Support backtrace after method calls

GitHub: fix #2902, #2917

The current implementation traverses stack to retrieve backtrace. But
stack will be changed when some operations are occurred. It means that
backtrace may be broken after some operations.

This change (1) saves the minimum information to retrieve backtrace when
exception is raised and (2) restores backtrace from the minimum
information when backtrace is needed. It reduces overhead for creating
backtrace Ruby objects.

The space for the minimum information is reused by multiple
exceptions. So memory allocation isn't occurred for each exception.

5 files changed, 307 insertions, 25 deletions

diff --git a/src/backtrace.c b/src/backtrace.c
index 80a5e2935..b581083df 100644
--- a/src/backtrace.c
+++ b/src/backtrace.c
@@ -14,6 +14,15 @@
 #include <mruby/error.h>
 #include <mruby/numeric.h>
 
+struct backtrace_location_raw {
+  int i;
+  int lineno;
+  const char *filename;
+  mrb_sym method_id;
+  const char *sep;
+  struct RClass *klass;
+};
+
 struct backtrace_location {
   int i;
   int lineno;
@@ -23,6 +32,7 @@ struct backtrace_location {
   const char *class_name;
 };
 
+typedef void (*each_backtrace_func)(mrb_state*, struct backtrace_location_raw*, void*);
 typedef void (*output_stream_func)(mrb_state*, struct backtrace_location*, void*);
 
 #ifndef MRB_DISABLE_STDIO
@@ -89,7 +99,7 @@ get_backtrace_i(mrb_state *mrb, struct backtrace_location *loc, void *data)
 }
 
 static void
-output_backtrace(mrb_state *mrb, mrb_int ciidx, mrb_code *pc0, output_stream_func func, void *data)
+each_backtrace(mrb_state *mrb, mrb_int ciidx, mrb_code *pc0, each_backtrace_func func, void *data)
 {
   int i;
 
@@ -97,7 +107,7 @@ output_backtrace(mrb_state *mrb, mrb_int ciidx, mrb_code *pc0, output_stream_fun
     ciidx = 10; /* ciidx is broken... */
 
   for (i = ciidx; i >= 0; i--) {
-    struct backtrace_location loc;
+    struct backtrace_location_raw loc;
     mrb_callinfo *ci;
     mrb_irep *irep;
     mrb_code *pc;
@@ -134,13 +144,40 @@ output_backtrace(mrb_state *mrb, mrb_int ciidx, mrb_code *pc0, output_stream_fun
       loc.filename = "(unknown)";
     }
 
-    loc.method = mrb_sym2name(mrb, ci->mid);
-    loc.class_name = mrb_class_name(mrb, ci->proc->target_class);
+    loc.method_id = ci->mid;
+    loc.klass = ci->proc->target_class;
     loc.i = i;
     func(mrb, &loc, data);
   }
 }
 
+struct output_backtrace_args {
+  output_stream_func func;
+  void *data;
+};
+
+static void
+output_backtrace_i(mrb_state *mrb, struct backtrace_location_raw *loc_raw, void *data)
+{
+  struct backtrace_location loc;
+  struct output_backtrace_args *args = data;
+
+  loc.i          = loc_raw->i;
+  loc.lineno     = loc_raw->lineno;
+  loc.filename   = loc_raw->filename;
+  loc.method     = mrb_sym2name(mrb, loc_raw->method_id);
+  loc.sep        = loc_raw->sep;
+  loc.class_name = mrb_class_name(mrb, loc_raw->klass);
+
+  args->func(mrb, &loc, args->data);
+}
+
+static void
+output_backtrace(mrb_state *mrb, mrb_int ciidx, mrb_code *pc0, output_stream_func func, void *data)
+{
+  each_backtrace(mrb, ciidx, pc0, output_backtrace_i, data);
+}
+
 static void
 exc_output_backtrace(mrb_state *mrb, struct RObject *exc, output_stream_func func, void *stream)
 {
@@ -167,18 +204,76 @@ exc_output_backtrace(mrb_state *mrb, struct RObject *exc, output_stream_func fun
 
