DRA829V: SCICLIENT_ROM_MSG_CERT_AUTH_FAIL

Bill Morgan said:

ti-fs-firmware-j721e-gp.bin

That file isn't signed.  It needs to be signed even on the GP device.  You can use <pdk>/packages/ti/drv/sciclient/soc/V1/tifs.bin instead.

Using tifs.bin I'm still getting SCICLIENT_ROM_MSG_CERT_AUTH_FAIL.

Here is the response on line 145 of sciclient_rom.c:

  pLocalRespHdr = 0x2A485004 -> (
    type = 34821 = 0x8805,
    host = 255 = 0xFF,
    seq = 0 = 0x0,
    flags = 16777215 = 0x00FFFFFF,
    payload = 24 = 0x18)

Bill - I've asked a colleague to help out.  You should hear from her soon.

Bill,

There seems to be some issue with ti-fs-firmware-j721e-gp.bin - we are checking on that.

I tried \ti-processor-sdk-rtos-j721e-evm-07_02_00_06\pdk_jacinto_07_01_05_14\packages\ti\drv\sciclient\soc\V1\tifs.bin and its functional. Can you give that an try?

-Sujith

This is the sha256 of the tifs.bin. It still does not load.

$ sha256sum.exe /d/x/git/ti-processor-sdk-rtos-j721e-evm-07_02_00_06/pdk_jacinto_07_01_05_14/packages/ti/drv/sciclient/soc/V1/tifs.bin
b541d0e475fb54dd6d1ce4dd8e6d2c4c678e50cb79e3ecd5fd6098b4a403072b */d/x/git/ti-processor-sdk-rtos-j721e-evm-07_02_00_06/pdk_jacinto_07_01_05_14/packages/ti/drv/sciclient/soc/V1/tifs.bin

I'm still getting SCICLIENT_ROM_MSG_CERT_AUTH_FAIL. 

Is there additional setup required prior to calling Sciclient_loadFirmware?

I was off the assumption that TI's SBL was being used. The sha256 matches with the firmware that works fine for me

b541d0e475fb54dd6d1ce4dd8e6d2c4c678e50cb79e3ecd5fd6098b4a403072b pdk_jacinto_07_01_05_14/packages/ti/drv/sciclient/soc/V1/tifs.bin

Can you please describe the steps that you are doing?

Regards, Sujith

copy tifs.bin from QSPI NOR into a buffer then call Sciclient_loadFirmware

At what point are you trying to do this? Are you attempting to write a secondary-boot-loader or something similar?

Also, have you ensured contents of tifs is intact (post flashing and reading from QSPI)

Regards, Sujith

Sujith said:

At what point are you trying to do this? Are you attempting to write a secondary-boot-loader or something similar?

This is being done from our custom boot loader running on MCU R5.

Sujith said:

Also, have you ensured contents of tifs is intact (post flashing and reading from QSPI)

yes I downloaded the tifs data via jtag post flash and verified it matches the tifs.bin that was written.

What conditions are required to get a SCICLIENT_ROM_MSG_CERT_AUTH_PASS?

Sujith said:

The sha256 matches with the firmware that works fine for me

Are you testing with GP or HS device? Does it matter? I'm using GP.

Bill,

Can you ensure that the buffer used to hold tifs is allocated from MCU_MSRAM?

Regards, Sujith

Sujith said:

Can you ensure that the buffer used to hold tifs is allocated from MCU_MSRAM?

The buffer to hold tifs is located at 0x41c1bd38. 

Sujith -- can you check the address used in your working example?

Bill -- can you try aligning that address to a 4KB boundary?  That's purely a guess on my part that alignment might play into it, but hopefully a quick thing to try.

Brad Griffis said:

Bill -- can you try aligning that address to a 4KB boundary?  That's purely a guess on my part that alignment might play into it, but hopefully a quick thing to try.

The address is now : 0x41c1d000, it still didn't work.

Bill -- It looks like the exact address being used in the TI code is 0x41C80000.  I don't think that should make a difference, but can you please try that to be sure?

Unfortunately this has taken too long and I've been taken off this issue to work on other stuff.

Thanks for the update.  Sorry, I wish we could have helped you close the issue sooner.