Vida CEM swapping

[Arty]

That’s all I can say on the matter, lads)))

[Yariy]

Test the `ShimTSDiCE32` shim — a native 32‑bit DLL (C++, MSVC, Win32/x86) that exports the standard 14 J2534 v4.04 functions (`PassThruOpen` … `PassThruIoctl`) and passes them through to the actual Volvo DiCE driver — `TSDiCE32.dll`. The purpose of the shim is to selectively translate protocol identifiers that the DiCE driver does not understand: the J2534-2 protocol `CAN_PS` (`0x8004`) is replaced with the proprietary `LS-CAN` DiCE (`0x10000001`).


Call chain:

```
Application (J2534)
│ PassThru* (__stdcall)
▼
ShimTSDiCE32.dll (interception + translation of ProtocolID/Flags)
│ PassThru* (__stdcall)
▼
TSDiCE32.dll (real Volvo DiCE driver)
▼
Volvo DiCE equipment

Run the file register-shim.bat to copy the DLL and add the device “Volvo DiCE (ShimTSDiCE32)” to the registry.
In your program, choose the "Volvo DiCE (ShimTSDiCE32)" device.

[Arty]

Test the `ShimTSDiCE32` shim — a native 32‑bit DLL (C++, MSVC, Win32/x86) that exports the standard 14 J2534 v4.04 functions (`PassThruOpen` … `PassThruIoctl`) and passes them through to the actual Volvo DiCE driver — `TSDiCE32.dll`. The purpose of the shim is to selectively translate protocol identifiers that the DiCE driver does not understand: the J2534-2 protocol `CAN_PS` (`0x8004`) is replaced with the proprietary `LS-CAN` DiCE (`0x10000001`).


Call chain:

```
Application (J2534)
│ PassThru* (__stdcall)
▼
ShimTSDiCE32.dll (interception + translation of ProtocolID/Flags)
│ PassThru* (__stdcall)
▼
TSDiCE32.dll (real Volvo DiCE driver)
▼
Volvo DiCE equipment

Copy ShimTSDiCE32.dll next to the driver: C:\Program Files (x86)\DiCE\Tools\
Run the registry file register-shim.reg to add the device "Volvo DiCE (ShimTSDiCE32)".
In your program, choose the "Volvo DiCE (ShimTSDiCE32)" device.

I actually did something pretty much like that. Just a bridge. The main thing is that the software thinks it’s using DICE.

[Yariy]

I have the opposite situation)) — so that the program thinks it’s working with a J2534 adapter, but in reality it’s working with DiCE.

[RickHaleParker]

Code: Select all

                     Software poll  Hardware capture
Valid samples                 1024              1024
Invalid samples                  0                 0
Minimum (us)                84.030             0.107
Maximum (us)                90.087             0.213
Range (max-min)              6.057             0.107
Mean (us)                   87.077             0.171
MAD (us)                     1.945             0.000

[Yariy]

Code: Select all

                     Software poll  Hardware capture
Valid samples                 1024              1024
Invalid samples                  0                 0
Minimum (us)                84.030             0.107
Maximum (us)                90.087             0.213
Range (max-min)              6.057             0.107
Mean (us)                   87.077             0.171
MAD (us)                     1.945             0.000

Could you please explain what this is?

[RickHaleParker]

Code: Select all

                     Software poll  Hardware capture
Valid samples                 1024              1024
Invalid samples                  0                 0
Minimum (us)                84.030             0.107
Maximum (us)                90.087             0.213
Range (max-min)              6.057             0.107
Mean (us)                   87.077             0.171
MAD (us)                     1.945             0.000

Could you please explain what this is?

Working on using the Cortex-M7 PWM hardware timer peripherals and DMA as a co-processor.
The idea is to wean the jitters out of the latency measurements.

The above is a comparison between the current version of the "co-processor" vs the current the Software polling method in the Volvo CEM Cracker.

[vtl]

Working on using the Cortex-M7 PWM hardware timer peripherals and DMA as a co-processor.
The idea is to wean the jitters out of the latency measurements.

The above is a comparison between the current version of the "co-processor" vs the current the Software polling method in the Volvo CEM Cracker.

87 us in your table is the average response latency from CEM. One loop in software on Teensy is under 100 ns.

[RickHaleParker]

87 us in your table is the average response latency from CEM. One loop in software on Teensy is under 100 ns.

I need to find a different arming trigger for RX. The CAN transceiver mirroring is tripping this one up.

[vtl]

87 us in your table is the average response latency from CEM. One loop in software on Teensy is under 100 ns.

I need to find a different arming trigger for RX. The CAN transceiver mirroring is tripping this one up.

I don't think more precision will help. The problem is the slowest domain clock - the CAN-bus. 500 KHz hides all the variability happening on the 30 MHz (CEM) side. P1 CEMs are running at 4 MHZ - only 8 MCU clocks per 1 CAN clock, and they are much easier to crack.

If ARM code execution gives a lot of jitter then what may help is turning off the speculations in ARM core, so it would execute the code slower, but with linear timing. Speculative execution and power saving.