing store.
*
* @return true if consolidation occurred
*/
static wear_leveling_status_t wear_leveling_write_raw_multibyte(uint32_t address, const void *value, size_t length) {
const uint8_t * p = value;
write_log_entry_t log = LOG_ENTRY_MAKE_MULTIBYTE(address, length);
for (size_t i = 0; i < length; ++i) {
log.raw8[3 + i] = p[i];
}
// Write to the backing store. See the multi-byte log format in the documentation header at the top of the file.
wear_leveling_status_t status;
#if BACKING_STORE_WRITE_SIZE == 2
status = wear_leveling_append_raw(log.raw16[0]);
if (status != WEAR_LEVELING_SUCCESS) {
return status;
}
status = wear_leveling_append_raw(log.raw16[1]);
if (status != WEAR_LEVELING_SUCCESS) {
return status;
}
if (length > 1) {
status = wear_leveling_append_raw(log.raw16[2]);
if (status != WEAR_LEVELING_SUCCESS) {
return status;
}
}
if (length > 3) {
status = wear_leveling_append_raw(log.raw16[3]);
if (status != WEAR_LEVELING_SUCCESS) {
return status;
}
}
#elif BACKING_STORE_WRITE_SIZE == 4
status = wear_leveling_append_raw(log.raw32[0]);
if (status != WEAR_LEVELING_SUCCESS) {
return status;
}
if (length > 1) {
status = wear_leveling_append_raw(log.raw32[1]);
if (status != WEAR_LEVELING_SUCCESS) {
return status;
}
}
#elif BACKING_STORE_WRITE_SIZE == 8
status = wear_leveling_append_raw(log.raw64);
if (status != WEAR_LEVELING_SUCCESS) {
return status;
}
#endif
return status;
}
/**
* Handles the actual writing of logical data into the write log section of the backing store.
*/
static wear_leveling_status_t wear_leveling_write_raw(uint32_t address, const void *value, size_t length) {
const uint8_t * p = value;
size_t remaining = length;
wear_leveling_status_t status = WEAR_LEVELING_SUCCESS;
while (remaining > 0) {
#if BACKING_STORE_WRITE_SIZE == 2
// Small-write optimizations - uint16_t, 0 or 1, address is even, address <16384:
if (remaining >= 2 && address % 2 == 0 && address < 16384) {
const uint16_t v = ((uint16_t)p[1]) << 8 | p[0]; // don't just dereference a uint16_t here -- if unaligned it generates faults on some MCUs
if (v == 0 || v == 1) {
const write_log_entry_t log = LOG_ENTRY_MAKE_WORD_01(address, v);
status = wear_leveling_append_raw(log.raw16[0]);
if (status != WEAR_LEVELING_SUCCESS) {
// If consolidation occurred, then the cache has already been written to the consolidated area. No need to continue.
// If a failure occurred, pass it on.
return status;
}
remaining -= 2;
address += 2;
p += 2;
continue;
}
}
// Small-write optimizations - address<64:
if (address < 64) {
const write_log_entry_t log = LOG_ENTRY_MAKE_OPTIMIZED_64(address, *p);
status = wear_leveling_append_raw(log.raw16[0]);
if (status != WEAR_LEVELING_SUCCESS) {
// If consolidation occurred, then the cache has already been written to the consolidated area. No need to continue.
// If a failure occurred, pass it on.
return status;
}
remaining--;
address++;
p++;
continue;
}
#endif // BACKING_STORE_WRITE_SIZE == 2
const size_t this_length = remaining >= LOG_ENTRY_MULTIBYTE_MAX_BYTES ? LOG_ENTRY_MULTIBYTE_MAX_BYTES : remaining;
status = wear_leveling_write_raw_multibyte(address, p, this_length);
if (status != WEAR_LEVELING_SUCCESS) {
// If consolidation occurred, then the cache has already been written to the consolidated area. No need to continue.
// If a failure occurred, pass it on.
return status;
}
remaining -= this_length;
address += (uint32_t)this_length;
p += this_length;
}
return status;
}
/**
* "Replays" the write log from the backing store, updating the local cache with updated values.
