nsi crash looks strange (can't repro locally, and other pipelines from recent mr don't show it) and is not related to the changes
context of the crash seems kinda broken (ebp = 0)
could likely improve backtrace to not repeat itself with null frame pointer