Add source file name and line number to error messages.

Byte code has a map from byte-code offset to source-code line number,
used to give better error messages.

py/bc.h

 mp_obj_t mp_execute_byte_code(const byte *code, const mp_obj_t *args, uint n_args, uint n_state);
-bool mp_execute_byte_code_2(const byte **ip_in_out, mp_obj_t *fastn, mp_obj_t **sp_in_out);
+bool mp_execute_byte_code_2(const byte *code_info, const byte **ip_in_out, mp_obj_t *fastn, mp_obj_t **sp_in_out);

py/builtineval.c

@@ -28,6 +28,7 @@ static mp_obj_t mp_builtin_eval(mp_obj_t o_in) {
     qstr parse_exc_id;
     const char *parse_exc_msg;
     mp_parse_node_t pn = mp_parse(lex, MP_PARSE_EVAL_INPUT, &parse_exc_id, &parse_exc_msg);
+    qstr source_name = mp_lexer_source_name(lex);
     mp_lexer_free(lex);
 
     if (pn == MP_PARSE_NODE_NULL) {
@@ -36,7 +37,7 @@ static mp_obj_t mp_builtin_eval(mp_obj_t o_in) {
     }
 
     // compile the string
-    mp_obj_t module_fun = mp_compile(pn, false);
+    mp_obj_t module_fun = mp_compile(pn, source_name, false);
 
     if (module_fun == mp_const_none) {
         // TODO handle compile error correctly

py/builtinimport.c

@@ -51,6 +51,7 @@ mp_obj_t mp_builtin___import__(int n_args, mp_obj_t *args) {
     qstr parse_exc_id;
     const char *parse_exc_msg;
     mp_parse_node_t pn = mp_parse(lex, MP_PARSE_FILE_INPUT, &parse_exc_id, &parse_exc_msg);
+    qstr source_name = mp_lexer_source_name(lex);
     mp_lexer_free(lex);
 
     if (pn == MP_PARSE_NODE_NULL) {
@@ -61,7 +62,7 @@ mp_obj_t mp_builtin___import__(int n_args, mp_obj_t *args) {
     }
 
     // compile the imported script
-    mp_obj_t module_fun = mp_compile(pn, false);
+    mp_obj_t module_fun = mp_compile(pn, source_name, false);
 
     if (module_fun == mp_const_none) {
         // TODO handle compile error correctly

py/compile.c

@@ -2505,6 +2505,7 @@ void compile_node(compiler_t *comp, mp_parse_node_t pn) {
         }
     } else {
         mp_parse_node_struct_t *pns = (mp_parse_node_struct_t*)pn;
+        EMIT(set_line_number, pns->source_line);
         compile_function_t f = compile_function[MP_PARSE_NODE_STRUCT_KIND(pns)];
         if (f == NULL) {
             printf("node %u cannot be compiled\n", (uint)MP_PARSE_NODE_STRUCT_KIND(pns));
@@ -3024,7 +3025,7 @@ void compile_scope_compute_things(compiler_t *comp, scope_t *scope) {
     }
 }
 
-mp_obj_t mp_compile(mp_parse_node_t pn, bool is_repl) {
+mp_obj_t mp_compile(mp_parse_node_t pn, qstr source_file, bool is_repl) {
     compiler_t *comp = m_new(compiler_t, 1);
 
     comp->is_repl = is_repl;
@@ -3131,7 +3132,7 @@ mp_obj_t mp_compile(mp_parse_node_t pn, bool is_repl) {
 
                 default:
                     if (emit_bc == NULL) {
-                        emit_bc = emit_bc_new(max_num_labels);
+                        emit_bc = emit_bc_new(source_file, max_num_labels);
                     }
                     comp->emit = emit_bc;
                     comp->emit_method_table = &emit_bc_method_table;

py/compile.h

-mp_obj_t mp_compile(mp_parse_node_t pn, bool is_repl);
+mp_obj_t mp_compile(mp_parse_node_t pn, qstr source_file, bool is_repl);

