| Message ID | 20241031063004.69956-1-markus.burri@mt.com |
|---|---|
| Headers | show |
| Series | Input: matrix-keypad: Various performance improvements | expand |
Hi Conor, On Tue, Nov 05, 2024 at 06:22:36PM +0000, Conor Dooley wrote: > On Tue, Nov 05, 2024 at 02:03:19PM +0100, Markus Burri wrote: > > Add missing property 'gpio-activelow' to DT schema. > > What do you mean "missing property"? Why isn't it sufficient to mark the > GPIOs as GPIO_ACTIVE_LOW in the various -gpios properties? Unfortunately we do have "gpio-activelow" property already used in the driver since 2012 when DT support was added to the driver. This patch merely acknowledges that it is there. If DT maintainers wish to adjust known DTSes and switch to proper polarity annotation through gpio property I am all for it. > > > > > Signed-off-by: Markus Burri <markus.burri@mt.com> > > --- > > .../devicetree/bindings/input/gpio-matrix-keypad.yaml | 5 +++++ > > 1 file changed, 5 insertions(+) > > > > diff --git a/Documentation/devicetree/bindings/input/gpio-matrix-keypad.yaml b/Documentation/devicetree/bindings/input/gpio-matrix-keypad.yaml > > index 745652b..9ea66b3 100644 > > --- a/Documentation/devicetree/bindings/input/gpio-matrix-keypad.yaml > > +++ b/Documentation/devicetree/bindings/input/gpio-matrix-keypad.yaml > > @@ -51,6 +51,11 @@ properties: > > (Legacy property supported: "linux,wakeup") > > default: false > > > > + gpio-activelow: > > + type: boolean > > + description: The GPIOs are low active. > > + default: false > > What you want is a flag, not a boolean here btw. Flags you can check for > the presence of, booleans you cannot. The behavior is fixed. If the flag is true GPIO is assumed to be active low, otherwise (and in the absence of the flag) GPIO is assumed to be active high. > > > + > > debounce-delay-ms: > > description: Debounce interval in milliseconds. > > default: 0 > > -- > > 2.39.5 > > Thanks.
On Tue, Nov 05, 2024 at 02:03:22PM +0100, Markus Burri wrote: > For a setup where the matrix keypad is connected over a slow interface > (e.g. a gpio-expansion over i2c), the scan can take a longer time to read. > > Interrupts need to be disabled during scan. And therefore changes in this > period are not detected. > To improve this situation, scan the matrix again if the row state changed > during interrupts disabled. And what if the state changes yet again during the 2nd scan attempt? > > Signed-off-by: Markus Burri <markus.burri@mt.com> > --- > drivers/input/keyboard/matrix_keypad.c | 17 +++++++++++++++++ > 1 file changed, 17 insertions(+) > > diff --git a/drivers/input/keyboard/matrix_keypad.c b/drivers/input/keyboard/matrix_keypad.c > index 46873b8..93c8404 100644 > --- a/drivers/input/keyboard/matrix_keypad.c > +++ b/drivers/input/keyboard/matrix_keypad.c > @@ -126,6 +126,10 @@ static void matrix_keypad_scan(struct work_struct *work) > const unsigned short *keycodes = input_dev->keycode; > uint32_t new_state[MATRIX_MAX_COLS]; > int row, col, code; > + u32 init_row_state, new_row_state; > + > + /* read initial row state to detect changes between scan */ > + init_row_state = read_row_state(keypad); > > /* de-activate all columns for scanning */ > activate_all_cols(keypad, false); > @@ -174,6 +178,19 @@ static void matrix_keypad_scan(struct work_struct *work) > keypad->scan_pending = false; > enable_row_irqs(keypad); > spin_unlock_irq(&keypad->lock); > + > + /* read new row state and detect if value has changed */ > + new_row_state = read_row_state(keypad); > + if (init_row_state != new_row_state) { > + spin_lock_irq(&keypad->lock); > + if (likely(!keypad->scan_pending && !keypad->stopped)) { > + disable_row_irqs(keypad); > + keypad->scan_pending = true; > + schedule_delayed_work(&keypad->work, > + msecs_to_jiffies(keypad->debounce_ms)); > + } > + spin_unlock_irq(&keypad->lock); > + } > } > > static irqreturn_t matrix_keypad_interrupt(int irq, void *id) > -- > 2.39.5 >
On Tue, Nov 05, 2024 at 11:14:32AM -0800, Dmitry Torokhov wrote: > Hi Conor, > > On Tue, Nov 05, 2024 at 06:22:36PM +0000, Conor Dooley wrote: > > On Tue, Nov 05, 2024 at 02:03:19PM +0100, Markus Burri wrote: > > > Add missing property 'gpio-activelow' to DT schema. > > > > What do you mean "missing property"? Why isn't it sufficient to mark the > > GPIOs as GPIO_ACTIVE_LOW in the various -gpios properties? > > Unfortunately we do have "gpio-activelow" property already used in the > driver since 2012 when DT support was added to the driver. This patch > merely acknowledges that it is there. > > If DT maintainers wish to adjust known DTSes and switch to proper > polarity annotation through gpio property I am all for it. If that's the rationale, it should be in the commit message. It's not my job to figure out people's intentions :) > > > Signed-off-by: Markus Burri <markus.burri@mt.com> > > > --- > > > .../devicetree/bindings/input/gpio-matrix-keypad.yaml | 5 +++++ > > > 1 file changed, 5 insertions(+) > > > > > > diff --git a/Documentation/devicetree/bindings/input/gpio-matrix-keypad.yaml b/Documentation/devicetree/bindings/input/gpio-matrix-keypad.yaml > > > index 745652b..9ea66b3 100644 > > > --- a/Documentation/devicetree/bindings/input/gpio-matrix-keypad.yaml > > > +++ b/Documentation/devicetree/bindings/input/gpio-matrix-keypad.yaml > > > @@ -51,6 +51,11 @@ properties: > > > (Legacy property supported: "linux,wakeup") > > > default: false > > > > > > + gpio-activelow: > > > + type: boolean > > > + description: The GPIOs are low active. > > > + default: false > > > > What you want is a flag, not a boolean here btw. Flags you can check for > > the presence of, booleans you cannot. > > The behavior is fixed. What do you mean by "fixed"? Corrected, unchangeable or something else? > If the flag is true GPIO is assumed to be active > low, otherwise (and in the absence of the flag) GPIO is assumed to be > active high. Flags cannot be false only true or absent, what you have described here is a boolean. > > > > > > + > > > debounce-delay-ms: > > > description: Debounce interval in milliseconds. > > > default: 0 > > > -- > > > 2.39.5 > > > > > Thanks. > > -- > Dmitry
This series is needed to avoid key-loss if the GPIOs are connected via a I2C GPIO Mux that introduces additonal latencies between key change and matrix scan. Patch 1, 2 and 6 in this series introduce a performance boost on its own and can be merged separately. Patch 3, 4 and 5 depend on each other. Markus Burri (6): Input: matrix_keypad - move gpio-row init to the init part Input: matrix_keypad - use fsleep for variable delay duration Input: matrix_keypad - add function for reading row state dt-bindings: input: matrix_keypad - add settle time after enable all columns Input: matrix_keypad - add settle time after enable all columns Input: matrix_keypad - detect change during scan .../bindings/input/gpio-matrix-keypad.txt | 2 + drivers/input/keyboard/matrix_keypad.c | 44 ++++++++++++++++--- 2 files changed, 39 insertions(+), 7 deletions(-)