Flashing custom firmware package onto uhk60 via agent80 not working.

[kareltucek]

I have a custom local firmware release on a tagged branch: uhk-firmware-11.2.0-stuck_keys_hunt_v2.tar.gz (5801ac64a70342c92f1daf2ee483d81a10fafff8), and I am trying to flash it.

Neither right nor left half gets flashed. Only trackball does. Actually, trackball gets flashed every single time, although given that subsequent flashes should have same checksums, it probably shouldn't get flashed second time around.

Right and left stay on 443e5df59f2ae66a0ee424618080ac4a239c5e51, which is a couple commits ago.

Agent flash log: flog.txt

Any idea (@ert78gb) what is wrong? Could you look into it?

[ert78gb]

The attached log contains only 1 runs of the firmware upgrade process.

[DeviceService] "Trackball" firmware version: 11.2.0
[DeviceService] "Trackball" current firmware checksum: 8e70d90f3a4e83cf959af8900658f456
[DeviceService] "Trackball" new firmware checksum: 00000000000000000000000000000000
[UhkOperations] Start flashing "Trackball" module firmware

based on these lines Agent correctly upgrade the Trackball firmware because the 2 checksum are different.

[ert78gb]

I am investigating it. The following is my hardware information. I already flashed my uhk60v2 keyboard with your firmware

There are 2 sub structure.
The moduleInfos that contains details information of an UHK module. Left half also a module.
The rightModuleInfo contains the right half information.

The 3 is the Trackball module ID. It is available in both sub structure.

 "hardwareModules": {
    "moduleInfos": [
      {
        "module": {
          "id": 1,
          "name": "Left keyboard half",
          "slotId": 1,
          "i2cAddress": 16,
          "firmwareUpgradeSupported": true
        },
        "info": {
          "moduleProtocolVersion": "4.3.0",
          "firmwareVersion": "11.2.0",
          "firmwareGitRepo": "UltimateHackingKeyboard/firmware-uhk80",
          "firmwareGitTag": "89e7f92",
          "firmwareChecksum": "9a9e8063c2bd1ca3674bd742b9aa455b"
        }
      },
      {
        "module": {
          "id": 3,
          "name": "Trackball",
          "configPath": "/add-on/trackball",
          "slotId": 3,
          "i2cAddress": 48,
          "firmwareUpgradeSupported": true
        },
        "info": {
          "moduleProtocolVersion": "4.3.0",
          "firmwareVersion": "11.2.0",
          "firmwareGitRepo": "UltimateHackingKeyboard/firmware-uhk80",
          "firmwareGitTag": "stuck_keys_hunt_v2",
          "firmwareChecksum": "8e70d90f3a4e83cf959af8900658f456"
        }
      }
    ],
    "rightModuleInfo": {
      "firmwareVersion": "11.2.0",
      "deviceProtocolVersion": "4.11.0",
      "moduleProtocolVersion": "4.3.0",
      "userConfigVersion": "8.3.0",
      "hardwareConfigVersion": "1.0.0",
      "smartMacrosVersion": "2.5.0",
      "firmwareGitRepo": "UltimateHackingKeyboard/firmware-uhk80",
      "firmwareGitTag": "28b254f",
      "firmwareChecksum": "951c0587f7d6b2283abc60e41b040964",
      "modules": {
        "1": {
          "md5": "9a9e8063c2bd1ca3674bd742b9aa455b"
        },
        "3": {
          "md5": "4b5b5b92c17ddb96b3a75f2c6573eaf1"
        }
      }
    }
  }

The hardwareModules. moduleInfos.3.firmwareChecksum: 8e70d90f3a4e83cf959af8900658f456The rightModuleInfo.modules.3.md5: 4b5b5b92c17ddb96b3a75f2c6573eaf1`
The 2 value are different. (I have to unify the naming)

We use the following USB command to collect these data
hardwareModules. moduleInfos.3.firmwareChecksum => 0x0e 0x03 0x03

• GetModuleProperty USB command 0x0e
• module ID 0x03
• protocolVersions property 0x03

rightModuleInfo.modules.3.md5 => 0x00 0x08 0x03

• GetProperty USB command 0x00
• FirmwareChecksum property 0x08
• module ID 0x03

Agent compares these 2 checksum and if they are different then flashes the module.
The original story #1663

[mondalaci]

Is this a duplicate of #2435 ?

[ert78gb]

Yes

[kareltucek]

Partially I think.

This looks like Agent concludes that the (trackball) checksums are different every single time, and therefore it always tries to flash only the module and never triggers the scenario when it allows to reflash everything even when checksums match.

Furthermore, the zero checksum in the log above is pretty suspicious. If it is present in the firmware build package, then that clearly is firmware build machinery fail.

[ert78gb]

This looks like Agent concludes that the (trackball) checksums are different every single time, and therefore it always tries to flash only the module and never triggers the scenario when it allows to reflash everything even when checksums match.

exactly

Furthermore, the zero checksum in the log above is pretty suspicious. If it is present in the firmware build package, then that clearly is firmware build machinery fail.

the first step off the firmware build is create a build with 0 checksum. And the 2nd step/run uses the calculated checksum of the first build.