X25519Kyber768Draft00 named group not being picked up by boringssl static

[jmcrawford45]

I am setting
System.setProperty("jdk.tls.namedGroups", "X25519Kyber768Draft00"); on 2.0.65.Final
for my application, but when I trace the code to setCurvesList0, I get false instead of true. I would expect this named group to be configured when requested, which is consistent with what I get when running e.g. ./bssl client -curves X25519Kyber768Draft00 -connect localhost:4433 -debug. Is there some additional property I need to set to get X25519Kyber768Draft00 recognized as a supported group?

[normanmaurer]

I don't think we respect this property in our implement... Maybe something we should do tho

[jmcrawford45]

@normanmaurer I see you added a related commit for this in https://github.com/netty/netty/pull/11660/files I'm curious what else would need to be added as it seems the referenced PR adds this support. I also see the property parsed and passed all the way down to the native entry point, but setCurvesList0 fails on "X25519Kyber768Draft00", but passes on all other supported bssl NamedGroups.

[normanmaurer]

Oh so I already implemented it 🤣... maybe support is not present in the chromium-stable branch of boringssl ?

[jmcrawford45]

I'm able to spin up a server and client with chromium-stable@{'3 months ago'} and use them with e.g. ./bssl client -curves X25519Kyber768Draft00 -connect localhost:4433 -debug

[normanmaurer]

I can have a look but it might take me a while ...

[jmcrawford45]

Thank you so much, please let me know if I can provide any additional context. I've been debugging this for a while, and my best guess at this point is that the ssl context created by Openssl.java excludes Kyber from the supported curves list field which causes the setCurve call to fail.

[jmcrawford45]

@normanmaurer I'm checking back in on this issue, please let me know if there's any way I can help with debugging.

[normanmaurer]

@jmcrawford45 unfortunately I had not time yet to investigate :/

[jmcrawford45]

@normanmaurer I'm checking back in on this issue, please let me know if there's any way I can help with debugging. Given the recent NIST standardization, there is a growing interest in the ability to configure these algorithms.