make objectid effects total

vtjnash · vtjnash · commit 841ebc66badb · 2023-11-10T20:52:09.000Z
Avoids the need to rehash most dictionaries on reload.

System image data size increase is about 109MB -&gt; 112MB, since there are
about 130k Arrays, 75k CodeInstances, 30k Methods, 80k TypeMapEntries,
75k MethodInstance, and 36k Core.Bindings, and other mutable objects.
diff --git a/base/reflection.jl b/base/reflection.jl
@@ -720,18 +720,10 @@ If `x === y` then `objectid(x) == objectid(y)`, and usually when `x !== y`, `obj
 
 See also [`hash`](@ref), [`IdDict`](@ref).
 """
-function objectid(x)
-    # objectid is foldable iff it isn't a pointer.
-    if isidentityfree(typeof(x))
-        return _foldable_objectid(x)
-    end
-    return _objectid(x)
-end
-function _foldable_objectid(@nospecialize(x))
-    @_foldable_meta
-    _objectid(x)
+function objectid(@nospecialize(x))
+    @_total_meta
+    return ccall(:jl_object_id, UInt, (Any,), x)
 end
-_objectid(@nospecialize(x)) = ccall(:jl_object_id, UInt, (Any,), x)
 
 """
     isdispatchtuple(T)
diff --git a/src/builtins.c b/src/builtins.c
@@ -344,6 +344,9 @@ static uintptr_t type_object_id_(jl_value_t *v, jl_varidx_t *env) JL_NOTSAFEPOIN
             i++;
             pe = pe->prev;
         }
+        uintptr_t bits = jl_astaggedvalue(v)->header;
+        if (bits & GC_IN_IMAGE)
+            return ((uintptr_t*)v)[-2];
         return inthash((uintptr_t)v);
     }
     if (tv == jl_uniontype_type) {
@@ -432,50 +435,56 @@ static uintptr_t immut_id_(jl_datatype_t *dt, jl_value_t *v, uintptr_t h) JL_NOT
     return h;
 }
 
-static uintptr_t NOINLINE jl_object_id__cold(jl_datatype_t *dt, jl_value_t *v) JL_NOTSAFEPOINT
+static uintptr_t NOINLINE jl_object_id__cold(uintptr_t tv, jl_value_t *v) JL_NOTSAFEPOINT
 {
-    if (dt == jl_simplevector_type)
-        return hash_svec((jl_svec_t*)v);
-    if (dt == jl_datatype_type) {
-        jl_datatype_t *dtv = (jl_datatype_t*)v;
-        uintptr_t h = ~dtv->name->hash;
-        return bitmix(h, hash_svec(dtv->parameters));
-    }
-    if (dt == jl_string_type) {
+    jl_datatype_t *dt = (jl_datatype_t*)jl_to_typeof(tv);
+    if (dt->name->mutabl) {
+        if (dt == jl_string_type) {
 #ifdef _P64
-        return memhash_seed(jl_string_data(v), jl_string_len(v), 0xedc3b677);
+            return memhash_seed(jl_string_data(v), jl_string_len(v), 0xedc3b677);
 #else
-        return memhash32_seed(jl_string_data(v), jl_string_len(v), 0xedc3b677);
+            return memhash32_seed(jl_string_data(v), jl_string_len(v), 0xedc3b677);
 #endif
-    }
-    if (dt == jl_module_type) {
-        jl_module_t *m = (jl_module_t*)v;
-        return m->hash;
-    }
-    if (dt->name->mutabl)
+        }
+        if (dt == jl_simplevector_type)
+            return hash_svec((jl_svec_t*)v);
+        if (dt == jl_datatype_type) {
+            jl_datatype_t *dtv = (jl_datatype_t*)v;
+            uintptr_t h = ~dtv->name->hash;
+            return bitmix(h, hash_svec(dtv->parameters));
+        }
+        if (dt == jl_module_type) {
+            jl_module_t *m = (jl_module_t*)v;
+            return m->hash;
+        }
+        uintptr_t bits = jl_astaggedvalue(v)->header;
+        if (bits & GC_IN_IMAGE)
+            return ((uintptr_t*)v)[-2];
         return inthash((uintptr_t)v);
+    }
     return immut_id_(dt, v, dt->hash);
 }
 
