Explicitly store a module's location (#55963)

Revise wants to know what file a module's `module` definition is in.
Currently it does this by looking at the source location for the
implicitly generated `eval` method. This is terrible for two reasons:

1. The method may not exist if the module is a baremodule (which is not
particularly common, which is probably why we haven't seen it).
2. The fact that the implicitly generated `eval` method has this
location information is an implementation detail that I'd like to get
rid of (#55949).

This PR adds explicit file/line info to `Module`, so that Revise doesn't
have to use the hack anymore.

base/reflection.jl

@@ -76,6 +76,17 @@ function fullname(m::Module)
     return (fullname(mp)..., mn)
 end
 
+"""
+    moduleloc(m::Module) -> LineNumberNode
+
+Get the location of the `module` definition.
+"""
+function moduleloc(m::Module)
+    line = Ref{Int32}(0)
+    file = ccall(:jl_module_getloc, Ref{Symbol}, (Any, Ref{Int32}), m, line)
+    return LineNumberNode(Int(line[]), file)
+end
+
 """
     names(x::Module; all::Bool=false, imported::Bool=false, usings::Bool=false) -> Vector{Symbol}
 

src/jl_exported_funcs.inc

@@ -316,6 +316,7 @@
     XX(jl_module_name) \
     XX(jl_module_names) \
     XX(jl_module_parent) \
+    XX(jl_module_getloc) \
     XX(jl_module_public) \
     XX(jl_module_public_p) \
     XX(jl_module_use) \

src/julia.h

@@ -710,6 +710,8 @@ typedef struct _jl_module_t {
     struct _jl_module_t *parent;
     _Atomic(jl_svec_t*) bindings;
     _Atomic(jl_genericmemory_t*) bindingkeyset; // index lookup by name into bindings
+    jl_sym_t *file;
+    int32_t line;
     // hidden fields:
     arraylist_t usings;  // modules with all bindings potentially imported
     jl_uuid_t build_id;

src/module.c

@@ -52,6 +52,8 @@ JL_DLLEXPORT jl_module_t *jl_new_module_(jl_sym_t *name, jl_module_t *parent, ui
     m->compile = -1;
     m->infer = -1;
     m->max_methods = -1;
+    m->file = name; // Using the name as a placeholder is better than nothing
+    m->line = 0;
     m->hash = parent == NULL ? bitmix(name->hash, jl_module_type->hash) :
         bitmix(name->hash, parent->hash);
     JL_MUTEX_INIT(&m->lock, "module->lock");
@@ -1179,6 +1181,14 @@ jl_module_t *jl_module_root(jl_module_t *m)
     }
 }
 
+JL_DLLEXPORT jl_sym_t *jl_module_getloc(jl_module_t *m, int32_t *line)
+{
+    if (line) {
+        *line = m->line;
+    }
+    return m->file;
+}
+
 JL_DLLEXPORT jl_uuid_t jl_module_build_id(jl_module_t *m) { return m->build_id; }
 JL_DLLEXPORT jl_uuid_t jl_module_uuid(jl_module_t* m) { return m->uuid; }
 

src/staticdata.c

@@ -1259,6 +1259,9 @@ static void jl_write_module(jl_serializer_state *s, uintptr_t item, jl_module_t
     jl_atomic_store_relaxed(&newm->bindingkeyset, NULL);
     arraylist_push(&s->relocs_list, (void*)(reloc_offset + offsetof(jl_module_t, bindingkeyset)));
     arraylist_push(&s->relocs_list, (void*)backref_id(s, jl_atomic_load_relaxed(&m->bindingkeyset), s->link_ids_relocs));
+    newm->file = NULL;
+    arraylist_push(&s->relocs_list, (void*)(reloc_offset + offsetof(jl_module_t, file)));
+    arraylist_push(&s->relocs_list, (void*)backref_id(s, m->file, s->link_ids_relocs));
 
     // write out the usings list
     memset(&newm->usings._space, 0, sizeof(newm->usings._space));

src/toplevel.c

@@ -213,6 +213,10 @@ static jl_value_t *jl_eval_module_expr(jl_module_t *parent_module, jl_expr_t *ex
         form = NULL;
     }
 
+    newm->file = jl_symbol(filename);
+    jl_gc_wb_knownold(newm, newm->file);
+    newm->line = lineno;
+
     for (int i = 0; i < jl_array_nrows(exprs); i++) {
         // process toplevel form
         ct->world_age = jl_atomic_load_acquire(&jl_world_counter);