py/emit.h

@@ -23,6 +23,7 @@ typedef struct _emit_method_table_t {
     bool (*last_emit_was_return_value)(emit_t *emit);
     int (*get_stack_size)(emit_t *emit);
     void (*set_stack_size)(emit_t *emit, int size);
+    void (*set_line_number)(emit_t *emit, int line);
 
     void (*load_id)(emit_t *emit, qstr qstr);
     void (*store_id)(emit_t *emit, qstr qstr);
@@ -119,7 +120,7 @@ extern const emit_method_table_t emit_native_thumb_method_table;
 emit_t *emit_pass1_new(qstr qstr___class__);
 void emit_pass1_free(emit_t *emit);
 emit_t *emit_cpython_new(uint max_num_labels);
-emit_t *emit_bc_new(uint max_num_labels);
+emit_t *emit_bc_new(qstr source_file, uint max_num_labels);
 emit_t *emit_native_x64_new(uint max_num_labels);
 emit_t *emit_native_thumb_new(uint max_num_labels);
 

py/emitcpy.c

@@ -68,6 +68,9 @@ static void emit_cpy_set_stack_size(emit_t *emit, int size) {
     emit->stack_size = size;
 }
 
+static void emit_cpy_set_source_line(emit_t *emit, int source_line) {
+}
+
 static void emit_cpy_load_id(emit_t *emit, qstr qstr) {
     emit_common_load_id(emit, &emit_cpython_method_table, emit->scope, qstr);
 }
@@ -798,6 +801,7 @@ const emit_method_table_t emit_cpython_method_table = {
     emit_cpy_last_emit_was_return_value,
     emit_cpy_get_stack_size,
     emit_cpy_set_stack_size,
+    emit_cpy_set_source_line,
 
     emit_cpy_load_id,
     emit_cpy_store_id,

py/emitnative.c

@@ -281,6 +281,9 @@ static void emit_native_set_stack_size(emit_t *emit, int size) {
     emit->stack_size = size;
 }
 
+static void emit_native_set_source_line(emit_t *emit, int source_line) {
+}
+
 static void adjust_stack(emit_t *emit, int stack_size_delta) {
     emit->stack_size += stack_size_delta;
     assert(emit->stack_size >= 0);
@@ -1228,6 +1231,7 @@ const emit_method_table_t EXPORT_FUN(method_table) = {
     emit_native_last_emit_was_return_value,
     emit_native_get_stack_size,
     emit_native_set_stack_size,
+    emit_native_set_source_line,
 
     emit_native_load_id,
     emit_native_store_id,

py/emitpass1.c

@@ -103,6 +103,7 @@ const emit_method_table_t emit_pass1_method_table = {
     (void*)emit_pass1_dummy,
     (void*)emit_pass1_dummy,
     (void*)emit_pass1_dummy,
+    (void*)emit_pass1_dummy,
 
     emit_pass1_load_id,
     emit_pass1_store_id,

py/lexer.c

@@ -3,6 +3,7 @@
 
 #include <stdint.h>
 #include <stdio.h>
+#include <string.h>
 #include <assert.h>
 
 #include "misc.h"
@@ -14,7 +15,7 @@
 // don't know if that's intentional or not, but we don't allow it
 
 struct _mp_lexer_t {
-    const char *name;           // name of source
+    qstr source_name;           // name of source
     void *stream_data;          // data for stream
     mp_lexer_stream_next_char_t stream_next_char;   // stream callback to get next char
     mp_lexer_stream_close_t stream_close;           // stream callback to free
@@ -49,7 +50,7 @@ bool str_strn_equal(const char *str, const char *strn, int len) {
 }
 
