Need to support hardware variants with multiple possible ECC locations #221

shawaj · 2023-02-12T19:39:22Z

It turns out that Syncrobit miners have several possible locations for the ECC key on the i2c bus...

We should add handling for this scenario

pritamghanghas · 2023-02-17T06:40:00Z

mfr throws exec error if we try to read a slot that is not programmed before.

root@78eef93a415c:/opt# ./python-dependencies/hm_pyhelper/gateway_mfr_aarch64 --device 'ecc://i2c-7:96?slot=5' key
Error: ecc608 error

Caused by:
    ecc error ExecError

Between i2cdetect and cycling through the slots, we can probably find out the correct combo. Assuming that multiple slots were not programmed.

shawaj · 2023-02-17T07:25:37Z

@pritamghanghas sadly I've found that sometimes multiple are programmed. See output here...

NebraLtd/helium-syncrobit#2 (comment)

This is a quick and hacky fix to the issue described here NebraLtd/helium-syncrobit#2 (comment) The idea is to allow a device level environment variable in balenaCloud to override the key location in the assigned variant without needing to define a new variant. This is useful for testing new devices or in instances like this Syncrobit one to provide a quick fix until we have implemented a "proper" fix to this issue. This additionally changes the following: - we hard code the gateway_mfr_rs version in the [test](https://github.com/NebraLtd/hm-pyhelper/blob/master/.github/workflows/publish-to-pypi-test.yml) and [production](https://github.com/NebraLtd/hm-pyhelper/blob/master/.github/workflows/publish-to-pypi.yml) python package build actions, so there is no need to support old versions of gateway-mfr-rs - due to removing the above, there is also no need to keep the KEY_STORAGE_BUS key in the hardware variants as this is not used anywhere else that I'm aware of **(we should double check this in config, diag, miner, packet forwarder and multiplexer containers just to be 100% sure)** - [Used as a fallback in diag](https://github.com/NebraLtd/hm-diag/blob/23c66980f75568a4cecb18c5375a08cd67bff550/hw_diag/utilities/hardware.py#L200) but not required Relates-to: NebraLtd/helium-syncrobit#2 Relates-to: #221 Relates-to: #222

shawaj · 2023-03-11T20:46:01Z

Also, currently the i2c address on the bus is hard coded in diagnostics - https://github.com/NebraLtd/hm-diag/blob/master/hw_diag/utilities/hardware.py#L235

This address is 0x60 / decimal 96 which is in the swarm key uri already for example ecc://i2c-1:96?slot=0

We should change this also at same time

Ref - NebraLtd/hm-diag#500

Also see https://nebraltd.slack.com/archives/C024BNQ1Y6T/p1678578139710139

This is a quick and hacky fix to the issue described here NebraLtd/helium-syncrobit#2 (comment) The idea is to allow a device level environment variable `SWARM_KEY_URI_OVERRIDE` in balenaCloud to override the key location in the assigned variant without needing to define a new variant. This is useful for testing new devices or in instances like this Syncrobit one to provide a quick fix until we have implemented a "proper" fix to this issue. This additionally changes the following: - we hard code the gateway_mfr_rs version in the [test](https://github.com/NebraLtd/hm-pyhelper/blob/master/.github/workflows/publish-to-pypi-test.yml) and [production](https://github.com/NebraLtd/hm-pyhelper/blob/master/.github/workflows/publish-to-pypi.yml) python package build actions, so there is no need to support old versions of gateway-mfr-rs - due to removing the above, there is also no need to keep the KEY_STORAGE_BUS key in the hardware variants as this is not used anywhere else that I'm aware of **(we should double check this in config, diag, miner, packet forwarder and multiplexer containers just to be 100% sure)** - [Used as a fallback in diag](https://github.com/NebraLtd/hm-diag/blob/23c66980f75568a4cecb18c5375a08cd67bff550/hw_diag/utilities/hardware.py#L200) but not required Relates-to: NebraLtd/helium-syncrobit#2 Relates-to: #221 Relates-to: #222

shawaj · 2023-03-12T00:53:42Z

This will also need significant updates in:

hm-miner
hm-block-tracker

This is a quick and hacky fix to the issue described here NebraLtd/helium-syncrobit#2 (comment) The idea is to allow a device level environment variable `SWARM_KEY_URI_OVERRIDE` in balenaCloud to override the key location in the assigned variant without needing to define a new variant. This is useful for testing new devices or in instances like this Syncrobit one to provide a quick fix until we have implemented a "proper" fix to this issue. This additionally changes the following: - we hard code the gateway_mfr_rs version in the [test](https://github.com/NebraLtd/hm-pyhelper/blob/master/.github/workflows/publish-to-pypi-test.yml) and [production](https://github.com/NebraLtd/hm-pyhelper/blob/master/.github/workflows/publish-to-pypi.yml) python package build actions, so there is no need to support old versions of gateway-mfr-rs - due to removing the above, there is also no need to keep the KEY_STORAGE_BUS key in the hardware variants as this is not used anywhere else that I'm aware of **(we should double check this in config, diag, miner, packet forwarder and multiplexer containers just to be 100% sure)** - [Used as a fallback in diag](https://github.com/NebraLtd/hm-diag/blob/23c66980f75568a4cecb18c5375a08cd67bff550/hw_diag/utilities/hardware.py#L200) but not required Relates-to: NebraLtd/helium-syncrobit#2 Relates-to: #221 Relates-to: #222

