Bug Report: Tivaware 2.1.0.12573 USBHCDPipeRead

Hi Jim,

sorry but I can't understand where is the problem and where the solution.

In fact the source you posted is exactly the same I found in Tivaware 2.1.0.12573.

Does It mean that the problem is already solved in this version ?

I am experiencing this issue when I connect a memory stick  and immediately extract it, before it is enumerated.

Doing so, the function USBHCDPipeRead ALWAYS hangs in the same infinite loop that you described in your post...

I have tried to force the exit from that loop but, from that moment on, the USB stick is no more recognized and enumerated.

Thank you in advance

Mario

I am using:

- EK-TM4C129XL Launchpad

- Tivaware 2.1.0.12573

- usb_host_msc project from Tivaware

Mario,

The way that I solved the problem was to add

ui32RemainingBytes = 0;

inside of the if statement that checked the EVENTs.

It now looks like this:

if(g_sUSBHCD.ui32IntEvents & (INT_EVENT_DISCONNECT |
INT_EVENT_VBUS_ERR |
INT_EVENT_POWER_FAULT))
{
//
// Set the pipe state to error.
//
g_sUSBHCD.psUSBOUTPipes[ui32PipeIdx].iState = ePipeError;

//
// Needs to be set to exit out of large while loop.
//
ui32RemainingBytes = 0;

break;
}

This allows the DISCONNECT event to break out of both the interior and exterior loops.

Thanks,
Jim

Jim,

thank you for the quick reply.

I have applied your correction and now I can see the program flow exiting the internal and external loop.

That is good but the problem is that the USB Host is no more working. When I insert the memory stick again, it is not recognized and enumerated.

In this situation the function USBHCDPipeRead() is continuously involved and it triggers the event INT_EVENT_DISCONNECT (set together with INT_EVENT_SOF).

The field g_sUSBHCD.psUSBINPipes[ui32PipeIdx].iState is set every other time at ePipeReadDMA or ePipeReading.

My feeling is that this problem concerns the DMA, but I'm not sure.

Any suggestions?

Thanks

Mario

Hi Mario,

Did u get solution to this problem. I am Facing Same issue during USB Ejection controller get struck in USBHCDPipeRead() infinite loop.

Tried JIM Solution by ui32RemainingBytes = 0; in ePipeError If check But after doing this my device is Always getting USBDevice Mode not Host/None mode.

(According to my application I need USBOTG(HOST/DEVICE/NoNE) all the modes switching)


Regards
Vidushi

Hi Vidushi.

The only way I found to bypass the problem is the following:

In the UsbMain() function in UsbHost.c module I have modified the case STATE_NO_DEVICE adding a check on timeout error and performing a USB re-init whenever this occurs. Below the new source code 

switch(g_eState)
{
   //
  // A previously connected device has been disconnected.

  //
  case STATE_NO_DEVICE:
  {
    if (g_eUIState == STATE_UNKNOWN_DEVICE)
    {
      UARTprintf("\nUnknown device disconnected.\n");
    }
#if defined(USB_)   // MG Patch : USB halted after fast pen-drive insert/extract operation
    else if (g_eUIState == STATE_TIMEOUT_DEVICE)
    {
      USBHMSCDriveClose(g_psMSCInstance);
      UsbInit();
      SysCtlDelay((g_ui32SysClock / 3) / 5);    // 200 ms delay
    }
#endif
    else
    {
      UARTprintf("\nMass storage device disconnected.\n");
    }

    ulPrompt = 1;
    break;
  }

Regards

Mario

Hi Mario,

Are you talking about usbHostEnum.c?

i didnt get USBHOst.c file in ti:\usblib folder however USBHost.h exists. please clarify.

Regards

Vidushi 

Hi Vidushi,

sorry for the mistake. I forgot that I renamed the module and the function while porting the source code from the demo program which I started from.
The demo program is "TivaWare_C_Series-2.1.0.12573\examples\boards\ek-tm4c1294xl\usb_host_msc.c".
The module is msc_host_msc.c and the function is ReadLine().

However, it is the USB control function to be called by the main loop of the application.

Regards
Mario

Hi Mario,

I am not using any sample code. Doing in my own project can you please tell me the change in your API level.
initially i take reference of your qslogger code in stellaris.

Regards
Vidushi

Hi Mario,

Please Suggest me temporary solution for this problem till your new library is under development  as it is lingering from long time tried lot of solutions but unable to resolve this matter.

Regards

Vidushi

Hi Vidushi,

unfortunately, after many tests, I have seen that also my solution doesn't cover every case of USB stuck.
After a lot of days (months?) spent to try to solve this issue, we have decided to migrate our application to another MCU from a different manufacturer.
We are very frustrated, more than anything else, for the lack of support from TI. After the ugly experience of Stellaris family MCUs, I would expect a great support on this new products, but I was wrong.
You just have to look at the last version of TivaWare (2.1.0.12573) that was released on FEB-7-2014(!) even if the forums are full of bug reports.
When will TI release a new debugged version?
Well, I can't wait anymore...

Regards
Mario

Hello Mario

It is rather unfortunate losing customer(s) based on the fact that we did not do our best in supporting each one at the time of crisis. In this case it was USB. The issue of the hang has been fixed (and we did make a premature patch release on the forum) but did not follow up with a timely TivaWare release. Having said that the next one shall be out soon, once we have the release notes completed with the bug fixes clearly documented.

I wish you the best and hope that the new TivaWare may help in case you try evaluating TM4C12x devices.

Regards
Amit

Hello Amit,

can you tell me where can I find the patch you are speaking about?

Regards

Mario

Hello Amit,

I have been working on the same problem for some time, so can you send me a link to the new release, if it's ready. Preferably with the changes that have been made to any source files and which particular bugs does it fix.

Thank you.

Regards,

Stamen.

Hello Stamen,

The patch has been pushed into the latest TivaWare release 2.1.1.71

Regards
Amit