 void mp_token_show(const mp_token_t *tok) {
-    printf("(%s:%d:%d) kind:%d str:%p len:%d", tok->src_name, tok->src_line, tok->src_column, tok->kind, tok->str, tok->len);
+    printf("(%d:%d) kind:%d str:%p len:%d", tok->src_line, tok->src_column, tok->kind, tok->str, tok->len);
     if (tok->str != NULL && tok->len > 0) {
         const char *i = tok->str;
         const char *j = i + tok->len;
@@ -292,7 +293,6 @@ static void mp_lexer_next_token_into(mp_lexer_t *lex, mp_token_t *tok, bool firs
             next_char(lex);
             if (!is_physical_newline(lex)) {
                 // SyntaxError: unexpected character after line continuation character
-                tok->src_name = lex->name;
                 tok->src_line = lex->line;
                 tok->src_column = lex->column;
                 tok->kind = MP_TOKEN_BAD_LINE_CONTINUATION;
@@ -309,7 +309,6 @@ static void mp_lexer_next_token_into(mp_lexer_t *lex, mp_token_t *tok, bool firs
     }
 
     // set token source information
-    tok->src_name = lex->name;
     tok->src_line = lex->line;
     tok->src_column = lex->column;
 
@@ -594,7 +593,7 @@ static void mp_lexer_next_token_into(mp_lexer_t *lex, mp_token_t *tok, bool firs
 mp_lexer_t *mp_lexer_new(const char *src_name, void *stream_data, mp_lexer_stream_next_char_t stream_next_char, mp_lexer_stream_close_t stream_close) {
     mp_lexer_t *lex = m_new(mp_lexer_t, 1);
 
-    lex->name = src_name; // TODO do we need to strdup this?
+    lex->source_name = qstr_from_strn_copy(src_name, strlen(src_name));
     lex->stream_data = stream_data;
     lex->stream_next_char = stream_next_char;
     lex->stream_close = stream_close;
@@ -642,6 +641,10 @@ void mp_lexer_free(mp_lexer_t *lex) {
     }
 }
 
+qstr mp_lexer_source_name(mp_lexer_t *lex) {
+    return lex->source_name;
+}
+
 void mp_lexer_to_next(mp_lexer_t *lex) {
     mp_lexer_next_token_into(lex, &lex->tok_cur, false);
 }
@@ -677,11 +680,11 @@ bool mp_lexer_opt_str(mp_lexer_t *lex, const char *str) {
 */
 
 bool mp_lexer_show_error_pythonic_prefix(mp_lexer_t *lex) {
-    printf("  File \"%s\", line %d column %d\n", lex->tok_cur.src_name, lex->tok_cur.src_line, lex->tok_cur.src_column);
+    printf("  File \"%s\", line %d column %d\n", qstr_str(lex->source_name), lex->tok_cur.src_line, lex->tok_cur.src_column);
     return false;
 }
 
 bool mp_lexer_show_error_pythonic(mp_lexer_t *lex, const char *msg) {
-    printf("  File \"%s\", line %d column %d\n%s\n", lex->tok_cur.src_name, lex->tok_cur.src_line, lex->tok_cur.src_column, msg);
+    printf("  File \"%s\", line %d column %d\n%s\n", qstr_str(lex->source_name), lex->tok_cur.src_line, lex->tok_cur.src_column, msg);
     return false;
 }

py/lexer.h

@@ -105,7 +105,6 @@ typedef enum _mp_token_kind_t {
 } mp_token_kind_t;
 
 typedef struct _mp_token_t {
-    const char *src_name;       // name of source
     uint src_line;              // source line
     uint src_column;            // source column
 
