Results 298 comments of David Benjamin

> There is such a reason. It is an easy metrics to have 0 leaks at shutdown. If you have some, you have to deal with them - whether they...

+1 to @bernd-edlinger. This PR will make OpenSSL's API dangerous and is incompatible with having an algorithm-generic signing API. This was why we didn't wire it up to EVP_PKEY_sign at...

@t8m Please see https://github.com/openssl/openssl/pull/3409 for the previous discussion. The reason you all choose to use `EVP_DigestSign` was for consistency with your existing APIs. You could add a new API if...

I guess then the question is what problem we're trying to solve. You already have an API for `EVP_PKEY_sign_data` and streaming `EVP_PKEY_sign_data`, as well as other APIs designed around that...

> Actually, this assumption about tbs is false, or rather, conditional to other functions being called, such as EVP_PKEY_CTX_set_signature_md() (and for RSA specifically, also requires setting a padding other than...

> But that doesn't change the current design of EVP_PKEY_sign(). Hmm. Is this an OpenSSL 3.x regression? At least in the OpenSSL 1.1.x days, `EVP_PKEY_sign` was always SignDigest and never...

A few TLSWG threads to call to your attention. As we've been reviewing the documents, we noticed a few hiccups in parts of the protocol: https://mailarchive.ietf.org/arch/msg/tls/6y8wTv8Q_IPM-PCcbCAmDOYg6bM/ https://mailarchive.ietf.org/arch/msg/tls/3IEIzc2ssdkZbyYEXIvz6t3ZhbI/ https://mailarchive.ietf.org/arch/msg/tls/_ku3-YDcroNmG_QKZsYTtqYzC0M/ https://mailarchive.ietf.org/arch/msg/tls/GYX_teYy5CTFiGCBgbQJQwv_Fj4/ (Thoughts...

> When SSL master key is exchanged using plain rsa encryption, the special RSA_OAEP1_WITH_TLS_PADDING mode is used to prevent Bleichenbacher's attack on PKCS #1 v1.5 RSA padding. I'm guessing this...

Yup, by removing the SIGILL probing as I've been suggesting. :-) BoringSSL does not have any SIGILL code and we've never run into any problems of that nature. The expectation...

@levitte Are you envisioning that CPython would dynamically find these files at runtime and parse them? Or that this would be a build-time process? If runtime, isn't this just a...