make note of subkernel id, and destination at some point

nac3artiq/demo/embedding_map.py

@@ -1,11 +1,18 @@
 class EmbeddingMap:
-    def __init__(self):
+    def __init__(self, old_embedding_map=None):
         self.object_inverse_map = {}
         self.object_map = {}
         self.string_map = {}
         self.string_reverse_map = {}
         self.function_map = {}
         self.attributes_writeback = []
+        self.subkernel_message_map = {}
+
+        if not old_embedding_map is None:
+            for key, obj_ref in old_embedding_map.subkernels().items():
+                self.object_map[key] = obj_ref
+                obj_id = id(obj_ref)
+                self.object_inverse_map[obj_id] = key
 
         # preallocate exception names
         self.preallocate_runtime_exception_names(["RuntimeError",
@@ -64,3 +71,23 @@ class EmbeddingMap:
     def retrieve_str(self, key):
         return self.string_map[key]
 
+    def subkernels(self):
+        subkernels = {}
+        for k, v in self.object_map.items():
+            if getattr(v, "__artiq_destination__") is not None:
+                subkernels[k] = v
+        return subkernels
+    
+    def subkernel_messages(self):
+        messages = {}
+        for msg_id in self.subkernel_message_map.values():
+            messages[msg_id] = self.retrieve_object(msg_id)
+        return messages
+
+    def subkernel_messages_unpaired(self):
+        unpaired = []
+        for msg_id in self.subkernel_message_map.values():
+            msg_obj = self.retrieve_object(msg_id)
+            if msg_obj.send_loc is None or msg_obj.recv_loc is None:
+                unpaired.append(msg_obj)
+        return unpaired

nac3artiq/demo/min_artiq.py

@@ -130,8 +130,19 @@ def kernel(function_or_method):
         @wraps(function_or_method)
         def run_on_core(*args, **kwargs):
             raise RuntimeError("Kernel functions need explicit core.run()")
+    run_on_core.__artiq_kernel__ = True
+    run_on_core.__artiq_destination__ = None
     return run_on_core
 
+def subkernel(function_or_method, destination=0):
+    """Decorates a function or method to be executed on a satellite core device."""
+    _register_function(function_or_method)
+    @wraps(function_or_method)
+    def run_on_core(*args, **kwargs):
+        raise RuntimeError("Subkernels cannot be called by the host")
+    run_on_core.__artiq_kernel__ = True
+    run_on_core.__artiq_destination__ = destination
+    return function_or_method
 
 def portable(function):
     """Decorates a function or method to be executed on the same device (host/core device) as the caller."""

nac3artiq/src/lib.rs

@@ -196,6 +196,7 @@ impl Nac3 {
                             decorator_list.iter().any(|decorator| {
                                 if let ExprKind::Name { id, .. } = decorator.node {
                                     id.to_string() == "kernel"
+                                        || id.to_string() == "subkernel"
                                         || id.to_string() == "portable"
                                         || id.to_string() == "rpc"
                                 } else {
@@ -212,7 +213,7 @@ impl Nac3 {
                     decorator_list.iter().any(|decorator| {
                         if let ExprKind::Name { id, .. } = decorator.node {
                             let id = id.to_string();
-                            id == "extern" || id == "portable" || id == "kernel" || id == "rpc"
+                            id == "extern" || id == "portable" || id == "kernel" || id == "subkernel" || id == "rpc"
                         } else {
                             false
                         }
@@ -414,6 +415,7 @@ impl Nac3 {
         let mut module_to_resolver_cache: HashMap<u64, _> = HashMap::new();
 
