Re: [PATCH 2/2] Add new force feedback driver for Mayflash game controller adapters.

[Benjamin Tissoires <benjamin.tissoires@redhat.com>]

On Oct 31 2016 or thereabouts, Marcel Hasler wrote:
> Hi
> 
> 2016-10-31 12:17 GMT+01:00 Benjamin Tissoires <benjamin.tissoires@redhat.com>:
> > On Oct 30 2016 or thereabouts, Marcel Hasler wrote:
> >> Add a new module named hid-mf that implements force feedback for game controller adapters
> >> manufactured by Mayflash. Currently only the PS3 adapter is supported, other adapters still need
> >> to be tested.
> >>
> >> Signed-off-by: Marcel Hasler <mahasler@gmail.com>
> >> ---
> >>  drivers/hid/Kconfig    |   8 +++
> >>  drivers/hid/Makefile   |   1 +
> >>  drivers/hid/hid-core.c |   1 +
> >>  drivers/hid/hid-mf.c   | 165 +++++++++++++++++++++++++++++++++++++++++++++++++
> >>  4 files changed, 175 insertions(+)
> >>  create mode 100644 drivers/hid/hid-mf.c
> >>
> >> diff --git a/drivers/hid/Kconfig b/drivers/hid/Kconfig
> >> index cd4599c..1530d28 100644
> >> --- a/drivers/hid/Kconfig
> >> +++ b/drivers/hid/Kconfig
> >> @@ -512,6 +512,14 @@ config HID_MAGICMOUSE
> >>       Say Y here if you want support for the multi-touch features of the
> >>       Apple Wireless "Magic" Mouse and the Apple Wireless "Magic" Trackpad.
> >>
> >> +config HID_MAYFLASH
> >> +     tristate "Mayflash game controller adapter force feedback"
> >> +     depends on HID
> >> +     select INPUT_FF_MEMLESS
> >> +     ---help---
> >> +     Say Y here if you have HJZ Mayflash PS3 game controller adapters
> >> +     and want to enable force feedback support.
> >> +
> >>  config HID_MICROSOFT
> >>       tristate "Microsoft non-fully HID-compliant devices"
> >>       depends on HID
> >> diff --git a/drivers/hid/Makefile b/drivers/hid/Makefile
> >> index 86b2b57..c0453f1 100644
> >> --- a/drivers/hid/Makefile
> >> +++ b/drivers/hid/Makefile
> >> @@ -58,6 +58,7 @@ obj-$(CONFIG_HID_LOGITECH)  += hid-logitech.o
> >>  obj-$(CONFIG_HID_LOGITECH_DJ)        += hid-logitech-dj.o
> >>  obj-$(CONFIG_HID_LOGITECH_HIDPP)     += hid-logitech-hidpp.o
> >>  obj-$(CONFIG_HID_MAGICMOUSE) += hid-magicmouse.o
> >> +obj-$(CONFIG_HID_MAYFLASH)   += hid-mf.o
> >>  obj-$(CONFIG_HID_MICROSOFT)  += hid-microsoft.o
> >>  obj-$(CONFIG_HID_MONTEREY)   += hid-monterey.o
> >>  obj-$(CONFIG_HID_MULTITOUCH) += hid-multitouch.o
> >> diff --git a/drivers/hid/hid-core.c b/drivers/hid/hid-core.c
> >> index 2b89c70..8470e22 100644
> >> --- a/drivers/hid/hid-core.c
> >> +++ b/drivers/hid/hid-core.c
> >> @@ -1883,6 +1883,7 @@ static const struct hid_device_id hid_have_special_driver[] = {
> >>       { HID_USB_DEVICE(USB_VENDOR_ID_DELCOM, USB_DEVICE_ID_DELCOM_VISUAL_IND) },
> >>       { HID_USB_DEVICE(USB_VENDOR_ID_DRAGONRISE, 0x0006) },
> >>       { HID_USB_DEVICE(USB_VENDOR_ID_DRAGONRISE, 0x0011) },
> >> +     { HID_USB_DEVICE(USB_VENDOR_ID_DRAGONRISE, USB_DEVICE_ID_DRAGONRISE_PS3) },
> >>       { HID_USB_DEVICE(USB_VENDOR_ID_DREAM_CHEEKY, USB_DEVICE_ID_DREAM_CHEEKY_WN) },
> >>       { HID_USB_DEVICE(USB_VENDOR_ID_DREAM_CHEEKY, USB_DEVICE_ID_DREAM_CHEEKY_FA) },
> >>       { HID_BLUETOOTH_DEVICE(USB_VENDOR_ID_ELECOM, USB_DEVICE_ID_ELECOM_BM084) },
> >> diff --git a/drivers/hid/hid-mf.c b/drivers/hid/hid-mf.c
> >> new file mode 100644
> >> index 0000000..6791ce7
> >> --- /dev/null
> >> +++ b/drivers/hid/hid-mf.c
> >> @@ -0,0 +1,165 @@
> >> +/*
> >> + * Force feedback support for Mayflash game controller adapters.
> >> + *
> >> + * These devices are manufactured by Mayflash but identify themselves
> >> + * using the vendor ID of DragonRise Inc.
> >> + *
> >> + * Tested with:
> >> + * 0079:1801 "DragonRise Inc. Mayflash PS3 Game Controller Adapter"
