Fix assorted bugs in ecpg's macro mechanism.

The code associated with EXEC SQL DEFINE was unreadable and full of
bugs, notably:

* It'd attempt to free a non-malloced string if the ecpg program
tries to redefine a macro that was defined on the command line.

* Possible memory stomp if user writes "-D=foo".

* Undef'ing or redefining a macro defined on the command line would
change the state visible to the next file, when multiple files are
specified on the command line.  (While possibly that could have been
an intentional choice, the code clearly intends to revert to the
original macro state; it's just failing to consider this interaction.)

* Missing "break" in defining a new macro meant that redefinition
of an existing name would cause an extra entry to be added to the
definition list.  While not immediately harmful, a subsequent undef
would result in the prior entry becoming visible again.

* The interactions with input buffering are subtle and were entirely
undocumented.

It's not that surprising that we hadn't noticed these bugs,
because there was no test coverage at all of either the -D
command line switch or multiple input files.  This patch adds
such coverage (in a rather hacky way I guess).

In addition to the code bugs, the user documentation was confused
about whether the -D switch defines a C macro or an ecpg one, and
it failed to mention that you can write "-Dsymbol=value".

These problems are old, so back-patch to all supported branches.

Discussion: https://postgr.es/m/998011.1713217712@sss.pgh.pa.us

doc/src/sgml/ecpg.sgml

@@ -5755,6 +5755,14 @@ EXEC SQL UPDATE Tbl SET col = MYNUMBER;
     embedded SQL query because in this case the embedded SQL precompiler is not
     able to see this declaration.
    </para>
+
+   <para>
+    If multiple input files are named on the <command>ecpg</command>
+    preprocessor's command line, the effects of <literal>EXEC SQL
+    DEFINE</literal> and <literal>EXEC SQL UNDEF</literal> do not carry
+    across files: each file starts with only the symbols defined
+    by <option>-D</option> switches on the command line.
+   </para>
   </sect2>
 
   <sect2 id="ecpg-ifdef">

doc/src/sgml/ref/ecpg-ref.sgml

@@ -93,10 +93,12 @@ PostgreSQL documentation
     </varlistentry>
 
     <varlistentry>
-     <term><option>-D <replaceable>symbol</replaceable></option></term>
+     <term><option>-D <replaceable>symbol</replaceable>[=<replaceable>value</replaceable>]</option></term>
      <listitem>
       <para>
-       Define a C preprocessor symbol.
+       Define a preprocessor symbol, equivalently to the <command>EXEC SQL
+       DEFINE</command> directive.  If no <replaceable>value</replaceable> is
+       specified, the symbol is defined with the value <literal>1</literal>.
       </para>
      </listitem>
     </varlistentry>

src/interfaces/ecpg/preproc/ecpg.c

@@ -82,35 +82,46 @@ add_include_path(char *path)
 	}
 }
 
+/*
+ * Process a command line -D switch
+ */
 static void
 add_preprocessor_define(char *define)
 {
-	struct _defines *pd = defines;
+	/* copy the argument to avoid relying on argv storage */
-	char	   *ptr,
+	char	   *define_copy = mm_strdup(define);
-			   *define_copy = mm_strdup(define);
+	char	   *ptr;
+	struct _defines *newdef;
 
-	defines = mm_alloc(sizeof(struct _defines));
+	newdef = mm_alloc(sizeof(struct _defines));
 
 	/* look for = sign */
 	ptr = strchr(define_copy, '=');
 	if (ptr != NULL)
 	{
+		/* symbol has a value */
 		char	   *tmp;
 
-		/* symbol has a value */
+		/* strip any spaces between name and '=' */
-		for (tmp = ptr - 1; *tmp == ' '; tmp--);
+		for (tmp = ptr - 1; tmp >= define_copy && *tmp == ' '; tmp--);
 		tmp[1] = '\0';
-		defines->olddef = define_copy;
+
-		defines->newdef = ptr + 1;
+		/*
+		 * Note we don't bother to separately malloc cmdvalue; it will never
+		 * be freed so that's not necessary.
+		 */
+		newdef->cmdvalue = ptr + 1;
 	}
 	else
 	{
-		defines->olddef = define_copy;
+		/* define it as "1"; again no need to malloc it */
-		defines->newdef = mm_strdup("1");
+		newdef->cmdvalue = "1";
 	}
-	defines->pertinent = true;
+	newdef->name = define_copy;
-	defines->used = NULL;
+	newdef->value = mm_strdup(newdef->cmdvalue);
-	defines->next = pd;
+	newdef->used = NULL;
+	newdef->next = defines;
+	defines = newdef;
 }
 
