Hi Barnabas, Thanks for your suggestions. On Thu, Jan 7, 2021 at 9:14 AM Barnabás Pőcze <pobrn@xxxxxxxxxxxxxx> wrote: > > Hi > > > I have just a couple minor comments. > > > 2021. január 2., szombat 23:30 keltezéssel, Roderick Colenbrander írta: > > > From: Roderick Colenbrander roderick.colenbrander@xxxxxxxx > > > > Implement support for PlayStation DualSense gamepad in USB mode. > > Support features include buttons and sticks, which adhere to the > > Linux gamepad spec. > > > > Signed-off-by: Roderick Colenbrander roderick.colenbrander@xxxxxxxx > > [...] > > +/* Common gamepad buttons across DualShock 3 / 4 and DualSense. > > + * Note: for device with a touchpad, touchpad button is not included > > + * as it will be part of the touchpad device. > > + */ > > +static const int ps_gamepad_buttons[] = { > > + BTN_WEST, /* Square */ > > + BTN_NORTH, /* Triangle */ > > + BTN_EAST, /* Circle */ > > + BTN_SOUTH, /* Cross */ > > + BTN_TL, /* L1 */ > > + BTN_TR, /* R1 */ > > + BTN_TL2, /* L2 */ > > + BTN_TR2, /* R2 */ > > + BTN_SELECT, /* Create (PS5) / Share (PS4) */ > > + BTN_START, /* Option */ > > + BTN_THUMBL, /* L3 */ > > + BTN_THUMBR, /* R3 */ > > + BTN_MODE, /* PS Home */ > > +}; > > + > > +static const struct {int x; int y; } ps_gamepad_hat_mapping[] = { > > + {0, -1}, {1, -1}, {1, 0}, {1, 1}, {0, 1}, {-1, 1}, {-1, 0}, {-1, -1}, > > + {0, 0} > > +}; > > I believe the preferred way is to have a comma after each array/enum/etc. element > unless it is a terminating entry. > > > > + > > +static struct input_dev *ps_allocate_input_dev(struct hid_device *hdev, const char *name_suffix) > > +{ > > + struct input_dev *input_dev; > > + > > + input_dev = devm_input_allocate_device(&hdev->dev); > > + if (!input_dev) > > + return ERR_PTR(-ENOMEM); > > + > > + input_dev->id.bustype = hdev->bus; > > + input_dev->id.vendor = hdev->vendor; > > + input_dev->id.product = hdev->product; > > + input_dev->id.version = hdev->version; > > + input_dev->uniq = hdev->uniq; > > + > > + if (name_suffix) { > > + input_dev->name = devm_kasprintf(&hdev->dev, GFP_KERNEL, "%s %s", hdev->name, > > + name_suffix); > > + if (!input_dev->name) > > + return ERR_PTR(-ENOMEM); > > + } else { > > + input_dev->name = hdev->name; > > + } > > + > > + input_set_drvdata(input_dev, hdev); > > + > > + return input_dev; > > +} > > + > > +static struct input_dev *ps_gamepad_create(struct hid_device *hdev) > > +{ > > + struct input_dev *gamepad; > > + unsigned int i; > > + int ret; > > + > > + gamepad = ps_allocate_input_dev(hdev, NULL); > > + if (IS_ERR(gamepad)) > > + return ERR_PTR(-ENOMEM); > > I know that at the moment ENOMEM is the only possible error, but I believe > `return ERR_CAST(gamepad);` would be better. (Or even just `return gamepad;`.) > > > > + > > + input_set_abs_params(gamepad, ABS_X, 0, 255, 0, 0); > > + input_set_abs_params(gamepad, ABS_Y, 0, 255, 0, 0); > > + input_set_abs_params(gamepad, ABS_Z, 0, 255, 0, 0); > > + input_set_abs_params(gamepad, ABS_RX, 0, 255, 0, 0); > > + input_set_abs_params(gamepad, ABS_RY, 0, 255, 0, 0); > > + input_set_abs_params(gamepad, ABS_RZ, 0, 255, 0, 0); > > + > > + input_set_abs_params(gamepad, ABS_HAT0X, -1, 1, 0, 0); > > + input_set_abs_params(gamepad, ABS_HAT0Y, -1, 1, 0, 0); > > + > > + for (i = 0; i < ARRAY_SIZE(ps_gamepad_buttons); i++) > > + input_set_capability(gamepad, EV_KEY, ps_gamepad_buttons[i]); > > + > > + ret = input_register_device(gamepad); > > + if (ret) > > + return ERR_PTR(ret); > > + > > + return gamepad; > > +} > > + > > +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. 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; > > +} > > + > > +static struct ps_device *dualsense_create(struct hid_device *hdev) > > +{ > > + struct dualsense *ds; > > + int ret; > > + > > + ds = devm_kzalloc(&hdev->dev, sizeof(*ds), GFP_KERNEL); > > + if (!ds) > > + return ERR_PTR(-ENOMEM); > > + > > + /* Patch version to allow userspace to distinguish between > > + * hid-generic vs hid-playstation axis and button mapping. > > + */ > > + hdev->version |= HID_PLAYSTATION_VERSION_PATCH; > > + > > + ds->base.hdev = hdev; > > + ds->base.parse_report = dualsense_parse_report; > > + hid_set_drvdata(hdev, ds); > > + > > + ds->gamepad = ps_gamepad_create(hdev); > > + if (IS_ERR(ds->gamepad)) { > > + ret = PTR_ERR(ds->gamepad); > > + goto err; > > + } > > + > > + return &ds->base; > > + > > +err: > > + return ERR_PTR(ret); > > +} > > [...] > > +static int ps_probe(struct hid_device *hdev, const struct hid_device_id *id) > > +{ > > + struct ps_device *dev; > > + int ret; > > + > > + ret = hid_parse(hdev); > > + if (ret) { > > + hid_err(hdev, "parse failed\n"); > > + return ret; > > + } > > + > > + ret = hid_hw_start(hdev, HID_CONNECT_HIDRAW); > > + if (ret) { > > + hid_err(hdev, "hw start failed\n"); > > + return ret; > > + } > > + > > + ret = hid_hw_open(hdev); > > + if (ret) { > > + hid_err(hdev, "hw open failed\n"); > > + goto err_stop; > > + } > > + > > + if (hdev->product == USB_DEVICE_ID_SONY_PS5_CONTROLLER) { > > I'm still not fully seeing the purpose of this `if`. The probe should not be > called for devices not in the id_table, so this seems to me to be a long way > of writing `if (true)`. Or am I missing something? It is not used. It more there for the future, when we will add DualShock 4 and perhaps some other devices here. > > > + dev = dualsense_create(hdev); > > + if (IS_ERR(dev)) { > > + hid_err(hdev, "Failed to create dualsense.\n"); > > I think it'd be preferable if all log messages would either be lowercase or > uppercase, not a mix of both. Same for punctuation. This applies to all patches. > > > > + ret = PTR_ERR(dev); > > + goto err_close; > > + } > > + } > > + > > + return ret; > > + > > +err_close: > > + hid_hw_close(hdev); > > +err_stop: > > + hid_hw_stop(hdev); > > + return ret; > > +} > > [...] > > > Regards, > Barnabás Pőcze Regards, Roderick Colenbrander