> >> + *
> >> + * The following adapters probably work too, but need to be tested:
> >> + * 0079:1800 "DragonRise Inc. Mayflash WIIU Game Controller Adapter"
> >> + * 0079:1843 "DragonRise Inc. Mayflash GameCube Game Controller Adapter"
> >> + *
> >> + * Copyright (c) 2016 Marcel Hasler <mahasler@gmail.com>
> >> + */
> >> +
> >> +/*
> >> + * This program is free software; you can redistribute it and/or modify
> >> + * it under the terms of the GNU General Public License as published by
> >> + * the Free Software Foundation; either version 2 of the License, or
> >> + * (at your option) any later version.
> >> + *
> >> + * This program is distributed in the hope that it will be useful,
> >> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> >> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> >> + * GNU General Public License for more details.
> >> + */
> >> +
> >> +#include <linux/input.h>
> >> +#include <linux/slab.h>
> >> +#include <linux/hid.h>
> >> +#include <linux/module.h>
> >> +
> >> +#include "hid-ids.h"
> >> +
> >> +struct mf_device {
> >> +     struct hid_report *report;
> >> +};
> >> +
> >> +static int mf_play(struct input_dev *dev, void *data, struct ff_effect *effect)
> >> +{
> >> +     struct hid_device *hid = input_get_drvdata(dev);
> >> +     struct mf_device *mf = data;
> >> +     int strong, weak;
> >> +
> >> +     strong = effect->u.rumble.strong_magnitude;
> >> +     weak = effect->u.rumble.weak_magnitude;
> >> +
> >> +     dbg_hid("Called with 0x%04x 0x%04x.\n", strong, weak);
> >> +
> >> +     strong = strong * 0xff / 0xffff;
> >> +     weak = weak * 0xff / 0xffff;
> >> +
> >> +     dbg_hid("Running with 0x%02x 0x%02x.\n", strong, weak);
> >> +
> >> +     mf->report->field[0]->value[0] = weak;
> >> +     mf->report->field[0]->value[1] = strong;
> >> +     hid_hw_request(hid, mf->report, HID_REQ_SET_REPORT);
> >> +
> >> +     return 0;
> >> +}
> >> +
> >> +static int mf_init(struct hid_device *hid)
> >> +{
> >> +     struct mf_device *mf;
> >> +
> >> +     struct list_head *report_list =
> >> +                     &hid->report_enum[HID_OUTPUT_REPORT].report_list;
> >> +
> >> +     struct list_head *report_ptr;
> >> +     struct hid_report *report;
> >> +
> >> +     struct hid_input *input =
> >> +                     list_first_entry(&hid->inputs, struct hid_input, list);
> >> +
> >> +     struct input_dev *dev;
> >> +
> >> +     int error;
> >> +
> >> +     /* Setup each of the four inputs */
> >> +     list_for_each(report_ptr, report_list) {
> >> +             report = list_entry(report_ptr, struct hid_report, list);
> >> +
> >> +             if (report->maxfield < 1 || report->field[0]->report_count < 2) {
> >> +                     hid_err(hid, "Invalid report, this should never happen!\n");
> >> +                     return -ENODEV;
> >> +             }
> >> +
> >> +             if (input == NULL) {
> >> +                     hid_err(hid, "Missing input, this should never happen!\n");
> >> +                     return -ENODEV;
> >> +             }
> >> +
> >> +             dev = input->input;
> >> +             input = list_next_entry(input, list);
> >
> > Wouldn't it make more sense to use .input_configured instead of this
> > after-the-fact loop?
> >
> >> +
> >> +             mf = kzalloc(sizeof(struct mf_device), GFP_KERNEL);
> >
> > I might not be fully awoken, but where is the corresponding kfree?
> > If you do not want to manually free the allocated memory, you can use
> > devm_kzalloc, as long as there is no races when removing the device.
> >
> 
> The corresponding kfree for mf in this case is in ml_ff_destroy().

