* Fix USER_PRINT on avr/atsam
* Update tmk_core/common/arm_atsam/_print.h
Co-authored-by: Ryan <fauxpark@gmail.com>
Co-authored-by: Ryan <fauxpark@gmail.com>
* Fix keycode mappings for via and ensure they don't change within protocol
* Update keycodes
* Fix broken keyboards
* added the missing keycodes found in via
* Remove invalid keycodes
Co-authored-by: David Hoelscher <infinityis@users.noreply.github.com>
* Modified tmk_core/rules.mk to avoid linking errors
Added -fcommon flag to avoid linking errors due to multiple variable definitions. Though this is neither a definitive nor good solution, proper changes and use of extern keyword to avoid those multiple definitions must be made
* Comment updated
* Initial refactor of ARM SLEEP_LED to enable more platforms
* fix build issues
* Disable SLEEP_LED for boards with no caps lock code
* Enable GPT14 for boards with caps lock code and SLEEP_LED enabled
* Enable GPT for boards with caps lock code and SLEEP_LED enabled
* tmk_core/common/report.h: define `enum mouse_buttons` in terms of `#define MOUSE_BTN_MASK()`
* tmk_core/common/action.c: collapse multiple `case KC_MS_BTN[1-8]:` into single `MOUSE_BTN_MASK(action.key.code - KC_MS_BTN1)`
We all love tapping on our keyboards but this is taking the piss.
This saves ~134 bytes on my ATmega32.
* Remove GCC check from debug
* Remove platform logic from common.mk
* Refactor platform logic within print.h
* restore debug.c format
* headers
* Rename function pointer type
* review comments
* Update tmk_core/common/printf.c
Co-authored-by: Nick Brassel <nick@tzarc.org>
* Format
Co-authored-by: Nick Brassel <nick@tzarc.org>
* Fix DEBUG_MATRIX_SCAN_RATE on chibiOS when console is enabled
* update type in dprintf
Co-authored-by: Ryan <fauxpark@gmail.com>
Co-authored-by: Ryan <fauxpark@gmail.com>
* resolve race condition between suspend and wake in LUFA
* avoid multiple calls to suspend_power_down() / suspend_wakeup_init()
* Remove duplicate suspend_power_down_kb() call
* pause on wakeup to wait for USB state to settle
* need the repeated suspend_power_down() (that's where the sleep is)
* more efficient implementation
* fine tune the pause after sending wakeup
* speculative chibios version of pause-after-wake
* make wakeup delay configurable, and adjust value
* better location for wakeup delay
* Add support for 8 buttons to mouse report
This includes support for 8 buttons in mousekeys. However, this does move the keys around due to the fact that the last mousekey keycode is already 0xFF, so any past that would not work with register_code and the like, breaking them for tap hold keys, encoders, and other features.
* Update mouse key docs
* Add changes based on feedback
* Fix VUSB report size comment
Because drashna red gud
* Fix typo in action.c
* Fix IS_MOUSE_BUTTON check
* Change start range for mousekeys so that the end is 0xFF properly
* condense mousekeys check
* fix matrix_io_delay() timing in quantum/matrix.c
* Updated comments explaining the need for matrix_io_delay() in quantum/matrix.c
* fix matrix_io_delay() timing in quantum/split_common/matrix.c
* Update quantum/matrix.c
Co-authored-by: Ryan <fauxpark@gmail.com>
* Update quantum/split_common/matrix.c
Co-authored-by: Ryan <fauxpark@gmail.com>
* Update quantum/matrix.c
Co-authored-by: Ryan <fauxpark@gmail.com>
* Update quantum/split_common/matrix.c
Co-authored-by: Ryan <fauxpark@gmail.com>
* add waitOutputPinValid() and wait_cpuclock() into quantum/quantum.h and tmk_core/common/wait.h
* add matrix_output_select_delay() and matrix_output_unselect_delay()
* fix quantum/matrix_common.c, tmk_core/common/matrix.h
* fix tmk_core/common/wait.h
* fix quantum/quantum.h, tmk_core/common/wait.h
* waitOutputPinValid() rename to waitInputPinDelay() in quantum/quantum.h.
