使用virtex4，xps_ll_temac， 进行网络处理遇到一个问题： 使用uboot，下面的函数在没有使用数据cache时，报文接收正确， 但是当使用数据cache时，接收异常，打印出调试信息 Received a frame of 67174464 bytes
Received a frame of 67174464 bytes
Received a frame of 67174464 bytes
明显是接收的长度有问题， edk给的sgdma的例子都是使用中断dma方式，在中断中判断 接收状态， 是否使能数据cache后，必须判断中断状态，sgdma的状态才是准确的？ int
eth_rx(void)
{
u32 RecvFrameLength;
int Status;
XLlDma_Bd *RxBdPtr;
u32 BdCount;
XLlDma_BdRing *RxRingPtr = &XLlDma_mGetRxRing(&LlDma); BdCount = XLlDma_BdRingFromHw(RxRingPtr, XLLDMA_ALL_BDS, &RxBdPtr); /* If there are any buffer descriptors ready to be received from the
* DMA then send the buffer up the network stack
*/
if (BdCount > 0) { RecvFrameLength = XLlDma_mBdRead(RxBdPtr, XLLDMA_BD_USR4_OFFSET); debug_printf("Received a frame of %d bytes\r\n", RecvFrameLength); NetReceive((uchar *)RxFrame, RecvFrameLength); /*
* Return the RxBD back to the channel for later allocation. Free the
* exact number we just post processed.
*/
Status = XLlDma_BdRingFree(RxRingPtr, 1, RxBdPtr);
if (Status != XST_SUCCESS) {
debug_printf("Error freeing up RxBD\n\r");
return 0;
} /*
* Allocate 1 RxBD. Note that TEMAC utilizes an in-place allocation
* scheme. The returned Bd1Ptr will point to a free BD in the memory
* segment setup with the call to XLlTemac_SgSetSpace()
*/
Status = XLlDma_BdRingAlloc(RxRingPtr, 1, &RxBdPtr);
if (Status != XST_SUCCESS) {
debug_printf("Error allocating RxBD");
return 0;
}
/*
* Setup the BD. The BD template used in the call to XLlTemac_SgSetSpace()
* set the "last" field of all RxBDs. Therefore we are not required to
* issue a XLlDma_Bd_mSetLast(Bd1Ptr) here.
*/
XLlDma_mBdSetBufAddr(RxBdPtr, &RxFrame);
XLlDma_mBdSetLength(RxBdPtr, sizeof(RxFrame));
XLlDma_mBdSetStsCtrl(RxBdPtr, XLLDMA_BD_STSCTRL_SOP_MASK | XLLDMA_BD_STSCTRL_EOP_MASK);
/*
* Enqueue to HW again so it can receive another frame
*/
Status = XLlDma_BdRingToHw(RxRingPtr, 1, RxBdPtr);
if (Status != XST_SUCCESS) {
debug_printf("Error committing RxBD to HW\n\r");
return 0;
}
return 1;
} else {
return 0;
}
}
| 
