Issue with LS1046A Custom Board Under Temperature Stress Test

[hyeonbin]

Hello,

I have designed a custom board using the LS1046A processor and conducted a temperature chamber test. However, when the board temperature reaches around 55°C and the CPU core temperature is approximately 70°C, the board's ASLEEP LED blinks periodically at about 0.5-second intervals, and the system stops functioning.

Additionally, when I connect a console cable, it appears that the system is attempting to reboot repeatedly, as shown in the log below. However, when the temperature decreases, the system resumes normal operation.

I used the LS1046A chip with the highest available temperature rating and referenced the FRWY board for my design.

Are there any debugging steps you would recommend?

(Log attached below)

[2025-03-17 11:39:33.155] NOTICE: Fixed DDR on board
[2025-03-17 11:39:33.592]
[2025-03-17 11:39:33.592] NOTICE: 4 GB DDR4, 64-bit, CL=15, ECC on
[2025-03-17 11:39:33.592] NOTICE: BL2: v1.5(release):LSDK-20.12-dirty
[2025-03-17 11:39:33.592] NOTICE: BL2: Built : 10:47:11, Feb 7 2023
[2025-03-17 11:39:33.867] NOTICE: BL2: Booting BL31NOTICE: BL31: v1.5(release):LSDK-20.12-dirty
[2025-03-17 11:39:33.867] NOTICE: BL31: Built : 10:47:11, Feb 7 2023
[2025-03-17 11:39:33.867] NOTICE: Welcome to LS1046 BL31 Phase
[2025-03-17 11:39:33.867]
[2025-03-17 11:39:33.867]
[2025-03-17 11:39:33.867] U-Boot 2020.04-dirty (Feb 06 2023 - 22:20:05 +0900)
[2025-03-17 11:39:33.946]
[2025-03-17 11:39:33.946] SoC: LS1046A Rev1.0 (0x87070110)
[2025-03-17 11:39:33.946] Clock Configuration:
[2025-03-17 11:39:33.946] CPU0(A72):1600 MHz CPU1(A72):1600 MHz CPU2(A72):1600 MHz
[2025-03-17 11:39:33.946] CPU3(A72):1600 MHz
[2025-03-17 11:39:33.946] Bus: 600 MHz DDR: 2100 MT/s FMAN: 700 MHz
[2025-03-17 11:39:33.946] Reset Configuration Word (RCW):
[2025-03-17 11:39:33.946] 00000000: 0c150010 0e000000 00000000 00000000
[2025-03-17 11:39:33.946] 00000010: 11338888 00400016 40025000 c1000000
[2025-03-17 11:39:33.946] 00000020: 00000000 00000000 00000000 00038800
[2025-03-17 11:39:33.946] 00000030: 20044100 24003000 00000096 00000001
[2025-03-17 11:39:33.946] Model: LS1046A FRWY Board
[2025-03-17 11:39:33.946] Board: LS1046AFRWY-SATURN-HURA, Rev: B, boot from QSPI
[2025-03-17 11:39:33.946] SD1_CLK1 = 100.00MHZ, SD1_CLK2 = 100.00MHZ
[2025-03-17 11:39:33.946] DRAM: 3.9 GiB (DDR4, 64-bit, CL=15, ECC on)
[2025-03-17 11:39:33.955] SEC0: RNG instantiated
[2025-03-17 11:39:34.016] select_i2c_ch_pca9547: Cannot find udev for a bus 0
[2025-03-17 11:39:34.016] Using SERDES1 Protocol: 4403 (0x1133)
[2025-03-17 11:39:34.016] Using SERDES2 Protocol: 34952 (0x8888)
[2025-03-17 11:39:34.016] NAND: 0 MiB
[2025-03-17 11:39:34.016] MMC: FSL_SDHC: 0
[2025-03-17 11:39:34.016] Loading Environment from SPI Flash... SF: Detected mt25qu512a with page size 256 Bytes, erase size 64 KiB, total 64 MiB
[2025-03-17 11:39:34.016] OK
[2025-03-17 11:39:34.016] EEPROM: Read failed.
[2025-03-17 11:39:34.016] In: serial
[2025-03-17 11:39:34.016] Out: serial
[2025-03-17 11:39:34.016] Err: serial