@@ -129,6 +128,7 @@ mp_lexer_t *mp_lexer_new(const char *src_name, void *stream_data, mp_lexer_strea
 mp_lexer_t *mp_lexer_new_from_str_len(const char *src_name, const char *str, uint len, uint free_len);
 
 void mp_lexer_free(mp_lexer_t *lex);
+qstr mp_lexer_source_name(mp_lexer_t *lex);
 void mp_lexer_to_next(mp_lexer_t *lex);
 const mp_token_t *mp_lexer_cur(const mp_lexer_t *lex);
 bool mp_lexer_is_kind(mp_lexer_t *lex, mp_token_kind_t kind);

py/obj.h

@@ -273,6 +273,8 @@ machine_int_t mp_obj_int_get_checked(mp_obj_t self_in);
 // exception
 extern const mp_obj_type_t exception_type;
 qstr mp_obj_exception_get_type(mp_obj_t self_in);
+void mp_obj_exception_set_source_info(mp_obj_t self_in, qstr file, machine_uint_t line);
+void mp_obj_exception_get_source_info(mp_obj_t self_in, qstr *file, machine_uint_t *line);
 
 // str
 extern const mp_obj_type_t str_type;

py/objexcept.c

@@ -17,6 +17,8 @@
 // have args tuple (or otherwise have it as NULL).
 typedef struct mp_obj_exception_t {
     mp_obj_base_t base;
+    qstr source_file;
+    machine_uint_t source_line;
     qstr id;
     qstr msg;
     mp_obj_tuple_t args;
@@ -87,6 +89,8 @@ mp_obj_t mp_obj_new_exception_msg_varg(qstr id, const char *fmt, ...) {
     // make exception object
     mp_obj_exception_t *o = m_new_obj_var(mp_obj_exception_t, mp_obj_t*, 0);
     o->base.type = &exception_type;
+    o->source_file = 0;
+    o->source_line = 0;
     o->id = id;
     o->args.len = 0;
     if (fmt == NULL) {
@@ -109,3 +113,23 @@ qstr mp_obj_exception_get_type(mp_obj_t self_in) {
     mp_obj_exception_t *self = self_in;
     return self->id;
 }
+
+void mp_obj_exception_set_source_info(mp_obj_t self_in, qstr file, machine_uint_t line) {
+    assert(MP_OBJ_IS_TYPE(self_in, &exception_type));
+    mp_obj_exception_t *self = self_in;
+    // TODO make a list of file/line pairs for the traceback
+    // for now, just keep the first one
+    if (file != 0 && self->source_file == 0) {
+        self->source_file = file;
+    }
+    if (line != 0 && self->source_line == 0) {
+        self->source_line = line;
+    }
+}
+
+void mp_obj_exception_get_source_info(mp_obj_t self_in, qstr *file, machine_uint_t *line) {
+    assert(MP_OBJ_IS_TYPE(self_in, &exception_type));
+    mp_obj_exception_t *self = self_in;
+    *file = self->source_file;
+    *line = self->source_line;
+}

py/objgenerator.c

@@ -58,6 +58,7 @@ mp_obj_t mp_obj_new_gen_wrap(uint n_locals, uint n_stack, mp_obj_t fun) {
 
 typedef struct _mp_obj_gen_instance_t {
     mp_obj_base_t base;
+    const byte *code_info;
     const byte *ip;
     mp_obj_t *sp;
     uint n_state;
@@ -74,7 +75,7 @@ mp_obj_t gen_instance_getiter(mp_obj_t self_in) {
 
