Hi, I am using user-level context switching with linux makecontext and swapcontext for a discrete event simulation to schedule several user-level threads using only one pthread. Before, I was using several pthreads and BDWGC was working perfectly fine. Now, I get a segfault in GC_push_all_eager, when a GC_gcollect is triggered. For each user-level thread, a stack is allocated - not with GC_malloc, but with normal malloc. Could this segfault happen, because these user-level stacks are not known to the BDWGC and how could I possibly solve this issue?