 #define ECPG_GETOPT_LONG_REGRESSION		1
@@ -347,6 +358,8 @@ main(int argc, char *const argv[])
 			{
 				struct cursor *ptr;
 				struct _defines *defptr;
+				struct _defines *prevdefptr;
+				struct _defines *nextdefptr;
 				struct typedefs *typeptr;
 				struct declared_list *list;
 
@@ -384,28 +397,28 @@ main(int argc, char *const argv[])
 					free(this);
 				}
 
-				/* remove non-pertinent old defines as well */
+				/* restore defines to their command-line state */
-				while (defines && !defines->pertinent)
+				prevdefptr = NULL;
+				for (defptr = defines; defptr != NULL; defptr = nextdefptr)
 				{
-					defptr = defines;
+					nextdefptr = defptr->next;
-					defines = defines->next;
+					if (defptr->cmdvalue != NULL)
-
-					free(defptr->newdef);
-					free(defptr->olddef);
-					free(defptr);
-				}
-
-				for (defptr = defines; defptr != NULL; defptr = defptr->next)
-				{
-					struct _defines *this = defptr->next;
-
-					if (this && !this->pertinent)
 					{
-						defptr->next = this->next;
+						/* keep it, resetting the value */
-
+						free(defptr->value);
-						free(this->newdef);
+						defptr->value = mm_strdup(defptr->cmdvalue);
-						free(this->olddef);
+						prevdefptr = defptr;
-						free(this);
+					}
+					else
+					{
+						/* remove it */
+						if (prevdefptr != NULL)
+							prevdefptr->next = nextdefptr;
+						else
+							defines = nextdefptr;
+						free(defptr->name);
+						free(defptr->value);
+						free(defptr);
 					}
 				}
 

src/interfaces/ecpg/preproc/pgc.l

@@ -69,7 +69,14 @@ char *token_start;
 static int state_before_str_start;
 static int state_before_str_stop;
 
-struct _yy_buffer
+/*
+ * State for handling include files and macro expansion.  We use a new
+ * flex input buffer for each level of include or macro, and create a
+ * struct _yy_buffer to remember the previous level.  There is not a struct
+ * for the currently active input source; that state is kept in the global
+ * variables YY_CURRENT_BUFFER, yylineno, and input_filename.
+ */
+static struct _yy_buffer
 {
 	YY_BUFFER_STATE		buffer;
 	long				lineno;
@@ -77,8 +84,6 @@ struct _yy_buffer
 	struct _yy_buffer  *next;
 } *yy_buffer = NULL;
 
-static char *old;
-
 /*
  * Vars for handling ifdef/elif/endif constructs.  preproc_tos is the current
  * nesting depth of such constructs, and stacked_if_value[preproc_tos] is the
@@ -426,6 +431,8 @@ cppline			{space}*#([^i][A-Za-z]*|{if}|{ifdef}|{ifndef}|{import})((\/\*[^*/]*\*+
 
 %{
 		/* code to execute during start of each call of yylex() */
+		char *newdefsymbol = NULL;
+
 		token_start = NULL;
 %}
 
@@ -961,6 +968,7 @@ cppline			{space}*#([^i][A-Za-z]*|{if}|{ifdef}|{ifndef}|{import})((\/\*[^*/]*\*+
 				}
 
 {identifier}	{
+					/* First check to see if it's a define symbol to expand */
 					if (!isdefine())
 					{
 						int		kwvalue;
@@ -1153,17 +1161,23 @@ cppline			{space}*#([^i][A-Za-z]*|{if}|{ifdef}|{ifndef}|{import})((\/\*[^*/]*\*+
 					yytext[i+1] = '\0';
 
 
-					for (ptr = defines; ptr != NULL; ptr2 = ptr, ptr = ptr->next)
+					/* Find and unset any matching define; should be only 1 */
+					for (ptr = defines; ptr; ptr2 = ptr, ptr = ptr->next)
 					{
-						if (strcmp(yytext, ptr->olddef) == 0)
+						if (strcmp(yytext, ptr->name) == 0)
 						{
-							if (ptr2 == NULL)
+							free(ptr->value);
-								defines = ptr->next;
+							ptr->value = NULL;
-							else
+							/* We cannot forget it if there's a cmdvalue */
-								ptr2->next = ptr->next;
+							if (ptr->cmdvalue == NULL)
-							free(ptr->newdef);
+							{
-							free(ptr->olddef);
+								if (ptr2 == NULL)
-							free(ptr);
+									defines = ptr->next;
+								else
+									ptr2->next = ptr->next;
+								free(ptr->name);
+								free(ptr);
+							}
 							break;
 						}
 					}
@@ -1368,11 +1382,17 @@ cppline			{space}*#([^i][A-Za-z]*|{if}|{ifdef}|{ifndef}|{import})((\/\*[^*/]*\*+
 							;
 						yytext[i+1] = '\0';
 