* waitOutputPinValid() rename to waitInputPinDelay() in quantum/matrix_common.c
* update tmk_core/common/wait.h
* update comment in quantum/matrix.c, quantum/split_common/matrix.c
* update quantum/quantum.h: Make more margin in the GPIO_INPUT_PIN_DELAY default value.
Co-authored-by: Ryan <fauxpark@gmail.com>
* Allow recording of the last matrix activity time, to simplify implementation of display timeouts and the like.
* Add requested changes from code review.
* Simplify split matrix last changed.
* Add proper prototypes for Tap-Hold Per Key functions
* Fix handwired/tennie default keymap
* Remove unneeded references
* Fix tapping term per key check in space cadet
* Pre-emptive fix for tap dance
* Fix marksard/leftover30
* Replace hard coded tapping term with define
* Add target 'build-for-compare' to `build_keyboard.mk`
The `build-for-compare` target provides an easy way to check the md5 checksum of the generated binary.
You can easily see if there is any change in the generated binaries between the two versions, as in the example below.
```
$ git checkout 0.11.0
M build_keyboard.mk
M tmk_core/rules.mk
Note: checking out '0.11.0'.
HEAD is now at c66df1664 2020 November 28 Breaking Changes Update (#11053)
$ make helix:all:build-for-compare | grep ^MD5
MD5 (.build/helix_rev2_default.hex) = 5c3606562c944bb4d18832e601b45d4a
MD5 (.build/helix_rev2_edvorakjp.hex) = 9e43d13d389d518ba7e99cd7337e28d6
MD5 (.build/helix_rev2_five_rows.hex) = 8bcb61c2fd5d237c2997f2fa007d4934
MD5 (.build/helix_rev2_five_rows_jis.hex) = b97cd818d52f73ca2d4e78c86d90a791
MD5 (.build/helix_rev2_froggy.hex) = c492172364188f4e2918b10bf0f3a0a6
MD5 (.build/helix_rev2_froggy_106.hex) = b0861fd735a8f81881a8c02730641a2b
MD5 (.build/helix_rev2_led_test.hex) = 5c97d982a5da5cfb3dacb28a8934b81d
MD5 (.build/helix_rev2_xulkal.hex) = 01f603dc46bcf9094d7e106831d8f5b1
MD5 (.build/helix_rev2_yshrsmz.hex) = 5a008bca2d0c5790a151c02834c529ba
$ git checkout 0.11.1
M build_keyboard.mk
M tmk_core/rules.mk
Previous HEAD position was c66df1664 2020 November 28 Breaking Changes Update (#11053)
HEAD is now at cc08e3082 nix-shell: add milc dependency (#11086)
$ make helix:all:build-for-compare | grep ^MD5
MD5 (.build/helix_rev2_default.hex) = 5c3606562c944bb4d18832e601b45d4a
MD5 (.build/helix_rev2_edvorakjp.hex) = 9e43d13d389d518ba7e99cd7337e28d6
MD5 (.build/helix_rev2_five_rows.hex) = 8bcb61c2fd5d237c2997f2fa007d4934
MD5 (.build/helix_rev2_five_rows_jis.hex) = b97cd818d52f73ca2d4e78c86d90a791
MD5 (.build/helix_rev2_froggy.hex) = c492172364188f4e2918b10bf0f3a0a6
MD5 (.build/helix_rev2_froggy_106.hex) = b0861fd735a8f81881a8c02730641a2b
MD5 (.build/helix_rev2_led_test.hex) = 5c97d982a5da5cfb3dacb28a8934b81d
MD5 (.build/helix_rev2_xulkal.hex) = d848383adfd7463b138c6da179cf1436
MD5 (.build/helix_rev2_yshrsmz.hex) = 5a008bca2d0c5790a151c02834c529ba
```
* make builds reproducable by default
* update build_keyboard.mk: remove 'build-for-compare' target
* GNU make (3.81) on macOS 10.14(Mojave) does not have the 'undefine' directive.
* Adopted fauxpark's suggestion.
* Update tmk_core/rules.mk
Co-authored-by: Ryan <fauxpark@gmail.com>
* update tmk_core/rules.mk
* fix tmk_core/rules.mk
Co-authored-by: Zach White <skullydazed@gmail.com>
Co-authored-by: Ryan <fauxpark@gmail.com>
* Update ADB impelmentation in tmk_core to recent version.
