Do you still get this issue with udev enabled?

(In reply to Gabriel Jacobo from comment #1)
> Do you still get this issue with udev enabled?

Can you elaborate on what you mean by udev enabled?
I do have udev rules to make sure the joystick properties are properly set (i.e. readable with user and not root). But even with those I still needed the patch [makes sense to me, the if statement requires both a key AND an axis to consider it a joystick, and the rudder pedals don't have any buttons so it would never be recognized as such])

I mean compile SDL with udev support built in. On Ubuntu/Debian you make sure that support is available with:

    sudo apt-get libudev-dev

Then configure;make;sudo make install

The IsJoystick function now trusts what comes from udev without checking it, so there's a good chance that if udev thinks your device is a joystick, it will be treated as such by SDL.

Also, this will only work with the HG version of SDL (not the precompiled 2.0.1 release).

just recompiled 2.0.1 and made sure I had libudev
( /usr/local/include/SDL2:/* #undef HAVE_LIBUDEV_H */)

libsdl did not recognize the joystick without buttons (even when I put it my patch)


I reverted back to the old build (left it in a separate folder). Except now I am worried that if I try to run configure again it'll use libudev. How can I make sure it does not?

Use SDL from HG! *Not* the 2.0.1 release. You can use udevadm to verify that the system reports your joystick as such (look for the ID_JOYSTICK string).

To disable udev entirely you configure with:

configure --disable-libudev

Created attachment 1423 [details]
output of udevadm info for js1

Created attachment 1424 [details]
output of udevadm info for js0

Alright, I checked out this Changeset:
7974 (5d12bfc720e1) Added a few notes from Ryan's email to README-linux.txt
User:
Gabriel Jacobo <gabomdq@gmail.com>
Date:
2013-11-12 11:31:13 -0300 (7 hours)
Parent:
7973 (96f2f61be1c6) Linux README explaining build vs runtime, and build deps for Ubuntu 13.04



Here is a snippet from the config log

configure:20806: checking libudev.h usability
configure:20806: gcc -c -g -O2 -Iinclude -I/home/zouhair/SDL2.0/include    -Iinclude -I/home/zouhair/SDL2.0/include  conftest.c >&5
configure:20806: $? = 0
configure:20806: result: yes
configure:20806: checking libudev.h presence
configure:20806: gcc -E  -Iinclude -I/home/zouhair/SDL2.0/include  conftest.c
configure:20806: $? = 0
configure:20806: result: yes
configure:20806: checking for libudev.h
configure:20806: result: yes


Still same problem; all my other devices show up except for the rudders.
I've included the output of:
udevadm info -a -p  $(udevadm info -q path -n /dev/input/js0) > stick.txt
udevadm info -a -p  $(udevadm info -q path -n /dev/input/js1) > rudders.txt

Try formatting the output like below...if ID_INPUT_JOYSTICK=1 is present, SDL will pick up the joystick.

udevadm info --query=all --name=input/js0
P: /devices/pci0000:00/0000:00:1a.0/usb1/1-1/1-1.2/1-1.2:1.0/input/input16/js0
N: input/js0
S: input/by-id/usb-©Microsoft_Xbox_360_Wireless_Receiver_for_Windows_FF3FFD30-joystick
S: input/by-path/pci-0000:00:1a.0-usb-0:1.2:1.0-joystick
E: DEVLINKS=/dev/input/by-id/usb-©Microsoft_Xbox_360_Wireless_Receiver_for_Windows_FF3FFD30-joystick /dev/input/by-path/pci-0000:00:1a.0-usb-0:1.2:1.0-joystick
E: DEVNAME=/dev/input/js0
E: DEVPATH=/devices/pci0000:00/0000:00:1a.0/usb1/1-1/1-1.2/1-1.2:1.0/input/input16/js0
E: ID_BUS=usb
E: ID_FOR_SEAT=input-pci-0000_00_1a_0-usb-0_1_2_1_0
E: ID_INPUT=1
E: ID_INPUT_JOYSTICK=1
E: ID_MODEL=Xbox_360_Wireless_Receiver_for_Windows
E: ID_MODEL_ENC=Xbox\x20360\x20Wireless\x20Receiver\x20for\x20Windows
E: ID_MODEL_ID=0719
E: ID_PATH=pci-0000:00:1a.0-usb-0:1.2:1.0
E: ID_PATH_TAG=pci-0000_00_1a_0-usb-0_1_2_1_0
E: ID_REVISION=0100
E: ID_SERIAL=©Microsoft_Xbox_360_Wireless_Receiver_for_Windows_FF3FFD30
E: ID_SERIAL_SHORT=FF3FFD30
E: ID_TYPE=generic
E: ID_USB_DRIVER=xpad
E: ID_USB_INTERFACES=:ff5d81:ff5d82:
E: ID_USB_INTERFACE_NUM=00
E: ID_VENDOR=©Microsoft
E: ID_VENDOR_ENC=©Microsoft
E: ID_VENDOR_ID=045e
E: MAJOR=13
E: MINOR=0
E: SUBSYSTEM=input
E: TAGS=:seat:uaccess:
E: USEC_INITIALIZED=834746

looks like the Saitek Rudder pedals have ID_INPUT but no ID_INPUT_JOYSTICK.

I should point out that jstest-gtk does detect just fine.