         let mut rpc_ids = vec![];
+        let mut subkernel_ids = vec![];
         for (stmt, path, module) in &self.top_levels {
             let py_module: &PyAny = module.extract(py)?;
             let module_id: u64 = id_fn.call1((py_module,))?.extract()?;
@@ -481,6 +483,10 @@ impl Nac3 {
                     if decorator_list.iter().any(|decorator| matches!(decorator.node, ExprKind::Name { id, .. } if id == "rpc".into())) {
                         store_fun.call1(py, (def_id.0.into_py(py), module.getattr(py, name.to_string().as_str()).unwrap())).unwrap();
                         rpc_ids.push((None, def_id));
+                    } else if decorator_list.iter().any(|decorator| matches!(decorator.node, ExprKind::Name { id, .. } if id == "subkernel".into())) {
+                        store_fun.call1(py, (def_id.0.into_py(py), module.getattr(py, name.to_string().as_str()).unwrap())).unwrap();
+                        let dest = decorator_list.SOMETHING
+                        subkernel_rpc_ids.push((None, def_id, dest));
                     }
                 }
                 StmtKind::ClassDef { name, body, .. } => {
@@ -496,6 +502,15 @@ impl Nac3 {
                                     )));
                                 }
                                 rpc_ids.push((Some((class_obj.clone(), *name)), def_id));
+                            } else if decorator_list.iter().any(|decorator| matches!(decorator.node, ExprKind::Name { id, .. } if id == "subkernel".into())) {
+                                if name == &"__init__".into() {
+                                    return Err(CompileError::new_err(format!(
+                                        "compilation failed\n----------\nThe constructor of class {} should not be decorated with subkernel decorator (at {})",
+                                        class_name, stmt.location
+                                    )));
+                                }
+                                let dest = decorator_list.iter(). // figure out what's in the decorator list
+                                subkernel_ids.push((Some((class_obj.clone(), *name)), def_id, dest));
                             }
                         }
                     }

