Thread (42 messages) 42 messages, 6 authors, 2021-01-11

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

From: Barnabás Pőcze <hidden>
Date: 2021-01-08 11:54:24

Hi


2021. január 8., péntek 8:12 keltezéssel, Roderick Colenbrander írta:
[...]
quoted
quoted
+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.

quoted
+     /* 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
quoted
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.
quoted
quoted
+     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
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help