-						for (defptr = defines;
+						/* Does a definition exist? */
-							 defptr != NULL &&
+						for (defptr = defines; defptr; defptr = defptr->next)
-							 strcmp(yytext, defptr->olddef) != 0;
+						{
-							 defptr = defptr->next)
+							if (strcmp(yytext, defptr->name) == 0)
-							/* skip */ ;
+							{
+								/* Found it, but is it currently undefined? */
+								if (defptr->value == NULL)
+									defptr = NULL; /* pretend it's not found */
+								break;
+							}
+						}
 
 						this_active = (defptr ? ifcond : !ifcond);
 						stacked_if_value[preproc_tos].active =
@@ -1393,7 +1413,7 @@ cppline			{space}*#([^i][A-Za-z]*|{if}|{ifdef}|{ifndef}|{import})((\/\*[^*/]*\*+
 				yyterminate();
 			}
 <def_ident>{identifier} {
-				old = mm_strdup(yytext);
+				newdefsymbol = mm_strdup(yytext);
 				BEGIN(def);
 				startlit();
 			}
@@ -1402,26 +1422,31 @@ cppline			{space}*#([^i][A-Za-z]*|{if}|{ifdef}|{ifndef}|{import})((\/\*[^*/]*\*+
 				yyterminate();
 			}
 <def>{space}*";"	{
-						struct _defines *ptr, *this;
+						struct _defines *ptr;
 
+						/* Does it already exist? */
 						for (ptr = defines; ptr != NULL; ptr = ptr->next)
 						{
-							 if (strcmp(old, ptr->olddef) == 0)
+							if (strcmp(newdefsymbol, ptr->name) == 0)
-							 {
+							{
-								free(ptr->newdef);
+								free(ptr->value);
-								ptr->newdef = mm_strdup(literalbuf);
+								ptr->value = mm_strdup(literalbuf);
-							 }
+								/* Don't leak newdefsymbol */
+								free(newdefsymbol);
+								break;
+							}
 						}
 						if (ptr == NULL)
 						{
-							this = (struct _defines *) mm_alloc(sizeof(struct _defines));
+							/* Not present, make a new entry */
-
+							ptr = (struct _defines *) mm_alloc(sizeof(struct _defines));
-							/* initial definition */
+
-							this->olddef = old;
+							ptr->name = newdefsymbol;
-							this->newdef = mm_strdup(literalbuf);
+							ptr->value = mm_strdup(literalbuf);
-							this->next = defines;
+							ptr->cmdvalue = NULL;
-							this->used = NULL;
+							ptr->used = NULL;
-							defines = this;
+							ptr->next = defines;
+							defines = ptr;
 						}
 
 						BEGIN(C);
@@ -1438,6 +1463,7 @@ cppline			{space}*#([^i][A-Za-z]*|{if}|{ifdef}|{ifndef}|{import})((\/\*[^*/]*\*+
 <<EOF>>				{
 					if (yy_buffer == NULL)
 					{
+						/* No more input */
 						if ( preproc_tos > 0 )
 						{
 							preproc_tos = 0;
@@ -1447,16 +1473,20 @@ cppline			{space}*#([^i][A-Za-z]*|{if}|{ifdef}|{ifndef}|{import})((\/\*[^*/]*\*+
 					}
 					else
 					{
+						/* Revert to previous input source */
 						struct _yy_buffer *yb = yy_buffer;
 						int i;
 						struct _defines *ptr;
 
+						/* Check to see if we are exiting a macro value */
 						for (ptr = defines; ptr; ptr = ptr->next)
+						{
 							if (ptr->used == yy_buffer)
 							{
 								ptr->used = NULL;
-								break;
+								break; /* there can't be multiple matches */
 							}
+						}
 
 						if (yyin != NULL)
 							fclose(yyin);
@@ -1681,15 +1711,24 @@ ecpg_isspace(char ch)
 	return false;
 }
 
-static bool isdefine(void)
+/*
+ * If yytext matches a define symbol, begin scanning the symbol's value
+ * and return true
+ */
+static bool
+isdefine(void)
 {
 	struct _defines *ptr;
 
