nRF9160 DK Ethereum Ecdsa "s problem"

RE: nRF9160 DK Ethereum Ecdsa "s problem"

Gioele — Sat, 17 Sep 2022 00:09:38 GMT

We understood that the problem is related to the ECDSA provided by Nordic that doesn't take in input Keccak256.

RE: nRF9160 DK Ethereum Ecdsa "s problem"

dejans — Wed, 07 Sep 2022 16:44:33 GMT

Hi,

I have made an internal inquiry on this topic.
Since you try to implement Ethereum ECDSA, use keccak256 with secp256k1, and since it is hash algorithm we do not currently provide, we cannot provide full support related to your question.

I would suggest that you engage more with Ethereum crypto and developer communities.
There is a list of software resources implementing keccak.

Best regards,
Dejan

RE: nRF9160 DK Ethereum Ecdsa "s problem"

Gioele — Mon, 05 Sep 2022 10:07:25 GMT

We are using deterministic ECDSA in both. For some values the ECDSA result is wrong related to the "s value".
We noticed that when we make the Keccak256 of a standard number or a text the "r and s values" are correct, but when we make the keccack256 of RLP only the "r value" is correct.

RE: nRF9160 DK Ethereum Ecdsa "s problem"

dejans — Mon, 05 Sep 2022 09:28:41 GMT

Hi,

Do you get the same signature every time you run your python script? Do you use deterministic ECDSA in your python project?

Best regards,
Dejan

RE: nRF9160 DK Ethereum Ecdsa "s problem"

Gioele — Thu, 01 Sep 2022 15:25:12 GMT

static uint8_t keccak256[32] = {
0x99, 0x68, 0x3d, 0xfd, 0x16, 0xa3, 0x65, 0xb2,
0xf1, 0x45, 0x82, 0x59, 0xd7, 0xaf, 0x30, 0x6d,
0xb5, 0x74, 0xa3, 0x4b, 0xe7, 0xf8, 0x6d, 0x21,
0x02, 0xca, 0xb0, 0x47, 0xb0, 0xee, 0x4c, 0xf3
};

static uint8_t m_priv_key[32] = {
0x1d, 0x68, 0xd6, 0x4b, 0xa9, 0x19, 0x60, 0x41,
0x1e, 0x4b, 0xb8, 0x5d, 0xd7, 0xd4, 0x61, 0x02,
0x2f, 0x70, 0x6c, 0x6d, 0xbe, 0x67, 0x6b, 0xf1,
0x3b, 0xdb, 0xfc, 0x3f, 0x7e, 0x1c, 0x13, 0x38
};



status = psa_import_key(&key_attributes, m_priv_key, sizeof(m_priv_key), &keypair_handle);
if (status != PSA_SUCCESS) {
LOG_INF("psa_import_private_key failed! (Error: %d)", status);
return APP_ERROR;
}


status = psa_sign_hash(keypair_handle,
PSA_ALG_DETERMINISTIC_ECDSA(PSA_ALG_SHA_256),
keccak256,
sizeof(m_plain_text),
m_signature,
sizeof(m_signature),
&output_len);
if (status != PSA_SUCCESS) {
LOG_INF("psa_sign_hash failed! (Error: %d)", status);
return APP_ERROR;
}

This is the code of the ECDSA example by Nordic and this gives us an incorrect "s value".

signed_tx = web3.eth.account.signTransaction(tx, private_key)
tx = {
'nonce': nonce,
'to': to_acc,
'value': web3.toWei(1, 'ether'),
'gas': 2000000,
'gasPrice': web3.toWei('50', 'gwei')
}

private_key = '1d68d64ba91960411e4bb85dd7d461022f706c6dbe676bf13bdbfc3f7e1c1338'

This is the code in python that give us the correct "s value".

RE: nRF9160 DK Ethereum Ecdsa "s problem"

dejans — Thu, 01 Sep 2022 11:35:32 GMT

Hi,

Thank you for this information.

I would appreciate it if you could answer following 2 questions.

[quote user=""]We think that there is a mistake with the procedure to obtain the "s value".[/quote]

Could you provide more information about the procedure that you used in both cases?

