MI_DmaCopy*

C Specification

#include <nitro/mi.h>
  
  void MI_DmaCopy32( u32 dmaNo, const void* src, void* dest, u32 size );
  void MI_DmaCopy16( u32 dmaNo, const void* src, void* dest, u32 size );
  
  void MI_DmaCopy32Async( u32 dmaNo, const void* src, void* dest, u32 size,
                                               MIDmaCallback callback, void* arg );
  void MI_DmaCopy16Async( u32 dmaNo, const void* src, void* dest, u32 size,
                                               MIDmaCallback callback, void* arg );

Arguments

dmaNo	DMA channel used.
`src`	The transfer source address.
`dest`	The transfer destination address.
`size`	Transfer size.
`callback`	Callback when DMA ends.
`arg`	Callback argument when DMA ends.

Return Values

None.

Description

Uses DMA to copy.

MI_DmaCopy16() and MI_DmaCopy16Async() copy in 16-bit units. Both the transfer source address and the transfer destination address must be 2-byte aligned. MI_DmaCopy32() and MI_DmaCopy32Async() copy in 32-bit units. Both the transfer source address and the transfer destination address must be 4-byte aligned.

MI_DmaCopy16() and MI_DmaCopy32() wait within the function until DMA completes. MI_DmaCopy16Async() and MI_DmaCopy32Async() call a callback function when DMA ends. The callback, callback, is a MIDmaCallback type (a void type function which takes one void* argument). callback is called from the system DMA interrupt handler and therefore is called while interrupts are prohibited.

It is necessary to understand how to use Async-type APIs in order to obtain maximum functionality.

The following is an example of a function name Call_BackGroundJob_with_DMA() that is placed within TCM.

The following processes are performed:

Data transfer using DMA within the function begins.
When the DMA ends, TRUE is entered in isDmaFinished in callback().
isDmaFinished is polled as an end flag and BackGroundJob() can be called continuously until DMA ends.

However, the following must be considered to have this process proceed as intended:

If isDmaFinished is not taken inside TCM, then stalling will occur during the while(!isDmaFinished) check.
Stalling will occur if BackGroundJob() accesses the ARM bus.
In many cases, the stack will also need to be allocated within TCM. (The SDK default has the stack being secured within DTCM.)
If this function is located in main memory, stalling will occur when DMA starts. Therefore, BackGroundJob() cannot be called continuously.

//--- sample code ( This code must be in ITCM )
void Call_BackGroundJob_with_DMA(void)
{
    vu32 isDmaFinished = FALSE;

    MI_DmaCopy32Async( dmaNo, srcArea, destArea, dataSize, callback, (void*)&isDmaFinished );
    while( !isDmaFinished ){ BackGroundJob(); }

}

void callback( void* arg )
{
    *(vu32*)arg = TRUE;
}

Internal Operation

Uses the following IO registers: DMAn source register (0x40000B0 + 12n), DMAn destination register (0x40000B4 + 12n), and DMAn control register (0x40000B8 + 12n). n is the DMA channel used.)

Revision History

03/08/2005 Standardized the Japanese term for "interrupt".
12/22/2004 Added a description regarding callback.
12/01/2003 Initial version

CONFIDENTIAL