Also I have this in my /etc/udev/rules.d/:
# Saitek Pro Flight Rudder Pedals
KERNEL=="event*", ATTRS{idProduct}=="0763", ATTRS{idVendor}=="06a3", MODE="0666"
KERNEL=="event*", ATTRS{idProduct}=="0764", ATTRS{idVendor}=="06a3", MODE="0666"


$udevadm info --query=all --name=input/js1
P: /devices/pci0000:00/0000:00:1a.0/usb1/1-1/1-1.5/1-1.5:1.0/input/input6/js1
N: input/js1
E: DEVNAME=/dev/input/js1
E: DEVPATH=/devices/pci0000:00/0000:00:1a.0/usb1/1-1/1-1.5/1-1.5:1.0/input/input6/js1
E: ID_BUS=usb
E: ID_INPUT=1
E: ID_MODEL=Saitek_Pro_Flight_Combat_Rudder_Pedals
E: ID_MODEL_ENC=Saitek\x20Pro\x20Flight\x20Combat\x20Rudder\x20Pedals
E: ID_MODEL_ID=0764
E: ID_PATH=pci-0000:00:1a.0-usb-0:1.5:1.0
E: ID_PATH_TAG=pci-0000_00_1a_0-usb-0_1_5_1_0
E: ID_REVISION=0201
E: ID_SERIAL=Saitek_Saitek_Pro_Flight_Combat_Rudder_Pedals
E: ID_TYPE=hid
E: ID_USB_DRIVER=usbhid
E: ID_USB_INTERFACES=:030000:
E: ID_USB_INTERFACE_NUM=00
E: ID_VENDOR=Saitek
E: ID_VENDOR_ENC=Saitek
E: ID_VENDOR_ID=06a3
E: MAJOR=13
E: MINOR=1
E: SUBSYSTEM=input
E: UDEV_LOG=3
E: USEC_INITIALIZED=9070173

So, the fix for this appears to be a udev rule to add the env variable ID_INPUT_JOYSTICK? Try with...


KERNEL=="event*", ATTRS{idProduct}=="0763", ATTRS{idVendor}=="06a3", MODE="0666", ENV{ID_INPUT_JOYSTICK}="1"
KERNEL=="event*", ATTRS{idProduct}=="0764", ATTRS{idVendor}=="06a3", MODE="0666", ENV{ID_INPUT_JOYSTICK}="1"

Let me know if this worked, I'd like to add a few notes to the README-linux.txt file if this is indeed the correct fix.

Hey,

`udevadm info --query=all --name=input/js1` now shows ID_INPUT_JOYSTICK=1 as per dev-rule (Initially it did not work but after `sudo udevadm trigger`)

However, it actually broke my test program (I had a hard coded number of maximum joysticks). So I fell back to the testjoystick.c in the source tree. This is what I get when I compile it against 2.0.1 (without libudev and with my patch):

INFO: There are 4 joysticks attached
INFO: Joystick 0: Thustmaster Joystick - HOTAS Warthog
INFO:        axes: 2
INFO:       balls: 0
INFO:        hats: 1
INFO:     buttons: 19
INFO: instance id: 0
INFO:        guid: 030000004f0400000204000011010000
INFO: Joystick 1: Saitek Saitek Pro Flight Combat Rudder Pedals
INFO:        axes: 3
INFO:       balls: 0
INFO:        hats: 0
INFO:     buttons: 0
INFO: instance id: 1
etc...

When I do ./testjoystick 1 the rudder pedals work fine. And my test program as well as my end user application work just fine.


Now if I compile testjoystick with the head of SDL on HG and with libudev; this is what I get as an output:

INFO: There are 25 joysticks attached
INFO: Joystick 0: Power Button
INFO:        axes: 0
INFO:       balls: 0
INFO:        hats: 0
INFO:     buttons: 0
INFO: instance id: 0
INFO:        guid: 19000000506f77657220427574746f00
INFO: Joystick 1: Video Bus
INFO:        axes: 0
INFO:       balls: 0
INFO:        hats: 0
INFO:     buttons: 0
INFO: instance id: 1
INFO:        guid: 19000000566964656f20427573000000
INFO: Joystick 2: Power Button
.....
INFO: Joystick 8: Saitek Saitek Pro Flight Combat Rudder Pedals
INFO:        axes: 3
INFO:       balls: 0
INFO:        hats: 0
INFO:     buttons: 0
INFO: instance id: 8
etc....

using ./testjoystick 8 shows working pedals; and although my test application is broken (fixed it now)  my end user application is fine.

Anyway, things are working again; but it's odd that I have "25" joysticks now. Could that be because of the udev rule?

It looks like something in your udev rules is adding ID_INPUT_JOYSTICK to everything! And given the code in HG now trusts completely what comes from udev, you get that scenario where every device is a joystick.

This is what's in my udev/rules.d/ related to INPUT_JOYSTICK

I thought the ATTRS with idProduct and idVendor would filter to only apply
to the rudder pedals?

# Saitek Pro Flight Rudder Pedals

KERNEL=="event*", ATTRS{idProduct}=="0763", ATTRS{idVendor}=="06a3",
MODE="0666", ENV{ID_INPUT_JOYSTICK}="1"
KERNEL=="event*", ATTRS{idProduct}=="0764", ATTRS{idVendor}=="06a3",
MODE="0666", ENV{ID_INPUT_JOYSTICK}="1"

I've added a few troubleshooting notes here: https://hg.libsdl.org/SDL/rev/3122e7af0bf5

That's as much as we can do on this, I think.