HelloWorld on EVM6472

Hello Sajesh,

I can see that this requires CCSv4.2.0. I am already using CCSv4.1.2 that came with the EVM module. 

The OOBDemo_BIOS6 project builds and works with the current version of CCSv4.1.2. Please let me know if it is okay to update software baseline on v4.1.2 with the one you suggested.

Thanks,

Arun

Arun,

The BIOS MCSDK 1.0 release is a broad market production release intended for C6472, C6474, and C6457. Yes, please upgrade the software that came with the EVM module with BIOS MCSDK 1.0 and the corresponding CCS 4.2.0 releases.

To bring up the out-of-box demo that is a part of BIOS MCSDK 1.0, please follow the guide at:
http://processors.wiki.ti.com/index.php/BIOS_MCSDK_1.0_Getting_Started_Guide

To begin development with this release, please consult the users guide at:
http://processors.wiki.ti.com/index.php/BIOS_MCSDK_1.0_User_Guide

Raj

Hi Raj,

Can you tell me specifically what you mean by "please upgrade the software that came with the EVM module with BIOS MCSDK 1.0 and the corresponding CCS 4.2.0 releases."

What are the corresponding CCS4.2.0 releases?

Thank you very much,

Arun

Hi Raj,

Is there a CCSv4.2.0 limited version for EVM6472? Or should I purchase one(full-blown version)? I was kind of expecting to work with the CCS that came with the EVM for now.

Thanks,

Arun

Thank you very much for your prompt replies. It works now.

Arun

Hi Sajesh,

I am trying to run the UDP helloworld example on 6472 core B.  I am using CCSv4.2.1 & mcsdk_1_00_00_08.

Is this example only meant to run on core 0.? I modified the application code check to "if (platform_get_coreid() != 1)" to no avail.

Thanks,

Arun

Arun,

This example should be run on core 0 only.

Please see the link http://e2e.ti.com/support/embedded/f/355/p/86502/303985.aspx#303985 for more details.

Regards

Sajesh

Hi Sajesh,

Thanks for the link. I changed the nimu_eth.c to have it compare to core-B in a couple of places.

Now, in the application console log, I see:

Running on core 1

TCP/IP Stack 'Hello World!' Application

Registration of the EMAC Successful, waiting for link up ..
Network Added: If-1:10.21.41.128
Port 0 Link Status: 100Mb/s Full Duplex on PHY 24

PLATFORM_MAC_TYPE_EFUSE - I am unable to ping the core 1. I tried following another suggestion on adding an entry into the ARP table. I step through to see that the MAC address does not change in case of core 0/1. I tried adding this entry to ARP. Still ping fails.

PLATFORM_MAC_TYPE_EEPROM - ping fails on core 1.

I want to verify the UDP application works on all cores.

Thanks,

Arun

Arun,

Are you running your UDP on both core 0 and core 1 or on core 1 only? If you are running on core 1 only, you need to set the master core flag to TRUE for core 1 when calling emac_open() API in the nimu driver, and you also need to set the packet channle number allocated by the EMAC driver (in this case TX/RX channel number is 0)  when constructing the TX packet descriptor for Core 1 in the nimu driver.

Please refer to http://e2e.ti.com/support/embedded/f/355/p/86502/303985.aspx#303985 for more details.

Regards,

Hao.

Ho Hao,

Currently I am only on Core 1. It helped when I made the below change like you suggested.

    pkt_desc.PktChannel  = 0;

What do you suggest when running on all cores?

pkt_desc.PktChannel  = coreNum;  // leave this at coreNum???

Thanks,

Arun

Arun,

Please refer to EMAC driver API header, pdk_c64x_1_00_00_0x\packages\ti\drv\emac\emac_drv.h:

typedef struct EMAC_OPEN_CONFIG_INFO_tag
{
    Bool                            master_core_flag;
    /**< True: master core configuration                                    */
    Bool                            mdio_flag;
    /**< True: configure MDIO                                               */
    Uint32                          phy_addr;
    /**< Physical layer transceiver address mapped to the EMAC port
         valid address range is 0 - (EMAC_MAX_NUM_PHY_ADDR-1).
         if a valid address is specified, EMAC driver will check the link
         status only on the application specifid phy address; if an invalid
         phy_addr is specified, EMAC driver will search the phy address that
         can be used by itself                                              */
    Bool                            loop_back;
    /**< True: packets loop back                                            */
    Uint32                          num_of_tx_pkt_desc;
    /**< Total # of pkt desc initialized for all the TX chans in TX queues  */
    Uint32                          num_of_rx_pkt_desc;
    /**< Total # of pkt desc initialized for all the RX chans in RX queues  */
    Uint32                          max_pkt_size;
    /**< Maximum size of the packet in bytes                                */
    Uint32                          num_of_chans;
   /**< Total number of TX/RX channels configured for this core             */
    EMAC_CHAN_MAC_ADDR_T*           p_chan_mac_addr;
   /**< Application needs to provide the MAC address configurations;
        EMAC driver will allocate the channels if available and
        return the channel info to the application                          */
    EMAC_RX_PKT_CALLBACK_FN_T*      rx_pkt_cb;
    /**< EMAC RX call back function                                         */
    EMAC_ALLOC_PKT_CALLBACK_FN_T*   alloc_pkt_cb;
    /**< EMAC allocate packet call back function                            */
    EMAC_FREE_PKT_CALLBACK_FN_T*    free_pkt_cb;
    /**< EMAC free packet call back function                                */

} EMAC_OPEN_CONFIG_INFO_T;

NIMU driver needs to save the channel number allocated and use it for the TX packet descriptor.

Regards,

Hao.

Hi Hao, ive posted my queries relating to helloWorld in NDK 2.2 before too.

Im facing a issue here:

I too am currently using the Spectrum Digital TMS320TCI6488 EVM board, and facing issues like the helloWorld application hanging around the isFpgaReady() code.

How do i change the platform_utils library for this EVM board shipped to me. I believe the Platform_utils library built in MCSDK 1.0 is for evm light.

How do i now get to run the helloWorld NDK application on this board ? Pls help.

Also when i comment some part of the code under EVM_init() which is used to initialize the FPGA, the program gets through, but i cannot PING the EVM board.

The console prints:

TCP/IP Stack 'Hello World!' Application

emac_init: core 0, port 0, total number of channels/MAC addresses: 1/1

MAC addresses configured for channel 0:

00-24-BA-3A-0E-F2

SGMII reset successful........

SGMII config successful........

emac_open core 0 port 0 successfully

Registration of the EMAC Successful, waiting for link up ..

Network Added: If-1:192.168.128.201

Port 0 Link Status: 100Mb/s Full Duplex on PHY 1

Please do help.

Hi Arun,

Were you able to ping your evm board from an external host machine by placing the NDK under Core 0.

Which example are you running...Please let me know.

Im using MCSDK 1.0 having NDK 2.2 and using the helloWorld applicaiton, but i fail to PING the ethernet interface from external host machine.

Hao...Do put in your queries and let me know if there is something like EVM lite since the platform libraries built have the name: platform_utils_evmc6474l.lib.

What does that "l" stand for since the platform_init() fails on my EVM board by Spectrum Digital TCI6488.

Pls help.

Santosh,

Please see my post in another thread regarding this problem:

http://e2e.ti.com/support/embedded/f/355/p/86502/315157.aspx#315157

"l" means the low cost lite version of the EVM which only has one DSP device on the board.

Regards,

Hao.