 	/* is it a define? */
 	for (ptr = defines; ptr; ptr = ptr->next)
 	{
-		if (strcmp(yytext, ptr->olddef) == 0 && ptr->used == NULL)
+		/* notice we do not match anything being actively expanded */
+		if (strcmp(yytext, ptr->name) == 0 &&
+			ptr->value != NULL &&
+			ptr->used == NULL)
 		{
+			/* Save state associated with the current buffer */
 			struct _yy_buffer *yb;
 
 			yb = mm_alloc(sizeof(struct _yy_buffer));
@@ -1698,10 +1737,17 @@ static bool isdefine(void)
 			yb->lineno = yylineno;
 			yb->filename = mm_strdup(input_filename);
 			yb->next = yy_buffer;
+			yy_buffer = yb;
 
-			ptr->used = yy_buffer = yb;
+			/* Mark symbol as being actively expanded */
+			ptr->used = yb;
 
-			yy_scan_string(ptr->newdef);
+			/*
+			 * We use yy_scan_string which will copy the value, so there's
+			 * no need to worry about a possible undef happening while we
+			 * are still scanning it.
+			 */
+			yy_scan_string(ptr->value);
 			return true;
 		}
 	}
@@ -1709,7 +1755,12 @@ static bool isdefine(void)
 	return false;
 }
 
-static bool isinformixdefine(void)
+/*
+ * Handle replacement of INFORMIX built-in defines.  This works just
+ * like isdefine() except for the source of the string to scan.
+ */
+static bool
+isinformixdefine(void)
 {
 	const char *new = NULL;
 

src/interfaces/ecpg/preproc/type.h

@@ -158,13 +158,25 @@ struct typedefs
 	struct typedefs *next;
 };
 
+/*
+ * Info about a defined symbol (macro), coming from a -D command line switch
+ * or a define command in the program.  These are stored in a simple list.
+ * Because ecpg supports compiling multiple files per run, we have to remember
+ * the command-line definitions and be able to revert to those; this motivates
+ * storing cmdvalue separately from value.
+ * name and value are separately-malloc'd strings; cmdvalue typically isn't.
+ * used is NULL unless we are currently expanding the macro, in which case
+ * it points to the buffer before the one scanning the macro; we reset it
+ * to NULL upon returning to that buffer.  This is used to prevent recursive
+ * expansion of the macro.
+ */
 struct _defines
 {
-	char	   *olddef;
+	char	   *name;			/* symbol's name */
-	char	   *newdef;
+	char	   *value;			/* current value, or NULL if undefined */
-	int			pertinent;
+	const char *cmdvalue;		/* value set on command line, or NULL */
-	void	   *used;
+	void	   *used;			/* buffer pointer, or NULL */
-	struct _defines *next;
+	struct _defines *next;		/* list link */
 };
 
 /* This is a linked list of the variable names and types. */

src/interfaces/ecpg/test/expected/sql-define.c

@@ -6,6 +6,21 @@
 /* End of automatic include section */
 #define ECPGdebug(X,Y) ECPGdebug((X)+100,(Y))
 
+#line 1 "define_prelim.pgc"
+/*
+ * Test that the effects of these commands don't carry over to the next
+ * file named on the ecpg command line.
+ */
+
+
+/* Processed by ecpg (regression mode) */
+/* These include files are added by the preprocessor */
+#include <ecpglib.h>
+#include <ecpgerrno.h>
+#include <sqlca.h>
+/* End of automatic include section */
+#define ECPGdebug(X,Y) ECPGdebug((X)+100,(Y))
+
 #line 1 "define.pgc"
 
 #line 1 "sqlca.h"
@@ -195,11 +210,57 @@ if (sqlca.sqlcode < 0) sqlprint ( );}
 
    
 
+   /* test handling of a macro defined on the command line */
+   { ECPGdo(__LINE__, 0, 1, NULL, 0, ECPGst_normal, "select 123", ECPGt_EOIT, 
+	ECPGt_int,&(i),(long)1,(long)1,sizeof(int), 
+	ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L, ECPGt_EORT);
+#line 57 "define.pgc"
+
+if (sqlca.sqlcode < 0) sqlprint ( );}
+#line 57 "define.pgc"
+
+   printf("original CMDLINESYM: %d\n", i);
+
+   
+
+   { ECPGdo(__LINE__, 0, 1, NULL, 0, ECPGst_normal, "select 42", ECPGt_EOIT, 
+	ECPGt_int,&(i),(long)1,(long)1,sizeof(int), 
+	ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L, ECPGt_EORT);
+#line 62 "define.pgc"
+
+if (sqlca.sqlcode < 0) sqlprint ( );}
+#line 62 "define.pgc"
+
+   printf("redefined CMDLINESYM: %d\n", i);
+
+   
+
+   { ECPGdo(__LINE__, 0, 1, NULL, 0, ECPGst_normal, "select 43", ECPGt_EOIT, 
+	ECPGt_int,&(i),(long)1,(long)1,sizeof(int), 
+	ECPGt_NO_INDICATOR, NULL , 0L, 0L, 0L, ECPGt_EORT);
+#line 67 "define.pgc"
+
+if (sqlca.sqlcode < 0) sqlprint ( );}
+#line 67 "define.pgc"
+
+   printf("redefined CMDLINESYM: %d\n", i);
+
+   
+
+   
+           
+   
+
+   /* this macro should not have carried over from define_prelim.pgc */
+   
+           
+   
+
    { ECPGdisconnect(__LINE__, "CURRENT");
-#line 56 "define.pgc"
+#line 81 "define.pgc"
 
