Bug#757356: Apple keyboard: Scan code event (EV_MSC) not generated when the EV_KEY event is generated by hid-apple.c

[Salvatore Bonaccorso <carnil@debian.org>]

Hi Vincent,

On Wed, May 26, 2021 at 12:02:12PM +0200, Vincent Lefevre wrote:
> Control: retitle -1 Apple keyboard: Scan code event (EV_MSC) not generated when the EV_KEY event is generated by hid-apple.c
> Control: tags -1 patch
> 
> On 2021-05-26 10:39:11 +0200, Vincent Lefevre wrote:
> > And the cursor keys. Actually, all the keys corresponding to
> > 
> > static const struct applespi_key_translation applespi_fn_codes[] = {
> >         { KEY_BACKSPACE, KEY_DELETE },
> >         { KEY_ENTER,    KEY_INSERT },
> >         { KEY_F1,       KEY_BRIGHTNESSDOWN,     APPLE_FLAG_FKEY },
> >         { KEY_F2,       KEY_BRIGHTNESSUP,       APPLE_FLAG_FKEY },
> >         { KEY_F3,       KEY_SCALE,              APPLE_FLAG_FKEY },
> >         { KEY_F4,       KEY_DASHBOARD,          APPLE_FLAG_FKEY },
> >         { KEY_F5,       KEY_KBDILLUMDOWN,       APPLE_FLAG_FKEY },
> >         { KEY_F6,       KEY_KBDILLUMUP,         APPLE_FLAG_FKEY },
> >         { KEY_F7,       KEY_PREVIOUSSONG,       APPLE_FLAG_FKEY },
> >         { KEY_F8,       KEY_PLAYPAUSE,          APPLE_FLAG_FKEY },
> >         { KEY_F9,       KEY_NEXTSONG,           APPLE_FLAG_FKEY },
> >         { KEY_F10,      KEY_MUTE,               APPLE_FLAG_FKEY },
> >         { KEY_F11,      KEY_VOLUMEDOWN,         APPLE_FLAG_FKEY },
> >         { KEY_F12,      KEY_VOLUMEUP,           APPLE_FLAG_FKEY },
> >         { KEY_RIGHT,    KEY_END },
> >         { KEY_LEFT,     KEY_HOME },
> >         { KEY_DOWN,     KEY_PAGEDOWN },
> >         { KEY_UP,       KEY_PAGEUP },
> >         { }
> > };
> > 
> > in drivers/input/keyboard/applespi.c.
> > 
> > Just in case, in /etc/modprobe.d/hid_apple.conf, I have
> > 
> > options hid_apple fnmode=2
> > options hid_apple iso_layout=0
> 
> But since I'm using hid_apple, I should have taken
> drivers/hid/hid-apple.c, which has the same kind of code:
> 
> static const struct apple_key_translation apple_fn_keys[] = {
>         { KEY_BACKSPACE, KEY_DELETE },
>         { KEY_ENTER,    KEY_INSERT },
>         { KEY_F1,       KEY_BRIGHTNESSDOWN, APPLE_FLAG_FKEY },
>         { KEY_F2,       KEY_BRIGHTNESSUP,   APPLE_FLAG_FKEY },
>         { KEY_F3,       KEY_SCALE,          APPLE_FLAG_FKEY },
>         { KEY_F4,       KEY_DASHBOARD,      APPLE_FLAG_FKEY },
>         { KEY_F5,       KEY_KBDILLUMDOWN,   APPLE_FLAG_FKEY },
>         { KEY_F6,       KEY_KBDILLUMUP,     APPLE_FLAG_FKEY },
>         { KEY_F7,       KEY_PREVIOUSSONG,   APPLE_FLAG_FKEY },
>         { KEY_F8,       KEY_PLAYPAUSE,      APPLE_FLAG_FKEY },
>         { KEY_F9,       KEY_NEXTSONG,       APPLE_FLAG_FKEY },
>         { KEY_F10,      KEY_MUTE,           APPLE_FLAG_FKEY },
>         { KEY_F11,      KEY_VOLUMEDOWN,     APPLE_FLAG_FKEY },
>         { KEY_F12,      KEY_VOLUMEUP,       APPLE_FLAG_FKEY },
>         { KEY_UP,       KEY_PAGEUP },
>         { KEY_DOWN,     KEY_PAGEDOWN },
>         { KEY_LEFT,     KEY_HOME },
>         { KEY_RIGHT,    KEY_END },
>         { }
> };
> 
> In the conditions from hidinput_apple_event(), the only ones that
> should match according to my settings are
> 
>         if (usage->code == fn_keycode) {
> 
> and
> 
>         if (fnmode) {
> 
> and these are the keys (when trans is true, for fnmode) for which I do
> not get a scan code event. Said otherwise, if hidinput_apple_event()
> returns 1, I do not get a scan code event. There are input_event()
> calls, but I suppose that they will just generate an EV_KEY event,
> and EV_MSC is the one that is missing.
> 
> Note: in hid-apple.c, apple_event() calls hidinput_apple_event(), and
> one has
> 
> static struct hid_driver apple_driver = {
> [...]
>         .event = apple_event,
> [...]
> };
> module_hid_driver(apple_driver);
> 
> I forgot that there was
> 
>   https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=757356#35
> 
> from Daniel Lin, with a patch, in 2017. I've looked at this patch
> (but have not tried it), and it adds an additional EV_MSC event
> when hidinput_apple_event() has to generate an EV_KEY event. So
> I confirm that should solve this issue and I'm adding the patch
> tag (I don't know whether the patch needs an update, though).

Can you, time permitting, starting from there (and needed refreshes)
try to confirm if the patch solves the issue on top of 5.10.40? If so,
next step would be to propose the change/report the bug at least, to
upstream, get_maintainers.pl would suggest to report it to:

Jiri Kosina <jikos@kernel.org> (maintainer:HID CORE LAYER)
Benjamin Tissoires <benjamin.tissoires@redhat.com> (maintainer:HID CORE LAYER)
linux-input@vger.kernel.org (open list:HID CORE LAYER)
linux-kernel@vger.kernel.org (open list)

Regards,
Salvatore