Pcked from tmk_keyboard repository revision: 48d696443857512d45f9a7329e0dd0a76345860f
* Restore convenient ADB functions used in QMK port.
* Do cformat.
* Split dmac_desc declaration and definition
According to the official documentation[1] gcc 10 is more strict about
correct extern usage.
I've had to move the definition of dmac_desc and dmac_desc_wb from
i2c_master.h to the corresponding .c file. This could be an issue if
anyone includes the i2c_master.h file without liking with the object
file.
[1]: https://gcc.gnu.org/gcc-10/porting_to.html
* Remove the keymap_config definition from keymaps
The keymap_config def was conflicting with the one found in
tmk_core/common/magic.c. Declaring it extern in magic.c breaks a bunch
of keyboard that rely on that declaration (like the ergodox). Instead
I've removed the one found in the keymap.c of the massdrop alt.
The same change will have to be made to other keyboards.
- Set all other ChibiOS defaults to 'off', when not targeting Proton-C
- Modified all existing F303 boards to point at the QMK_PROTON_C to ensure repeatable binary output
- Modified version.h generation so that SKIP_VERSION=yes generates the same output
* Branch point for 2020 November 28 Breaking Change
* Remove matrix_col_t to allow MATRIX_ROWS > 32 (#10183)
* Add support for soft serial to ATmega32U2 (#10204)
* Change MIDI velocity implementation to allow direct control of velocity value (#9940)
* Add ability to build a subset of all keyboards based on platform.
* Actually use eeprom_driver_init().
* Make bootloader_jump weak for ChibiOS. (#10417)
* Joystick 16-bit support (#10439)
* Per-encoder resolutions (#10259)
* Share button state from mousekey to pointing_device (#10179)
* Add hotfix for chibios keyboards not wake (#10088)
* Add advanced/efficient RGB Matrix Indicators (#8564)
* Naming change.
* Support for STM32 GPIOF,G,H,I,J,K (#10206)
* Add milc as a dependency and remove the installed milc (#10563)
* ChibiOS upgrade: early init conversions (#10214)
* ChibiOS upgrade: configuration file migrator (#9952)
* Haptic and solenoid cleanup (#9700)
* XD75 cleanup (#10524)
* OLED display update interval support (#10388)
* Add definition based on currently-selected serial driver. (#10716)
* New feature: Retro Tapping per key (#10622)
* Allow for modification of output RGB values when using rgblight/rgb_matrix. (#10638)
* Add housekeeping task callbacks so that keyboards/keymaps are capable of executing code for each main loop iteration. (#10530)
* Rescale both ChibiOS and AVR backlighting.
* Reduce Helix keyboard build variation (#8669)
* Minor change to behavior allowing display updates to continue between task ticks (#10750)
* Some GPIO manipulations in matrix.c change to atomic. (#10491)
* qmk cformat (#10767)
* [Keyboard] Update the Speedo firmware for v3.0 (#10657)
* Maartenwut/Maarten namechange to evyd13/Evy (#10274)
* [quantum] combine repeated lines of code (#10837)
* Add step sequencer feature (#9703)
* aeboards/ext65 refactor (#10820)
* Refactor xelus/dawn60 for Rev2 later (#10584)
* add DEBUG_MATRIX_SCAN_RATE_ENABLE to common_features.mk (#10824)
* [Core] Added `add_oneshot_mods` & `del_oneshot_mods` (#10549)
* update chibios os usb for the otg driver (#8893)
* Remove HD44780 References, Part 4 (#10735)
* [Keyboard] Add Valor FRL TKL (+refactor) (#10512)
* Fix cursor position bug in oled_write_raw functions (#10800)
* Fixup version.h writing when using SKIP_VERSION=yes (#10972)
* Allow for certain code in the codebase assuming length of string. (#10974)
* Add AT90USB support for serial.c (#10706)
* Auto shift: support repeats and early registration (#9826)
* Rename ledmatrix.h to match .c file (#7949)
* Split RGB_MATRIX_ENABLE into _ENABLE and _DRIVER (#10231)
* Split LED_MATRIX_ENABLE into _ENABLE and _DRIVER (#10840)
* Merge point for 2020 Nov 28 Breaking Change
* Update massdrop/alt keyboard to fix Caps Lock LED flickering.
