[bitcoindev] Revisiting secp256r1 signatures (i.e. P256, mobile HSM support)

[bitcoindev] Revisiting secp256r1 signatures (i.e. P256, mobile HSM support)

[Josh Doman]

[-- Attachment #1.1: Type: text/plain, Size: 3342 bytes --]

A brief search on gnusha.org <https://gnusha.org/pi/bitcoindev/?q=secp256r1> 
suggests that it's been over 10 years since the Bitcoin community last 
discussed adding secp256r1 support (also known as P256). The most in-depth 
discussions I found were on BitcoinTalk in 2011 
<https://bitcointalk.org/?topic=2699.0> and 2013 
<https://bitcointalk.org/index.php?topic=151120.0>.

Since then, P256 has gained widespread adoption across the modern internet 
and on mobile. Most notably, millions of users now possess mobile devices 
capable of generating and storing private keys in secure enclaves (see 
Apple iCloud Keychain and Android Keystore). Millions of users might be 
able to immediately use this to start self-custodying bitcoin, except this 
hardware only supports P256 signatures, which is incompatible with the 
secp256k1 curve that Bitcoin currently uses.

Reading through old discussions, it appears that the primary concern the 
community had with P256 is the possibility of a NIST backdoor. Putting the 
likelihood of this aside, it seems reasonable to me that in 2025, users 
should at least have the option of using P256, if they wish. Native HSM 
support would significantly improve the onboarding experience for new 
users, increase the security and accessibility of hot wallets, and 
potentially reduce the cost of collaborative multisigs. Meanwhile, the 
community can continue to use secp256k1 as the ideal curve for private keys 
secured in cold storage.

At a technical level, Tapscript makes P256 mechanically straightforward to 
adopt, because it has built-in support for new types of public keys. For 
example, we could define a 33-byte public key as one requiring a P256 ECDSA 
signature, while continuing to use 32-bytes for keys requiring Schnorr 
signatures over secp256k1.

A secondary concern that I came across is that P256 can be 2-3x slower to 
validate than secp256k1. Assuming this cannot be improved, we can account 
for slower validation by doubling or tripling the validation weight cost 
for a P256 signature. Users can then pre-commit in their script to this 
additional weight or commit to it in the annex, as intended by BIP341 
<https://github.com/bitcoin/bips/blob/master/bip-0341.mediawiki>.

P256 support would grant apps the ability to use platform APIs to access 
the secure HSM on user's mobile devices, but alone, P256 is insufficient 
for non-custodial WebAuthn / passkey-based wallets. To verify a WebAuthn 
signature, we'd additionally need CSFS and CAT, so we can compute a 
WebAuthn message from a sighash and the necessary WebAuthn data on the 
stack. Alternatively, we could create a dedicated WebAuthn opcode to verify 
a WebAuthn message without enabling recursive covenants. Regardless, the 
ability to verify a P256 signature would be an important primitive.

In summary, *given the widespread hardware adoption and industry usage, is 
it worth revisiting adding P256 support to Bitcoin?*

Josh Doman

-- 
You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/bitcoindev/8fbe1fe3-425d-4056-8387-993f6ecc1been%40googlegroups.com.

[-- Attachment #1.2: Type: text/html, Size: 3721 bytes --]

Re: [bitcoindev] Revisiting secp256r1 signatures (i.e. P256, mobile HSM support)

['conduition' via Bitcoin Development Mailing List]

[-- Attachment #1.1.1: Type: text/plain, Size: 5873 bytes --]

Hey Josh, thanks for raising the idea.

While it's a neat premise, I think the timelines involved will not mesh well. It'd take several years to get community consensus (the fact it hasn't been discussed suggests not many people are interested), to build a high-quality implementation on-par with libsecp256k1, and to spec out and implement a soft fork.

By itself that'd be OK, but not when you consider other "new sig algo" projects happening in parallel: BIP360 and other post-quantum debates which will eventually lead to the addition of at least one new signature algorithm to consensus. By the time P256 is standardized and available, everyone will likely be moving towards post-quantum opcodes. It may well be obviated if a cryptographically relevant quantum computer comes out earlier than expected.