…227) * feat: allow override of key location and remove old version support This is a quick and hacky fix to the issue described here NebraLtd/helium-syncrobit#2 (comment) The idea is to allow a device level environment variable `SWARM_KEY_URI_OVERRIDE` in balenaCloud to override the key location in the assigned variant without needing to define a new variant. This is useful for testing new devices or in instances like this Syncrobit one to provide a quick fix until we have implemented a "proper" fix to this issue. This additionally changes the following: - we hard code the gateway_mfr_rs version in the [test](https://github.com/NebraLtd/hm-pyhelper/blob/master/.github/workflows/publish-to-pypi-test.yml) and [production](https://github.com/NebraLtd/hm-pyhelper/blob/master/.github/workflows/publish-to-pypi.yml) python package build actions, so there is no need to support old versions of gateway-mfr-rs - due to removing the above, there is also no need to keep the KEY_STORAGE_BUS key in the hardware variants as this is not used anywhere else that I'm aware of **(we should double check this in config, diag, miner, packet forwarder and multiplexer containers just to be 100% sure)** - [Used as a fallback in diag](https://github.com/NebraLtd/hm-diag/blob/23c66980f75568a4cecb18c5375a08cd67bff550/hw_diag/utilities/hardware.py#L200) but not required Relates-to: NebraLtd/helium-syncrobit#2 Relates-to: #221 Relates-to: #222

Syncrobit has multiple potential key locations, change to list and parse list if necessary Relates-to: #221 Relates-to: NebraLtd/helium-syncrobit#2 (comment) Supersedes: #222

shawaj · 2023-03-15T20:45:14Z

This PR is a partial fix for Syncrobit multi key issue:
#230

This will solve the vast majority of cases. However some will still need to be overridden with a SWARM_KEY_URI_OVERRIDE device variable currently.

In order to provide a complete fix, we still need to implement the following:

if we have definitions in hardware variants for multiple different slots on the same i2c address then:
- Check keys on public API. If one is found, then use that one
- If none of the available keys are found on public API, check the onboarding server
- If only one is found, use that one
- If multiple found on onboarding server, use the first valid one for onboarding

We also need to pythonise the miner start scripts (or portions of it) to override the keyslot in the sys.config. Or to publish the determined key as an environment variable that the miner can ingest... But that seems more convoluted.

Ref https://github.com/syncrobit/chameleonos/blob/main/board/common/overlay/etc/init.d/S31hwdetect

shawaj · 2023-03-15T21:12:38Z

Ref https://github.com/helium/onboarding-server

Ref https://onboarding.dewi.org/api/v2/hotspots/11n6ApbTNiNhCLf84Hm3mZGDyqNhP83dsj4Cvj2juTXuKD7byf

shawaj mentioned this issue Mar 11, 2023

feat: allow override of key location and remove old version support #227

Merged

4 tasks

shawaj mentioned this issue Mar 12, 2023

Add onboarding key location (specifically for Helium OG definition) #229

Closed

shawaj mentioned this issue Mar 15, 2023

feat: add support for multiple possible key locations #230

Merged

4 tasks

shawaj mentioned this issue Apr 7, 2023

possibility to combine get_ecc_location and get_onboarding_location #239

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Need to support hardware variants with multiple possible ECC locations #221

Need to support hardware variants with multiple possible ECC locations #221

shawaj commented Feb 12, 2023

pritamghanghas commented Feb 17, 2023

shawaj commented Feb 17, 2023

shawaj commented Mar 11, 2023 •

edited

Loading

shawaj commented Mar 12, 2023

shawaj commented Mar 15, 2023 •

edited

Loading

shawaj commented Mar 15, 2023

Need to support hardware variants with multiple possible ECC locations #221

Need to support hardware variants with multiple possible ECC locations #221

Comments

shawaj commented Feb 12, 2023

pritamghanghas commented Feb 17, 2023

shawaj commented Feb 17, 2023

shawaj commented Mar 11, 2023 • edited Loading

shawaj commented Mar 12, 2023

shawaj commented Mar 15, 2023 • edited Loading

shawaj commented Mar 15, 2023

shawaj commented Mar 11, 2023 •

edited

Loading

shawaj commented Mar 15, 2023 •

edited

Loading