TODO RC: Difference between revisions
Jump to navigation
Jump to search
No edit summary |
(Changes merged for v4.16) |
||
| Line 2: | Line 2: | ||
* struct ir_raw_event_ctrl member prev_ev. Investigate if this can be removed. |
* struct ir_raw_event_ctrl member prev_ev. Investigate if this can be removed. |
||
* if driver depends on RC_CORE, change this to select RC_CORE, depends on INPUT? OR use a patch like https://patchwork.linuxtv.org/patch/40601/ and remove "depends on RC_CORE". |
* if driver depends on RC_CORE, change this to select RC_CORE, depends on INPUT? OR use a patch like https://patchwork.linuxtv.org/patch/40601/ and remove "depends on RC_CORE". |
||
* RC device RC_DRIVER_IR_RAW_TX leaks dev->raw on rc_unregister_device()? |
|||
* Unplug iguanair with lirc chardev open causes kernel to go bang |
|||
* CEC repeat code needs to move into rc-core |
|||
* mceusb can do wakeups. Allow this to be configured |
* mceusb can do wakeups. Allow this to be configured |
||
* Can sir_ir and/or serial_ir use serdev |
* Can sir_ir and/or serial_ir use serdev |
||
| Line 15: | Line 12: | ||
** Further improvement: make the kernel mapping rc_proto:scancode -> keycode, not just scancode. This should reduce the number of possible conflicts. |
** Further improvement: make the kernel mapping rc_proto:scancode -> keycode, not just scancode. This should reduce the number of possible conflicts. |
||
* Compare lirc rc6 decoder. Can it handle multiple keypresses? What repeat/delay times does it use? See https://bugzilla.kernel.org/show_bug.cgi?id=48761 |
* Compare lirc rc6 decoder. Can it handle multiple keypresses? What repeat/delay times does it use? See https://bugzilla.kernel.org/show_bug.cgi?id=48761 |
||
* lirc_zilog rc core driver (Sean Young working on this) |
|||
* Once lirc_zilog is removed, remove lirc kapi. Example is here: https://git.linuxtv.org/syoung/media_tree.git/log/?h=lirc_scancodes |
|||
* Implement scancode reading and transmitting: |
|||
** It's very hard to tell what protocol a remote is using |
|||
** Kernel already has encoders, make sure they can be used for tx |
|||
** CEC might want to reuse this interface |
|||
** Patch series already exist for this |
|||
* MS MCE Keyboard keymap needs multiple protocols. Keymap needed to select it. https://www.mail-archive.com/linux-media@vger.kernel.org/msg115706.html |
* MS MCE Keyboard keymap needs multiple protocols. Keymap needed to select it. https://www.mail-archive.com/linux-media@vger.kernel.org/msg115706.html |
||
* ir-ctl should list available devices if run without any argument and allow to select device based on rc (based on /sys/class/rc/). |
* ir-ctl should list available devices if run without any argument and allow to select device based on rc (based on /sys/class/rc/). |
||
Revision as of 15:25, 14 December 2017
Minor bugs/improvement
- struct ir_raw_event_ctrl member prev_ev. Investigate if this can be removed.
- if driver depends on RC_CORE, change this to select RC_CORE, depends on INPUT? OR use a patch like https://patchwork.linuxtv.org/patch/40601/ and remove "depends on RC_CORE".
- mceusb can do wakeups. Allow this to be configured
- Can sir_ir and/or serial_ir use serdev
- iguanair_tx() can be simplified
- NEC32 mess (again)
- Rename lirc record -> lirc receive (e.g. ir-ctl).
Larger work
- Make it possible to load multiple keymaps with ir-keytable. It should detect if the new keymap conflicts with any already loaded keymap, and ensure the right protocols are selected.
- Further improvement: make the kernel mapping rc_proto:scancode -> keycode, not just scancode. This should reduce the number of possible conflicts.
- Compare lirc rc6 decoder. Can it handle multiple keypresses? What repeat/delay times does it use? See https://bugzilla.kernel.org/show_bug.cgi?id=48761
- MS MCE Keyboard keymap needs multiple protocols. Keymap needed to select it. https://www.mail-archive.com/linux-media@vger.kernel.org/msg115706.html
- ir-ctl should list available devices if run without any argument and allow to select device based on rc (based on /sys/class/rc/).
- Rework ir raw decoders; register should only happen when protocol enabled (so we don't get an MCE input device). Also if decoders not enabled, #ifdef a load of stuff out.