I'm not familiar with the WebAuthn standard, but surely its authors are themselves also thinking about post-quantum security. Perhaps if you want HSM support in the next decade, your best shot may be to research WebAuthn's post-quantum signature formats. Possible relevant reading. I'd suggest you research a way to make WebAuthn's signing flow compatible with the post-quantum sig verification opcodes being developed for bitcoin (or vice-versa, talk with Ethan Heilman and try to make the Bitcoin standards compatible with WebAuthn).

The next part is just for my own curiosity. 

I'm not sure relying on webauthn is a good idea in the first place. It's a standard suited for web-based authentication to centralized services, not for long-term cryptographic identities or ownership across distributed systems. I've never heard of any WebAuthn authenticator which gives users a deterministic backup seed of any kind, so how would users recover their bitcoins independently in this context? Could a hypothetical webauthn wallet handle something like BIP32 without upstream support in WebAuthn?

And how would multi-address wallets work? I'm imagining the webauthn wallet would need to generate a new credential every time the user wants to generate a new P256 address. Wouldn't that clutter the user's keychain?

regards,
conduition
On Tuesday, July 22nd, 2025 at 3:03 PM, Josh Doman <joshsdoman@gmail.com> wrote:

> A brief search on gnusha.org suggests that it's been over 10 years since the Bitcoin community last discussed adding secp256r1 support (also known as P256). The most in-depth discussions I found were on BitcoinTalk in 2011 and 2013.
> 

> Since then, P256 has gained widespread adoption across the modern internet and on mobile. Most notably, millions of users now possess mobile devices capable of generating and storing private keys in secure enclaves (see Apple iCloud Keychain and Android Keystore). Millions of users might be able to immediately use this to start self-custodying bitcoin, except this hardware only supports P256 signatures, which is incompatible with the secp256k1 curve that Bitcoin currently uses.
> 

> Reading through old discussions, it appears that the primary concern the community had with P256 is the possibility of a NIST backdoor. Putting the likelihood of this aside, it seems reasonable to me that in 2025, users should at least have the option of using P256, if they wish. Native HSM support would significantly improve the onboarding experience for new users, increase the security and accessibility of hot wallets, and potentially reduce the cost of collaborative multisigs. Meanwhile, the community can continue to use secp256k1 as the ideal curve for private keys secured in cold storage.
> 

> At a technical level, Tapscript makes P256 mechanically straightforward to adopt, because it has built-in support for new types of public keys. For example, we could define a 33-byte public key as one requiring a P256 ECDSA signature, while continuing to use 32-bytes for keys requiring Schnorr signatures over secp256k1.
> 

> A secondary concern that I came across is that P256 can be 2-3x slower to validate than secp256k1. Assuming this cannot be improved, we can account for slower validation by doubling or tripling the validation weight cost for a P256 signature. Users can then pre-commit in their script to this additional weight or commit to it in the annex, as intended by BIP341.
> 

> P256 support would grant apps the ability to use platform APIs to access the secure HSM on user's mobile devices, but alone, P256 is insufficient for non-custodial WebAuthn / passkey-based wallets. To verify a WebAuthn signature, we'd additionally need CSFS and CAT, so we can compute a WebAuthn message from a sighash and the necessary WebAuthn data on the stack. Alternatively, we could create a dedicated WebAuthn opcode to verify a WebAuthn message without enabling recursive covenants. Regardless, the ability to verify a P256 signature would be an important primitive.
> 

> In summary, given the widespread hardware adoption and industry usage, is it worth revisiting adding P256 support to Bitcoin?
> 

> 

> Josh Doman
> 

> --
> You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/bitcoindev/8fbe1fe3-425d-4056-8387-993f6ecc1been%40googlegroups.com.

-- 
You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/bitcoindev/GVA2lG8TET5qUPOU8JlHttBNLbTskqr_dbhzK9aOe_ZsPsNDyk2kWIPwmx3ngTchlGrsAIpubrVl24qmyqVgBvrZ8FT0POeomam2qB6-cqk%3D%40proton.me.

[-- Attachment #1.1.2.1: Type: text/html, Size: 8309 bytes --]

[-- Attachment #1.2: publickey - conduition@proton.me - 0x474891AD.asc --]
[-- Type: application/pgp-keys, Size: 649 bytes --]

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 343 bytes --]