 #ifndef MRB_DISABLE_STDIO
 
+static void
+print_backtrace(mrb_state *mrb, mrb_value backtrace)
+{
+  int i, n;
+  FILE *stream = stderr;
+
+  fprintf(stream, "trace:\n");
+
+  n = RARRAY_LEN(backtrace);
+  for (i = 0; i < n; i++) {
+    mrb_value entry = RARRAY_PTR(backtrace)[i];
+
+    fprintf(stream, "\t[%d] %.*s\n", i, RSTRING_LEN(entry), RSTRING_PTR(entry));
+  }
+}
+
+static void
+print_backtrace_saved(mrb_state *mrb)
+{
+  int i;
+  FILE *stream = stderr;
+
+  fprintf(stream, "trace:\n");
+  for (i = 0; i < mrb->backtrace.n; i++) {
+    mrb_backtrace_entry *entry;
+
+    entry = &(mrb->backtrace.entries[i]);
+    fprintf(stream, "\t[%d] %s:%d", i, entry->filename, entry->lineno);
+
+    if (entry->method_id != 0) {
+      const char *method_name;
+
+      method_name = mrb_sym2name(mrb, entry->method_id);
+      if (entry->klass) {
+        fprintf(stream, ":in %s%s%s",
+                mrb_class_name(mrb, entry->klass),
+                entry->sep,
+                method_name);
+      }
+      else {
+        fprintf(stream, ":in %s", method_name);
+      }
+    }
+
+    fprintf(stream, "\n");
+  }
+}
+
 MRB_API void
 mrb_print_backtrace(mrb_state *mrb)
 {
-  struct print_backtrace_args args;
+  mrb_value backtrace;
 
   if (!mrb->exc || mrb_obj_is_kind_of(mrb, mrb_obj_value(mrb->exc), E_SYSSTACK_ERROR)) {
     return;
   }
 
-  args.stream = stderr;
-  args.tracehead = TRUE;
-  exc_output_backtrace(mrb, mrb->exc, print_backtrace_i, (void*)&args);
+  backtrace = mrb_obj_iv_get(mrb, mrb->exc, mrb_intern_lit(mrb, "backtrace"));
+  if (!mrb_nil_p(backtrace)) {
+    print_backtrace(mrb, backtrace);
+  }
+  else if (mrb->backtrace.n > 0) {
+    print_backtrace_saved(mrb);
+  }
+  else {
+    struct print_backtrace_args args;
+    args.stream = stderr;
+    args.tracehead = TRUE;
+    exc_output_backtrace(mrb, mrb->exc, print_backtrace_i, (void*)&args);
+  }
 }
 
 #else
@@ -215,3 +310,112 @@ mrb_get_backtrace(mrb_state *mrb)
 
