Manuel Pégourié-Gonnard

@ronald-cron-arm Can you add appropriate labels? (So that this PR stops showing when searching for PRs that need triaging.) Thanks!

@gilles-peskine-arm A number of my thoughts here are actually 4.0 items - https://github.com/Mbed-TLS/mbedtls/pull/9236#discussion_r1749897761 Rename `MBEDTLS_CIPHER_NULL_CIPHER` to a TLS option (enabling integrity-only ciphersuites). - https://github.com/Mbed-TLS/mbedtls/pull/9236#discussion_r1749898802 Organise `mbedtls_config.h` into X.509 / TLS...

> The section links do not work in the rich diff display. Not sure if this is a real issue or just an issue with the tool. I tried a...

I can confirm that installing abi-dumper 1.2 locally, using the [same method as in the 16.04 Dockerfile](https://github.com/Mbed-TLS/mbedtls-test/blob/master/resources/docker_files/ubuntu-16.04/Dockerfile#L242) (but s/1.1/1.2/) solves the problem on my local Ubuntu 20.04 machine. We should...

FWIW, using `abi-dumper` 1.3 (aka current head of the master branch) works for me on my Ubuntu 24.04 machine. That is, with this version, `abi_check.py` finds compatibility issues between 2.28...

Hi @jforissier, did you also measure performance on real hardware? I tried it on two boards, based on Cortex-M0 and Cortex-M4 cores, and the performance regression compared to 2.25 (the...

FWIW, I've also done a few measurements on A-class (v7 and v8) and in both cases it looks like the change of the default window size was enough to compensate...

Thanks for the new data. This makes a lot of sense: we made our `exp_mod` function constant-time in 3.6, which is needed for the security of private operations: it should...

I'm labelling this "bug" and adding to the 3.6.1 EPIC because I don't think a 10x regression on RSA public key operations is OK. (And even more if for some...

See first draft of an attempt to reduce the regression: https://github.com/Mbed-TLS/mbedtls/pull/9281