 mp_obj_t gen_instance_iternext(mp_obj_t self_in) {
     mp_obj_gen_instance_t *self = self_in;
-    bool yield = mp_execute_byte_code_2(&self->ip, &self->state[self->n_state - 1], &self->sp);
+    bool yield = mp_execute_byte_code_2(self->code_info, &self->ip, &self->state[self->n_state - 1], &self->sp);
     if (yield) {
         return *self->sp;
     } else {
@@ -98,6 +99,7 @@ const mp_obj_type_t gen_instance_type = {
 mp_obj_t mp_obj_new_gen_instance(const byte *bytecode, uint n_state, int n_args, const mp_obj_t *args) {
     mp_obj_gen_instance_t *o = m_new_obj_var(mp_obj_gen_instance_t, mp_obj_t, n_state);
     o->base.type = &gen_instance_type;
+    o->code_info = bytecode;
     o->ip = bytecode;
     o->sp = &o->state[0] - 1; // sp points to top of stack, which starts off 1 below the state
     o->n_state = n_state;
@@ -111,6 +113,10 @@ mp_obj_t mp_obj_new_gen_instance(const byte *bytecode, uint n_state, int n_args,
     // prelude for making cells (closed over variables)
     // for now we just make sure there are no cells variables
     // need to work out how to implement closed over variables in generators
+
+    // get code info size
+    machine_uint_t code_info_size = bytecode[0] | (bytecode[1] << 8) | (bytecode[2] << 16) | (bytecode[3] << 24);
+    o->ip += code_info_size;
     assert(o->ip[0] == 0);
     o->ip += 1;
 

py/parse.c

@@ -80,7 +80,8 @@ static const rule_t *rules[] = {
 };
 
 typedef struct _rule_stack_t {
-    byte rule_id;
+    unsigned int src_line : 24;
+    unsigned int rule_id : 8;
     int32_t arg_i; // what should be the size and signedness?
 } rule_stack_t;
 
@@ -92,45 +93,54 @@ typedef struct _parser_t {
     uint result_stack_alloc;
     uint result_stack_top;
     mp_parse_node_t *result_stack;
+
+    mp_lexer_t *lexer;
 } parser_t;
 
-static void push_rule(parser_t *parser, const rule_t *rule, int arg_i) {
+static void push_rule(parser_t *parser, int src_line, const rule_t *rule, int arg_i) {
     if (parser->rule_stack_top >= parser->rule_stack_alloc) {
         parser->rule_stack = m_renew(rule_stack_t, parser->rule_stack, parser->rule_stack_alloc, parser->rule_stack_alloc * 2);
         parser->rule_stack_alloc *= 2;
     }
-    parser->rule_stack[parser->rule_stack_top].rule_id = rule->rule_id;
-    parser->rule_stack[parser->rule_stack_top].arg_i = arg_i;
-    parser->rule_stack_top += 1;
+    rule_stack_t *rs = &parser->rule_stack[parser->rule_stack_top++];
+    rs->src_line = src_line;
+    rs->rule_id = rule->rule_id;
+    rs->arg_i = arg_i;
 }
 
 static void push_rule_from_arg(parser_t *parser, uint arg) {
     assert((arg & RULE_ARG_KIND_MASK) == RULE_ARG_RULE || (arg & RULE_ARG_KIND_MASK) == RULE_ARG_OPT_RULE);
     uint rule_id = arg & RULE_ARG_ARG_MASK;
     assert(rule_id < RULE_maximum_number_of);
-    push_rule(parser, rules[rule_id], 0);
+    push_rule(parser, mp_lexer_cur(parser->lexer)->src_line, rules[rule_id], 0);
 }
 
-static void pop_rule(parser_t *parser, const rule_t **rule, uint *arg_i) {
+static void pop_rule(parser_t *parser, const rule_t **rule, uint *arg_i, uint *src_line) {
     parser->rule_stack_top -= 1;
     *rule = rules[parser->rule_stack[parser->rule_stack_top].rule_id];
     *arg_i = parser->rule_stack[parser->rule_stack_top].arg_i;
+    *src_line = parser->rule_stack[parser->rule_stack_top].src_line;
 }
 
 mp_parse_node_t mp_parse_node_new_leaf(machine_int_t kind, machine_int_t arg) {
     return (mp_parse_node_t)(kind | (arg << 4));
 }
 