* Fix Caps-lock LED during underglow-only for massdrop/ctrl
* Update keyboards/massdrop/ctrl/keymaps/default/keymap.c
* First pass
* Add config options to docs
* Update some wording
* Slight tidy up of backlight caps logic
* Init pin to correct state
* Move init location
* Reverse default state
The `KEYBOARD_SHARED_EP=yes` option was breaking the VIA support,
because the raw HID interface number in this case was 0 instead of 1,
and the VIA app depends on the exact interface number for raw HID.
Change the interface ordering to put the shared interface before the raw
HID interface if `KEYBOARD_SHARED_EP` is enabled, so that the raw HID
interface can keep its number.
* Output an error message if LINK_TIME_OPTIMIZATION_ENABLE is set but LTO_ENABLE is not.
* Update common.mk
Specify that LINK_TIME_OPTIMZATION_ENABLE has been renamed, not deprecated.
* Better handle LTO_ENABLE
Especially when calling from command line
* Replace LINK_TIME_OPTIMIZATION_ENABLE with LTO_ENABLE
* Remove long for LTO from show_options.mk
* Update vusb to match 3rd endpoint.
- With the addition of https://github.com/qmk/v-usb/pull/1 a 3rd endpoint (endpoint4) becomes available.
- We can assign mouse/extrakeys to that endpoint as its a desirable feature and leave rawhid and console to compete for the 2nd endpoint.
NOTE: The version of vusb.c in future branch is older than master. Just remember that it will need a #error if both raw_hid and console are enabled at the same time.
* Final Fixes
* Update tmk_core/protocol/vusb/vusb.c
* Update tmk_core/protocol/vusb/vusb.c
* Update tmk_core/protocol/vusb/usbconfig.h
* Update tmk_core/protocol/vusb/usbconfig.h
* Update tmk_core/protocol/vusb/usbconfig.h
* Update tmk_core/protocol/vusb/usbconfig.h
* Updated vusb submodule to latest commit
* Add `st-flash` flash target
Add support for flashing the firmware via the `st-flash` utility from
the STLink Tools package (https://github.com/stlink-org/stlink).
* Add `st-flash` to the `qmk flash -b` output
* Initialize Layer State on startup
Right now, on startup, the default layer state gets called and set, triggering the callback functions for the default layer state. However, the normal layer state never actually gets initialized. It's set to 0 directly, by default, but the callback functions are never actually called. This creates some inconsistency in the behavior for end users. This adds a simple "clear" that triggers the callback on startup. This should produce more consisten behavior between the two functions and layer masks.
* Stupid hack
* Fix type casting?
* Fix compile issues with magic is disabled
* add support for hid gamepad interface
add documentation for HID joystick
Add joystick_task to read analog axes values even when no key is pressed or release. update doc
Update docs/feature_joystick.md
Manage pin setup and read to maintain matrix scan after analog read
* Incorporates patches and changes to HID reporting
There are some patches provided by @a-chol incorporated on this commit,
and also some changes I made to the HID Report structure.
The most interesting is the one dealing with number of buttons: Linux
doesn't seem to care, but Windows requires the HID structure to be byte
aligned (that's in the spec). So if one declares 8/16/32... buttons they
should not have any issues, but this is what happens when you have 9
buttons:
```
bits |0|1|2|3|4|5|6|7|
|*|*|*|*|*|*|*|*| axis 0 (report size 8)
|*|*|*|*|*|*|*|*| ...
|*|*|*|*|*|*|*|*|
|*|*|*|*|*|*|*|*|
|*|*|*|*|*|*|*|*|
|*|*|*|*|*|*|*|*|
|*|*|*|*|*|*|*|*| axis 6
|*|*|*|*|*|*|*|*| first 8 buttons (report size 1)
|*| | | | | | | | last of 9 buttons, not aligned
```
So for that I added a conditonal that will add a number of reports with
size 1 to make sure it aligns to the next multiple of 8. Those reports
send dummy inputs that don't do anything aside from aligning the data.
Tested on Linux, Windows 10 and Street Fighter (where the joystick is
recognized as direct-input)
* Add save and restore of each pin used in reading joystick (AVR).
Allow output pin to be JS_VIRTUAL_AXIS if the axis is connected to Vcc
instead of an output pin from the MCU.