nac3artiq/src/subkernels.rs

@@ -0,0 +1,340 @@
+pub struct Subkernels {}
+
+impl Subkernels {
+    fn get_subkernel_builtins() -> Vec<Box<BuiltinFuncSpec>> {
+        vec![
+                Box::new(|primitives, unifier| {
+                    let arg_ty = unifier.get_fresh_var(Some("T".into()), None);
+                    (
+                        "subkernel_await".into(),
+                        FunSignature {
+                            args: vec![FuncArg {
+                                name: "subkernel".into(),
+                                ty: arg_ty.ty,
+                                default_value: None,
+                                is_vararg: false,
+                            },
+                            FuncArg {
+                                name: "timeout".into(),
+                                ty: primitives.int32,
+                                default_value: 0,
+                                is_vararg: false
+                            }],
+                            // todo: figure out how to return the type function returns 
+                            // or catch it in the type inferencer
+                            ret: primitives.none,
+                            vars: into_var_map([arg_ty]),
+                        },
+                        Arc::new(GenCall::new(Box::new(move |ctx, obj, fun, args, generator| {
+                            gen_subkernel_await(ctx, &obj, fun, &args, generator)?;
+    
+                            Ok(None)
+                        }))),
+                    )
+                }),
+                Box::new(|primitives, unifier| {
+                    let arg_ty = unifier.get_fresh_var(Some("T".into()), None);
+                    (
+                        "subkernel_preload".into(),
+                        FunSignature {
+                            args: vec![
+                                FuncArg {
+                                    name: "subkernel".into(),
+                                    ty: arg_ty.ty,
+                                    default_value: None,
+                                    is_vararg: false,
+                                }
+                            ],
+                            ret: primitives.none,
+                            vars: into_var_map([arg_ty]),
+                        },
+                        Arc::new(GenCall::new(Box::new(move |ctx, obj, fun, args, generator| {
+                            gen_subkernel_preload(ctx, &obj, fun, &args, generator)?;
+    
+                            Ok(None)
+                        }))),
+                    )
+                }),
+            ]
+    }
+
+    fn gen_subkernel_await<'ctx>(
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        obj: &Option<(Type, ValueEnum<'ctx>)>,
+        fun: (&FunSignature, DefinitionId),
+        args: &[(Option<StrRef>, ValueEnum<'ctx>)],
+        generator: &mut dyn CodeGenerator,
+    ) {
+        let sid_type = ctx.ctx.i32_type();
+        let timeout_type = ctx.ctx.i64_type();
+        assert!(matches!(args.len(), 1..=2));
+        let timeout = if args.len() == 1 {
+            timeout_type.const_zero().to_basic_value_enum(context, generator, obj_ty)?; // ?
+        } else { args[0].1.clone().to_basic_value_enum(context, generator, obj_ty)?; } // ?
+
+        let subkernel_await_finish = ctx.module.get_function("subkernel_await_finish").unwrap_or_else(|| {
+            ctx.module.add_function(
+                "subkernel_await_finish",
+                ctx.ctx.void_type().fn_type(&[sid_type.into(), timeout_type.into()], false),
+                None,
+            )
+        });
+
+        // call or invoke
+        // generate RPC for receiving return value depending on fun ret
+    }
+
+    fn gen_subkernel_preload<'ctx>(
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        obj: &Option<(Type, ValueEnum<'ctx>)>,
+        fun: (&FunSignature, DefinitionId),
+        args: &[(Option<StrRef>, ValueEnum<'ctx>)],
+        generator: &mut dyn CodeGenerator,
+    ) {
+        assert_eq!(args.len(), 1);
+        let sid_type = ctx.ctx.i32_type();
+        let dest_type = ctx.ctx.i8_type();
+        let run_type = ctx.ctx.bool_type();
+        let subkernel_load_run = ctx.module.get_function("subkernel_load_run").unwrap_or_else(|| {
+            ctx.module.add_function(
+                "subkernel_load_run",
+                ctx.ctx.void_type().fn_type(&[sid_type.into(), dest_type.into(), run_type.into()], false),
+                None,
+            )
+        });
+
+        let subkernel_id = int32.const_int(fun.1 .0 as u64, false);
+        let destination = int32.const_int(fun.? as u64, false); // TODO
+        ctx.builder
+            .build_call_or_invoke(subkernel_load_run, &[subkernel_id.into(), destination.into(), run_type.const_zero()], "subkernel.preload")
+            .unwrap();
+    }
+
+    fn subkernel_callback_fn<'ctx>(
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        obj: Option<(Type, ValueEnum<'ctx>)>,
