2026-07-27, 12:30:48 +0200, Sabrina Dubroca wrote: > This should never happen. We need to fix whatever path leads to that > invalid condition.
Your diff is right. Tested on net (53658c6f3682), stock config: without it the test in 2/2 oopses in the AEAD walk, with it applied there is no oops and no regression in the tls selftests. Agreed, my patch was treating the symptom. Sorry about the verbosity of v1. Happy to send a v2 with your diff if you'd like me to carry it, or leave it to you. Should Fixes: stay at fe1e81d4f73b? Thanks, chanyoung