Fix joystick report id
Fix broken v-usb hid joystick interface. Make it more resilient to unusual settings (none multiple of eight button count, 0 buttons or 0 axes)
Correct adc reading for multiple axes. Piecewise range conversion for uncentered raw value range. Input, output and ground pin configuration per axis.
Documentation fixes
* Fix port addressing for joystick analog read
* The other required set of changes
As per the PR, the changes still holding it up.
Add onekey for testing.
Fix ARM builds.
Fix device descriptor when either axes or buttons is zero.
Add compile-time check for at least one axis or button.
Move definition to try to fix conflict.
PR review comments.
qmk cformat
* avoid float functions to compute range mapping for axis adc reading
* Remove V-USB support for now. Updated docs accordingly.
* Update tmk_core/protocol/lufa/lufa.c
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Update tmk_core/protocol/usb_descriptor.c
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Update tmk_core/protocol/usb_descriptor.c
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Update tmk_core/protocol/usb_descriptor.c
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Add support for joystick adc reading for stm32 MCUs. Fix joystick hid report sending for chibios
* Fix HID joystick report sending for ChibiOS.
Add one analog axis to the onekey:joystick keymap.
Fix pin state save and restore during joystick analog read for STM32
MCUs.
* Update tmk_core/protocol/chibios/usb_main.c
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Update tmk_core/protocol/lufa/lufa.c
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Add missing mcuconf.h and halconf.h to onekey:joystick keymap.
Add suggested fixes from PR.
* Switch saveState and restoreState signature to use pin_t type.
onekey:joystick : add a second axis, virtual and programmatically animated.
* Update docs/feature_joystick.md
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Update docs/feature_joystick.md
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Add PR corrections
* Remove halconf.h and mcuconf.h from onekey keymaps
* Change ADC_PIN to A0
Co-authored-by: achol <allecooll@hotmail.com>
Co-authored-by: José Júnior <jose.junior@gmail.com>
Co-authored-by: a-chol <achol@notamail.com>
Co-authored-by: Nick Brassel <nick@tzarc.org>
Co-authored-by: Ryan <fauxpark@gmail.com>
* Disable NKRO on V-USB controllers
* not _currently_ supported text
Co-authored-by: Ryan <fauxpark@gmail.com>
Co-authored-by: Ryan <fauxpark@gmail.com>
* Re-fix the dual-bank bootloader stuff.
* Use wait_ms() instead of using nop's for a delay, as ChibiOS is actually running at the time of bootloader jump.
Mousekey scrolling should have a separate repeat variable
to keep track of scrolling acceleration, instead of being
tied to mouse movement scolling in mousekeys. The send function
should record when the last movement was made since this is
when movement is actually sent. Doing this fixes the bug where
the initial press of a mousekey scroll button causes a double scroll.
Signed-off-by: Daniel Hong <daniel.hong@live.com>
* Initial work for consolidation of board files and default ChibiOS configs.
* Migrate F401/F411 black pills for testing.
* Add early init bootloader jump flag.
* Add support for I2C in order to use i2c_scanner keymap.
* Add F401/F411 HSE bypass to get things booting.
* Exempt "hooked" ChibiOS conf files from updater script.
* Fix up ordering for bootloader_defs file check.
* Match previous $(KEYBOARD_PATHS) value for Proton-C, updated for all board configs.
Fixes the handling for the oneshot cleanup, so it only cleans up if it is active. It should not cleanup of SHO is off (eg using a normal oneshot key), nor if it's actively pressed or used.
Previous behavior BROKE swap hand key.
* I2C_TIMEOUT is not defined on arm teensy
* Work round teensy having different ChibiOS config options
* Stash OLED conf files
* update comment
* update comment
* Remove stm32 alias to allow teensy alt mode
Process mouse movement in the keymap before it is sent to the host. Example uses
include filtering noise, adding acceleration, and automatically activating a
layer. To use, define the following function in your keymap:
void ps2_mouse_moved_user(report_mouse_t *mouse_report);
With this change, when ps2_mouse is disabled, mousekeys works as usual. With
ps2_mouse enabled, mousekeys button state is shared with ps2_mouse for clicking,
dragging, and scrolling, mousekeys clicks are produced by ps2_mouse only, and
mouskeys button state is transferred to mousekeys without generating clicks to
enable mousekeys dragging.
