The da850-evm baseboard (BB) and its UI board both have tca6416 IO expanders. They are bootstrapped to different I2C addresses so they can be used concurrently. The expander on the UI board is currently used to enable/disable the peripherals that are available on the UI board. In addition to this functionality the expander is also connected to 8 pushbuttons. The expander on the baseboard is not currently used; it is connected to deep sleep enable, sw reset, a push button, some switches and LEDs. This proposed patch series enables the push buttons and switches on the UI and BB expanders using the gpio-keys polling mode patch by Alexander Clouter. Some work was performed to test irq-based gpio-keys support on the expanders (a WIP patch can be posted on request) but I believe that it is not possible to use irq-based gpio-keys on IO expanders for arm systems at this time. The attempt started when I noticed the patch of Alek Du and Alan Cox [1] which was recently committed [2]; a stab at integrating irq-based gpio-keys support based on that patch was attempted. I found that I either got a warning that the irq could not be mapped for the given gpio ; or, when N_IRQ was increased, a system freeze. >From what I have read (particularly the message by Grant Likely [3]) IRQs on IO expanders are not ready in ARM yet. I _think_ that the sparse IRQ rework by Thomas Gleixner [4] will resolve the blocker to irq-based gpio-keys support. In the meantime we have buttons and switches that we would like to excersise in our prototyping development. The patch to convert this series to irq-based gpio-keys will be straighforward once the support in arch/arm is there. There is an existing tca6416-keypad driver with polling support which I did not employ because it isn't possible to keep the gpio's used for peripheral enable/disable on the UI board or the LEDs on the baseboard registered while simultaneously registering the pushbuttons or switches as a tca6416-keypad instance. I tested this patch series using evtest on the resulting /dev/input/eventN devices and also on the event node of a non-polling gpio-keys instance to ensure that irq-based input handling is not broken by the introduction of the polling-mode gpio-keys patch. The non-polling instance creation and registration is not included in this series since it uses one of the boot-mode DIP switches and woult not (I think) be suitable for mainline. Disclaimer: I'm not an expert in irq's or gpio-keys; this is, in fact, my first proposed feature. Please feel free to correct me -- I welcome the chance to learn from your expertise. [1] http://article.gmane.org/gmane.linux.kernel/1052551 [2] http://article.gmane.org/gmane.linux.kernel.commits.head/260919 [3] http://www.mail-archive.com/devicetree-discuss@xxxxxxxxxxxxxxxx/msg01974.html [4] http://article.gmane.org/gmane.linux.kernel.cross-arch/7786 Alexander Clouter (1): input: gpio_keys: polling mode support Ben Gardiner (3): da850-evm: add UI Expander pushbuttons da850-evm: extract defines for SEL{A,B,C} pins in UI expander da850-evm: add baseboard GPIO expander buttons, switches and LEDs arch/arm/mach-davinci/Kconfig | 3 + arch/arm/mach-davinci/board-da850-evm.c | 347 +++++++++++++++++++++++++++++-- drivers/input/keyboard/gpio_keys.c | 120 +++++++++-- include/linux/gpio_keys.h | 1 + 4 files changed, 440 insertions(+), 31 deletions(-) --- Changes since v2: * register a single input device for switches and keys on the baseboard since there is no benefit to separate devices with different polling intervals (Dmitry Torokhov) * use static array intialization and range intialization for platform data structure to minimize the amount of runtime intialization needed: (Sekhar Nori) * Use the da850_evm variable name prefix for static symbols in board-da850-evm.c Changes since v1: * use locally defined functions that are no-ops/error checkers when INPUT_POLLDEV is not defined. * disable polling mode support when input-polldev is a module and gpio_keys is builtin * set INPUT_POLLDEV default for DA850_EVM machine, but don't select it unconditionally * adding note to description about why tca6416-keypad was not used -- 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
