[gawk-diffs] [SCM] gawk branch, feature/fix-ns-memleak, created. gawk-4.1.0-3602-gc25c7d9

[Arnold Robbins]

This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "gawk".

The branch, feature/fix-ns-memleak has been created
        at  c25c7d9359451ec91c4290a84b17185fa32f167f (commit)

- Log -----------------------------------------------------------------
http://git.sv.gnu.org/cgit/gawk.git/commit/?id=c25c7d9359451ec91c4290a84b17185fa32f167f

commit c25c7d9359451ec91c4290a84b17185fa32f167f
Author: Arnold D. Robbins <address@hidden>
Date:   Fri Jan 18 14:39:08 2019 +0200

    Start fixing memory leaks in namespace management.

diff --git a/ChangeLog b/ChangeLog
index 2c3a932..e1df2e1 100755
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,19 @@
+2019-01-18         Arnold D. Robbins     <address@hidden>
+
+       Start fixing memory leaks related to namespaces.
+
+       * awkgram.y (Grammar): At simple_variable -> NAME, qualify names
+       before lookup/install.
+       (next_sourcefile): Check if need to free current_namespace before
+       assigning to it.
+       (yylex): Fix length of string to be dup'ed before returning
+       NAME or FUNC_CALL.
+       (set_namespace): Always free ns->lextok, adjust memory allocations
+       appropriately.
+       * main.c (main): Check if need to free current_namespace before
+       assigning to it.
+       * profile.c (pp_namespace): Ditto. (This may not be necessary.)
+
 2019-01-14         Andrew J. Schorr      <address@hidden>
 
        * builtin.c (do_typeof): Fix memory leak when populating the
diff --git a/awkgram.c b/awkgram.c
index aa86581..de2647a 100644
--- a/awkgram.c
+++ b/awkgram.c
@@ -700,8 +700,8 @@ static const yytype_uint16 yyrline[] =
     1826,  1828,  1830,  1832,  1834,  1839,  1843,  1875,  1883,  1889,
     1895,  1908,  1909,  1910,  1915,  1920,  1924,  1928,  1943,  1964,
     1969,  2006,  2048,  2049,  2055,  2056,  2061,  2063,  2070,  2087,
-    2104,  2106,  2113,  2118,  2126,  2136,  2148,  2157,  2161,  2166,
-    2170,  2174,  2178,  2183,  2184,  2188,  2192,  2196
+    2104,  2106,  2113,  2118,  2148,  2158,  2170,  2179,  2183,  2188,
+    2192,  2196,  2200,  2205,  2206,  2210,  2214,  2218
 };
 #endif
 
@@ -4341,28 +4341,50 @@ regular_print:
   case 193:
 #line 2119 "awkgram.y" /* yacc.c:1645  */
     {
+               // FIXME: Code  here and at function call
+               // should be moved into a function
                char *var_name = (yyvsp[0])->lextok;
 
