Tiva USB HID Touch

Greg Ford said:

BUT the end result of an actual touch event is banishment to the FaultISR().

You may analyze the place on which HardFault (FaultISR()) occurs, using calling stack view of the debugger (on Keil). But I believe it is caused because the touch screen should return an input report of more than 3 bytes, expected for boot mouse.

Anyway, to support the touch screen, you have to write another HID host driver, modifying usbhhidmouse.c.
C:\ti\TivaWare_C_Series-2.0.1.11577\usblib\host\usbhhidmouse.c

Before starting modification, the target touch screen should be examined on a PC well, to fix the required feature for modification.

Using a USB sniffer on a PC, read out descriptors of the touch screen (including report descriptor). These sniffers get the descriptors at enumeration.

USBlyzer (33days free trial) http://www.usblyzer.com/
USBTrace (15days free trial) http://www.sysnucleus.com/

To capture enumeration at plug in,
- Run USBlyzer, enable "Capture->Capture Hot-plugged" menu, and plug in your device.
OR
- After starting USBTrace, click the "Capture Hot Plugged Devices" menu or toolbar button and "start capture". And then, plug in your device.

These sniffers display Descriptors in readable format. Save the Descriptors into a file.
Zip and post it, so that we can discuss on the descriptors.

Tsuneo

In addition to the great suggestions already posted, we have an app note that is often useful in diagnosing faults.

http://www.ti.com/lit/an/spma043/spma043.pdf

Dexter

7446.USB Input Device.zip

Tsuneo,

Thanks for the help.  Attached is the descriptor for the touch panel. 

Although I use the IAR tool chain, I was able to confirm the cause of the Hard Fault.  The HIDDriverOpen(...) function returns successfully, but a call back function is not registered; therefore any subsequent call to pfnCallback(...) causes the fault.  Which in turn indicates that you are spot on in your assessment of the overall issue; An HID dost driver is required.

There is a large amount of information where do I begin?

Umm, the fact didn't go like my above speculation.

The outline of the mouse detection and initialization of its instance is as follows,

In usb_host_mouse.c
(1) When the host detects device connection, USBHCDEvents() callback is called with USB_EVENT_CONNECTED
(2) USBHCDEvents() checks the interface triad (class, subclass, protocol) fields
when the triad matches to boot mouse (0x03, 0x01, 0x02),
(3) STATE_MOUSE_INIT is assigned to g_eUSBState
(4) main loop catches STATE_MOUSE_INIT,
(5) USBHMouseInit() initializes the moust device instance (g_psMouseInstance).


usb_host_mouse.c

void
USBHCDEvents(void *pvData)
{
    ...
    switch(pEventInfo->ui32Event)
    {
        //
        // New mouse detected.
        //
        case USB_EVENT_CONNECTED:       // <------ (1)
        {
            //
            // See if this is a HID Mouse.
            //
                                        // <------ (2)
            if((USBHCDDevClass(pEventInfo->ui32Instance, 0) == USB_CLASS_HID) &&
               (USBHCDDevProtocol(pEventInfo->ui32Instance, 0) ==
                USB_HID_PROTOCOL_MOUSE))
            {
                ...
                //
                // Proceed to the STATE_MOUSE_INIT state so that the main loop
                // can finish initialized the mouse since USBHMouseInit()
                // cannot be called from within a callback.
                //
                g_eUSBState = STATE_MOUSE_INIT;    // <------ (3)
            }
            break;
        }

int
main(void)
{
    ...
    while(1)
    {
        ...
        switch(g_eUSBState)
        {
            //
            // This state is entered when the mouse is first detected.
            //
            case STATE_MOUSE_INIT:                  // <------ (4)
            {
                //
                // Initialize the newly connected mouse.
                //
                USBHMouseInit(g_psMouseInstance);   // <------ (5)


Thanking to your post of the descriptors,
the interface triad of the touch panel is (0x03, 0x00, 0x00)

Interface Descriptor 0/0 HID, 2 Endpoints
0 bLength            1 09h  
1 bDescriptorType    1 04h Interface
2 bInterfaceNumber   1 00h  
3 bAlternateSetting  1 00h  
4 bNumEndpoints      1 02h  
5 bInterfaceClass    1 03h HID     <-------
6 bInterfaceSubClass 1 00h         <-------
7 bInterfaceProtocol 1 00h         <-------
8 iInterface         1 03h "EndPoint1 Interrupt Pipe"

It doesn't match to boot mouse.
And then, the device doesn't pass the check of above (2)
The sequence, (3) and after, doesn't occur.
In this way, the device instance isn't initialized.


If you would like to support this touch panel, you have to write a new driver, based on usbhhidmouse.c
\TivaWare_C_Series-2.0.1.11577\usblib\host\usbhhidmouse.c
It won't be so much modification from the original usbhhidmouse.c
Do you try a custom driver?

Tsuneo

1. tDeviceDescriptor * pDesc = &g_psMouseInstance->psHIDInstance->psDevice->sDeviceDescriptor; This causes a compiler error

On the usb_host_mouse.c,

unsigned long USBHCDDevVIDPID(unsigned long ulInstance);
#define TOUCH_VIDPID (0x14E1 << 16 | 0x6000)

void
USBHCDEvents(void *pvData)
{
    ...
    switch(pEventInfo->ulEvent)
    {
        //
        // New mouse detected.
        //
        case USB_EVENT_CONNECTED:
        {
            //
            // See if this is a HID Mouse.
            //

      // <---- comment these lines
//            if((USBHCDDevClass(pEventInfo->ulInstance, 0) == USB_CLASS_HID) &&
//               (USBHCDDevProtocol(pEventInfo->ulInstance, 0) ==
//               USB_HID_PROTOCOL_MOUSE))

      // <---- add these lines
            //
            // See if this is the touch panel.
            //
            if( USBHCDDevVIDPID(pEventInfo->ulInstance) == TOUCH_VIDPID )
            {

2. It appears that the critical information for success is found in the device descriptor so I used the debugger to stop in the HIDDriverOpen(...) and the device information appears to be garbage

Can you tell me how your new the values hidden within the input descriptor?