Co-authored-by: Drashna Jaelre <drashna@live.com>
Co-authored-by: Drashna Jaelre <drashna@live.com>
* init
* add RETRO_TAP; tap anyway after TAP_TERM, if no interruption
* RETRO_TAP works for other types of taps
* revert to upstream/master
* explain this fork in readme
* use one readme.md file instaed
* fix the error if NO_ACTION_ONESHOT is defined
* restore readme.md to upstream master
Co-authored-by: Tsan-Kuang Lee <tsan.kuang.lee@gmail.com>
* Branch point for 2020 May 30 Breaking Change
* Migrate `ACTION_LAYER_TOGGLE` to `TG()` (#8954)
* Migrate `ACTION_MODS_ONESHOT` to `OSM()` (#8957)
* Migrate `ACTION_DEFAULT_LAYER_SET` to `DF()` (#8958)
* Migrate `ACTION_LAYER_MODS` to `LM()` (#8959)
* Migrate `ACTION_MODS_TAP_KEY` to `MT()` (#8968)
* Convert V-USB usbdrv to a submodule (#8321)
* Unify Tap Hold functions and documentation (#8348)
* Changing board names to prevent confusion (#8412)
* Move the Keyboardio Model01 to a keyboardio/ subdir (#8499)
* Move spaceman keyboards (#8830)
* Migrate miscellaneous `fn_actions` entries (#8977)
* Migrate `ACTION_MODS_KEY` to chained mod keycodes (#8979)
* Organizing my keyboards (plaid, tartan, ergoinu) (#8537)
* Refactor Lily58 to use split_common (#6260)
* Refactor zinc to use split_common (#7114)
* Add a message if bin/qmk doesn't work (#9000)
* Fix conflicting types for 'tfp_printf' (#8269)
* Fixed RGB_DISABLE_AFTER_TIMEOUT to be seconds based & small internals cleanup (#6480)
* Refactor and updates to TKC1800 code (#8472)
* Switch to qmk forks for everything (#9019)
* audio refactor: replace deprecated PLAY_NOTE_ARRAY (#8484)
* Audio enable corrections (2/3) (#8903)
* Split HHKB to ANSI and JP layouts and Add VIA support for each (#8582)
* Audio enable corrections (Part 4) (#8974)
* Fix typo from PR7114 (#9171)
* Augment future branch Changelogs (#8978)
* Revert "Branch point for 2020 May 30 Breaking Change"
This commits add the SH_OS keycode, which works similarly to one shot
layers:
* while pressed, the keyboard is swapped
* if no keys were pressed while it was pressed, the next key press is
swapped
SH_OS also supports chaining with one shot layers:
OSL(x) + SH_OS + key interprets the key press on the oneshot layer.
The ONESHOT_TIMEOUT setting used by one shot keys and layers is also
used by oneshot swap hands. In the above chaining scenario the timeout
of the oneshot layer is reset when swap hands is activated.
Resolves#2682
* Improve security by eliminating the use of well-known names.
* Add an additional $ so the shell expands $TMP1 and $TMP2
Co-authored-by: Joel Challis <git@zvecr.com>
Co-authored-by: Joel Challis <git@zvecr.com>
* Refactor to use mpaland/printf
* trim firmware size
* remove keymap changes
* run clang format
* Fixup after rebase
* fix up git-submodule command for printf
* Add support for hardware and board initialisation overrides.
* qmk cformat.
* Add some documentation.
* Docs clarity.
* Make early_hardware_init_pre a no-op for now, until migrations occur.
* Doco update
* Make distinction between keyboard and ChibiOS board in docs
* Doc anchors.
* Update tmk_core/protocol/chibios/main.c
Co-Authored-By: Joel Challis <git@zvecr.com>
* Rework bootloader entry to be off by default, allow opting-in.
Co-authored-by: Joel Challis <git@zvecr.com>
* tmk_core/common: Fixing TIMER_DIFF macro to calculate difference correctly after the timer wraps.
Let's go through an example, using the following macro:
If the first timer read is 0xe4 and the second one is 0x32, the timer wrapped.
If the timer would have had more bits, it's new value would have been 0x132,
and the correct difference in time is 0x132 - 0xe4 = 0x4e
old code TIMER_DIFF_8(0x32, 0xe4) = 0xff - 0xe4 + 0x32 = 0x4d, which is wrong.
new code TIMER_DIFF_8(0x32, 0xe4) = 0xff + 1 - 0xe4 + 0x32 = 0x4e, which is correct.
