USB Integrated DMA Controller on EK-TM4C129XL

Update,

I have actually worked out what seems to be sort of a hack. It might work but it doesn't seem like it is the right thing to do. In usbhandler.c in USB Library I modified the code as such:

void USB0DeviceIntHandler(void) {
	uint32_t ui32Status;
	uint32_t ui32DMAChIntStatus = 0;
	//
	// Get the controller interrupt status.
	//
	ui32Status = MAP_USBIntStatusControl(USB0_BASE);
	ui32DMAChIntStatus = USBDMAChannelIntStatus(USB0_BASE);
	if (ui32DMAChIntStatus) {
		SysTickDisable();
	}

	//
	// Call the internal handler.
	//
	USBDeviceIntHandlerInternal(0, ui32Status);
}

The systickdisable call is to stop a timer I have that I am using to time the data transfer. I am using DMA Channel 3 and for some reason if I use the condition USB_DMA_INT_CH3 I do not enter the if statement. For some reason I am also unable to put a watch expression on ui32DMAChIntStatus, which is quite frustrating. Right now my timer is telling me it takes around 400uS to transmit 8192 bytes of data over USB HS, which is believable but I need to know for sure.

Hi again Amit,

I was able to fix my issue with adding a watch expression for the status of the register. It looks to be the case that there is a discrepancy in the way the USB_DMAINTR register is indexed and the way that the tokens for determining the interrupt that was triggered. I have attached some screenshots below showing the issue.

Regards,

Bryan

Hello Amit,

Good suggestions, but the problem with the watchpoint was not that it was reporting zero. It seems there is an issue related to CCS that doesn't allow the setting of watchpoint's for certain expressions. It said that it was unable to recognize the expression so I moved the variable to be a global variable within usbhandler.c and I was able to see the variable and confirm the issue.

It would be nice to have a breakpoint on the line of code right before USBDMAInterruptStatus, but the USB handler actually gets called quite often during the setup of the device and it causes strange issues with passing descriptors to windows when I spend too much time stepping through and waiting for the interrupt to be related to DMA.

Cheers,

Bryan

Hello Amit,

Here is a screenshot of the issue with assigning a watchpoint to the value of the DMA Interrupt register:

Cheers,

Bryan

Hello All,

I have revisited this code because it seems to have stopped behaving reliably. Sometimes when I initiate USB DMA Transfers, I am not ever interrupted when the DMA Transfer has completed. It might be the case that the interrupt status register for DMA transfers is getting cleared by a function that I am not aware of.

Is there a way to register an interrupt handler to specifically deal with USB DMA Interrupts?

Right now my interrupt handling for USB DMA Interrupts is handled within the following function below:

void USB0DeviceIntHandler(void) {
	uint32_t ui32Status;
	uint32_t ui32DMAChIntStatus = 0;

	//
	// Get the controller interrupt status.
	//
	ui32Status = MAP_USBIntStatusControl(USB0_BASE);

	// Set pending transmission as complete
	ui32DMAChIntStatus = USBDMAChannelIntStatus(USB0_BASE);
	if (ui32DMAChIntStatus)
	{
		usbDMADisable();
	}

	//
	// Call the internal handler.
	//
	USBDeviceIntHandlerInternal(0, ui32Status);
}

Any help is greatly appreciated!

Thanks,

Bryan

Hello Amit,

Do you know if there is a way to check the device's registers to see if the DMA Transfer has completed? That is all I really need to know. Really I would like to know whether there is any other way besides using the USBDMAChannelIntStatus function to check if a configured DMA transfer is complete.

Cheers,

Bryan

When a DMA channel completes, XFERMODE  (bits 0,1 & 2) of the associated DMA channel control word go to zero. This is a direct indication that the DMA channel has stopped. In document number spms433b this is detailed on page 708.

Is that any use to you Bryan?

Oops.... I read TM4C1294.