Page MenuHomePhabricator

[identity] Add RPC for secondary device login
ClosedPublic

Authored by bartek on Jan 29 2024, 4:39 AM.
Tags
None
Referenced Files
Unknown Object (File)
Tue, Nov 12, 12:15 PM
Unknown Object (File)
Thu, Nov 7, 2:19 AM
Unknown Object (File)
Mon, Nov 4, 5:50 AM
Unknown Object (File)
Fri, Nov 1, 11:58 PM
Unknown Object (File)
Fri, Nov 1, 11:57 PM
Unknown Object (File)
Fri, Nov 1, 11:57 PM
Unknown Object (File)
Fri, Nov 1, 11:56 PM
Unknown Object (File)
Fri, Nov 1, 11:34 PM
Subscribers

Details

Summary

This diff adds a RPC that is called by secondary device, after receiving information from primary device that it has been added to the device list.

This RPC serves two purposes:

  • Perform a device key upload for newly added secondary device
  • Mint an access token for the secondary device

The implementation is done in the next diff

Depends on D10800

Test Plan
  • Identity service and grpc_cliends compile
  • Flow for grpc-web codegen

Diff Detail

Repository
rCOMM Comm
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

bartek held this revision as a draft.
bartek published this revision for review.Jan 29 2024, 5:29 AM
ashoat requested changes to this revision.Jan 29 2024, 11:21 AM

Concerned that we're short-circuiting the proper "Existing Device Log-in" protocol here

shared/protos/identity_unauth.proto
29 ↗(On Diff #36257)

Would this RPC ever be used in a different case that isn't secondary device auth? I'm wondering if we should pick a different name. For instance, UploadKeysForRegisteredDeviceAndLogIn.

I spent some time looking at the whitepaper, and I can't find any other examples of "New Device and Comm run Key Upload (5.2.2) and Existing Device Log-in (5.3.6)" occurring, so I think we can probably leave it as-is, unless you prefer the alternate name I proposed.

218–220 ↗(On Diff #36257)

In the whitepaper, I think this request is described as "New Device and Comm run Key Upload (5.2.2) and Existing Device Log-in (5.3.6)".

I think you're taking a slightly different approach here. Instead of a proper "Existing Device Log-in", you're authing the user just based on their DeviceKeyUpload.

I don't think this is as safe. By skipping the nonce and timestamp steps, you make the system hypothetically vulnerable to replay attacks.

223–227 ↗(On Diff #36257)

Realized that this message is repeated many times in this file:

  • RegistrationFinishResponse
  • OpaqueLoginFinishResponse
  • WalletLoginResponse
  • This case: SecondaryDeviceLoginResponse

Is there a reason we're repeating it so many times? Does it make it easier to make changes later if we need to?

This revision now requires changes to proceed.Jan 29 2024, 11:21 AM
bartek added inline comments.
shared/protos/identity_unauth.proto
29 ↗(On Diff #36257)

Glad you raised this. Yes, this RPC is intended to be called in this only specific case (QR code auth).
My name isn't perfect but I didn't have any better idea. I'll follow your proposal.

218–220 ↗(On Diff #36257)

This is my bad, I treated our password/wallet login RPCs as "Existing Device Log-in (5.3.6)" but now looking at LogInWalletUser RPC implementation I see that it indeed validates the nonce.

223–227 ↗(On Diff #36257)

Yeah I noticed this pattern too, so I followed it.
This probably can be merged into one message unless I'm missing something, cc @varun.
If so, I can create a task and do this. But I prefer doing it separately from this stack, for clarity.

Renamed the RPC; added request field for the nonce/timestamp challenge

shared/protos/identity_unauth.proto
218 ↗(On Diff #36611)

Not sure of the best name for this field.

Assuming the RPC for vending the challenge will be handled separately. Proto changes and grpc-web changes look good to me – resigning so that somebody else can take a look at the Rust

shared/protos/identity_unauth.proto
218 ↗(On Diff #36613)

For this to work, don't we need another RPC to vend the challenge?

varun added inline comments.
shared/protos/identity_unauth.proto
223–227 ↗(On Diff #36257)

i think we can merge them

This revision is now accepted and ready to land.Feb 6 2024, 7:14 AM
shared/protos/identity_unauth.proto
224–228 ↗(On Diff #36613)

we can remove this now or create a new task to replace this with the message introduced in D11061