-JL_DLLEXPORT inline uintptr_t jl_object_id_(jl_value_t *tv, jl_value_t *v) JL_NOTSAFEPOINT
+JL_DLLEXPORT inline uintptr_t jl_object_id_(uintptr_t tv, jl_value_t *v) JL_NOTSAFEPOINT
 {
-    jl_datatype_t *dt = (jl_datatype_t*)tv;
-    if (dt == jl_symbol_type)
+    if (tv == jl_symbol_tag << 4) {
         return ((jl_sym_t*)v)->hash;
-    if (dt == jl_typename_type)
-        return ((jl_typename_t*)v)->hash;
-    if (dt == jl_datatype_type) {
+    }
+    else if (tv == jl_datatype_tag << 4) {
         jl_datatype_t *dtv = (jl_datatype_t*)v;
         if (dtv->isconcretetype)
             return dtv->hash;
     }
-    return jl_object_id__cold(dt, v);
+    else if (tv == (uintptr_t)jl_typename_type) {
+        return ((jl_typename_t*)v)->hash;
+    }
+    return jl_object_id__cold(tv, v);
 }
 
 
 JL_DLLEXPORT uintptr_t jl_object_id(jl_value_t *v) JL_NOTSAFEPOINT
 {
-    return jl_object_id_(jl_typeof(v), v);
+    return jl_object_id_(jl_typetagof(v), v);
 }
 
 // eq hash table --------------------------------------------------------------
diff --git a/src/ccall.cpp b/src/ccall.cpp
@@ -1880,7 +1880,7 @@ static jl_cgval_t emit_ccall(jl_codectx_t &ctx, jl_value_t **args, size_t nargs)
             if (!val.isghost && !val.ispointer())
                 val = value_to_pointer(ctx, val);
             Value *args[] = {
-                emit_typeof(ctx, val),
+                emit_typeof(ctx, val, false, true),
                 val.isghost ? ConstantPointerNull::get(T_pint8_derived) :
                     ctx.builder.CreateBitCast(
                         decay_derived(ctx, data_pointer(ctx, val)),
diff --git a/src/codegen.cpp b/src/codegen.cpp
@@ -1165,7 +1165,7 @@ static const auto jlapplytype_func = new JuliaFunction<>{
 static const auto jl_object_id__func = new JuliaFunction<TypeFnContextAndSizeT>{
     XSTR(jl_object_id_),
     [](LLVMContext &C, Type *T_size) { return FunctionType::get(T_size,
-            {JuliaType::get_prjlvalue_ty(C), PointerType::get(getInt8Ty(C), AddressSpace::Derived)}, false); },
+            {T_size, PointerType::get(getInt8Ty(C), AddressSpace::Derived)}, false); },
     nullptr,
 };
 static const auto setjmp_func = new JuliaFunction<TypeFnContextAndTriple>{
diff --git a/src/iddict.c b/src/iddict.c
@@ -5,7 +5,7 @@
 // compute empirical max-probe for a given size
 #define max_probe(size) ((size) <= 1024 ? 16 : (size) >> 6)
 
-#define keyhash(k) jl_object_id_(jl_typeof(k), k)
+#define keyhash(k) jl_object_id_(jl_typetagof(k), k)
 #define h2index(hv, sz) (size_t)(((hv) & ((sz)-1)) * 2)
 
 static inline int jl_table_assign_bp(jl_genericmemory_t **pa, jl_value_t *key, jl_value_t *val);
diff --git a/src/julia_internal.h b/src/julia_internal.h
@@ -1370,7 +1370,7 @@ JL_DLLEXPORT int jl_stored_inline(jl_value_t *el_type);
 JL_DLLEXPORT jl_value_t *(jl_array_data_owner)(jl_array_t *a);
 JL_DLLEXPORT jl_array_t *jl_array_copy(jl_array_t *ary);
 
-JL_DLLEXPORT uintptr_t jl_object_id_(jl_value_t *tv, jl_value_t *v) JL_NOTSAFEPOINT;
+JL_DLLEXPORT uintptr_t jl_object_id_(uintptr_t tv, jl_value_t *v) JL_NOTSAFEPOINT;
 JL_DLLEXPORT void jl_set_next_task(jl_task_t *task) JL_NOTSAFEPOINT;
 
 // -- synchronization utilities -- //
diff --git a/src/staticdata.c b/src/staticdata.c
@@ -500,8 +500,6 @@ typedef struct {
     int8_t incremental;
 } jl_serializer_state;
 
-static jl_value_t *jl_idtable_type = NULL;
-static jl_typename_t *jl_idtable_typename = NULL;
 static jl_value_t *jl_bigint_type = NULL;
 static int gmp_limb_size = 0;
 static jl_sym_t *jl_docmeta_sym = NULL;
@@ -1252,24 +1250,35 @@ static void jl_write_values(jl_serializer_state *s) JL_GC_DISABLED
         assert(!(s->incremental && jl_object_in_image(v)));
         jl_datatype_t *t = (jl_datatype_t*)jl_typeof(v);
         assert((!jl_is_datatype_singleton(t) || t->instance == v) && "detected singleton construction corruption");
+        int mutabl = t->name->mutabl;
         ios_t *f = s->s;
         if (t->smalltag) {
             if (t->layout->npointers == 0 || t == jl_string_type) {
-                if (jl_datatype_nfields(t) == 0 || t->name->mutabl == 0 || t == jl_string_type) {
+                if (jl_datatype_nfields(t) == 0 || mutabl == 0 || t == jl_string_type) {
                     f = s->const_data;
                 }
             }
         }
 
