gh-117494: extract the Instruction Sequence data structure into a separate file (#117496)

Author: iritkatriel

Include/internal/pycore_compile.h

@@ -9,6 +9,7 @@ extern "C" {
 #endif
 
 #include "pycore_symtable.h"  // _Py_SourceLocation
+#include "pycore_instruction_sequence.h"
 
 struct _arena;   // Type defined in pycore_pyarena.h
 struct _mod;     // Type defined in pycore_ast.h
@@ -37,38 +38,6 @@ extern int _PyAST_Optimize(
     int optimize,
     int ff_features);
 
-typedef struct {
-    int h_label;
-    int h_startdepth;
-    int h_preserve_lasti;
-} _PyCompile_ExceptHandlerInfo;
-
-typedef struct {
-    int i_opcode;
-    int i_oparg;
-    _Py_SourceLocation i_loc;
-    _PyCompile_ExceptHandlerInfo i_except_handler_info;
-
-    /* Used by the assembler */
-    int i_target;
-    int i_offset;
-} _PyCompile_Instruction;
-
-typedef struct {
-    _PyCompile_Instruction *s_instrs;
-    int s_allocated;
-    int s_used;
-
-    int *s_labelmap;       /* label id --> instr offset */
-    int s_labelmap_size;
-    int s_next_free_label; /* next free label id */
-} _PyCompile_InstructionSequence;
-
-int _PyCompile_InstructionSequence_UseLabel(_PyCompile_InstructionSequence *seq, int lbl);
-int _PyCompile_InstructionSequence_Addop(_PyCompile_InstructionSequence *seq,
-                                         int opcode, int oparg,
-                                         _Py_SourceLocation loc);
-int _PyCompile_InstructionSequence_ApplyLabelMap(_PyCompile_InstructionSequence *seq);
 
 typedef struct {
     PyObject *u_name;

Include/internal/pycore_flowgraph.h

@@ -8,16 +8,13 @@ extern "C" {
 #  error "this header requires Py_BUILD_CORE define"
 #endif
 
-#include "pycore_opcode_utils.h"
 #include "pycore_compile.h"
-
-typedef struct {
-    int id;
-} _PyCfgJumpTargetLabel;
+#include "pycore_instruction_sequence.h"
+#include "pycore_opcode_utils.h"
 
 struct _PyCfgBuilder;
 
-int _PyCfgBuilder_UseLabel(struct _PyCfgBuilder *g, _PyCfgJumpTargetLabel lbl);
+int _PyCfgBuilder_UseLabel(struct _PyCfgBuilder *g, _PyJumpTargetLabel lbl);
 int _PyCfgBuilder_Addop(struct _PyCfgBuilder *g, int opcode, int oparg, _Py_SourceLocation loc);
 
 struct _PyCfgBuilder* _PyCfgBuilder_New(void);
@@ -27,14 +24,14 @@ int _PyCfgBuilder_CheckSize(struct _PyCfgBuilder* g);
 int _PyCfg_OptimizeCodeUnit(struct _PyCfgBuilder *g, PyObject *consts, PyObject *const_cache,
                             int nlocals, int nparams, int firstlineno);
 
-int _PyCfg_ToInstructionSequence(struct _PyCfgBuilder *g, _PyCompile_InstructionSequence *seq);
+int _PyCfg_ToInstructionSequence(struct _PyCfgBuilder *g, _PyInstructionSequence *seq);
 int _PyCfg_OptimizedCfgToInstructionSequence(struct _PyCfgBuilder *g, _PyCompile_CodeUnitMetadata *umd,
                                              int code_flags, int *stackdepth, int *nlocalsplus,
-                                             _PyCompile_InstructionSequence *seq);
+                                             _PyInstructionSequence *seq);
 
 PyCodeObject *
 _PyAssemble_MakeCodeObject(_PyCompile_CodeUnitMetadata *u, PyObject *const_cache,
-                           PyObject *consts, int maxdepth, _PyCompile_InstructionSequence *instrs,
+                           PyObject *consts, int maxdepth, _PyInstructionSequence *instrs,
                            int nlocalsplus, int code_flags, PyObject *filename);
 