+               bool is_all_upper = true;
+               char *cp;
+               for (cp = var_name; *cp != '\0'; cp++) {
+                       if (! isupper(*cp)) {
+                               is_all_upper = false;
+                               break;
+                       }
+               }
+
+               if (current_namespace != awk_namespace && strchr(var_name, ':') 
== NULL && ! is_all_upper) {
+                       size_t len = strlen(current_namespace) + 2 + 
strlen(var_name) + 1;
+                       char *buf;
+
+                       emalloc(buf, char *, len, "simple_variable");
+                       sprintf(buf, "%s::%s", current_namespace, var_name);
+
+                       efree((void *) (yyvsp[0])->lextok);
+                       (yyvsp[0])->lextok = buf;
+               }
+
                (yyvsp[0])->opcode = Op_push;
-               (yyvsp[0])->memory = variable((yyvsp[0])->source_line, 
var_name, Node_var_new);
+               (yyvsp[0])->memory = variable((yyvsp[0])->source_line, 
(yyvsp[0])->lextok, Node_var_new);
                (yyval) = list_create((yyvsp[0]));
          }
-#line 4351 "awkgram.c" /* yacc.c:1645  */
+#line 4373 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 194:
-#line 2127 "awkgram.y" /* yacc.c:1645  */
+#line 2149 "awkgram.y" /* yacc.c:1645  */
     {
                char *arr = (yyvsp[-1])->lextok;
                (yyvsp[-1])->memory = variable((yyvsp[-1])->source_line, arr, 
Node_var_new);
                (yyvsp[-1])->opcode = Op_push_array;
                (yyval) = list_prepend((yyvsp[0]), (yyvsp[-1]));
          }
-#line 4362 "awkgram.c" /* yacc.c:1645  */
+#line 4384 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 195:
-#line 2137 "awkgram.y" /* yacc.c:1645  */
+#line 2159 "awkgram.y" /* yacc.c:1645  */
     {
                INSTRUCTION *ip = (yyvsp[0])->nexti;
                if (ip->opcode == Op_push
@@ -4374,85 +4396,85 @@ regular_print:
                } else
                        (yyval) = (yyvsp[0]);
          }
-#line 4378 "awkgram.c" /* yacc.c:1645  */
+#line 4400 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 196:
-#line 2149 "awkgram.y" /* yacc.c:1645  */
+#line 2171 "awkgram.y" /* yacc.c:1645  */
     {
                (yyval) = list_append((yyvsp[-1]), (yyvsp[-2]));
                if ((yyvsp[0]) != NULL)
                        mk_assignment((yyvsp[-1]), NULL, (yyvsp[0]));
          }
-#line 4388 "awkgram.c" /* yacc.c:1645  */
+#line 4410 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 197:
-#line 2158 "awkgram.y" /* yacc.c:1645  */
+#line 2180 "awkgram.y" /* yacc.c:1645  */
     {
                (yyvsp[0])->opcode = Op_postincrement;
          }
-#line 4396 "awkgram.c" /* yacc.c:1645  */
+#line 4418 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 198:
-#line 2162 "awkgram.y" /* yacc.c:1645  */
+#line 2184 "awkgram.y" /* yacc.c:1645  */
     {
                (yyvsp[0])->opcode = Op_postdecrement;
          }
-#line 4404 "awkgram.c" /* yacc.c:1645  */
+#line 4426 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 199:
-#line 2166 "awkgram.y" /* yacc.c:1645  */
+#line 2188 "awkgram.y" /* yacc.c:1645  */
     { (yyval) = NULL; }
-#line 4410 "awkgram.c" /* yacc.c:1645  */
+#line 4432 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 200:
-#line 2170 "awkgram.y" /* yacc.c:1645  */
+#line 2192 "awkgram.y" /* yacc.c:1645  */
     { (yyval) = (yyvsp[0]); }
-#line 4416 "awkgram.c" /* yacc.c:1645  */
+#line 4438 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 201:
-#line 2174 "awkgram.y" /* yacc.c:1645  */
+#line 2196 "awkgram.y" /* yacc.c:1645  */
     { (yyval) = (yyvsp[0]); yyerrok; }
-#line 4422 "awkgram.c" /* yacc.c:1645  */
+#line 4444 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 202:
-#line 2178 "awkgram.y" /* yacc.c:1645  */
+#line 2200 "awkgram.y" /* yacc.c:1645  */
     { yyerrok; }
-#line 4428 "awkgram.c" /* yacc.c:1645  */
+#line 4450 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 203:
-#line 2183 "awkgram.y" /* yacc.c:1645  */
+#line 2205 "awkgram.y" /* yacc.c:1645  */
     { (yyval) = NULL; }
-#line 4434 "awkgram.c" /* yacc.c:1645  */
+#line 4456 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 205:
-#line 2188 "awkgram.y" /* yacc.c:1645  */
+#line 2210 "awkgram.y" /* yacc.c:1645  */
     { yyerrok; }
-#line 4440 "awkgram.c" /* yacc.c:1645  */
+#line 4462 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 206:
-#line 2192 "awkgram.y" /* yacc.c:1645  */
+#line 2214 "awkgram.y" /* yacc.c:1645  */
     { (yyval) = (yyvsp[0]); yyerrok; }
-#line 4446 "awkgram.c" /* yacc.c:1645  */
+#line 4468 "awkgram.c" /* yacc.c:1645  */
     break;
 
   case 207:
-#line 2196 "awkgram.y" /* yacc.c:1645  */
+#line 2218 "awkgram.y" /* yacc.c:1645  */
     { (yyval) = (yyvsp[0]); yyerrok; }
-#line 4452 "awkgram.c" /* yacc.c:1645  */
+#line 4474 "awkgram.c" /* yacc.c:1645  */
     break;
 
 
-#line 4456 "awkgram.c" /* yacc.c:1645  */
+#line 4478 "awkgram.c" /* yacc.c:1645  */
       default: break;
     }
   /* User semantic actions sometimes alter yychar, and that requires
@@ -4679,7 +4701,7 @@ yyreturn:
 #endif
   return yyresult;
 }
-#line 2198 "awkgram.y" /* yacc.c:1903  */
+#line 2220 "awkgram.y" /* yacc.c:1903  */
 
 
 struct token {
@@ -5534,6 +5556,8 @@ next_sourcefile()
                sourceline = 0;
                source = NULL;
                lasttok = 0;
+               if (current_namespace != awk_namespace)
+                       efree((char *) current_namespace);
                current_namespace = awk_namespace;
        }
 }
@@ -6943,7 +6967,7 @@ make_instruction:
                return lasttok = class;
        }
 out:
-       tokkey = estrdup(tokstart, tok - tokstart);
+       tokkey = estrdup(tokstart, tok - tokstart - 1);
        if (*lexptr == '(') {
                yylval = bcalloc(Op_token, 2, sourceline);
                yylval->lextok = tokkey;
@@ -9284,15 +9308,17 @@ set_namespace(INSTRUCTION *ns, INSTRUCTION *comment)
        }
 
