Bug 197974 – [WebAuthN] Allow authenticators that support both CTAP and U2F to try U2F if CTAP fails in authenticatorGetAssertion

• Component: WebKit Misc.
• Status: RESOLVED FIXED

[Jiewen Tan]

Allow authenticators that support both CTAP and U2F to try U2F if CTAP fails in authenticatorGetAssertion.

[Radar WebKit Bug Importer]

<rdar://problem/50879746>

[Radar WebKit Bug Importer]

<rdar://problem/50879747>

[Jiewen Tan]

Created attachment 370100 [details]
Patch

[Jiewen Tan]

Comment on attachment 370100 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=370100&action=review

> Source/WebKit/ChangeLog:16
> +        to ask a potential U2F credential once a kCtap2ErrInvalidCredential error is returned regarding to the first CTAP request,

I should probably relax all conditions to always downgrade to U2F whenever there is no responses to the first CTAP request.

[Jiewen Tan]

Comment on attachment 370100 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=370100&action=review

> Source/WebKit/UIProcess/WebAuthentication/fido/CtapHidAuthenticator.cpp:93
> +        auto error = getResponseCode(data);

I should probably relax all conditions to always downgrade to U2F whenever there is no responses to the first CTAP request.

[Jiewen Tan]

Comment on attachment 370100 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=370100&action=review

>> Source/WebKit/UIProcess/WebAuthentication/fido/CtapHidAuthenticator.cpp:93
>> +        auto error = getResponseCode(data);
> 
> I should probably relax all conditions to always downgrade to U2F whenever there is no responses to the first CTAP request.

The reason behinds this decision is that CTAP2_ERR_NO_CREDENTIALS is recommended in the spec: https://fidoalliance.org/specs/fido-v2.0-ps-20190130/fido-client-to-authenticator-protocol-v2.0-ps-20190130.html#authenticatorGetAssertion. On contrary, I get CTAP2_ERR_INVALID_CREDENTIAL from some keys. Therefore, we should probably not check any error.

Also, bad user agents might somehow prefer U2F over CTAP, and therefore relaying parties might not be aware of it and supply an appid to indicate a U2F credential should be used instead. Therefore, we should omit that check as well.

[Jiewen Tan]

Created attachment 370163 [details]
Patch

[Brent Fulgham]

Comment on attachment 370163 [details]
Patch

Looks good. r=me

[Jiewen Tan]

Created attachment 370165 [details]
Patch

[Jiewen Tan]

(In reply to Brent Fulgham from comment #8)
> Comment on attachment 370163 [details]
> Patch
> 
> Looks good. r=me

Sorry, Brent. I modified a test and uploaded a new patch which obsolete your r+ flag. Could you r+ it again?

[Brent Fulgham]

Comment on attachment 370165 [details]
Patch

R=me

[Jiewen Tan]

(In reply to Brent Fulgham from comment #11)
> Comment on attachment 370165 [details]
> Patch
> 
> R=me

Thanks Brent for r+ this patch.

[WebKit Commit Bot]

Comment on attachment 370165 [details]
Patch

Clearing flags on attachment: 370165

Committed r245500: <https://trac.webkit.org/changeset/245500>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.