*/
static wear_leveling_status_t wear_leveling_playback_log(void) {
wl_dprintf("Playback write log\n");
wear_leveling_status_t status = WEAR_LEVELING_SUCCESS;
bool cancel_playback = false;
uint32_t address = (WEAR_LEVELING_LOGICAL_SIZE) + 8; // +8 due to the FNV1a_64 of the consolidated area
while (!cancel_playback && address < (WEAR_LEVELING_BACKING_SIZE)) {
backing_store_int_t value;
bool ok = backing_store_read(address, &value);
if (!ok) {
wl_dprintf("Failed to load from backing store, skipping playback of write log\n");
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
break;
}
if (value == 0) {
wl_dprintf("Found empty slot, no more log entries\n");
cancel_playback = true;
break;
}
// If we got a nonzero value, then we need to increment the address to ensure next write occurs at next location
address += (BACKING_STORE_WRITE_SIZE);
// Read from the write log
write_log_entry_t log;
#if BACKING_STORE_WRITE_SIZE == 2
log.raw16[0] = value;
#elif BACKING_STORE_WRITE_SIZE == 4
log.raw32[0] = value;
#elif BACKING_STORE_WRITE_SIZE == 8
log.raw64 = value;
#endif
switch (LOG_ENTRY_GET_TYPE(log)) {
case LOG_ENTRY_TYPE_MULTIBYTE: {
#if BACKING_STORE_WRITE_SIZE == 2
ok = backing_store_read(address, &log.raw16[1]);
if (!ok) {
wl_dprintf("Failed to load from backing store, skipping playback of write log\n");
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
break;
}
address += (BACKING_STORE_WRITE_SIZE);
#endif // BACKING_STORE_WRITE_SIZE == 2
const uint32_t a = LOG_ENTRY_MULTIBYTE_GET_ADDRESS(log);
const uint8_t l = LOG_ENTRY_MULTIBYTE_GET_LENGTH(log);
if (a + l > (WEAR_LEVELING_LOGICAL_SIZE)) {
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
break;
}
#if BACKING_STORE_WRITE_SIZE == 2
if (l > 1) {
ok = backing_store_read(address, &log.raw16[2]);
if (!ok) {
wl_dprintf("Failed to load from backing store, skipping playback of write log\n");
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
break;
}
address += (BACKING_STORE_WRITE_SIZE);
}
if (l > 3) {
ok = backing_store_read(address, &log.raw16[3]);
if (!ok) {
wl_dprintf("Failed to load from backing store, skipping playback of write log\n");
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
break;
}
address += (BACKING_STORE_WRITE_SIZE);
}
#elif BACKING_STORE_WRITE_SIZE == 4
if (l > 1) {
ok = backing_store_read(address, &log.raw32[1]);
if (!ok) {
wl_dprintf("Failed to load from backing store, skipping playback of write log\n");
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
break;
}
address += (BACKING_STORE_WRITE_SIZE);
}
#endif
memcpy(&wear_leveling.cache[a], &log.raw8[3], l);
} break;
#if BACKING_STORE_WRITE_SIZE == 2
case LOG_ENTRY_TYPE_OPTIMIZED_64: {
const uint32_t a = LOG_ENTRY_OPTIMIZED_64_GET_ADDRESS(log);
const uint8_t v = LOG_ENTRY_OPTIMIZED_64_GET_VALUE(log);
if (a >= (WEAR_LEVELING_LOGICAL_SIZE)) {
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
break;
}
wear_leveling.cache[a] = v;
} break;
case LOG_ENTRY_TYPE_WORD_01: {
const uint32_t a = LOG_ENTRY_WORD_01_GET_ADDRESS(log);
const uint8_t v = LOG_ENTRY_WORD_01_GET_VALUE(log);
if (a + 1 >= (WEAR_LEVELING_LOGICAL_SIZE)) {
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
break;
}
wear_leveling.cache[a + 0] = v;
wear_leveling.cache[a + 1] = 0;
} break;
#endif // BACKING_STORE_WRITE_SIZE == 2
default: {
cancel_playback = true;
status = WEAR_LEVELING_FAILED;
} break;
}
}
// We've reached the end of the log, so we're at the new write location
wear_leveling.write_address = address;
if (status == WEAR_LEVELING_FAILED) {
// If we had a failure during readback, assume we're corrupted -- force a consolidation with the data we already have
status = wear_leveling_consolidate_force();
} else {
// Consolidate the cache + write log if required
status = wear_leveling_consolidate_if_needed();
}
return status;
}
/**
* Wear-leveling initialization
*/
wear_leveling_status_t wear_leveling_init(void) {
wl_dprintf("Init\n");
// Reset the cache
wear_leveling_clear_cache();
// Initialise the backing store
if (!backing_store_init()) {
// If it failed, clear the cache and return with failure
wear_leveling_clear_cache();
return WEAR_LEVELING_FAILED;
}
// Read the previous consolidated values, then replay the existing write log so that the cache has the "live" values
wear_leveling_status_t status = wear_leveling_read_consolidated();
if (status == WEAR_LEVELING_FAILED) {
// If it failed, clear the cache and return with failure
wear_leveling_clear_cache();
return status;
}
status = wear_leveling_playback_log();
if (status == WEAR_LEVELING_FAILED) {
// If it failed, clear the cache and return with failure
wear_leveling_clear_cache();
return status;
}
return status;
}
/**
* Wear-leveling erase.
