i.MX8: random DRAM problems when using SCFW compiled with DEBUG and MONITOR enabled

[Niebel-TQ]

When using SCFW (based on 1.15 version of porting kit) compiled for our board with DEBUG and MONITOR enabled we saw random program crashes especially when using applications related to VPU or GPU.

When trying to isolate the problem we used a testing setup using glmark2-es-wayland and memtester.

We could reproduce this on several systems with 4 GB of RAM using:

```
glmark2-es-wayland --run-forever &
memtester 3G 100
```

After a while memtester spits out errors without a reproducible pattern. This happens with / without cooling. The same boards runs for hours with the same SCFW compiled without DEBUG.

@danny_john 

[Bio_TICFSL]

Hello,

Yes we saw this behavior in previous BSP, so please tell us which version of BSP are you using? This is a custom board or EVK? in order to reproduced please send us your code test to check it

Thanks

[Niebel-TQ]

- SCU based on 1.15 - other parts of BSP:

• linux-imx based on 5.10,
• atf-imx based 2.2
• yocto hardknott (meta-freescale + meta-tq)

Software runs on TQMa8QM SOM mounted on a custom Mainboard

[Niebel-TQ]

on SOM with 4GB RAM to reproduce
(needs up to 30 minutes)
```
glmark2-es2-wayland --run-forever &
memtester 3G 100
```
or (triggered after some minutes) 
```
glmark2-es2-wayland --run-forever &
stressapptest -M 2500 -s 3000 -C 4 -i 4 -m 4
```

CPU governernor: ondemand, SCU receives lots of requests to scale voltages for booth Cortex-A Clusters

[Bio_TICFSL]

Hello,

You need at least 6.6.3 BSP please download. BTW we are not able to see the issue with 5.15 BSP the EVK is running for an hour, maybe you need to check with your vendor.

Regards