Sebastien Marie

I could reproduce under OpenBSD with the following gdb backtrace: ``` Thread 1 received signal SIGSEGV, Segmentation fault. 0x00000ab4d6f705dc in std::__1::__cxx_atomic_load (__a=0x58, __order=std::__1::memory_order_seq_cst) at /usr/include/c++/v1/atomic:1009 1009 return __c11_atomic_load(const_cast(&__a->__a_value), static_cast(__order)); (gdb)...

@Skytrias it could be related in the sens that both case are due to `entity_from_expr()` returning `nullptr` because the entity is `Ast_ProcLit`.

I am reproducing on OpenBSD too. It segfaults due to failing assert [`GB_ASSERT(proc->kind == Entity_Procedure)`](https://github.com/odin-lang/Odin/blob/fd415f0b45ad7e07a72fc3938c9a22387f3556dc/src/llvm_backend_general.cpp#L1273). ``` Thread 1 received signal SIGSEGV, Segmentation fault. 0x0000014da01ba035 in lb_set_nested_type_name_ir_mangled_name (e=0x15050e80720, p=0x0) at src/llvm_backend_general.c...

correcting myself: it failed because of segfault in the assert line: `proc` is `NULL` do deferencing it (for testing the assert) fails.

it segfaults at the same place in #1570. are you able to test the diff for workaround it (it requires recompiling odin binary) ?

thanks for testing it. so it might be something else

Setting an alignment (using `LLVMSetAlignment()`) makes llvm to generate a different opcode to access the memory. Before, it used `movaps` (Move Aligned Packed Single-Precision Floating-Point Values) and failed because the...

FYI, I migrated the repository to https://codeberg.org/semarie/sysclean

closing due to no response.

I am the original reported of the problem (on the public chat). I just want to add that I am still running bupstash on OpenBSD, on several hosts, several times...