   return ary;
 }
+
+void
+mrb_free_backtrace(mrb_state *mrb)
+{
+  mrb->backtrace.exc = 0;
+  mrb->backtrace.n = 0;
+  mrb->backtrace.n_allocated = 0;
+  mrb_free(mrb, mrb->backtrace.entries);
+}
+
+static void
+save_backtrace_i(mrb_state *mrb,
+                 struct backtrace_location_raw *loc_raw,
+                 void *data)
+{
+  mrb_backtrace_entry *entry;
+
+  if (loc_raw->i >= mrb->backtrace.n_allocated) {
+    int new_n_allocated;
+    if (mrb->backtrace.n_allocated == 0) {
+      new_n_allocated = 8;
+    }
+    else {
+      new_n_allocated = mrb->backtrace.n_allocated * 2;
+    }
+    mrb->backtrace.entries =
+      mrb_realloc(mrb,
+                  mrb->backtrace.entries,
+                  sizeof(mrb_backtrace_entry) * new_n_allocated);
+    mrb->backtrace.n_allocated = new_n_allocated;
+  }
+
+  entry = &mrb->backtrace.entries[mrb->backtrace.n];
+  entry->filename  = loc_raw->filename;
+  entry->lineno    = loc_raw->lineno;
+  entry->klass     = loc_raw->klass;
+  entry->sep       = loc_raw->sep;
+  entry->method_id = loc_raw->method_id;
+
+  mrb->backtrace.n++;
+}
+
+void
+mrb_save_backtrace(mrb_state *mrb)
+{
+  mrb_value lastpc;
+  mrb_code *code;
+  mrb_int ciidx;
+
+  mrb->backtrace.n = 0;
+  mrb->backtrace.exc = 0;
+
+  if (!mrb->exc)
+    return;
+
+  mrb->backtrace.exc = mrb->exc;
+
+  lastpc = mrb_obj_iv_get(mrb, mrb->exc, mrb_intern_lit(mrb, "lastpc"));
+  if (mrb_nil_p(lastpc)) {
+    code = NULL;
+  }
+  else {
+    code = (mrb_code*)mrb_cptr(lastpc);
+  }
+
+  ciidx = mrb_fixnum(mrb_obj_iv_get(mrb, mrb->exc, mrb_intern_lit(mrb, "ciidx")));
+
+  each_backtrace(mrb, ciidx, code, save_backtrace_i, NULL);
+}
+
+mrb_value
+mrb_restore_backtrace(mrb_state *mrb)
+{
+  int i;
+  mrb_value backtrace;
+
+  backtrace = mrb_ary_new(mrb);
+  for (i = 0; i < mrb->backtrace.n; i++) {
+    int ai;
+    mrb_backtrace_entry *entry;
+    mrb_value mrb_entry;
+
+    ai = mrb_gc_arena_save(mrb);
+    entry = &(mrb->backtrace.entries[i]);
+
+    mrb_entry = mrb_str_new_cstr(mrb, entry->filename);
+    mrb_str_cat_lit(mrb, mrb_entry, ":");
+    mrb_str_concat(mrb, mrb_entry,
+                   mrb_fixnum_to_str(mrb,
+                                     mrb_fixnum_value(entry->lineno),
+                                     10));
+    if (entry->method_id != 0) {
+      mrb_str_cat_lit(mrb, mrb_entry, ":in ");
+
+      if (entry->klass) {
+        mrb_str_cat_cstr(mrb, mrb_entry, mrb_class_name(mrb, entry->klass));
+        mrb_str_cat_cstr(mrb, mrb_entry, entry->sep);
+      }
+
+      mrb_str_cat_cstr(mrb, mrb_entry, mrb_sym2name(mrb, entry->method_id));
+    }
+
+    mrb_ary_push(mrb, backtrace, mrb_entry);
+
+    mrb_gc_arena_restore(mrb, ai);
+  }
+
+  return backtrace;
+}
diff --git a/src/error.c b/src/error.c
index 15a969d93..14e4ab4d3 100644
--- a/src/error.c
+++ b/src/error.c
@@ -174,6 +174,42 @@ exc_inspect(mrb_state *mrb, mrb_value exc)
   return str;
 }
 
+void mrb_save_backtrace(mrb_state *mrb);
+mrb_value mrb_restore_backtrace(mrb_state *mrb);
+
+static mrb_value
+exc_get_backtrace(mrb_state *mrb, mrb_value exc)
+{
+  mrb_sym attr_name;
+  mrb_value backtrace;
+
+  attr_name = mrb_intern_lit(mrb, "backtrace");
+  backtrace = mrb_iv_get(mrb, exc, attr_name);
+  if (mrb_nil_p(backtrace)) {
+    if (mrb_obj_ptr(exc) == mrb->backtrace.exc && mrb->backtrace.n > 0) {
+      backtrace = mrb_restore_backtrace(mrb);
+      mrb->backtrace.n = 0;
+      mrb->backtrace.exc = 0;
+    }
+    else {
+      backtrace = mrb_exc_backtrace(mrb, exc);
+    }
+    mrb_iv_set(mrb, exc, attr_name, backtrace);
+  }
+
+  return backtrace;
+}
+
+static mrb_value
+exc_set_backtrace(mrb_state *mrb, mrb_value exc)
+{
+  mrb_value backtrace;
+
+  mrb_get_args(mrb, "o", &backtrace);
+  mrb_iv_set(mrb, exc, mrb_intern_lit(mrb, "backtrace"), backtrace);
+
+  return backtrace;
+}
 
 static void
 exc_debug_info(mrb_state *mrb, struct RObject *exc)
@@ -202,12 +238,52 @@ exc_debug_info(mrb_state *mrb, struct RObject *exc)
   }
 }
 