+        fun: (&FunSignature, DefinitionId),
+        args: Vec<(Option<StrRef>, ValueEnum<'ctx>)>,
+        generator: &mut dyn CodeGenerator,
+    ) -> Result<Option<BasicValueEnum<'ctx>>, String> {
+        let int8 = ctx.ctx.i8_type();
+        let int32 = ctx.ctx.i32_type();
+        let size_type = generator.get_size_type(ctx.ctx);
+        let ptr_type = int8.ptr_type(AddressSpace::default());
+        let tag_ptr_type = ctx.ctx.struct_type(&[ptr_type.into(), size_type.into()], false);
+    
+        let subkernel_id = int32.const_int(fun.1 .0 as u64, false);
+        let destination = int32.const_int(fun.? as u64, false); // TODO
+
+        // -- start the subkernel
+        let sid_type = ctx.ctx.i32_type();
+        let dest_type = ctx.ctx.i8_type();
+        let run_type = ctx.ctx.i1_type();
+        let subkernel_start = ctx.module.get_function("subkernel_load_run").unwrap_or_else(|| {
+            ctx.module.add_function(
+                "subkernel_load_run",
+                ctx.ctx.void_type().fn_type(
+                    &[
+                        ctx.ctx.void_type().fn_type(&[sid_type.into(), dest_type.into(), run_type.into()], false),
+                    ],
+                    false,
+                ),
+                None,
+            )
+        });
+        ctx.builder
+            .build_call_or_invoke(subkernel_start, &[subkernel_id.into(), destination.into(), run_type.const_int(1, false)], "subkernel.run")
+            .unwrap();
+        // -- setup rpc tags
+        let mut tag = Vec::new();
+        if obj.is_some() {
+            tag.push(b'O');
+        }
+        for arg in &fun.0.args {
+            gen_rpc_tag(ctx, arg.ty, &mut tag)?;
+        }
+        tag.push(b':');
+        gen_rpc_tag(ctx, fun.0.ret, &mut tag)?;
+    
+        let mut hasher = DefaultHasher::new();
+        tag.hash(&mut hasher);
+        let hash = format!("{}", hasher.finish());
+    
+        let tag_ptr = ctx
+            .module
+            .get_global(hash.as_str())
+            .unwrap_or_else(|| {
+                let tag_arr_ptr = ctx.module.add_global(
+                    int8.array_type(tag.len() as u32),
+                    None,
+                    format!("tagptr{}", fun.1 .0).as_str(),
+                );
+                tag_arr_ptr.set_initializer(&int8.const_array(
+                    &tag.iter().map(|v| int8.const_int(u64::from(*v), false)).collect::<Vec<_>>(),
+                ));
+                tag_arr_ptr.set_linkage(Linkage::Private);
+                let tag_ptr = ctx.module.add_global(tag_ptr_type, None, &hash);
+                tag_ptr.set_linkage(Linkage::Private);
+                tag_ptr.set_initializer(&ctx.ctx.const_struct(
+                    &[
+                        tag_arr_ptr.as_pointer_value().const_cast(ptr_type).into(),
+                        size_type.const_int(tag.len() as u64, false).into(),
+                    ],
+                    false,
+                ));
+                tag_ptr
+            })
+            .as_pointer_value();
+    
+        let arg_length = args.len() + usize::from(obj.is_some());
+    
+        let stackptr = call_stacksave(ctx, Some("rpc.stack"));
+        let args_ptr = ctx
+            .builder
+            .build_array_alloca(
+                ptr_type,
+                ctx.ctx.i32_type().const_int(arg_length as u64, false),
+                "argptr",
+            )
+            .unwrap();
+    
+        // -- rpc args handling
+        let mut keys = fun.0.args.clone();
+        let mut mapping = HashMap::new();
+        for (key, value) in args {
+            mapping.insert(key.unwrap_or_else(|| keys.remove(0).name), value);
+        }
+        // default value handling
+        for k in keys {
+            mapping
+                .insert(k.name, ctx.gen_symbol_val(generator, &k.default_value.unwrap(), k.ty).into());
+        }
+        // reorder the parameters
+        let mut real_params = fun
+            .0
+            .args
+            .iter()
+            .map(|arg| {
+                mapping
+                    .remove(&arg.name)
+                    .unwrap()
+                    .to_basic_value_enum(ctx, generator, arg.ty)
+                    .map(|llvm_val| (llvm_val, arg.ty))
+            })
+            .collect::<Result<Vec<(_, _)>, _>>()?;
+        if let Some(obj) = obj {
+            if let ValueEnum::Static(obj_val) = obj.1 {
+                real_params.insert(0, (obj_val.get_const_obj(ctx, generator), obj.0));
+            } else {
+                // should be an error here...
+                panic!("only host object is allowed");
+            }
+        }
+    
+        for (i, (arg, arg_ty)) in real_params.iter().enumerate() {
+            let arg_slot = format_rpc_arg(generator, ctx, (*arg, *arg_ty, i));
+            let arg_ptr = unsafe {
+                ctx.builder.build_gep(
+                    args_ptr,
+                    &[int32.const_int(i as u64, false)],
+                    &format!("rpc.arg{i}"),