-        // realign stream to expected gc alignment (16 bytes)
+        // realign stream to expected gc alignment (16 bytes) after tag
         uintptr_t skip_header_pos = ios_pos(f) + sizeof(jl_taggedvalue_t);
+        uintptr_t object_id_expected = mutabl &&
+                 t != jl_datatype_type &&
+                 t != jl_typename_type &&
+                 t != jl_string_type &&
+                 t != jl_simplevector_type &&
+                 t != jl_module_type;
+        if (object_id_expected)
+            skip_header_pos += sizeof(size_t);
         write_padding(f, LLT_ALIGN(skip_header_pos, 16) - skip_header_pos);
 
         // write header
+        if (object_id_expected)
+            write_uint(f, jl_object_id(v));
         if (s->incremental && jl_needs_serialization(s, (jl_value_t*)t) && needs_uniquing((jl_value_t*)t))
             arraylist_push(&s->uniquing_types, (void*)(uintptr_t)(ios_pos(f)|1));
         if (f == s->const_data)
-            write_uint(s->const_data, ((uintptr_t)t->smalltag << 4) | GC_OLD_MARKED);
+            write_uint(s->const_data, ((uintptr_t)t->smalltag << 4) | GC_OLD_MARKED | GC_IN_IMAGE);
         else
             write_gctaggedfield(s, t);
         size_t reloc_offset = ios_pos(f);
@@ -1717,11 +1726,6 @@ static void jl_write_values(jl_serializer_state *s) JL_GC_DISABLED
                     arraylist_push(&s->fixup_objs, (void*)reloc_offset);
                 }
             }
-            else if (((jl_datatype_t*)(jl_typeof(v)))->name == jl_idtable_typename) {
-                assert(f == s->s);
-                // will need to rehash this, later (after types are fully constructed)
-                arraylist_push(&s->fixup_objs, (void*)reloc_offset);
-            }
             else if (jl_is_genericmemoryref(v)) {
                 assert(f == s->s);
                 record_memoryref(s, reloc_offset, *(jl_genericmemoryref_t*)v);
@@ -2581,8 +2585,6 @@ static void jl_save_system_image_to_stream(ios_t *f, jl_array_t *mod_array,
             }
         }
     }
-    jl_idtable_type = jl_base_module ? jl_get_global(jl_base_module, jl_symbol("IdDict")) : NULL;
-    jl_idtable_typename = jl_base_module ? ((jl_datatype_t*)jl_unwrap_unionall((jl_value_t*)jl_idtable_type))->name : NULL;
     jl_bigint_type = jl_base_module ? jl_get_global(jl_base_module, jl_symbol("BigInt")) : NULL;
     if (jl_bigint_type) {
         gmp_limb_size = jl_unbox_long(jl_get_global((jl_module_t*)jl_get_global(jl_base_module, jl_symbol("GMP")),
@@ -3408,12 +3410,7 @@ static void jl_restore_system_image_from_stream_(ios_t *f, jl_image_t *image, jl
             }
         }
         else {
-            // rehash IdDict
-            //assert(((jl_datatype_t*)(jl_typeof(obj)))->name == jl_idtable_typename);
-            jl_genericmemory_t **a = (jl_genericmemory_t**)obj;
-            assert(jl_typetagis(*a, jl_memory_any_type));
-            *a = jl_idtable_rehash(*a, (*a)->length);
-            jl_gc_wb(obj, *a);
+            abort();
         }
     }
     // Now pick up the globalref binding pointer field
diff --git a/test/precompile.jl b/test/precompile.jl
@@ -288,6 +288,9 @@ precompile_test_harness(false) do dir
               a_vec_inline = Pair{Int,Any}[]
               push!(a_vec_inline, 1=>2, 3=>4)
               a_mat_inline = reshape(a_vec_inline, (1, 2))
+
+              oid_vec_int = objectid(a_vec_int)
+              oid_mat_int = objectid(a_mat_int)
           end
           """)
     # Issue #12623
@@ -371,6 +374,10 @@ precompile_test_harness(false) do dir
         @test Foo.a_mat_inline == Pair{Int,Any}[1=>2 3=>4]
         Foo.a_mat_inline[1, 2] = 5=>6
         @test Foo.a_vec_inline[2] === Pair{Int,Any}(5, 6)
+
+        @test objectid(Foo.a_vec_int) === Foo.oid_vec_int
+        @test objectid(Foo.a_mat_int) === Foo.oid_mat_int
+        @test Foo.oid_vec_int !== Foo.oid_mat_int
     end
 
     @eval begin function ccallable_test()