-int num_parse_nodes_allocated = 0;
-mp_parse_node_struct_t *parse_node_new_struct(int rule_id, int num_args) {
+//int num_parse_nodes_allocated = 0;
+mp_parse_node_struct_t *parse_node_new_struct(int src_line, int rule_id, int num_args) {
     mp_parse_node_struct_t *pn = m_new_obj_var(mp_parse_node_struct_t, mp_parse_node_t, num_args);
-    pn->source = 0; // TODO
+    pn->source_line = src_line;
     pn->kind_num_nodes = (rule_id & 0xff) | (num_args << 8);
-    num_parse_nodes_allocated += 1;
+    //num_parse_nodes_allocated += 1;
     return pn;
 }
 
 void mp_parse_node_show(mp_parse_node_t pn, int indent) {
+    if (MP_PARSE_NODE_IS_STRUCT(pn)) {
+        printf("[% 4d] ", (int)((mp_parse_node_struct_t*)pn)->source_line);
+    } else {
+        printf("       ");
+    }
     for (int i = 0; i < indent; i++) {
         printf(" ");
     }
@@ -258,8 +268,8 @@ static void push_result_token(parser_t *parser, const mp_lexer_t *lex) {
     push_result_node(parser, pn);
 }
 
-static void push_result_rule(parser_t *parser, const rule_t *rule, int num_args) {
-    mp_parse_node_struct_t *pn = parse_node_new_struct(rule->rule_id, num_args);
+static void push_result_rule(parser_t *parser, int src_line, const rule_t *rule, int num_args) {
+    mp_parse_node_struct_t *pn = parse_node_new_struct(src_line, rule->rule_id, num_args);
     for (int i = num_args; i > 0; i--) {
         pn->nodes[i - 1] = pop_result(parser);
     }
@@ -280,6 +290,8 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
     parser->result_stack_top = 0;
     parser->result_stack = m_new(mp_parse_node_t, parser->result_stack_alloc);
 
+    parser->lexer = lex;
+
     // work out the top-level rule to use, and push it on the stack
     int top_level_rule;
     switch (input_kind) {
@@ -287,13 +299,14 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
         case MP_PARSE_EVAL_INPUT: top_level_rule = RULE_eval_input; break;
         default: top_level_rule = RULE_file_input;
     }
-    push_rule(parser, rules[top_level_rule], 0);
+    push_rule(parser, mp_lexer_cur(lex)->src_line, rules[top_level_rule], 0);
 
     // parse!
 
-    uint n, i;
+    uint n, i; // state for the current rule
+    uint rule_src_line; // source line for the first token matched by the current rule
     bool backtrack = false;
-    const rule_t *rule;
+    const rule_t *rule = NULL;
     mp_token_kind_t tok_kind;
     bool emit_rule;
     bool had_trailing_sep;
@@ -304,7 +317,7 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
             break;
         }
 
-        pop_rule(parser, &rule, &i);
+        pop_rule(parser, &rule, &i, &rule_src_line);
         n = rule->act & RULE_ACT_ARG_MASK;
 
         /*
@@ -333,8 +346,8 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
                             }
                             break;
                         case RULE_ARG_RULE:
-                            push_rule(parser, rule, i + 1);
-                            push_rule_from_arg(parser, rule->arg[i]);
+                            push_rule(parser, rule_src_line, rule, i + 1); // save this or-rule
+                            push_rule_from_arg(parser, rule->arg[i]); // push child of or-rule
                             goto next_rule;
                         default:
                             assert(0);
@@ -398,14 +411,9 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
                             }
                             break;
                         case RULE_ARG_RULE:
-                            //if (i + 1 < n) {
-                                push_rule(parser, rule, i + 1);
-                            //}
-                            push_rule_from_arg(parser, rule->arg[i]);
-                            goto next_rule;
                         case RULE_ARG_OPT_RULE:
-                            push_rule(parser, rule, i + 1);
-                            push_rule_from_arg(parser, rule->arg[i]);
+                            push_rule(parser, rule_src_line, rule, i + 1); // save this and-rule
+                            push_rule_from_arg(parser, rule->arg[i]); // push child of and-rule
                             goto next_rule;
                         default:
                             assert(0);
@@ -462,9 +470,9 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
                 }
                 //printf("done and %s n=%d i=%d notnil=%d\n", rule->rule_name, n, i, num_not_nil);
                 if (emit_rule) {
-                    push_result_rule(parser, rule, i);
+                    push_result_rule(parser, rule_src_line, rule, i);
                 } else if (num_not_nil == 0) {
-                    push_result_rule(parser, rule, i); // needed for, eg, atom_paren, testlist_comp_3b
+                    push_result_rule(parser, rule_src_line, rule, i); // needed for, eg, atom_paren, testlist_comp_3b
                     //result_stack_show(parser);
                     //assert(0);
                 } else if (num_not_nil == 1) {
@@ -478,7 +486,7 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
                     }
                     push_result_node(parser, pn);
                 } else {
-                    push_result_rule(parser, rule, i);
+                    push_result_rule(parser, rule_src_line, rule, i);
                 }
                 break;
 