+                )
+            }
+            .unwrap();
+            ctx.builder.build_store(arg_ptr, arg_slot).unwrap();
+        }
+    
+        // send the message
+        let subkernel_send = ctx.module.get_function("subkernel_send_message").unwrap_or_else(|| {
+            ctx.module.add_function(
+                "subkernel_send_message",
+                ctx.ctx.void_type().fn_type(
+                    &[
+                        int32.into(),
+                        tag_ptr_type.ptr_type(AddressSpace::default()).into(),
+                        ptr_type.ptr_type(AddressSpace::default()).into(),
+                    ],
+                    false,
+                ),
+                None,
+            )
+        });
+        ctx.builder
+            .build_call_or_invoke(subkernel_send, &[service_id.into(), tag_ptr.into(), args_ptr.into()], "rpc.send")
+            .unwrap();
+        // reclaim stack space used by arguments
+        call_stackrestore(ctx, stackptr);
+    }
+
+    pub fn subkernel_codegen_callback() -> Arc<GenCall> {
+        Arc::new(GenCall::new(Box::new(|ctx, obj, fun, args, generator| {
+            subkernel_codegen_callback_fn(ctx, obj, fun, args, generator)
+        })))
+    }
+
+    fn subkernel_recv_message<'ctx>(
+        ctx: &mut CodeGenContext<'ctx, '_>,
+        obj: Option<(Type, ValueEnum<'ctx>)>,
+        fun: (&FunSignature, DefinitionId),
+        args: Vec<(Option<StrRef>, ValueEnum<'ctx>)>,
+        generator: &mut dyn CodeGenerator,
+    ) -> Result<Option<BasicValueEnum<'ctx>>, String>  {
+        // -- receive value:
+        // T result = {
+        //   void *ret_ptr = alloca(sizeof(T));
+        //   void *ptr = ret_ptr;
+        //   loop: int size = rpc_recv(ptr);
+        //   // Non-zero: Provide `size` bytes of extra storage for variable-length data.
+        //   if(size) { ptr = alloca(size); goto loop; }
+        //   else *(T*)ret_ptr
+        // }
+        let rpc_recv = ctx.module.get_function("rpc_recv").unwrap_or_else(|| {
+            ctx.module.add_function("rpc_recv", int32.fn_type(&[ptr_type.into()], false), None)
+        });
+    
+        if ctx.unifier.unioned(fun.0.ret, ctx.primitives.none) {
+            ctx.build_call_or_invoke(rpc_recv, &[ptr_type.const_null().into()], "rpc_recv");
+            return Ok(None);
+        }
+    
+        let prehead_bb = ctx.builder.get_insert_block().unwrap();
+        let current_function = prehead_bb.get_parent().unwrap();
+        let head_bb = ctx.ctx.append_basic_block(current_function, "rpc.head");
+        let alloc_bb = ctx.ctx.append_basic_block(current_function, "rpc.continue");
+        let tail_bb = ctx.ctx.append_basic_block(current_function, "rpc.tail");
+    
+        let ret_ty = ctx.get_llvm_abi_type(generator, fun.0.ret);
+        let need_load = !ret_ty.is_pointer_type();
+        let slot = ctx.builder.build_alloca(ret_ty, "rpc.ret.slot").unwrap();
+        let slotgen = ctx.builder.build_bitcast(slot, ptr_type, "rpc.ret.ptr").unwrap();
+        ctx.builder.build_unconditional_branch(head_bb).unwrap();
+        ctx.builder.position_at_end(head_bb);
+    
+        let phi = ctx.builder.build_phi(ptr_type, "rpc.ptr").unwrap();
+        phi.add_incoming(&[(&slotgen, prehead_bb)]);
+        let alloc_size = ctx
+            .build_call_or_invoke(rpc_recv, &[phi.as_basic_value()], "rpc.size.next")
+            .unwrap()
+            .into_int_value();
+        let is_done = ctx
+            .builder
+            .build_int_compare(inkwell::IntPredicate::EQ, int32.const_zero(), alloc_size, "rpc.done")
+            .unwrap();
+    
+        ctx.builder.build_conditional_branch(is_done, tail_bb, alloc_bb).unwrap();
+        ctx.builder.position_at_end(alloc_bb);
+    
+        let alloc_ptr = ctx.builder.build_array_alloca(ptr_type, alloc_size, "rpc.alloc").unwrap();
+        let alloc_ptr = ctx.builder.build_bitcast(alloc_ptr, ptr_type, "rpc.alloc.ptr").unwrap();
+        phi.add_incoming(&[(&alloc_ptr, alloc_bb)]);
+        ctx.builder.build_unconditional_branch(head_bb).unwrap();
+    
+        ctx.builder.position_at_end(tail_bb);
+    
+        let result = ctx.builder.build_load(slot, "rpc.result").unwrap();
+        if need_load {
+            call_stackrestore(ctx, stackptr);
+        }
+        Ok(Some(result))
+    }
+}

nac3core/src/toplevel/composer.rs

@@ -2057,6 +2057,13 @@ impl TopLevelComposer {
                         instance_to_symbol.insert(String::new(), simple_name.to_string());
                         continue;
                     }
+                    if !decorator_list.is_empty()
+                        && matches!(&decorator_list[0].node,
+                            ast::ExprKind::Name{ id, .. } if id == &"subkernel".into())
+                    {
+                        instance_to_symbol.insert(String::new(), simple_name.to_string());
+                        continue;
+                    }
 
                     let fun_body = body
                         .into_iter()