+static void
+set_backtrace(mrb_state *mrb, mrb_value info, mrb_value bt)
+{
+  mrb_funcall(mrb, info, "set_backtrace", 1, bt);
+}
+
+static mrb_bool
+have_backtrace(mrb_state *mrb, struct RObject *exc)
+{
+  return !mrb_nil_p(mrb_obj_iv_get(mrb, exc, mrb_intern_lit(mrb, "backtrace")));
+}
+
+void
+mrb_exc_set(mrb_state *mrb, mrb_value exc)
+{
+  if (!mrb->gc.out_of_memory && mrb->backtrace.n > 0) {
+    mrb_value target_exc = mrb_nil_value();
+    if ((mrb->exc && !have_backtrace(mrb, mrb->exc))) {
+      target_exc = mrb_obj_value(mrb->exc);
+    }
+    else if (!mrb_nil_p(exc) && mrb_obj_ptr(exc) == mrb->backtrace.exc) {
+      target_exc = exc;
+    }
+    if (!mrb_nil_p(target_exc)) {
+      mrb_value backtrace;
+      backtrace = mrb_restore_backtrace(mrb);
+      set_backtrace(mrb, target_exc, backtrace);
+    }
+  }
+
+  mrb->backtrace.n = 0;
+  if (mrb_nil_p(exc)) {
+    mrb->exc = 0;
+  }
+  else {
+    mrb->exc = mrb_obj_ptr(exc);
+  }
+}
+
 MRB_API mrb_noreturn void
 mrb_exc_raise(mrb_state *mrb, mrb_value exc)
 {
-  mrb->exc = mrb_obj_ptr(exc);
+  mrb_exc_set(mrb, exc);
   if (!mrb->gc.out_of_memory) {
     exc_debug_info(mrb, mrb->exc);
+    mrb_save_backtrace(mrb);
   }
   if (!mrb->jmp) {
     mrb_p(mrb, exc);
@@ -337,12 +413,6 @@ mrb_bug(mrb_state *mrb, const char *fmt, ...)
   exit(EXIT_FAILURE);
 }
 
-static void
-set_backtrace(mrb_state *mrb, mrb_value info, mrb_value bt)
-{
-  mrb_funcall(mrb, info, "set_backtrace", 1, bt);
-}
-
 static mrb_value
 make_exception(mrb_state *mrb, int argc, const mrb_value *argv, mrb_bool isstr)
 {
@@ -449,7 +519,8 @@ mrb_init_exception(mrb_state *mrb)
   mrb_define_method(mrb, exception, "to_s",            exc_to_s,          MRB_ARGS_NONE());
   mrb_define_method(mrb, exception, "message",         exc_message,       MRB_ARGS_NONE());
   mrb_define_method(mrb, exception, "inspect",         exc_inspect,       MRB_ARGS_NONE());
-  mrb_define_method(mrb, exception, "backtrace",       mrb_exc_backtrace, MRB_ARGS_NONE());
+  mrb_define_method(mrb, exception, "backtrace",       exc_get_backtrace, MRB_ARGS_NONE());
+  mrb_define_method(mrb, exception, "set_backtrace",   exc_set_backtrace, MRB_ARGS_REQ(1));
 
   mrb->eStandardError_class = mrb_define_class(mrb, "StandardError", mrb->eException_class); /* 15.2.23 */
   runtime_error = mrb_define_class(mrb, "RuntimeError", mrb->eStandardError_class);          /* 15.2.28 */
diff --git a/src/load.c b/src/load.c
index da88f0d3a..0b47f2b89 100644
--- a/src/load.c
+++ b/src/load.c
@@ -614,10 +614,12 @@ mrb_read_irep(mrb_state *mrb, const uint8_t *bin)
   return read_irep(mrb, bin, flags);
 }
 
+void mrb_exc_set(mrb_state *mrb, mrb_value exc);
+
 static void
 irep_error(mrb_state *mrb)
 {
-  mrb->exc = mrb_obj_ptr(mrb_exc_new_str_lit(mrb, E_SCRIPT_ERROR, "irep load error"));
+  mrb_exc_set(mrb, mrb_exc_new_str_lit(mrb, E_SCRIPT_ERROR, "irep load error"));
 }
 
 MRB_API mrb_value
diff --git a/src/state.c b/src/state.c
index c8c0658e4..898e77336 100644
--- a/src/state.c
+++ b/src/state.c
@@ -215,6 +215,8 @@ mrb_str_pool(mrb_state *mrb, mrb_value str)
   return mrb_obj_value(ns);
 }
 
+void mrb_free_backtrace(mrb_state *mrb);
+
 MRB_API void
 mrb_free_context(mrb_state *mrb, struct mrb_context *c)
 {
@@ -242,6 +244,7 @@ mrb_close(mrb_state *mrb)
 