 if (sqlca.sqlcode < 0) sqlprint ( );}
-#line 56 "define.pgc"
+#line 81 "define.pgc"
 
    return 0;
 }

src/interfaces/ecpg/test/expected/sql-define.stderr

@@ -48,5 +48,29 @@
 [NO_PID]: sqlca: code: 0, state: 00000
 [NO_PID]: ecpg_process_output on line 53: OK: SET
 [NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_execute on line 57: query: select 123; with 0 parameter(s) on connection ecpg1_regression
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_execute on line 57: using PQexec
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_process_output on line 57: correctly got 1 tuples with 1 fields
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_get_data on line 57: RESULT: 123 offset: -1; array: no
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_execute on line 62: query: select 42; with 0 parameter(s) on connection ecpg1_regression
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_execute on line 62: using PQexec
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_process_output on line 62: correctly got 1 tuples with 1 fields
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_get_data on line 62: RESULT: 42 offset: -1; array: no
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_execute on line 67: query: select 43; with 0 parameter(s) on connection ecpg1_regression
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_execute on line 67: using PQexec
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_process_output on line 67: correctly got 1 tuples with 1 fields
+[NO_PID]: sqlca: code: 0, state: 00000
+[NO_PID]: ecpg_get_data on line 67: RESULT: 43 offset: -1; array: no
+[NO_PID]: sqlca: code: 0, state: 00000
 [NO_PID]: ecpg_finish: connection ecpg1_regression closed
 [NO_PID]: sqlca: code: 0, state: 00000

src/interfaces/ecpg/test/expected/sql-define.stdout

@@ -1 +1,4 @@
 i: 1, s: 29-abcdef
+original CMDLINESYM: 123
+redefined CMDLINESYM: 42
+redefined CMDLINESYM: 43

src/interfaces/ecpg/test/sql/Makefile

@@ -32,5 +32,8 @@ TESTS = array array.c \
 
 all: $(TESTS)
 
+define.c: define.pgc define_prelim.pgc $(ECPG_TEST_DEPENDENCIES)
+	$(ECPG) -DCMDLINESYM=123 -o $@ $(srcdir)/define_prelim.pgc $<
+
 oldexec.c: oldexec.pgc $(ECPG_TEST_DEPENDENCIES)
 	$(ECPG) -r questionmarks -o $@ $<

src/interfaces/ecpg/test/sql/define.pgc

@@ -53,6 +53,31 @@ int main(void)
    exec sql SET TIMEZONE TO TZVAR;
    exec sql endif;
 
+   /* test handling of a macro defined on the command line */
+   exec sql select CMDLINESYM INTO :i;
+   printf("original CMDLINESYM: %d\n", i);
+
+   exec sql define CMDLINESYM 42;
+
+   exec sql select CMDLINESYM INTO :i;
+   printf("redefined CMDLINESYM: %d\n", i);
+
+   exec sql define CMDLINESYM 43;
+
+   exec sql select CMDLINESYM INTO :i;
+   printf("redefined CMDLINESYM: %d\n", i);
+
+   exec sql undef CMDLINESYM;
+
+   exec sql ifdef CMDLINESYM;
+   exec sql insert into test values (NUMBER, 'no string');
+   exec sql endif;
+
+   /* this macro should not have carried over from define_prelim.pgc */
+   exec sql ifdef NONCMDLINESYM;
+   exec sql insert into test values (NUMBER, 'no string');
+   exec sql endif;
+
    exec sql disconnect;
    return 0;
 }

src/interfaces/ecpg/test/sql/define_prelim.pgc

@@ -0,0 +1,6 @@
+/*
+ * Test that the effects of these commands don't carry over to the next
+ * file named on the ecpg command line.
+ */
+exec sql define CMDLINESYM 999;
+exec sql define NONCMDLINESYM 1234;