K, thanks. I should definitively have taken more attention while reading
the patch.

> 
> >> +             if (!mf)
> >> +                     return -ENOMEM;
> >> +
> >> +             set_bit(FF_RUMBLE, dev->ffbit);
> >> +
> >> +             error = input_ff_create_memless(dev, mf, mf_play);
> >> +             if (error) {
> >> +                     kfree(mf);
> >> +                     return error;
> >> +             }
> >> +
> >> +             mf->report = report;
> >> +             mf->report->field[0]->value[0] = 0x00;
> >> +             mf->report->field[0]->value[1] = 0x00;
> >> +             hid_hw_request(hid, mf->report, HID_REQ_SET_REPORT);
> >> +     }
> >> +
> >> +     hid_info(hid, "Force feedback for HJZ Mayflash game controller "
> >> +                   "adapters by Marcel Hasler <mahasler@gmail.com>\n");
> >> +
> >> +     return 0;
> >> +}
> >> +
> >> +static int mf_probe(struct hid_device *hdev, const struct hid_device_id *id)
> >> +{
> >> +     int error;
> >> +
> >> +     dev_dbg(&hdev->dev, "Mayflash HID hardware probe...\n");
> >> +
> >> +     /* Split device into four inputs */
> >> +     hdev->quirks |= HID_QUIRK_MULTI_INPUT;
> >
> > Looks like the entry in the previous patch was not required apparently
> > :)
> >
> >> +
> >> +     error = hid_parse(hdev);
> >> +     if (error) {
> >> +             hid_err(hdev, "HID parse failed.\n");
> >> +             return error;
> >> +     }
> >> +
> >> +     error = hid_hw_start(hdev, HID_CONNECT_DEFAULT & ~HID_CONNECT_FF);
> >> +     if (error) {
> >> +             hid_err(hdev, "HID hw start failed\n");
> >> +             return error;
> >> +     }
> >> +
> >> +     error = mf_init(hdev);
> >
> > This seems completely racy. hid_hw_start() exports the input nodes to
> > user-space and the ff initialization is done after. It would be much
> > better to initialize the ff part in .input_configured when the device
> > hasn't been registered to user space yet.
> >
> > Cheers,
> > Benjamin
> >
> >> +     if (error) {
> >> +             hid_err(hdev, "Force feedback init failed.\n");
> >> +             hid_hw_stop(hdev);
> >> +             return error;
> >> +     }
> >> +
> >> +     return 0;
> >> +}
> >> +
> >> +static const struct hid_device_id mf_devices[] = {
> >> +     { HID_USB_DEVICE(USB_VENDOR_ID_DRAGONRISE, USB_DEVICE_ID_DRAGONRISE_PS3),  },
> >> +     { }
> >> +};
> >> +MODULE_DEVICE_TABLE(hid, mf_devices);
> >> +
> >> +static struct hid_driver mf_driver = {
> >> +     .name = "hid_mf",
> >> +     .id_table = mf_devices,
> >> +     .probe = mf_probe,
> >> +};
> >> +module_hid_driver(mf_driver);
> >> +
> >> +MODULE_LICENSE("GPL");
> >> --
> >> 2.10.1
> >>
> 
> Thanks for the hints. I tried your idea with .input_configured but
> there's a bit of a problem with that. For each of the four gamepads
> two input device nodes are created, one for the buttons/axes and one
> for force feedback. Unfortunately systemd only assigns the first nodes
> the uaccess tag. Now, I looked at the other ff drivers and essentially
> followed their example. None of these use .input_configured, but
> instead just use the very first hid_input and make ff available
> through that. This works and actually has the desired effect that
> normal users can use force feedback without having to be in the
> 'input' group (which as I understand is deprecated).

I haven't taken much attention to the ff implementation both in the
kernel and in userspace. After your explanation. I am not sure I want to
spend more time on it though :)

> 
> So while I completely agree that using .input_configured would be the
> proper thing to do, I think either all ff drivers should use it (and
> systemd should be fixed to make those input nodes accessible) or
> hid-mf should do what all others do (at least for the time being).

Yes, please stick to what other kernel drivers do, and hope for the
best. If I understand correctly, you'll need to remove the for loop and
only check for the first available input node, right?

Cheers,
Benjamin

> 
> Best regards
> Marcel