   /* free */
   mrb_gc_free_gv(mrb);
+  mrb_free_backtrace(mrb);
   mrb_free_context(mrb, mrb->root_c);
   mrb_free_symtbl(mrb);
   mrb_alloca_free(mrb);
diff --git a/src/vm.c b/src/vm.c
index 391646017..b2149b250 100644
--- a/src/vm.c
+++ b/src/vm.c
@@ -258,6 +258,8 @@ cipop(mrb_state *mrb)
   c->ci--;
 }
 
+void mrb_exc_set(mrb_state *mrb, mrb_value exc);
+
 static void
 ecall(mrb_state *mrb, int i)
 {
@@ -669,7 +671,7 @@ localjump_error(mrb_state *mrb, localjump_error_kind kind)
   mrb_str_cat(mrb, msg, lead, sizeof(lead) - 1);
   mrb_str_cat(mrb, msg, kind_str[kind], kind_str_len[kind]);
   exc = mrb_exc_new_str(mrb, E_LOCALJUMP_ERROR, msg);
-  mrb->exc = mrb_obj_ptr(exc);
+  mrb_exc_set(mrb, exc);
 }
 
 static void
@@ -688,7 +690,7 @@ argnum_error(mrb_state *mrb, mrb_int num)
                   mrb_fixnum_value(mrb->c->ci->argc), mrb_fixnum_value(num));
   }
   exc = mrb_exc_new_str(mrb, E_ARGUMENT_ERROR, str);
-  mrb->exc = mrb_obj_ptr(exc);
+  mrb_exc_set(mrb, exc);
 }
 
 #define ERR_PC_SET(mrb, pc) mrb->c->ci->err = pc;
@@ -1007,7 +1009,7 @@ RETRY_TRY_BLOCK:
 
     CASE(OP_RAISE) {
       /* A      raise(R(A)) */
-      mrb->exc = mrb_obj_ptr(regs[GETARG_A(i)]);
+      mrb_exc_set(mrb, regs[GETARG_A(i)]);
       goto L_RAISE;
     }
 
@@ -1241,7 +1243,7 @@ RETRY_TRY_BLOCK:
         mrb_value exc;
 
         exc = mrb_exc_new_str_lit(mrb, E_NOMETHOD_ERROR, "super called outside of method");
-        mrb->exc = mrb_obj_ptr(exc);
+        mrb_exc_set(mrb, exc);
         goto L_RAISE;
       }
       recv = regs[0];
@@ -1331,7 +1333,7 @@ RETRY_TRY_BLOCK:
           mrb_value exc;
 
           exc = mrb_exc_new_str_lit(mrb, E_NOMETHOD_ERROR, "super called outside of method");
-          mrb->exc = mrb_obj_ptr(exc);
+          mrb_exc_set(mrb, exc);
           goto L_RAISE;
         }
         stack = e->stack + 1;
@@ -1561,7 +1563,7 @@ RETRY_TRY_BLOCK:
             }
             if (mrb->c->prev->ci == mrb->c->prev->cibase) {
               mrb_value exc = mrb_exc_new_str_lit(mrb, E_FIBER_ERROR, "double resume");
-              mrb->exc = mrb_obj_ptr(exc);
+              mrb_exc_set(mrb, exc);
               goto L_RAISE;
             }
             /* automatic yield at the end */
@@ -2323,7 +2325,7 @@ RETRY_TRY_BLOCK:
       /* A      R(A) := target_class */
       if (!mrb->c->ci->target_class) {
         mrb_value exc = mrb_exc_new_str_lit(mrb, E_TYPE_ERROR, "no target class or module");
-        mrb->exc = mrb_obj_ptr(exc);
+        mrb_exc_set(mrb, exc);
         goto L_RAISE;
       }
       regs[GETARG_A(i)] = mrb_obj_value(mrb->c->ci->target_class);
@@ -2381,7 +2383,7 @@ RETRY_TRY_BLOCK:
       else {
         exc = mrb_exc_new_str(mrb, E_LOCALJUMP_ERROR, msg);
       }
-      mrb->exc = mrb_obj_ptr(exc);
+      mrb_exc_set(mrb, exc);
       goto L_RAISE;
     }
   }