About the last commit:

The function usGenerateProtocolChecksum() now needs a parameter that indicates the length of a packet.
The function xNetworkInterfaceOutput() called a function called xCheckLoopback(0, which is not defined. Added this parameter to vGMACGenerateChecksum() as well.

/bot run checks

The EMAC task was started with a fixed high task-priority. In other drivers a macro is defined:

  #define niEMAC_HANDLER_TASK_PRIORITY     configMAX_PRIORITIES - 1

Users can override this macro in FreeRTOSIPConfig.h.

A second change: The driver didn't work on SAME70 Rev-B. The reason was that mentioned part has 6 priority queues in its EMAC.
Those queues are also initialised now:

    /* Note that SAME70 REV B had 6 priority queues. */
    gmac_set_tx_priority_queue(p_hw, (uint32_t)&gs_tx_desc_null, GMAC_QUE_3);
    gmac_set_tx_priority_queue(p_hw, (uint32_t)&gs_tx_desc_null, GMAC_QUE_4);
    gmac_set_tx_priority_queue(p_hw, (uint32_t)&gs_tx_desc_null, GMAC_QUE_5);

The changes are tested on both SAME70 Rev-A and Rev-B.

Hein

About the commit "phyHandling.c : Make the right choice between slow and frequent polling of the Ethernet PHY"

When the PHY Link Status is low, one wants to poll the status frequently. But once the PHY is connected, frequent polling of the PHY would be disturbing, and a slow polling is preferred.

In order to check the Link Status, the mask was compared with a const phyBMSR_LINK_STATUS, which is incorrect, mentioned macro is a register bit.

-    if( ( pxPhyObject->ulLinkStatusMask & phyBMSR_LINK_STATUS ) != 0 )
+    if( ( pxPhyObject->ulLinkStatusMask & ( ulBitMask >> 1 ) ) != 0 )

Closing this PR #57 because a new PR #78 (v2) will take it over.
Hein