This also gives a chance for a smart compiler to optimize the code using normal
integer overflow.
For example on AVR, the following C code:
uint8_t __attribute__ ((noinline)) test(uint8_t current_timer, uint8_t start_timer)
{
return TIMER_DIFF_8(current_timer, start_timer);
}
With the original code, it gets translated to the following list of instructions:
00004c6e <test>:
4c6e: 98 2f mov r25, r24
4c70: 86 1b sub r24, r22
4c72: 96 17 cp r25, r22
4c74: 08 f4 brcc .+2 ; 0x4c78 <test+0xa>
4c76: 81 50 subi r24, 0x01 ; 1
4c78: 08 95 ret
But with this commit, it gets translated to a single instruction:
00004c40 <test>:
4c40: 86 1b sub r24, r22
4c42: 08 95 ret
This unfortunately doesn't always work so nicely, for example the following C code:
int __attribute__ ((noinline)) test(uint8_t current_timer, uint8_t start_timer)
{
return TIMER_DIFF_8(current_timer, start_timer);
}
(Note: return type changed to int)
With the original code it gets translated to:
00004c6e <test>:
4c6e: 28 2f mov r18, r24
4c70: 30 e0 ldi r19, 0x00 ; 0
4c72: 46 2f mov r20, r22
4c74: 50 e0 ldi r21, 0x00 ; 0
4c76: 86 17 cp r24, r22
4c78: 20 f0 brcs .+8 ; 0x4c82 <test+0x14>
4c7a: c9 01 movw r24, r18
4c7c: 84 1b sub r24, r20
4c7e: 95 0b sbc r25, r21
4c80: 08 95 ret
4c82: c9 01 movw r24, r18
4c84: 84 1b sub r24, r20
4c86: 95 0b sbc r25, r21
4c88: 81 50 subi r24, 0x01 ; 1
4c8a: 9f 4f sbci r25, 0xFF ; 255
4c8c: 08 95 ret
Wth this commit it gets translated to:
00004c40 <test>:
4c40: 28 2f mov r18, r24
4c42: 30 e0 ldi r19, 0x00 ; 0
4c44: 46 2f mov r20, r22
4c46: 50 e0 ldi r21, 0x00 ; 0
4c48: 86 17 cp r24, r22
4c4a: 20 f0 brcs .+8 ; 0x4c54 <test+0x14>
4c4c: c9 01 movw r24, r18
4c4e: 84 1b sub r24, r20
4c50: 95 0b sbc r25, r21
4c52: 08 95 ret
4c54: c9 01 movw r24, r18
4c56: 84 1b sub r24, r20
4c58: 95 0b sbc r25, r21
4c5a: 93 95 inc r25
4c5c: 08 95 ret
There is not much performance improvement in this case, however at least with this
commit it functions correctly.
Note: The following commit will improve compiler output for the latter example.
* tmk_core/common: Improve code generation for TIMER_DIFF* macros
Because of integer promotion the compiler is having a hard time generating
efficient code to calculate TIMER_DIFF* macros in some situations.
In the below example, the return value is "int", and this is causing the
trouble.
Example C code:
int __attribute__ ((noinline)) test(uint8_t current_timer, uint8_t start_timer)
{
return TIMER_DIFF_8(current_timer, start_timer);
}
BEFORE: (with -Os)
00004c40 <test>:
4c40: 28 2f mov r18, r24
4c42: 30 e0 ldi r19, 0x00 ; 0
4c44: 46 2f mov r20, r22
4c46: 50 e0 ldi r21, 0x00 ; 0
4c48: 86 17 cp r24, r22
4c4a: 20 f0 brcs .+8 ; 0x4c54 <test+0x14>
4c4c: c9 01 movw r24, r18
4c4e: 84 1b sub r24, r20
4c50: 95 0b sbc r25, r21
4c52: 08 95 ret
4c54: c9 01 movw r24, r18
4c56: 84 1b sub r24, r20
4c58: 95 0b sbc r25, r21
4c5a: 93 95 inc r25
4c5c: 08 95 ret
AFTER: (with -Os)
00004c40 <test>:
4c40: 86 1b sub r24, r22
4c42: 90 e0 ldi r25, 0x00 ; 0
4c44: 08 95 ret
Note: the example is showing -Os but improvements can be seen at all optimization levels,
including -O0. We never use -O0, but I tested it to make sure that no extra code is
generated in that case.OA
* quantum/debounce: Fix custom wrapping timers in eager_pr and eager_pk debounce algorithms
Please see the below simulated sequence of events:
Column A is the 16-bit value returned by read_timer();
Column B is the value returned by custom_wrap_timer_read();
Column C is the original code: (timer_read() % MAX_DEBOUNCE)
A, B, C
65530, 19, 30
65531, 20, 31
65532, 21, 32
65533, 22, 33
65534, 23, 34
65535, 24, 35
0 25, 0
1, 26, 1
2, 27, 2
3, 28, 3
4, 29, 4
5, 30, 5
read_timer() wraps about every 1.09 seconds, and so debouncing might
fail at these times without this commit.