[2025-03-17 11:39:34.016] Net: SF: Detected mt25qu512a with page size 256 Bytes, erase size 64 KiB, total 64 MiB
[2025-03-17 11:39:34.016] Fman1: Uploading microcode version 106.4.18
[2025-03-17 11:39:34.088] PCIe1: pcie@3400000 Root Complex: no link
[2025-03-17 11:39:34.088] PCIe2: pcie@3500000 disabled
[2025-03-17 11:39:34.088] PCIe3: pcie@3600000 disabled
[2025-03-17 11:39:34.088] FM1@DTSEC5, FM1@DTSEC6, FM1@TGEC1
[2025-03-17 11:39:34.088] Error: FM1@TGEC1 address not set.
[2025-03-17 11:39:34.088] , FM1@TGEC2
[2025-03-17 11:39:34.088] Error: FM1@TGEC2 address not set.
[2025-03-17 11:39:34.088]
[2025-03-17 11:39:34.088] Hit any key to stop autoboot: 3 2 1 0
[2025-03-17 11:39:37.144] switch to partitions #0, OK
[2025-03-17 11:39:37.160] mmc0(part 0) is current device
[2025-03-17 11:39:37.176] Scanning mmc 0:1...
[2025-03-17 11:39:37.326] Scanning mmc 0:2...
[2025-03-17 11:39:37.367] Found U-Boot script /ls1046afrwy_boot.scr
[2025-03-17 11:39:38.551] 17635632 bytes read in 1159 ms (14.5 MiB/s)
[2025-03-17 11:39:38.551] do_fpgaload: fpga_load 17635632 bytes in 2591 ms
[2025-03-17 11:39:41.181] 968 bytes read in 17 ms (54.7 KiB/s)
[2025-03-17 11:39:41.181] ## Executing script at 80000000
[2025-03-17 11:39:43.519] 35828224 bytes read in 2333 ms (14.6 MiB/s)
[2025-03-17 11:39:43.558] 31419 bytes read in 17 ms (1.8 MiB/s)
[2025-03-17 11:39:43.798] NOTICE: Fixed DDR on board
[2025-03-17 11:39:44.095] NOTICE: Fixed DDR on board
[2025-03-17 11:39:44.451] NOTICE: Fixed DDR on board
[2025-03-17 11:39:44.861] NOTICE: Fixed DDR on board
[2025-03-17 11:39:45.243] NOTICE: Fixed DDR on board
[2025-03-17 11:39:45.658] NOTICE: Fixed DDR on board
[2025-03-17 11:39:46.047] NOTICE: Fixed DDR on board
[2025-03-17 11:39:46.464] NOTICE: Fixed DDR on board
[2025-03-17 11:39:46.878] NOTICE: Fixed DDR on board
[2025-03-17 11:39:47.282] NOTICE: Fixed DDR on board
[2025-03-17 11:39:47.704] NOTICE: Fixed DDR on board
[2025-03-17 11:39:48.103] NOTICE: Fixed DDR on board
[2025-03-17 11:39:48.530] NOTICE: Fixed DDR on board
[2025-03-17 11:39:48.939] NOTICE: Fixed DDR on board
[2025-03-17 11:39:49.346] NOTICE: Fixed DDR on board
[2025-03-17 11:39:49.754] NOTICE: Fixed DDR on board
[2025-03-17 11:39:50.162] NOTICE: Fixed DDR on board
[2025-03-17 11:39:50.575] NOTICE: Fixed DDR on board
[2025-03-17 11:39:50.991] NOTICE: Fixed DDR on board
[2025-03-17 11:39:51.415] NOTICE: Fixed DDR on board
[2025-03-17 11:39:51.825] NOTICE: Fixed DDR on board
[2025-03-17 11:40:31.984] NOTICE: Fixed DDR on board
[2025-03-17 11:40:32.374] NOTICE: Fixed DDR on board
[2025-03-17 11:40:32.780] NOTICE: Fixed DDR on board
[2025-03-17 11:40:33.190] NOTICE: Fixed DDR on board

[yipingwang]

The thermal management function is based on TMU (Thermal Monitoring Unit).
The driver sets two thresholds for management function. If the CPU temperature crosses the first one (75 C for LS2080. 85
C for other platforms), the driver will trigger CPU frequency limitation auto-scaling according to the temperature trend; If the
CPU temperature crosses the second one (85 C for LS2080, 95 C for other platforms, critical for core) the driver will shut down the system.