AM6442: AM6442: sbl_ospi_linux fast boot breaks R5F0_0 EtherCAT discovery when Linux/A53 boot code is enabled

Part Number: AM6442
Other Parts Discussed in Thread: SYSCONFIG

Dear TI Team,

We are encountering a critical issue where the real-time networking peripherals stop functioning whenever the A53/Linux boot flow is enabled within the SBL.

Scenario A – Standalone RTOS Boot (Working)

When we comment out the Linux image processing logic inside the SBL (App_loadLinuxImages() and App_runLinuxCpu()), both R5F0_0 and R5F1_0 boot successfully from OSPI flash.

In this configuration:

• Ethernet PHYs achieve proper physical link-up.

• The PLC is able to successfully discover the EtherCAT subdevice.

• EtherCAT communication functions as expected.

Scenario B – Linux Boot Enabled (Failure)

When the Linux parsing/jump logic is enabled in the SBL, U-Boot and Linux boot completely successfully on the A53.

However:

• EtherCAT discovery fails entirely from the PLC side.

• R5F1_0 application still appears to boot and run.

• Physical Ethernet link behavior is inconsistent/non-functional from the EtherCAT perspective.

Root Cause Investigation Already Performed

To isolate the issue, we have already verified and ruled out the following common conflict scenarios:

1. Linux/Kernel Hardware Resource Conflict (Ruled Out)

We booted Linux in an isolated configuration where the k3-r5-rproc driver is completely disabled.

Verification:

• /sys/class/remoteproc/ is completely empty.

• Live DT status checks confirm Linux does not own or access the relevant peripherals:

bus@f4000/i2c@20000000/status     -> disabled
bus@f4000/icssg@30080000/status   -> disabled

This confirms Linux is fully passive with respect to:

• ICSSG1

• I2C0

• Related EtherCAT peripherals

Despite this, EtherCAT discovery still fails whenever the Linux boot flow is enabled in SBL.

2. UART Console/Register Conflict (Ruled Out)

To avoid UART0 register conflicts with the A53/Linux console:

• UART logging has been completely disabled in SysConfig for both the R5F0_0 and R5F1_0 applications.

Could you please suggest what additional areas we should investigate or whether there are known dependencies when combining EtherCAT RTOS cores with A53/Linux boot through SBL?

We would appreciate any guidance or debugging recommendations.

Best regards,
Sonu Kumar