Re: [PATCH v2 01/13] HID: playstation: initial DualSense USB support.

["Barnabás Pőcze" <pobrn@protonmail.com>]

Hi


2021. január 8., péntek 8:12 keltezéssel, Roderick Colenbrander írta:

> [...]
> > > +static int dualsense_parse_report(struct ps_device *ps_dev, struct hid_report *report,
> > > +             u8 *data, int size)
> > > +{
> > > +     struct hid_device *hdev = ps_dev->hdev;
> > > +     struct dualsense *ds = container_of(ps_dev, struct dualsense, base);
> > > +     struct dualsense_input_report *ds_report;
> > > +     uint8_t value;
> > > +
> >
> > I think `size` should be checked somewhere around here.
> >
> >
> > > +     /* DualSense in USB uses the full HID report for reportID 1, but
> > > +      * Bluetooth uses a minimal HID report for reportID 1 and reports
> > > +      * the full report using reportID 49.
> > > +      */
> > > +     if (report->id == DS_INPUT_REPORT_USB && hdev->bus == BUS_USB) {
> > > +             ds_report = (struct dualsense_input_report *)&data[1];
> > > +     } else {
> > > +             hid_err(hdev, "Unhandled reportID=%d\n", report->id);
> > > +             return -1;
> > > +     }
> > > +
> > > +     input_report_abs(ds->gamepad, ABS_X,  ds_report->x);
> > > +     input_report_abs(ds->gamepad, ABS_Y,  ds_report->y);
> > > +     input_report_abs(ds->gamepad, ABS_RX, ds_report->rx);
> > > +     input_report_abs(ds->gamepad, ABS_RY, ds_report->ry);
> > > +     input_report_abs(ds->gamepad, ABS_Z,  ds_report->z);
> > > +     input_report_abs(ds->gamepad, ABS_RZ, ds_report->rz);
> > > +
> > > +     value = ds_report->buttons[0] & DS_BUTTONS0_HAT_SWITCH;
> > > +     if (value > 7)
> > > +             value = 8; /* center */
> >
> > This seems a bit flimsy to me, it relies on a different part of the code
> > being in a certain way that is not enforced by anything
>
> What do you mean with not enforced? I'm not saying I'm a big fan of
> the code, but HATs seem to work like this. The DualShock4/DualSense
> describe it in their HID descriptors with a logical minimum value of 0
> and a max value of 7.

What I really meant is that I think it would be better to somehow decrease
the chances of unintentionally breaking the code by providing stronger
guarantees that `ps_gamepad_hat_mapping[value]` will not be an out of bounds
access. E.g. checking with static_assert that it has 9 elements, using
ARRAY_SIZE() in some way in the condition, etc. I'm not trying to say it's
very probable, but I think it's easily preventable and thus it should be done.

By "not enforced", I meant that there are no compile time or even runtime guarantees
that the `ps_gamepad_hat_mapping` contains as many elements as expected by this
part of the code.


>
> The code is very similar to hid-input.c:
>
> static const struct {
> __s32 x;
> __s32 y;
> }  hid_hat_to_axis[] = {{ 0, 0}, { 0,-1}, { 1,-1}, { 1, 0}, { 1, 1}, {
> 0, 1}, {-1, 1}, {-1, 0}, {-1,-1}};
>
> int hat_dir = usage->hat_dir;
> if (!hat_dir)
>     hat_dir = (value - usage->hat_min) * 8 / (usage->hat_max -
> usage->hat_min + 1) + 1;
> if (hat_dir < 0 || hat_dir > 8) hat_dir = 0;
> input_event(input, usage->type, usage->code    , hid_hat_to_axis[hat_dir].x);
> input_event(input, usage->type, usage->code + 1, hid_hat_to_axis[hat_dir].y);
>
> Main difference seems to be that this code places {0, 0} at the start
> and adds a "+1" to avoid having to set the value to "8" when out of
> range.
>
>  I'd probably do something
> > like this:
> >
> > enum {
> >   HAT_DIR_W = 0,
> >   HAT_DIR_NW,
> >   ...
> >   HAT_DIR_SW,
> >   HAT_DIR_NONE,
> > };
> >
> > static const struct {int x; int y; } ps_gamepad_hat_mapping[] = {
> >   [HAT_DIR_W] = {0, -1},
> >   ...
> >   [HAT_DIR_NONE] = {0, 0},
> > };
> >
> > and then
> >
> > if (value >= ARRAY_SIZE(ps_gamepad_hat_mapping))
> >   value = HAT_DIR_NONE;
> >
> > Please consider it. By the way, are values 9..15 actually sent by the controller?
>
> See above. They are not sent. The Hat Switch in the report descriptor
> is reported with a logical minimum of 0 and a max of 8.
>
> >
> > > +     input_report_abs(ds->gamepad, ABS_HAT0X, ps_gamepad_hat_mapping[value].x);
> > > +     input_report_abs(ds->gamepad, ABS_HAT0Y, ps_gamepad_hat_mapping[value].y);
> > > +
> > > +     input_report_key(ds->gamepad, BTN_WEST,   ds_report->buttons[0] & DS_BUTTONS0_SQUARE);
> > > +     input_report_key(ds->gamepad, BTN_SOUTH,  ds_report->buttons[0] & DS_BUTTONS0_CROSS);
> > > +     input_report_key(ds->gamepad, BTN_EAST,   ds_report->buttons[0] & DS_BUTTONS0_CIRCLE);
> > > +     input_report_key(ds->gamepad, BTN_NORTH,  ds_report->buttons[0] & DS_BUTTONS0_TRIANGLE);
> > > +     input_report_key(ds->gamepad, BTN_TL,     ds_report->buttons[1] & DS_BUTTONS1_L1);
> > > +     input_report_key(ds->gamepad, BTN_TR,     ds_report->buttons[1] & DS_BUTTONS1_R1);
> > > +     input_report_key(ds->gamepad, BTN_TL2,    ds_report->buttons[1] & DS_BUTTONS1_L2);
> > > +     input_report_key(ds->gamepad, BTN_TR2,    ds_report->buttons[1] & DS_BUTTONS1_R2);
> > > +     input_report_key(ds->gamepad, BTN_SELECT, ds_report->buttons[1] & DS_BUTTONS1_CREATE);
> > > +     input_report_key(ds->gamepad, BTN_START,  ds_report->buttons[1] & DS_BUTTONS1_OPTIONS);
> > > +     input_report_key(ds->gamepad, BTN_THUMBL, ds_report->buttons[1] & DS_BUTTONS1_L3);
> > > +     input_report_key(ds->gamepad, BTN_THUMBR, ds_report->buttons[1] & DS_BUTTONS1_R3);
> > > +     input_report_key(ds->gamepad, BTN_MODE,   ds_report->buttons[2] & DS_BUTTONS2_PS_HOME);
> > > +     input_sync(ds->gamepad);
> > > +
> > > +     return 0;
> > > +}
> [...]


Regards,
Barnabás Pőcze