@@ -538,8 +546,8 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
                                 }
                                 break;
                             case RULE_ARG_RULE:
-                                push_rule(parser, rule, i + 1);
-                                push_rule_from_arg(parser, arg);
+                                push_rule(parser, rule_src_line, rule, i + 1); // save this list-rule
+                                push_rule_from_arg(parser, arg); // push child of list-rule
                                 goto next_rule;
                             default:
                                 assert(0);
@@ -559,13 +567,13 @@ mp_parse_node_t mp_parse(mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr
                     // list matched single item
                     if (had_trailing_sep) {
                         // if there was a trailing separator, make a list of a single item
-                        push_result_rule(parser, rule, i);
+                        push_result_rule(parser, rule_src_line, rule, i);
                     } else {
                         // just leave single item on stack (ie don't wrap in a list)
                     }
                 } else {
                     //printf("done list %s %d %d\n", rule->rule_name, n, i);
-                    push_result_rule(parser, rule, i);
+                    push_result_rule(parser, rule_src_line, rule, i);
                 }
                 break;
 

py/parse.h

@@ -28,7 +28,7 @@ struct _mp_lexer_t;
 typedef machine_uint_t mp_parse_node_t; // must be pointer size
 
 typedef struct _mp_parse_node_struct_t {
-    uint32_t source;            // file identifier, and line number
+    uint32_t source_line;       // line number in source file
     uint32_t kind_num_nodes;    // parse node kind, and number of nodes
     mp_parse_node_t nodes[];    // nodes
 } mp_parse_node_struct_t;

py/showbc.c

@@ -18,6 +18,10 @@
 void mp_show_byte_code(const byte *ip, int len) {
     const byte *ip_start = ip;
 
+    // get code info size
+    machine_uint_t code_info_size = ip[0] | (ip[1] << 8) | (ip[2] << 16) | (ip[3] << 24);
+    ip += code_info_size;
+
     // decode prelude
     {
         uint n_local = *ip++;

py/vm.c

@@ -38,8 +38,13 @@ mp_obj_t mp_execute_byte_code(const byte *code, const mp_obj_t *args, uint n_arg
         assert(i < 8);
         state[n_state - 1 - i] = args[i];
     }
+
     const byte *ip = code;
 
+    // get code info size
+    machine_uint_t code_info_size = ip[0] | (ip[1] << 8) | (ip[2] << 16) | (ip[3] << 24);
+    ip += code_info_size;
+
     // execute prelude to make any cells (closed over variables)
     {
         for (uint n_local = *ip++; n_local > 0; n_local--) {
@@ -53,7 +58,7 @@ mp_obj_t mp_execute_byte_code(const byte *code, const mp_obj_t *args, uint n_arg
     }
 
     // execute the byte code
-    if (mp_execute_byte_code_2(&ip, &state[n_state - 1], &sp)) {
+    if (mp_execute_byte_code_2(code, &ip, &state[n_state - 1], &sp)) {
         // it shouldn't yield
         assert(0);
     }
@@ -65,7 +70,7 @@ mp_obj_t mp_execute_byte_code(const byte *code, const mp_obj_t *args, uint n_arg
 
