AM5728: Kernel panic

Hi Rio Chan,

From what I understand, customer is using AM572x custom board. Do they use AM57x PSDK Linux v5.03? Can they reproduce this issue on AM572x board (EVM or IDK)? How many custom boards they have, on do they observe the same for all custom boards?

Yes, that might be RAM issue. You can run the memory diagnostic test (mem_test) to check if your RAM is correct.

External memory (DDR): DDR timing and leveling setting can be checked out using mem_test

Regards,
Pavel

Hi Pavel:

Thanks for quickly replying.

#1. This issue won't be happening on the TI EVM.

#2. They are using SDK5.02 based.

#3. However, the kernel crash will happen very quickly.

Sometimes, the kernel will be panic during the kernel boot stage. (Means: System crash before the Root user logging).

Sometimes, the kernel will be okay to boot completely, but, it will crash right after the user logging.

In this kind of condition, how to perform the memtest?

Any other way?

BR Rio

Rio,

Yes, I think you can run the mem_test diagnostic test In this kind of conditions.

You might also use u-boot memory tester and/or linux memtester.

For u-boot memory test:

u-boot/doc/README.memory-test
u-boot/common/memsize.c

For linux memtester:

Check also below docs:

There might be also some voltage/power HW issue. Please go through below docs also:

Regards,
Pavel

Hi Pavel:

Sorry, after double checking the DDR, the customer is using the same DDR that TI EVM used:

KINGSTON DDR3L 4Gb/256MX16 D2516EC4BXGGB

However, they don't have the JTAG on their PCB.

So, to download the RTOS for doing Mem_Tester is not possible.

Any better way to solve / find the root cause?

BR Rio

Hi Pavel:

Thanks for the hint.

Customer has tested the DDR during u-boot by this command:

ddr test 0x80000000 0xff000000

The memory will crash.

Is this related to the Address over 2G?

What's the max Memory test address that TI AM5728 EVM can reach?

Depends on the DDR Memory size, right? 

BR Rio

Hi Rio,

By default only 2GBytes of RAM can be access/tested in u-boot. If you have 4GBytes RAM, the upper 2GBytes are accessible only by Cortex-A15 MPU via LPAE. LPAE is enabled by default in linux kernel only.

Check below e2e thread for details:

Regards,
Pavel

Hi Pavel:

We double checked the customer's PCB, they are using the same with TI AM57 GP EVM DDR.

TI AM5728 GP EVM:

               KINGSTON D2516EC4BXGGB

CM’s PCB

               KINGSTON D2516EC4BXGGB

So, they are using the same size as 2GB DDR.

Any other possible root cause to cause this RAM fault issue while in the kernel?

BR Rio

Rio,

Rio Chan said:

We double checked the customer's PCB, they are using the same with TI AM57 GP EVM DDR.

TI AM5728 GP EVM:

               KINGSTON D2516EC4BXGGB

CM’s PCB

               KINGSTON D2516EC4BXGGB

So, they are using the same size as 2GB DDR.

Any other possible root cause to cause this RAM fault issue while in the kernel?

Even using the same external DDR3 chips as TI EVM, DDR settings need to be adjusted according to your custom board specific routing. But you will need JTAG interface for this.

Other possible root cause can be HW malfunction of your custom board. You can request schematics review through internal channels, please check below e2e threads for details:

Let me know if you have more questions.

Regards,
Pavel