QSPI boot failed on EVB Gen3 with SDK BSP39 and HSE v0.2.22 firmware

[hittzt]

Hi,

Recently, I tested qspi boot on EVB Gen3 board with SDK BSP39 and HSE v0.2.22 firmware.

But when I try to write fip.s32 image into qspi flash, the following error shows:

  root@s32g399aevb3:~# flashcp fip.s32-qspi /dev/mtd0
  While erasing blocks from 0x00000000-0x00180000 on /dev/mtd0: Connection timed out

And the whole test log is attached.

Would you please help to check the issue and give the reason?

Thanks,

Zhantao

[Daniel-Aguirre]

Hi,

We apologize for the delay from our side.

We have received the following update:

"

The root cause of this issue is bad hardware connection of the EVB board. I checked the configuration of EVB and RDB. They are the same. and I can successfully flash image to QSPI by flashcp command. So please check the connection between S32G and EVB board.

"

Please, let us know.

[hittzt]

Hi,

Woud you please help to give your EVB board's version or revision, our EVB platform board's REV info as below:

SCH-30081 REV B

700-30081 REV B

and the daughter board's CPU is S32G3.

The old EVB board revisions may be one of the reasons of the issue, but there is no way to get latest EVB boards.

Thanks,

Zhantao

[Daniel-Aguirre]

Hi,

We have received the following update:

"

The version of EVB3 used is SPF-50784_B.

The issue is caused by the EVB sockets, so it has no relation to the EVB version.

"

Please, let us know.

[Daniel-Aguirre]

Hi,

Thanks for your update. The revision you are providing seems to be related to the PLATEVB platform (or the motherboard), the daughter board should be the EVB one, can you help us share the revision of your daughter board?

Still, we will ask for which the version that was used for this situation.

Please, let us know.

[hittzt]

Hi,

I checked the daughter board's revision as following:

    SCH-32170 REV B2

    700-32170 REV A4

Thanks,

Zhantao

[hittzt]

Hi,

Thanks for you update.

I will try to test it again later as you suggested.

Thanks,

Zhantao

[Daniel-Aguirre]

Hi,

Since the EVB3 platform uses the provided heatsink+plate to provide the required pressure so that the different balls of the S32G3 touch the socket pads, can you help us confirm that you are able to run the "Qspi_Ip_Example_S32G399A_M7" example under S32DS?

Just to discard that the problem is not related to the chip not being able to sit on all the pads adequately. Also, if you have any RDB3 board available, since the chip is soldered, you should not see the pads problem, to confirm that indeed the problem is related to the BSP or HSE-FW.

We apologize for the inconvenience this may cause.

Please, let us know.

[hittzt]

Hi,

This issue not happens with BSP38 image, and just comes out after updated to BSP39 image.

And also there is no this issue on RDB3 board, so I think the issue should be caused due to the BSP39 update, especially in u-boot update.

Thanks,

Zhantao

[Daniel-Aguirre]

Hi,

We have received the following update:

"Could you please check if they have secure boot enabled."

Can you help us confirm if you have secure boot enabled under the scenario being described?

Please, let us know.

[hittzt]

Hi,

The secure boot is not enabled, but HSE firmware are built into the related fip.s32,

will it affect the qspi operation?

As the previous attached log shows, it is the qspi erasing blocks timeout, which casuing qspi boot failure.

Thanks,

Zhantao

[hittzt]

To make it more clear, I just apped the following two lines in project's conf/local.conf:

DISTRO_FEATURES:append = " hse"
NXP_FIRMWARE_LOCAL_DIR = "<path to HSE 0.2.22 firmware for g3>"

Thanks,

Zhantao

[Daniel-Aguirre]

Hi,

Thanks for your feedback. With this, we will continue looking into the behavior you are seeing.

Once we have any update, we will let you know.

We apologize for any delay.

Please ,let us know.

[Daniel-Aguirre]

Hi,

Thanks for your update. With this information, we will verify within the internal team if this behavior was already detected or not.

We apologize for any delay from our side.

Please, let us know.