Re: [PATCH v2 00/24] platform/x86: thinkpad_acpi: Refactor hotkey handling and add support for some new hotkeys

[Hans de Goede <hdegoede@redhat.com>]

Hi All,

On 4/24/24 2:28 PM, Hans de Goede wrote:
> Hi All,
> 
> Here is v2 of my patch-series to refactor thinkpad_acpi's hotkey handling
> and to add support for some new hotkeys on new models.
> 
> Changes in v2:
> - Some small code style tweaks in response to reviews
> - Add various Reviewed-by and Tested-by tags
> 
> Relevant parts of v1 cover-letter:
> 
> I have been very careful to not introduce any changes wrt support for
> the original ThinkPad models / hotkeys which use the hotkey_*_mask
> related code.
> 
> I've also done my best to avoid any *significant* functional change but
> there are still some functional changes, which should all not impact
> userspace compatibility:
> 
> 1. Adaptive keyboard special keys will now also send EV_MSC events with
>    the scancode, just like all the other hotkeys.
> 
> 2. Rely on the input core for KEY_RESERVED suppression. This means that
>    keys marked as KEY_RESERVED will still send EV_MSC evdev events when
>    pressed (which helps users with mapping them to non reserved KEY_FOO
>    values if desired).
> 
> 3. Align the keycodes for volume up/down/mute and brightness up/down with
>    those set by userspace through udev upstream's hwdb. Note these are all
>    for keys which are suppressed by hotkey_reserved_mask by default.
>    So this is only a functional change for users who override the default
>    hotkey-mask *and* who do not have udev's default hwdb installed.
> 
> 4. Suppress ACPI netlink event generation for unknown 0x1xxx hkey events to
>    avoid userspace starting to rely on the netlink events for new hotkeys
>    before these have been added to the keymap, only to have the netlink
>    events get disabled by the adding of the new hotkeys to the keymap.
> 
>    This should not cause a behavior change for existing models since all
>    currently known 0x1xxx events have a mapping.
> 
> Here is a quick breakdown of the patches in this series:
> 
> Patch 1 - 2: Fix a small locking issue on rmmod the only problem here
>    really is a lockdep warning, so I plan to route these fixes through
>    for-next together with the rest to keep things simple.
> 
> Patch 3 - 14: Do a bunch of cleanups and refactoring
> 
> Patch 15: Implements functional change no 4. I really kinda want to just
>    completely disable ACPI netlink event generation when also sending evdev
>    events instead of reporting these twice. But for the 0x11xx / 0x13xx
>    hkey events the kernel has send ACPI netlink events for years now. So
>    this disables ACPI netlink events for any new hotkeys going forward.
> 
> Patch 16 - 18: Refactor / cleanup reserved key handling
> 
> Patch 19: Actually move to sparse-keymaps
> 
> Patch 20: Update the keymap for 2 adaptive kbd Fn row keys
> 
> Patch 21 - 24: Mark's original patches adding support for the new Fn + N
>    key combo and for trackpoint doubletap slightly reworked to use
>    the new sparse-keymap.

I've added this series to my review-hans (soon to be for-next) branch now.

Regards,

Hans


> Hans de Goede (20):
>   platform/x86: thinkpad_acpi: Take hotkey_mutex during hotkey_exit()
>   platform/x86: thinkpad_acpi: Provide hotkey_poll_stop_sync() dummy
>   platform/x86: thinkpad_acpi: Drop setting send_/ignore_acpi_ev
>     defaults twice
>   platform/x86: thinkpad_acpi: Drop ignore_acpi_ev
>   platform/x86: thinkpad_acpi: Use tpacpi_input_send_key() in adaptive
>     kbd code
>   platform/x86: thinkpad_acpi: Do hkey to scancode translation later
>   platform/x86: thinkpad_acpi: Make tpacpi_driver_event() return if it
>     handled the event
>   platform/x86: thinkpad_acpi: Move adaptive kbd event handling to
>     tpacpi_driver_event()
>   platform/x86: thinkpad_acpi: Move special original hotkeys handling
>     out of switch-case
>   platform/x86: thinkpad_acpi: Move hotkey_user_mask check to
>     tpacpi_input_send_key()
>   platform/x86: thinkpad_acpi: Always call tpacpi_driver_event() for
>     hotkeys
>   platform/x86: thinkpad_acpi: Drop tpacpi_input_send_key_masked() and
>     hotkey_driver_event()
>   platform/x86: thinkpad_acpi: Move hkey > scancode mapping to
>     tpacpi_input_send_key()
>   platform/x86: thinkpad_acpi: Move tpacpi_driver_event() call to
>     tpacpi_input_send_key()
>   platform/x86: thinkpad_acpi: Do not send ACPI netlink events for
>     unknown hotkeys
>   platform/x86: thinkpad_acpi: Change hotkey_reserved_mask
>     initialization
>   platform/x86: thinkpad_acpi: Use correct keycodes for volume and
>     brightness keys
>   platform/x86: thinkpad_acpi: Drop KEY_RESERVED special handling
>   platform/x86: thinkpad_acpi: Switch to using sparse-keymap helpers
>   platform/x86: thinkpad_acpi: Add mappings for adaptive kbd
>     clipping-tool and cloud keys
> 
> Mark Pearson (4):
>   platform/x86: thinkpad_acpi: Simplify known_ev handling
>   platform/x86: thinkpad_acpi: Support for trackpoint doubletap
>   platform/x86: thinkpad_acpi: Support for system debug info hotkey
>   platform/x86: thinkpad_acpi: Support hotkey to disable trackpoint
>     doubletap
> 
>  drivers/platform/x86/thinkpad_acpi.c | 854 +++++++++++----------------
>  1 file changed, 353 insertions(+), 501 deletions(-)
>