Improving drivers for Logitech Driving Force Pro and G25 wheels

Improving drivers for Logitech Driving Force Pro and G25 wheels

[chris guirl]

I have a Logitech Driving Force Pro and I've been trying to figure out
how to make it work better in Linux. I have recently learned some
things about both wheels that would be nice to incorporate into the
driver. Most of this information was gained through a contact at
Logitech. I intend to investigate making some changes to the kernel to
improve the support for these wheels. Two of the main features I'd
like to incorporate are setting the wheel axis range to different
values so the full 900 degree range of the wheel can be used, and
providing separated axes for the gas and brake pedals.

So, where to start? I'm not very familiar with kernel hacking but I'll
begin doing some research on it soon. Can anyone on the list provide
me with some guidance? I've tried to find some driver hacking
documentation specific to my task but I haven't found much. Something
along the lines of a joystick or USB driver hacking tutorial would
help me a lot. I've also looked through a good bit of the Linux
joystick and USB HID code in the input section and I am unsure where
device specific code goes. I guess I have some more reading to do...
:) Thanks in advance for any tips.

Chris Guirl

Re: Improving drivers for Logitech Driving Force Pro and G25 wheels

[chris guirl]

On Fri, Mar 14, 2008 at 5:42 AM, Jiri Kosina <jkosina@suse.cz> wrote:
>  thanks for your mail, I have added Jean-Philippe to CC, who has sent me
>  some time ago debugging information for this wheel, but I unfortunately
>  didn't yet had time to look into that in more detail.

Which wheel, the DFP or the G25? Either way they are similar in many
ways so better support for one would improve support for the other.
I'm interested to see this information. There has been some discussion
of the G25 on the forums at vdrift.net.

>  Regarding the HID code -- currently, all device-specific stuff for HID
>  devices is implemented as quirks in hid code -- please see
>  drivers/hid/hid-*quirks* and drivers/hid/usbhid/hid-quirks*. This is not a
>  solution that works well in long-term though, and Jiri Slaby (also added
>  to CC) is currently rewriting the HID core code, so that it will work as a
>  proper bus, so individual drivers could work nicely with that. Once this
>  is done, all the current quirks will be converted to HID-bus drivers.

I've looked through the quirks code and I have some suggestions for
things that can be changed. First, the Vendor/product IDs of the
wheels are not present. When the devices power up they assume VID:PID
0x046d:0xc294 which there is a line for in
drivers/hid/usbhid/hid-quirks.c:

#define USB_DEVICE_ID_LOGITECH_WHEEL	0xc294

This is used by the Logitech Formula Force or Driving Force wheel, and
a legacy mode on the DFP & G25. To use the DFP or G25 in their native
modes, they must be sent a command [0xf8, 0x01] or [0xf8, 0x10]
respectively. Once these wheels receive this command, they simulate a
detach and reattach as 0x046d:0xc298 and 0x046d:0xc299 respectively.
These VID:PID combinations are not handled in the kernel at all. At
the very least they need to be assigned the same quirks as the Driving
Force.

When a Driving Force/Formula Force wheel is detected the device needs
to be probed with the native mode commands to see if it will switch
into DFP or G25 mode. If these modes work, then it is possible to send
the wheels some other commands to enable full rotation of the wheel.
By default only 200 degrees of rotation is available, but both the DFP
and G25 can be switched to allow up to 900 degrees of rotation. I
think there should be a user-land tool to change settings on the wheel
such as this.

>  Depends on how massive your changes are going to be, you could either
>  coordinate with Jiri so that no effort is wasted, or you can just send me
>  a patch with simple quirks addition, and it will get converted into hidbus
>  driver later.

If there is a rewrite due then it may be best to work with Jiri so I
can make my changes work with his, no sense in patching twice.

In addition to HID code, I think I'll need to make some changes to the
joydev driver, as some other things need to be addressed (such as
splitting the gas and brake axis up into separate, independent axes).
Currently I am able to send raw commands to my DFP to make it change
modes, but the /dev/input/jsX device disappears when I do this, so
some other tweaks may be needed on the joystick end of things.

I'll be out of town this weekend so not working on anything yet, but
please send any ideas along as I hope to get started next week.
Thanks.

Chris

Fwd: Improving drivers for Logitech Driving Force Pro and G25 wheels

[Richie Ward]

I own a G25, Can you write out some of the knowledge you have acquired?