        if (strcmp(ns->lextok, current_namespace) == 0)
-               efree(ns->lextok);
+               ;       // nothing to do
        else if (strcmp(ns->lextok, awk_namespace) == 0) {
-               efree(ns->lextok);
+               if (current_namespace != awk_namespace)
+                       efree((char *) current_namespace);
                current_namespace = awk_namespace;
        } else {
                if (current_namespace != awk_namespace)
                        efree((char *) current_namespace);
-               current_namespace = ns->lextok;
+               current_namespace = estrdup(ns->lextok, strlen(ns->lextok));
        }
+       efree(ns->lextok);
 
        // save info and push on front of list of namespaces seen
        INSTRUCTION *new_ns = instruction(Op_K_namespace);
diff --git a/awkgram.y b/awkgram.y
index 49ac3ef..86eca82 100644
--- a/awkgram.y
+++ b/awkgram.y
@@ -2117,10 +2117,32 @@ subscript_list
 simple_variable
        : NAME
          {
+               // FIXME: Code  here and at function call
+               // should be moved into a function
                char *var_name = $1->lextok;
 
+               bool is_all_upper = true;
+               char *cp;
+               for (cp = var_name; *cp != '\0'; cp++) {
+                       if (! isupper(*cp)) {
+                               is_all_upper = false;
+                               break;
+                       }
+               }
+
+               if (current_namespace != awk_namespace && strchr(var_name, ':') 
== NULL && ! is_all_upper) {
+                       size_t len = strlen(current_namespace) + 2 + 
strlen(var_name) + 1;
+                       char *buf;
+
+                       emalloc(buf, char *, len, "simple_variable");
+                       sprintf(buf, "%s::%s", current_namespace, var_name);
+
+                       efree((void *) $1->lextok);
+                       $1->lextok = buf;
+               }
+
                $1->opcode = Op_push;
-               $1->memory = variable($1->source_line, var_name, Node_var_new);
+               $1->memory = variable($1->source_line, $1->lextok, 
Node_var_new);
                $$ = list_create($1);
          }
        | NAME subscript_list
@@ -3049,6 +3071,8 @@ next_sourcefile()
                sourceline = 0;
                source = NULL;
                lasttok = 0;
+               if (current_namespace != awk_namespace)
+                       efree((char *) current_namespace);
                current_namespace = awk_namespace;
        }
 }
@@ -4458,7 +4482,7 @@ make_instruction:
                return lasttok = class;
        }
 out:
-       tokkey = estrdup(tokstart, tok - tokstart);
+       tokkey = estrdup(tokstart, tok - tokstart - 1);
        if (*lexptr == '(') {
                yylval = bcalloc(Op_token, 2, sourceline);
                yylval->lextok = tokkey;
@@ -6799,15 +6823,17 @@ set_namespace(INSTRUCTION *ns, INSTRUCTION *comment)
        }
 
        if (strcmp(ns->lextok, current_namespace) == 0)
-               efree(ns->lextok);
+               ;       // nothing to do
        else if (strcmp(ns->lextok, awk_namespace) == 0) {
-               efree(ns->lextok);
+               if (current_namespace != awk_namespace)
+                       efree((char *) current_namespace);
                current_namespace = awk_namespace;
        } else {
                if (current_namespace != awk_namespace)
                        efree((char *) current_namespace);
-               current_namespace = ns->lextok;
+               current_namespace = estrdup(ns->lextok, strlen(ns->lextok));
        }
+       efree(ns->lextok);
 
        // save info and push on front of list of namespaces seen
        INSTRUCTION *new_ns = instruction(Op_K_namespace);
diff --git a/main.c b/main.c
index 03a96f3..3324b69 100644
--- a/main.c
+++ b/main.c
@@ -520,6 +520,8 @@ main(int argc, char **argv)
                interpret(code_block);
 
        if (do_pretty_print) {
+               if (current_namespace != awk_namespace)
+                       efree((char *) current_namespace);
                current_namespace = awk_namespace;
                dump_prog(code_block);
                dump_funcs();
diff --git a/profile.c b/profile.c
index 3d954d3..e44f8be 100644
--- a/profile.c
+++ b/profile.c
@@ -2033,6 +2033,8 @@ pp_namespace(const char *name, INSTRUCTION *comment)
        if (strcmp(current_namespace, name) == 0)
                return;
 
+       if (current_namespace != awk_namespace)
+               efree((void *) current_namespace);
        current_namespace = name;
 
        if (do_profile)

-----------------------------------------------------------------------


hooks/post-receive
-- 
gawk