* Post-condition: any reads from the backing store directly after an erase operation must come back as zero.
*/
wear_leveling_status_t wear_leveling_erase(void) {
wl_dprintf("Erase\n");
// Unlock the backing store
backing_store_lock_status_t lock_status = wear_leveling_unlock();
if (lock_status == STATUS_FAILURE) {
wear_leveling_lock();
return WEAR_LEVELING_FAILED;
}
// Perform the erase
bool ret = backing_store_erase();
wear_leveling_clear_cache();
// Lock the backing store if we acquired the lock successfully
if (lock_status == STATUS_SUCCESS) {
ret &= (wear_leveling_lock() != STATUS_FAILURE);
}
return ret ? WEAR_LEVELING_SUCCESS : WEAR_LEVELING_FAILED;
}
/**
* Writes logical data into the backing store. Skips writes if there are no changes to values.
*/
wear_leveling_status_t wear_leveling_write(const uint32_t address, const void *value, size_t length) {
wl_assert(address + length <= (WEAR_LEVELING_LOGICAL_SIZE));
if (address + length > (WEAR_LEVELING_LOGICAL_SIZE)) {
return WEAR_LEVELING_FAILED;
}
wl_dprintf("Write ");
wl_dump(address, value, length);
// Skip write if there's no change compared to the current cached value
if (memcmp(value, &wear_leveling.cache[address], length) == 0) {
return true;
}
// Update the cache before writing to the backing store -- if we hit the end of the backing store during writes to the log then we'll force a consolidation in-line
memcpy(&wear_leveling.cache[address], value, length);
// Unlock the backing store
backing_store_lock_status_t lock_status = wear_leveling_unlock();
if (lock_status == STATUS_FAILURE) {
wear_leveling_lock();
return WEAR_LEVELING_FAILED;
}
// Perform the actual write
wear_leveling_status_t status = wear_leveling_write_raw(address, value, length);
switch (status) {
case WEAR_LEVELING_CONSOLIDATED:
case WEAR_LEVELING_FAILED:
// If the write triggered consolidation, or the write failed, then nothing else needs to occur.
break;
case WEAR_LEVELING_SUCCESS:
// Consolidate the cache + write log if required
status = wear_leveling_consolidate_if_needed();
break;
default:
// Unsure how we'd get here...
status = WEAR_LEVELING_FAILED;
break;
}
if (lock_status == STATUS_SUCCESS) {
if (wear_leveling_lock() == STATUS_FAILURE) {
status = WEAR_LEVELING_FAILED;
}
}
return status;
}
/**
* Reads logical data from the cache.
*/
wear_leveling_status_t wear_leveling_read(const uint32_t address, void *value, size_t length) {
wl_assert(address + length <= (WEAR_LEVELING_LOGICAL_SIZE));
if (address + length > (WEAR_LEVELING_LOGICAL_SIZE)) {
return WEAR_LEVELING_FAILED;
}
// Only need to copy from the cache
memcpy(value, &wear_leveling.cache[address], length);
wl_dprintf("Read ");
wl_dump(address, value, length);
return WEAR_LEVELING_SUCCESS;
}
/**
* Weak implementation of bulk read, drivers can implement more optimised implementations.
*/
__attribute__((weak)) bool backing_store_read_bulk(uint32_t address, backing_store_int_t *values, size_t item_count) {
for (size_t i = 0; i < item_count; ++i) {
if (!backing_store_read(address + (i * BACKING_STORE_WRITE_SIZE), &values[i])) {
return false;
}
}
return true;
}
/**
* Weak implementation of bulk write, drivers can implement more optimised implementations.
*/
__attribute__((weak)) bool backing_store_write_bulk(uint32_t address, backing_store_int_t *values, size_t item_count) {
for (size_t i = 0; i < item_count; ++i) {
if (!backing_store_write(address + (i * BACKING_STORE_WRITE_SIZE), values[i])) {
return false;
}
}
return true;
}