On Monday 20 April 2009, David Brownell wrote: > On Saturday 17 January 2009, David Brownell wrote: > > From: David Brownell <dbrownell@xxxxxxxxxxxxxxxxxxxxx> > > Subject: input: dm355evm_keys driver > > > > PING?? I was told this was in the input queue, but it does > not seem to be in > > http://git.kernel.org/?p=linux/kernel/git/dtor/input.git > > Current version of the patch is appended. > > - Dave > Hmm, still no response. What's up, input team?? Appended is a minor tweak to this code. I'm going to resend the patch again. Maybe one day, soon, I can take it out of my list of patches-due-to-hit-mainline ... - Dave =============== CUT HERE From: David Brownell <dbrownell@xxxxxxxxxxxxxxxxxxxxx> Convert the dm355evm keys driver to use IRQ threading instead of a private workqueue ... that mechanism was added to Linux after the driver was first written. Signed-off-by: David Brownell <dbrownell@xxxxxxxxxxxxxxxxxxxxx> --- drivers/input/keyboard/dm355evm_keys.c | 38 ++++++++++++++----------------- 1 file changed, 18 insertions(+), 20 deletions(-) --- a/drivers/input/keyboard/dm355evm_keys.c +++ b/drivers/input/keyboard/dm355evm_keys.c @@ -23,30 +23,16 @@ * pressed, or its autorepeat kicks in, an event is sent. This driver * read those events from the small (32 event) queue and reports them. * - * Because we communicate with the MSP430 using I2C, and all I2C calls - * in Linux sleep, we need to cons up a kind of threaded IRQ handler - * using a work_struct. The IRQ is active low, but we use it through - * the GPIO controller so we can trigger on falling edges. - * * Note that physically there can only be one of these devices. * * This driver was tested with firmware revision A4. */ struct dm355evm_keys { - struct work_struct work; struct input_dev *input; struct device *dev; int irq; }; -static irqreturn_t dm355evm_keys_irq(int irq, void *_keys) -{ - struct dm355evm_keys *keys = _keys; - - schedule_work(&keys->work); - return IRQ_HANDLED; -} - /* These initial keycodes can be remapped by dm355evm_setkeycode(). */ static struct { u16 event; @@ -110,13 +96,12 @@ static struct { { 0x3169, KEY_PAUSE, }, }; -static void dm355evm_keys_work(struct work_struct *work) +/* runs in an IRQ thread -- can (and will!) sleep */ +static irqreturn_t dm355evm_keys_irq(int irq, void *_keys) { - struct dm355evm_keys *keys; + struct dm355evm_keys *keys = _keys; int status; - keys = container_of(work, struct dm355evm_keys, work); - /* For simplicity we ignore INPUT_COUNT and just read * events until we get the "queue empty" indicator. * Reading INPUT_LOW decrements the count. @@ -184,6 +169,19 @@ static void dm355evm_keys_work(struct wo input_report_key(keys->input, keycode, 0); input_sync(keys->input); } + return IRQ_HANDLED; +} + +/* + * Because we communicate with the MSP430 using I2C, and all I2C calls + * in Linux sleep, we use a threaded IRQ handler. The IRQ itself is + * active low, but we go through the GPIO controller so we can trigger + * on falling edges and not worry about enabling/disabling the IRQ in + * the keypress handling path. + */ +static irqreturn_t dm355evm_keys_hardirq(int irq, void *_keys) +{ + return IRQ_WAKE_THREAD; } static int dm355evm_setkeycode(struct input_dev *dev, int index, int keycode) @@ -259,7 +257,6 @@ static int __devinit dm355evm_keys_probe if (status < 0) goto fail1; keys->irq = status; - INIT_WORK(&keys->work, dm355evm_keys_work); /* REVISIT: flush the event queue? */ @@ -269,7 +266,8 @@ static int __devinit dm355evm_keys_probe goto fail0; /* start reporting events */ - status = request_irq(keys->irq, dm355evm_keys_irq, + status = request_threaded_irq(keys->irq, + dm355evm_keys_hardirq, dm355evm_keys_irq, IRQF_TRIGGER_FALLING, dev_name(&pdev->dev), keys); if (status < 0) { -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