 On Fri, Mar 14, 2008 at 7:52 PM, chris guirl <thelusiv@gmail.com> wrote:
 > On Fri, Mar 14, 2008 at 5:42 AM, Jiri Kosina <jkosina@suse.cz> wrote:
 >  >  thanks for your mail, I have added Jean-Philippe to CC, who has sent me
 >  >  some time ago debugging information for this wheel, but I unfortunately
 >  >  didn't yet had time to look into that in more detail.
 >
 >  Which wheel, the DFP or the G25? Either way they are similar in many
 >  ways so better support for one would improve support for the other.
 >  I'm interested to see this information. There has been some discussion
 >  of the G25 on the forums at vdrift.net.
 >
 >  >  Regarding the HID code -- currently, all device-specific stuff for HID
 >  >  devices is implemented as quirks in hid code -- please see
 >  >  drivers/hid/hid-*quirks* and drivers/hid/usbhid/hid-quirks*.
This is not a
 >  >  solution that works well in long-term though, and Jiri Slaby (also added
 >  >  to CC) is currently rewriting the HID core code, so that it
will work as a
 >  >  proper bus, so individual drivers could work nicely with that. Once this
 >  >  is done, all the current quirks will be converted to HID-bus drivers.
 >
 >  I've looked through the quirks code and I have some suggestions for
 >  things that can be changed. First, the Vendor/product IDs of the
 >  wheels are not present. When the devices power up they assume VID:PID
 >  0x046d:0xc294 which there is a line for in
 >  drivers/hid/usbhid/hid-quirks.c:
 >
 >  #define USB_DEVICE_ID_LOGITECH_WHEEL    0xc294
 >
 >  This is used by the Logitech Formula Force or Driving Force wheel, and
 >  a legacy mode on the DFP & G25. To use the DFP or G25 in their native
 >  modes, they must be sent a command [0xf8, 0x01] or [0xf8, 0x10]
 >  respectively. Once these wheels receive this command, they simulate a
 >  detach and reattach as 0x046d:0xc298 and 0x046d:0xc299 respectively.
 >  These VID:PID combinations are not handled in the kernel at all. At
 >  the very least they need to be assigned the same quirks as the Driving
 >  Force.
 >
 >  When a Driving Force/Formula Force wheel is detected the device needs
 >  to be probed with the native mode commands to see if it will switch
 >  into DFP or G25 mode. If these modes work, then it is possible to send
 >  the wheels some other commands to enable full rotation of the wheel.
 >  By default only 200 degrees of rotation is available, but both the DFP
 >  and G25 can be switched to allow up to 900 degrees of rotation. I
 >  think there should be a user-land tool to change settings on the wheel
 >  such as this.
 >
 >  >  Depends on how massive your changes are going to be, you could either
 >  >  coordinate with Jiri so that no effort is wasted, or you can just send me
 >  >  a patch with simple quirks addition, and it will get converted
into hidbus
 >  >  driver later.
 >
 >  If there is a rewrite due then it may be best to work with Jiri so I
 >  can make my changes work with his, no sense in patching twice.
 >
 >  In addition to HID code, I think I'll need to make some changes to the
 >  joydev driver, as some other things need to be addressed (such as
 >  splitting the gas and brake axis up into separate, independent axes).
 >  Currently I am able to send raw commands to my DFP to make it change
 >  modes, but the /dev/input/jsX device disappears when I do this, so
 >  some other tweaks may be needed on the joystick end of things.
 >
 >  I'll be out of town this weekend so not working on anything yet, but
 >  please send any ideas along as I hope to get started next week.
 >  Thanks.
 >
 >  Chris
 >
 >
 > --
 >  To unsubscribe from this list: send the line "unsubscribe linux-input" in
 >  the body of a message to majordomo@vger.kernel.org
 >  More majordomo info at  http://vger.kernel.org/majordomo-info.html
 >



 --
 Thanks, Richie Ward



-- 
Thanks, Richie Ward

Re: Improving drivers for Logitech Driving Force Pro and G25 wheels

[chris guirl]

On Tue, Mar 18, 2008 at 12:30 AM, Richie Ward <richies@gmail.com> wrote:
> I own a G25, Can you write out some of the knowledge you have acquired?

First off a little info on the G25, and how it works on Windows (for
those unfamiliar). The device allows up to 900 degrees of wheel
rotation (with force feedback), it has a "H" style shifter, and it has
three foot pedals (gas, brake and clutch). It has a total of six axes
and nineteen buttons. How many axes/buttons are reported depends on
what mode the wheel is in. It also has other settings like the limits
of wheel rotation. The wheel can be set to only be allowed to turn as
little as 40 degrees. On Windows, Logitech Profiler allows the user to
change the wheel rotation range, along with other settings. When the
wheel starts up, the Logitech drivers on Windows send a command to the
device to put it into its native mode, so that all the axes/buttons
are present.

In Linux, when the wheel starts up, it is in legacy mode, and only
four axes and twelve buttons are available. I believe this causes the
clutch and shifter to not work as well, if I understand correctly.
Also, the device is limited to its default wheel rotation range of 200
degrees, and the user-oriented tools to change wheel settings do not
exist in Linux.

Another thing missing from all the steering wheel drivers for Linux is
the ability to separate the pedal axes. Basically, most steering
wheels act by default as a generic HID joystick device with two axes,
to ensure backwards-compatibility with games that only support
joysticks with two axes. These two axes are the steering axis, and a
combined brake/gas axis.

When the two pedal axes are combined as one, the pedals do not act
quite normally. When the brake is pressed by itself, the combined axis
goes negative; while the gas is pressed by itself, the combined axis
goes positive. When both pedals are pressed simultaneously they cancel
each other out. To fix this, the Linux steering wheel drivers need to
be able to change from combined to separated pedal axes with a
user-controllable tool similar to the Logitech Profiler in Windows.
For Logitech devices, the raw pedal data is available as a USB stream
unreported when acting as a standard HID device.

Through a contact at Logitech I have been given the information to fix
the deficiencies in the current Linux steering wheel drivers as
outlined above. I have posted a lot of this information already on the
message boards at VDrift.net. This thread [1] contains a Python tool
to send some commands to the DFP/G25. It was made for the G25 and
apparently works well for it, but this and other functionality needs
to be added to the kernel driver. It doesn't quite work for the DFP in
my experience because of some other issues (see my previous posts).

Much of the information I obtained was posted here [2]. I have some
other details and tidbits that might be useful. There are some other
threads on the VDrift forums with a good bit of information on the
force-feedback and other details for the G25 and other wheels, I'll be
happy to link it if anyone's interested.

I haven't had much time lately to start hacking on this, but hopefully
I will by this weekend at the latest.

Chris

[1] http://vdrift.net/Forum/viewtopic.php?t=866
[2] http://vdrift.net/Forum/viewtopic.php?p=7755#7755