finally

src/lib.rs

@@ -20,7 +20,7 @@ pub extern "C" fn run_asm(
     ly: u32,
     lz: u32,
     args_ptr: *const u64,
-) -> usize {
+) -> i32 {
     let kernel = match *OSX {
         true => {
             let mut lib_bytes: Vec<u8> = Vec::with_capacity(lib_sz as usize);

src/thread.rs

@@ -457,7 +457,7 @@ impl<'a> Thread<'a> {
                                 let (x, y, z) =
                                     (f16::from_bits(x), f16::from_bits(y), f16::from_bits(z));
                                 let ret = match op {
-                                    14 => Ok::<f16, Result<(), usize>>(f16::mul_add(x, y, z)),
+                                    14 => Ok::<f16, i32>(f16::mul_add(x, y, z)),
                                     15 => Ok(x + y),
                                     16 => Ok(x * y),
                                     17 => Ok(f16::min(x, y)),

src/work_group.rs

@@ -49,7 +49,7 @@ impl<'a> WorkGroup<'a> {
         };
     }
 
-    pub fn exec_waves(&mut self) -> Result<(), usize> {
+    pub fn exec_waves(&mut self) -> Result<(), i32> {
         let mut blocks = vec![];
         for z in 0..self.launch_bounds[2] {
             for y in 0..self.launch_bounds[1] {
@@ -78,7 +78,7 @@ impl<'a> WorkGroup<'a> {
         Ok(())
     }
 
-    fn exec_wave(&mut self, (wave_id, threads): (usize, &Vec<[u32; 3]>)) -> Result<(), usize> {
+    fn exec_wave(&mut self, (wave_id, threads): (usize, &Vec<[u32; 3]>)) -> Result<(), i32> {
         let wave_state = self.wave_state.get(&wave_id);
         let mut sds = match wave_state {
             Some(val) => val.6.clone(),
@@ -169,7 +169,7 @@ impl<'a> WorkGroup<'a> {
                     simm: None,
                     sgpr_co: &mut sgpr_co,
                 };
-                let ret = thread.interpret()?;
+                thread.interpret()?;
                 if thread.scalar {
                     pc = ((pc as isize) + 1 + (thread.pc_offset as isize)) as usize;
                     break;