 // fastn has items in reverse order (fastn[0] is local[0], fastn[-1] is local[1], etc)
 // sp points to bottom of stack which grows up
-bool mp_execute_byte_code_2(const byte **ip_in_out, mp_obj_t *fastn, mp_obj_t **sp_in_out) {
+bool mp_execute_byte_code_2(const byte *code_info, const byte **ip_in_out, mp_obj_t *fastn, mp_obj_t **sp_in_out) {
     // careful: be sure to declare volatile any variables read in the exception handler (written is ok, I think)
 
     const byte *ip = *ip_in_out;
@@ -79,12 +84,14 @@ bool mp_execute_byte_code_2(const byte **ip_in_out, mp_obj_t *fastn, mp_obj_t **
     volatile machine_uint_t currently_in_except_block = 0; // 0 or 1, to detect nested exceptions
     machine_uint_t exc_stack[8]; // on the exception stack we store (ip, sp | X) for each block, X = previous value of currently_in_except_block
     machine_uint_t *volatile exc_sp = &exc_stack[0] - 1; // stack grows up, exc_sp points to top of stack
+    const byte *volatile save_ip = ip; // this is so we can access ip in the exception handler without making ip volatile (which means the compiler can't keep it in a register in the main loop)
 
     // outer exception handling loop
     for (;;) {
         if (nlr_push(&nlr) == 0) {
             // loop to execute byte code
             for (;;) {
+                save_ip = ip;
                 int op = *ip++;
                 switch (op) {
                     case MP_BC_LOAD_CONST_FALSE:
@@ -518,6 +525,23 @@ bool mp_execute_byte_code_2(const byte **ip_in_out, mp_obj_t *fastn, mp_obj_t **
         } else {
             // exception occurred
 
+            // set file and line number that the exception occurred at
+            if (MP_OBJ_IS_TYPE(nlr.ret_val, &exception_type)) {
+                machine_uint_t code_info_size = code_info[0] | (code_info[1] << 8) | (code_info[2] << 16) | (code_info[3] << 24);
+                qstr = code_info[4] | (code_info[5] << 8) | (code_info[6] << 16) | (code_info[7] << 24);
+                machine_uint_t source_line = 1;
+                machine_uint_t bc = save_ip - code_info - code_info_size;
+                //printf("find %lu %d %d\n", bc, code_info[8], code_info[9]);
+                for (const byte* ci = code_info + 8; bc > ci[0]; ci += 2) {
+                    bc -= ci[0];
+                    source_line += ci[1];
+                    if (ci[0] == 0 && ci[1] == 0) {
+                        break;
+                    }
+                }
+                mp_obj_exception_set_source_info(nlr.ret_val, qstr, source_line);
+            }
+
             while (currently_in_except_block) {
                 // nested exception
 

stm/main.c

@@ -421,6 +421,7 @@ void do_repl(void) {
         qstr parse_exc_id;
         const char *parse_exc_msg;
         mp_parse_node_t pn = mp_parse(lex, MP_PARSE_SINGLE_INPUT, &parse_exc_id, &parse_exc_msg);
+        qstr source_name = mp_lexer_source_name(lex);
 
         if (pn == MP_PARSE_NODE_NULL) {
             // parse error
@@ -430,7 +431,7 @@ void do_repl(void) {
         } else {
             // parse okay
             mp_lexer_free(lex);
-            mp_obj_t module_fun = mp_compile(pn, true);
+            mp_obj_t module_fun = mp_compile(pn, source_name, true);
             if (module_fun != mp_const_none) {
                 nlr_buf_t nlr;
                 uint32_t start = sys_tick_counter;
@@ -465,6 +466,7 @@ bool do_file(const char *filename) {
     qstr parse_exc_id;
     const char *parse_exc_msg;