[quote user="dejans"]

Gioele said:
This is the correct output that comes from another Python Project:

Gioele said:
This is the output that comes from the ECDSA provided by Nordic.

[/quote]

Could you provide more information about the way in which you calculated the values you provided?

Best regards,
Dejan

RE: nRF9160 DK Ethereum Ecdsa "s problem"

Gioele — Wed, 31 Aug 2022 18:23:34 GMT

[quote userid="111786" url="~/f/nordic-q-a/91484/nrf9160-dk-ethereum-ecdsa-s-problem/384203"]What is our "s" that is wrong?
[/quote]

Correct

f86d0c850ba43b7400831e8480946fb6ba99b8a9f1330c1e83ec309c1dab81906e63880de0b6b3a7640000801ca0c4a47c3979ee3f169f2b1ad8b4754a0ea66297594af3b8d9e6f9a1dd01bd8152a0142e95021d0d979b1e49744421cbb0dbadcf48b3eba2b20957740f57a8e56d3b

Wrong

f86d0c850ba43b7400831e8480946fb6ba99b8a9f1330c1e83ec309c1dab81906e63880de0b6b3a7640000801ca0c4a47c3979ee3f169f2b1ad8b4754a0ea66297594af3b8d9e6f9a1dd01bd8152a0ebd16afde2f26864e1b68bbbde344f230cdf9432c3a5ee32685e4f352750d406

The first output comes from the Python program (web3 library), and the second output comes from the Nordic ECDSA example.

RE: nRF9160 DK Ethereum Ecdsa "s problem"

dejans — Wed, 31 Aug 2022 17:19:32 GMT

Hi,

Going back to my previous questions,

[quote user=""]We are comparing the results with signs of other projects and we see that our program computes the same sign but we ascertain that our "s" is wrong.[/quote]

What is our "s" that is wrong?

[quote user=""]The ECDSA sample doesn't provide the Keccak256 algorithm so we implemented it ourselves. We think that there is a mistake with the procedure to obtain the "s value".[/quote]

Where do you think is an error? What procedure do you refer to?

[quote user="Gioele"]This is the correct output that comes from another Python Project:[/quote]

[quote user="Gioele"]This is the output that comes from the ECDSA provided by Nordic.[/quote]

Could you explain in detail how you got both of these outputs?

Best regards,
Dejan

RE: nRF9160 DK Ethereum Ecdsa "s problem"

Gioele — Wed, 31 Aug 2022 15:29:07 GMT

[quote userid="111786" url="~/f/nordic-q-a/91484/nrf9160-dk-ethereum-ecdsa-s-problem/384188"]Could you elaborate on this?
[/quote]

When we talk about ECDSA we find out two important values that are r and s. These 2 values are appended to other characters to form the final output.

This is the correct output that comes from another Python Project:

f86d0c850ba43b7400831e8480946fb6ba99b8a9f1330c1e83ec309c1dab81906e63880de0b6b3a7640000801ca0c4a47c3979ee3f169f2b1ad8b4754a0ea66297594af3b8d9e6f9a1dd01bd8152a0142e95021d0d979b1e49744421cbb0dbadcf48b3eba2b20957740f57a8e56d3b

This is the output that comes from the ECDSA provided by Nordic.

f86d0c850ba43b7400831e8480946fb6ba99b8a9f1330c1e83ec309c1dab81906e63880de0b6b3a7640000801ca0c4a47c3979ee3f169f2b1ad8b4754a0ea66297594af3b8d9e6f9a1dd01bd8152a0ebd16afde2f26864e1b68bbbde344f230cdf9432c3a5ee32685e4f352750d406

There is something that doesn't work, in detail we can see that the problem is related to the "s value" that is different, and that is wrong.

RE: nRF9160 DK Ethereum Ecdsa "s problem"

dejans — Wed, 31 Aug 2022 15:16:12 GMT

Hi,

[quote user=""]We are comparing the results with signs of other projects and we see that our program computes the same sign but we ascertain that our "s" is wrong.[/quote]

Could you elaborate on this?

[quote user=""]We think that there is a mistake with the procedure to obtain the "s value".[/quote]

Could you elaborate on this as well?

Best regards,
Dejan