Rémi Tricot-Le Breton

Thanks for your quick reply. > The idea behind this is that if LVS is disabled, then there is no fwmark functionality to enable or disable, and so the error...

I'm not entirely sure yet but I managed to reproduce a crash and it seems to occur in the thread sanitizer library itself. The crash happens in our deinit_signals function...

According to OPENSSL_init_crypto's manpage (https://www.openssl.org/docs/man1.1.1/man3/OPENSSL_init_crypto.html), it is rarely required to call this function explicitely since it is already done during OpenSSL's start up. The only time when this function needs...

My bad I did not realize it was added lately since it's an old function. I assumed that since the reg-tests were ok with all the versions I tested then...

Thanks for your report, there indeed seems to be lots of unmanaged stuff related to OCSP when you start playing with the SSL cert CLI commands. Some of them are...

The patch already merged should fix this precise issue. We kept a previous OCSP certid instead of reconstructing it when the certificate is changed.

I tried reproducing your bug but I didn't manage to. That seems like a strange bug because in the two cases you describe, the runtime execution should be exactly the...

I never managed to reproduce it and tried to look for variables/buffer that could have concurrent accesses between different calls of the converter but could not find anything. You latest...

I managed to reproduce (by happy accident) the first crash you had in sess_build_logline. It's still a bit puzzling, I don't have a fix yet but at least I have...

The error message is indeed wrong and should be changed. I did not think about using the option on certificates that don't have OCSP URIs. The attached patch fixes the...