 #ifdef __cplusplus

Include/internal/pycore_instruction_sequence.h

@@ -0,0 +1,60 @@
+#ifndef Py_INTERNAL_INSTRUCTION_SEQUENCE_H
+#define Py_INTERNAL_INSTRUCTION_SEQUENCE_H
+
+#ifndef Py_BUILD_CORE
+#  error "this header requires Py_BUILD_CORE define"
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef struct {
+    int h_label;
+    int h_startdepth;
+    int h_preserve_lasti;
+} _PyExceptHandlerInfo;
+
+typedef struct {
+    int i_opcode;
+    int i_oparg;
+    _Py_SourceLocation i_loc;
+    _PyExceptHandlerInfo i_except_handler_info;
+
+    /* Temporary fields, used by the assembler and in instr_sequence_to_cfg */
+    int i_target;
+    int i_offset;
+} _PyInstruction;
+
+typedef struct {
+    _PyInstruction *s_instrs;
+    int s_allocated;
+    int s_used;
+
+    int s_next_free_label; /* next free label id */
+    /* Map of a label id to instruction offset (index into s_instrs).
+     * If s_labelmap is NULL, then each label id is the offset itself.
+     */
+    int *s_labelmap;       /* label id --> instr offset */
+    int s_labelmap_size;
+} _PyInstructionSequence;
+
+typedef struct {
+    int id;
+} _PyJumpTargetLabel;
+
+int _PyInstructionSequence_UseLabel(_PyInstructionSequence *seq, int lbl);
+int _PyInstructionSequence_Addop(_PyInstructionSequence *seq,
+                                 int opcode, int oparg,
+                                 _Py_SourceLocation loc);
+_PyJumpTargetLabel _PyInstructionSequence_NewLabel(_PyInstructionSequence *seq);
+int _PyInstructionSequence_ApplyLabelMap(_PyInstructionSequence *seq);
+int _PyInstructionSequence_InsertInstruction(_PyInstructionSequence *seq, int pos,
+                                             int opcode, int oparg, _Py_SourceLocation loc);
+void PyInstructionSequence_Fini(_PyInstructionSequence *seq);
+
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* !Py_INTERNAL_INSTRUCTION_SEQUENCE_H */

Makefile.pre.in

@@ -442,6 +442,7 @@ PYTHON_OBJS=	\
 		Python/initconfig.o \
 		Python/interpconfig.o \
 		Python/instrumentation.o \
+		Python/instruction_sequence.o \
 		Python/intrinsics.o \
 		Python/jit.o \
 		Python/legacy_tracing.o \
@@ -1170,6 +1171,7 @@ PYTHON_HEADERS= \
 		$(srcdir)/Include/internal/pycore_importdl.h \
 		$(srcdir)/Include/internal/pycore_initconfig.h \
 		$(srcdir)/Include/internal/pycore_instruments.h \
+		$(srcdir)/Include/internal/pycore_instruction_sequence.h \
 		$(srcdir)/Include/internal/pycore_interp.h \
 		$(srcdir)/Include/internal/pycore_intrinsics.h \
 		$(srcdir)/Include/internal/pycore_jit.h \
@@ -1800,7 +1802,7 @@ regen-sre:
 		$(srcdir)/Modules/_sre/sre_constants.h \
 		$(srcdir)/Modules/_sre/sre_targets.h
 
-Python/compile.o Python/symtable.o Python/ast_unparse.o Python/ast.o Python/future.o: $(srcdir)/Include/internal/pycore_ast.h
+Python/compile.o Python/symtable.o Python/ast_unparse.o Python/ast.o Python/future.o: $(srcdir)/Include/internal/pycore_ast.h $(srcdir)/Include/internal/pycore_ast.h
 
 Python/getplatform.o: $(srcdir)/Python/getplatform.c
 		$(CC) -c $(PY_CORE_CFLAGS) -DPLATFORM='"$(MACHDEP)"' -o $@ $(srcdir)/Python/getplatform.c
@@ -1935,8 +1937,12 @@ regen-uop-metadata:
 	    $(srcdir)/Include/internal/pycore_uop_metadata.h.new $(srcdir)/Python/bytecodes.c
 	$(UPDATE_FILE) $(srcdir)/Include/internal/pycore_uop_metadata.h $(srcdir)/Include/internal/pycore_uop_metadata.h.new
 
-
-Python/compile.o: $(srcdir)/Include/internal/pycore_opcode_metadata.h
+Python/compile.o Python/assemble.o Python/flowgraph.o Python/instruction_sequence.o: \
+                $(srcdir)/Include/internal/pycore_compile.h \
+                $(srcdir)/Include/internal/pycore_flowgraph.h \
+                $(srcdir)/Include/internal/pycore_instruction_sequence.h \
+                $(srcdir)/Include/internal/pycore_opcode_metadata.h \
+                $(srcdir)/Include/internal/pycore_opcode_utils.h
 
 Python/ceval.o: \
 		$(srcdir)/Python/ceval_macros.h \

Misc/NEWS.d/next/Core and Builtins/2024-04-04-13-42-59.gh-issue-117494.GPQH64.rst

@@ -0,0 +1 @@
+Refactored the instruction sequence data structure out of compile.c into instruction_sequence.c.

