diff --git a/arch/x86-64/loader-entry.asm b/arch/sysv_x86-64/loader-entry.asm
similarity index 100%
rename from arch/x86-64/loader-entry.asm
rename to arch/sysv_x86-64/loader-entry.asm
diff --git a/arch/sysv_x86-64/thread-entry.asm b/arch/sysv_x86-64/thread-entry.asm
new file mode 100644
index 0000000..6605efb
--- /dev/null
+++ b/arch/sysv_x86-64/thread-entry.asm
@@ -0,0 +1,42 @@
+
+bits 64
+
+section .text
+global _cia_clone
+
+; flags, &stack[-2], &parent_tid, &child_tid, 0
+
+; DESCRIPTION:
+;   Function to be called when using the _clone syscall
+;   We don't just call the _clone syscall directly because
+;   if we call the clone syscall wrapper that's defined in
+;   linux/sys, our stack will be bad and we won't be able
+;   to return from the syscall wrapper
+; PARAMETERS:
+;   rdi - flags
+;   rsi - stack_base
+;   rdx - parent_tid_ptr
+;   rcx - child_tid_ptr
+;   r8  - tls
+; RETURNS:
+;   i64 value
+;     0 if returning as a parent
+;     1 if returning as a child
+;     negative value if there was an error making the thread
+_cia_clone:
+    mov r10, rcx
+    sub rsi, 8
+    ; As a child thread we wanna return to the same place as the parent
+    mov rax, qword [rsp]
+    mov qword [rsi], rax
+    ; ; Copy child tid ptr
+    ; mov qword [rsi + 0], rcx
+    ; Call syscall right away, since the order of the first 5 arguments
+    ; matches with the argument order of the function
+    mov rax, 56 ; SYS_CLONE
+    syscall
+    ; Check to see if we're child
+    test eax, eax
+    jnz .exit
+.exit:
+    ret
diff --git a/arch/x86-64/chkstk.asm b/arch/x64_x86-64/chkstk.asm
similarity index 92%
rename from arch/x86-64/chkstk.asm
rename to arch/x64_x86-64/chkstk.asm
index d3dfb6d..fa03d93 100644
--- a/arch/x86-64/chkstk.asm
+++ b/arch/x64_x86-64/chkstk.asm
@@ -1,3 +1,4 @@
+
 bits 64
 
 segment .text
diff --git a/build.py b/build.py
index d05b757..5b00abd 100755
--- a/build.py
+++ b/build.py
@@ -75,16 +75,7 @@ if not os.path.exists('lib'):
 if not os.path.exists('bin'):
     os.mkdir('bin')
 
-loader_flags = [
-    '-Wl,-e,_dlstart',
-    '-Wl,--sort-section,alignment',
-    '-Wl,--sort-common',
-    '-Wl,--gc-sections',
-    '-Wl,--hash-style=both',
-    '-Wl,--no-undefined',
-    '-Wl,--exclude-libs=ALL'
-]
-
+target_abi = 'sysv'
 target_arch = 'x86-64'
 target_os = 'linux'
 
@@ -97,13 +88,24 @@ cc_includes.append('include')
 cc_includes.append('include/linux')
 
 # Build the dynamic loader
+loader_flags = [
+    '-Wl,-e,_dlstart',
+    '-Wl,--sort-section,alignment',
+    '-Wl,--sort-common',
+    '-Wl,--gc-sections',
+    '-Wl,--hash-style=both',
+    '-Wl,--no-undefined',
+    '-Wl,--exclude-libs=ALL'
+]
 print_step("Building lib/ld-cia.so\n")
-assemble_obj('bin/loader-entry.o', [f'arch/{target_arch}/loader-entry.asm'], ['-f "elf64"'])
+assemble_obj('bin/loader-entry.o', [f'arch/{target_abi}_{target_arch}/loader-entry.asm'], ['-f "elf64"'])
 compile_shared('lib/ld-cia.so', ['bin/loader-entry.o','loader/loader-self-reloc.c','loader/loader.c'], loader_flags)
 
