On Wed, 24 Mar 2021 14:04:13 +1100 Matt Youill <[email protected]> wrote: > So this seems to be caused by the variable in memory_pool.cc: > > const util::optional<MemoryPoolBackend> user_selected_backend = > UserSelectedBackend(); > > being (or becoming) garbage. > > For some reason, after a few Gandiva batch evaluations > user_selected_backend is no longer "jemalloc" but "system" (probably > actually just null because "system" is 0) and after a while it isn't > valid at all and crashes.
Hmm... perhaps you can try to set up a watchpoint on this variable under gdb? https://sourceware.org/gdb/current/onlinedocs/gdb/Set-Watchpoints.html Though I don't know if it would catch mutations due to memory corruption. Regards Antoine.