* quantum/debounce/eager_pr and eager_pk: modifications for code readability according to code review.
* quantum/debounce/eager_pr and eager_pk: modifications for code readability according to code review. (2)
* Add PS2_MOUSE_ROTATE to compensate for device orientation
* fixup! Add PS2_MOUSE_ROTATE to compensate for device orientation
* Reformat with IndentPPDirectives: AfterHash as per #6316
* Define NO_ACTION_MACRO/FUNCTION in header instead of makefile when LTO is enabled
Currently, boards and keymaps that define NO_ACTION_MACRO/FUNCTION unconditionally
will not compile with LTO_ENABLE (#8604). This fixes the issue by moving the
definitions from common.mk to action.h, which enables us to check for previous
definitions of those macros (this cannot be done in a makefile).
* Remove LTO checks in templates
Since now NO_ACTION_MACRO/FUNCTION are defined as needed in action.h (which is
included by quantum.h), checking for LTO in keyboard and user code is no
longer required.
* Update LTO_ENABLE docs
* Change _delay_ms/us() to wait_ms/us()
* Switch to platform-agnostic GPIO macros
* Add AVR spi_master and migrate Adafruit BLE code
* Set verbose back to false
* Add clock divisor, bit order and SPI mode configuration for init
* Add start and stop functions
* Move configuration of mode, endianness and speed to `spi_start()`
* Some breaks here would be good
* Default Adafruit BLE clock divisor to 4 (2MHz on the Feather 32U4)
* Remove mode and divisor enums
* Add some docs
* No hr at EOF
* Add links in sidebar
* rewrite usbhid feature on vusb
* Apply suggestions from code review
Co-Authored-By: Ryan <fauxpark@gmail.com>
* fix typo
* fix typo again
* Update tmk_core/protocol/vusb/vusb.c
Co-Authored-By: Ryan <fauxpark@gmail.com>
* clean up defines
Co-authored-by: Ryan <fauxpark@gmail.com>
* Update tmk_core/common/progmem.h
Co-Authored-By: Ryan <fauxpark@gmail.com>
* Update quantum/rgblight.c
Co-Authored-By: Ryan <fauxpark@gmail.com>
* fixed problem with implicit declaration in quantum/rgblight.c (#8381)
Co-authored-by: Ryan <fauxpark@gmail.com>
* Port over some AVR backlight logic to SLEEP_LED
* Port over some AVR backlight logic to SLEEP_LED - add timer 3
* Port over some AVR backlight logic to SLEEP_LED - clang format
* Enable SLEEP_LED within vusb protocol
* Add support for RAW endpoint for arm_atsam
This the excellent work from helluvamatt/qmk_firmware in bb6eeb93b.
* Reformat arm_atsam RAW endpoint code
Co-authored-by: Matt Schneeberger <helluvamatt@gmail.com>
* Improve process_record system
Code based on @colinta's
* Rename and better handle functions
* Fix incorrect function call to process_record_user
* Add documentation for post_process_record
* Add both get_event_keycode and get_record_keycode functions
And add some comments about these functions
* Update code format
* Cleanup merge artifacts
* Add support for Bootmagic lite when using SPLIT_HAND_PIN
* Deduplicate bootmagic_lite logic from within via
* Revert location of defaults so that user overrides still work for now
* Tidy up code slightly