 # Build the ciabatta
+print_step("Building lib/cia.a\n")
+assemble_obj('bin/thread-entry.o', [f'arch/{target_abi}_{target_arch}/thread-entry.asm'], ['-f "elf64"'])
 compile_obj('bin/cia.o', ['cia.c'])
-archive('lib/cia.a', ['bin/cia.o'])
+archive('lib/cia.a', ['bin/cia.o', 'bin/thread-entry.o'])
 
 # Build the test
-compile_exe('a', ['tests/hello.c', 'lib/cia.a'], ['-Wl,-dynamic-linker,lib/ld-cia.so'])
+compile_exe('a', ['tests/threaded.c', 'lib/cia.a'], ['-Wl,-dynamic-linker,lib/ld-cia.so'])
diff --git a/os/linux/tinyrt.c b/os/linux/tinyrt.c
index 8d41864..9c40c8d 100644
--- a/os/linux/tinyrt.c
+++ b/os/linux/tinyrt.c
@@ -1,6 +1,15 @@
 
 // See src/tinyrt.h file for the interface this file implements
 
+extern i64 _cia_clone(
+    u64 flags,
+    void *stack_base,
+    int *parent_tid,
+    int *child_tid,
+    void *tls,
+    u64 stack_size
+);
+
 _Noreturn static void _rt_program_exit(int code) {
     sys_exit(code);
 }
@@ -18,26 +27,27 @@ static _RT_Status _rt_thread_create(_RT_Thread *thread, void (*thread_fn)(void *
     if((i64)stack_base < 0) {
         return _RT_ERROR_GENERIC;
     }
-    u64 *stack = (void *)((u8 *)stack_base + stack_size);
-    stack[-1] = (u64)&&thread_return;
-    stack[-2] = 0;
+    void *stack = (u8*)stack_base + stack_size;
     // Create the new thread
     u64 flags = 0;
-    flags |= CLONE_CHILD_CLEARTID;
-    flags |= CLONE_PARENT_SETTID;
+    // flags |= CLONE_CHILD_CLEARTID;
+    // flags |= CLONE_PARENT_SETTID;
     flags |= CLONE_FS;
     flags |= CLONE_FILES;
     flags |= CLONE_SIGHAND;
     flags |= CLONE_THREAD;
     flags |= CLONE_VM;
-    int parent_tid = 0;
-    int child_tid = 0;
-    i64 cur_tid = sys_clone(flags, &stack[-2], &parent_tid, &child_tid, 0);
-thread_return:
-    if(cur_tid < 0) {
+    flags |= CLONE_SYSVSEM;
+    int *temp_permanent_storage = stack_base;
+    int *child_tid = &temp_permanent_storage[0];
+    int *parent_tid = &temp_permanent_storage[1];
+    *child_tid = 1;
+    *parent_tid = 0;
+    i64 ret = _cia_clone(flags, stack, parent_tid, child_tid, 0, stack_size);
+    if(ret < 0) {
         return _RT_ERROR_GENERIC;
     }
-    if(cur_tid == child_tid) {
+    if(!ret) {
         thread_fn(ctx);
     }
     return _RT_STATUS_OK;
diff --git a/tests/threaded.c b/tests/threaded.c
index 089b374..8623306 100644
--- a/tests/threaded.c
+++ b/tests/threaded.c
@@ -12,11 +12,14 @@ int thrd_func(void *arg) {
 }
 
 int main() {
-    {char string[] = "main thred: before\n";
+    {char string[] = "main thread: before!\n";
     fwrite(string, 1, sizeof string-1, stdout);}
+    
     thrd_t thrd;
     thrd_create(&thrd, thrd_func, NULL);
+    
     {char string[] = "main thread: after!\n";
     fwrite(string, 1, sizeof string-1, stdout);}
+    for(;;);
     return 0;
 }