PCbuild/_freeze_module.vcxproj

@@ -222,6 +222,7 @@
     <ClCompile Include="..\Python\import.c" />
     <ClCompile Include="..\Python\importdl.c" />
     <ClCompile Include="..\Python\initconfig.c" />
+    <ClCompile Include="..\Python\instruction_sequence.c" />
     <ClCompile Include="..\Python\interpconfig.c" />
     <ClCompile Include="..\Python\intrinsics.c" />
     <ClCompile Include="..\Python\instrumentation.c" />

PCbuild/_freeze_module.vcxproj.filters

@@ -229,6 +229,7 @@
     <ClCompile Include="..\Python\initconfig.c">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="..\Python\instruction_sequence.c">
     <ClCompile Include="..\Python\interpconfig.c">
       <Filter>Source Files</Filter>
     </ClCompile>

PCbuild/pythoncore.vcxproj

@@ -248,6 +248,7 @@
     <ClInclude Include="..\Include\internal\pycore_import.h" />
     <ClInclude Include="..\Include\internal\pycore_importdl.h" />
     <ClInclude Include="..\Include\internal\pycore_initconfig.h" />
+    <ClInclude Include="..\Include\internal\pycore_instruction_sequence.h" />
     <ClInclude Include="..\Include\internal\pycore_interp.h" />
     <ClInclude Include="..\Include\internal\pycore_intrinsics.h" />
     <ClInclude Include="..\Include\internal\pycore_jit.h" />
@@ -590,6 +591,7 @@
     <ClCompile Include="..\Python\initconfig.c" />
     <ClCompile Include="..\Python\interpconfig.c" />
     <ClCompile Include="..\Python\intrinsics.c" />
+    <ClCompile Include="..\Python\instruction_sequence.c" />
     <ClCompile Include="..\Python\instrumentation.c" />
     <ClCompile Include="..\Python\jit.c" />
     <ClCompile Include="..\Python\legacy_tracing.c" />

PCbuild/pythoncore.vcxproj.filters

@@ -663,6 +663,9 @@
     <ClInclude Include="..\Include\internal\pycore_initconfig.h">
       <Filter>Include\internal</Filter>
     </ClInclude>
+    <ClInclude Include="..\Include\internal\pycore_instruction_sequence.h">
+      <Filter>Include\internal</Filter>
+    </ClInclude>
     <ClInclude Include="..\Include\internal\pycore_interp.h">
       <Filter>Include\internal</Filter>
     </ClInclude>
@@ -1349,6 +1352,9 @@
     <ClCompile Include="..\Python\intrinsics.c">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="..\Python\instruction_sequence.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
     <ClCompile Include="..\Python\instrumentation.c">
       <Filter>Source Files</Filter>
     </ClCompile>

Python/assemble.c

@@ -3,6 +3,7 @@
 #include "Python.h"
 #include "pycore_code.h"            // write_location_entry_start()
 #include "pycore_compile.h"
+#include "pycore_instruction_sequence.h"
 #include "pycore_opcode_utils.h"    // IS_BACKWARDS_JUMP_OPCODE
 #include "pycore_opcode_metadata.h" // is_pseudo_target, _PyOpcode_Caches
 #include "pycore_symtable.h"        // _Py_SourceLocation
@@ -23,8 +24,8 @@
     }
 
 typedef _Py_SourceLocation location;
-typedef _PyCompile_Instruction instruction;
-typedef _PyCompile_InstructionSequence instr_sequence;
+typedef _PyInstruction instruction;
+typedef _PyInstructionSequence instr_sequence;
 
 static inline bool
 same_location(location a, location b)
@@ -132,7 +133,7 @@ assemble_emit_exception_table_item(struct assembler *a, int value, int msb)
 static int
 assemble_emit_exception_table_entry(struct assembler *a, int start, int end,
                                     int handler_offset,
-                                    _PyCompile_ExceptHandlerInfo *handler)
+                                    _PyExceptHandlerInfo *handler)
 {
     Py_ssize_t len = PyBytes_GET_SIZE(a->a_except_table);
     if (a->a_except_table_off + MAX_SIZE_OF_ENTRY >= len) {
@@ -158,7 +159,7 @@ static int
 assemble_exception_table(struct assembler *a, instr_sequence *instrs)
 {
     int ioffset = 0;
-    _PyCompile_ExceptHandlerInfo handler;
+    _PyExceptHandlerInfo handler;
     handler.h_label = -1;
     handler.h_startdepth = -1;
     handler.h_preserve_lasti = -1;
@@ -736,8 +737,7 @@ _PyAssemble_MakeCodeObject(_PyCompile_CodeUnitMetadata *umd, PyObject *const_cac
                            PyObject *consts, int maxdepth, instr_sequence *instrs,
                            int nlocalsplus, int code_flags, PyObject *filename)
 {
-
-    if (_PyCompile_InstructionSequence_ApplyLabelMap(instrs) < 0) {
+    if (_PyInstructionSequence_ApplyLabelMap(instrs) < 0) {
         return NULL;
     }
     if (resolve_unconditional_jumps(instrs) < 0) {