We are getting some errors when running the initrd flashing method with the UEFI bootloader compiled from source. When we use the binary bundled with L4T 35.6.1 it works fine. We would like to try to compile the exact same version as is bundled from source to rule out issues with our build setup.
The combos page does not mention 35.6.1 at all, could you tell us which git commit was used or tag it in the repository? (Combos · NVIDIA/edk2-nvidia Wiki · GitHub)
Thank you for your reply. We are using a custom carrier board and a BSP with a few modifications. I’ll try to build r35.6.0 (before we were using r35.6.0-updates). If that doesn’t work I’ll write a proper summary here of the error we face. It’s interesting because when flashing through flash.sh the system works and boots fine. It’s the recovery mode boot for the initrd flashing that seems to cause issues.
The suggested tag did not fix the problem. The error we are getting is:
��I/TC: Reserved shared memory is disabled
I/TC: Dynamic shared memory is enabled
I/TC: Normal World virtualization support is disabled
I/TC: Asynchronous notifications are disabled
ERROR: MPIDR 0x80000000: exception reason=0 syndrome=0xbe000000
ERROR: **************************************
ERROR: RAS Error in L2, ERRSELR_EL1=0x200:
ERROR: Status = 0xfc006612
ERROR: IERR = SCF to L2 Slave Error Read: 0x66
ERROR: SERR = Error response from slave: 0x12
ERROR: Overflow (there may be more errors) - Uncorrectable
ERROR: Uncorrectable (this is fatal)
ERROR: MISC0 = 0x80000000400000
ERROR: MISC1 = 0x20240000003
ERROR: ADDR = 0x3270000
ERROR: **************************************
ERROR: **************************************
ERROR: RAS Error in SCF_IOB, ERRSELR_EL1=0x401:
ERROR: Status = 0xf4009604
ERROR: IERR = CBB Interface Error: 0x96
ERROR: SERR = Assertion failure: 0x4
ERROR: Uncorrectable (this is fatal)
ERROR: MISC0 = 0x40
ERROR: MISC1 = 0xe64a4445e1
ERROR: ADDR = 0x3270000
ERROR: **************************************
ERROR: RAS error handled!
ERROR: sdei_dispatch_event returned -1
ERROR: MPIDR 0x80000000: exception reason=0 syndrome=0xbe000000
ERROR: **************************************
ERROR: RAS Error in L2, ERRSELR_EL1=0x200:
ERROR: Status = 0xfc006512
ERROR: IERR = SCF to L2 Slave Error Write: 0x65
ERROR: SERR = Error response from slave: 0x12
ERROR: Overflow (there may be more errors) - Uncorrectable
ERROR: Uncorrectable (this is fatal)
ERROR: MISC0 = 0x80000000200000
ERROR: MISC1 = 0x20ac0000004
ERROR: ADDR = 0x3270000
ERROR: **************************************
ERROR: **************************************
ERROR: RAS Error in SCF_IOB, ERRSELR_EL1=0x401:
ERROR: Status = 0xfc009604
ERROR: IERR = CBB Interface Error: 0x96
ERROR: SERR = Assertion failure: 0x4
ERROR: Overflow (there may be more errors) - Uncorrectable
ERROR: Uncorrectable (this is fatal)
ERROR: MISC0 = 0x40
ERROR: MISC1 = 0x12a4a404221
ERROR: ADDR = 0x3270000
ERROR: **************************************
To make support easier and isolate the problem I’m now switching to a clean BSP (35.6.1) and a Xavier NX devkit.
When I copy our build of uefi_jetson.bin and BOOTAA64.efi into the BSP, and try to boot into the initrd with the command sudo ./tools/kernel_flash/l4t_initrd_flash.sh --initrd jetson-xavier-nx-devkit mmcblk0p1 the same error happens. I have attached the serial output and flashing log. Could you please try to build r35.6.0-updates from source, and try the same on your side?
Please treat this with priority, because we are doing this upgrade such that we can handle the PCN with new memory chips in our production. We are running out of jetsons with the old memory chips.
We can reproduce the RAS error as yours on the devkit.
We are working on release the correct UEFI binary for L4T r35.6.1 and will check if there’s still RAS error after updating UEFI binary.