Jetson NX based custom designed board crashes during iPerf test

Hi! I am using a custom designed board based on Jetson NX. The board runs L4T 32.4.3 and has Qualcomm wireless chipset connected via SDIO interface.

I ran into an odd scenario where the board crashes when a test is run with following steps:

Step 1. Connect the custom board and a host PC to the same wireless network (5GHz).
Step 2. Set the PC as server and the board as client.
Step 3. Run the iPerf test.
Step 4. When the test is running, “cat” a file(apx 100MB in size) on the custom board.

Result:

  1. Throughput in iPerf drops continuously with higher packet loss percentage.
  2. Board hangs and reboots.

I have attached the logs as from the point of crash as well crashlogs (25.3 KB).

Can someone point me to what could cause this behaviour?

Did you ever configure any device tree setting for this chipset?

Hi. Below is the device tree setting we use. We recently added “nvidia,disable-rtpm;” to avoid clock disabling.

sdhci_wifi: sdhci@3440000{
compatible = “nvidia,tegra194-sdhci”;
dma-coherent;
max-clk-limit = <50000000>;
bus-width = <0x4>;
cap-mmc-highspeed;
cap-sd-highspeed;
sd-uhs-sdr104;
sd-uhs-sdr50;
sd-uhs-sdr25;
sd-uhs-sdr12;
mmc-ddr-1_8v;
mmc-hs200-1_8v;
only-1-8-v;
non-removable;
nvidia,disable-rtpm;
nvidia,vqmmc-always-on;
status = “okay”;
/delete-property/ cd-inverted ;
};

Just a question that is your board able to run jetson nano? Actually, we don’t have any SDIO wifi support case on xavier modules.

Also, is this issue every affected by that “nvidia,disable-rtpm”?

Hi. I need to check if I can use a nano SOM in place of NX.

With “nvidia,disable-rtpm;” change, some boards crash (not all the time) and some boards don’t when following the steps as described in the topic.

Can you please give some pointers on what is happening from this crash log?

[ 527.426247] mmc1: Timeout waiting for hardware interrupt.
[ 527.426391] sdhci: =========== REGISTER DUMP (mmc1)===========
[ 527.426499] sdhci: Sys addr: 0x00000000 | Version: 0x00000505
[ 527.426600] sdhci: Blk size: 0x00007002 | Blk cnt: 0x00000000
[ 527.426707] sdhci: Argument: 0x965ffc02 | Trn mode: 0x00000003
[ 527.426808] sdhci: Present: 0x01db0000 | Host ctl: 0x00000013
[ 527.426909] sdhci: Power: 0x0000000f | Blk gap: 0x00000000
[ 527.427011] sdhci: Wake-up: 0x00000000 | Clock: 0x00000007
[ 527.427112] sdhci: Timeout: 0x0000000b | Int stat: 0x00000000
[ 527.427214] sdhci: Int enab: 0x02ff000b | Sig enab: 0x02fc000b
[ 527.427316] sdhci: AC12 err: 0x00000000 | Slot int: 0x00000000
[ 527.427416] sdhci: Caps: 0x3f6cd08c | Caps_1: 0x18002f73
[ 527.427707] sdhci: Cmd: 0x0000353a | Max curr: 0x00000000
[ 527.428167] sdhci: Host ctl2: 0x00003008
[ 527.428464] sdhci: ADMA Err: 0x00000000 | ADMA Ptr: 0x00000000ffefe420
[ 527.428978] sdhci: ===========================================
[ 527.594029] CPU1: SError detected, daif=1c0, spsr=0xc000c5, mpidr=80000001, esr=be000000
[ 527.594079] CPU:0, Error:CBB-NOC@0x2300000,irq=478
[ 527.594086] **************************************
[ 527.594088] * For more Internal Decode Help
[ 527.594090] * http://nv/cbberr
[ 527.594091] * NVIDIA userID is required to access
[ 527.594093] **************************************
[ 527.594094] CPU:0, Error:CBB-NOC
[ 527.594096] Error Logger : 0
[ 527.594104] ErrLog0 : 0x80000008
[ 527.594107] Transaction Type : WR - Write, Incrementing
[ 527.594109] Error Code : SLV
[ 527.594111] Error Source : Target
[ 527.594113] Error Description : Target error detected by CBB slave
[ 527.594127] AXI2APB_4 bridge error: RDFIFOF - Read Response FIFO Full interrupt
[ 527.594129] AXI2APB_5 bridge error: SFIFONE - Status FIFO Not Empty interrupt
[ 527.594132] AXI2APB_5 bridge error: TIM - Timer(Timeout) interrupt
[ 527.594132] Packet header Lock : 0
[ 527.594134] Packet header Len1 : 0
[ 527.594135] NOC protocol version : version >= 2.7
[ 527.594137] ErrLog1 : 0x319c13
[ 527.594139] ErrLog2 : 0x0
[ 527.594140] RouteId : 0x319c13
[ 527.594143] InitFlow : ccroc_p2ps/I/ccroc_p2ps
[ 527.594145] Targflow : axis_satellite_axi2apb_p2pm/T/axis_satellite_axi2apb_p2pm
[ 527.594147] TargSubRange : 78
[ 527.594148] SeqId : 0
[ 527.594150] ErrLog3 : 0x2f
[ 527.594152] ErrLog4 : 0x0
[ 527.594170] Address : 0x344002f – /sdhci@3440000 + 0x2f
[ 527.594172] ErrLog5 : 0x489f851
[ 527.594173] Non-Modify : 0x1
[ 527.594175] AXI ID : 0x9
[ 527.594177] Master ID : CCPLEX
[ 527.594179] Security Group(GRPSEC): 0x7e
[ 527.594181] Cache : 0x1 – Device
[ 527.594183] Protection : 0x2 – Unprivileged, Non-Secure, Data Access
[ 527.594185] FALCONSEC : 0x0
[ 527.594187] Virtual Queuing Channel(VQC): 0x0
[ 527.594189] **************************************
[ 527.594320] CPU:0, Error:CBB-NOC@0x2300000,irq=478
[ 527.594323] **************************************

This log may not have direct connection. When sdmmc crashed, it will always give out that register dump.

The only error here is “mmc1: Timeout waiting for hardware interrupt”.

The rest CPU error is possibly due to we still try to access sdio device after the link is down.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.