V: enable stack usage guard

Tmain/broken-tagname-in-ectags-format.d/run.sh

@@ -7,7 +7,7 @@ CTAGS=$1
 BUILD_SUBDIR=$2
 stderr_tmp=${BUILD_SUBDIR}/stderr-actual.txt.tmp
 
-${CTAGS} --options=NONE -o - --language-force=CTagsSelfTest --verbose --output-format=e-ctags input.cst \
+${CTAGS} --options=NONE -o - --language-force=CTagsSelfTest --stack-limit=8192 --verbose --output-format=e-ctags input.cst \
 	 2> ${stderr_tmp}
 
 # externalSortTags invokes sort command, and it is logged to stderr.

Tmain/broken-tagname-in-ectags-format.d/stderr-expected.txt

@@ -1,5 +1,6 @@
 ctags: Notice: No options will be read from files or environment
   Option: --output-format=e-ctags
+Stack usage limit: 8192 bytes (8.0 KiB)
 Reading command line arguments
 CTagsSelfTest requires a memory stream for input
 OPENING input.cst as CTagsSelfTest language file [new,required]

Tmain/broken-tagname.d/run.sh

@@ -7,12 +7,13 @@ CTAGS=$1
 BUILD_SUBDIR=$2
 stderr_tmp=${BUILD_SUBDIR}/stderr-actual.txt.tmp
 
-${CTAGS} --options=NONE -o - --language-force=CTagsSelfTest --verbose input.cst \
+${CTAGS} --options=NONE -o - --language-force=CTagsSelfTest --stack-limit=8192 --verbose input.cst \
 	 2> ${stderr_tmp}
 
 # externalSortTags invokes sort command, and it is logged to stderr.
 # Delete the line for the comparison.
-sed -e '/^system ("sort -u")$/d' < ${stderr_tmp} 1>&2
+sed -e '/^system ("sort -u")$/d' \
+	< ${stderr_tmp} 1>&2
 rm ${stderr_tmp}
 
 exit $?

Tmain/broken-tagname.d/stderr-expected.txt

@@ -1,4 +1,5 @@
 ctags: Notice: No options will be read from files or environment
+Stack usage limit: 8192 bytes (8.0 KiB)
 Reading command line arguments
 CTagsSelfTest requires a memory stream for input
 OPENING input.cst as CTagsSelfTest language file [new,required]

Tmain/versioning.d/stdout-expected.txt

@@ -70,3 +70,4 @@ EXAMINE  TEST       base => sub {shared}
 Implementation specific status
 -------------------------------------------------------
 allow null tags: no
+stack guard: unsupported

configure.ac

@@ -506,6 +506,7 @@ fi
 
 AC_CHECK_HEADERS([direct.h dirent.h fcntl.h io.h stat.h types.h unistd.h])
 AC_CHECK_HEADERS([sys/dir.h sys/stat.h sys/types.h sys/wait.h])
+AC_CHECK_HEADERS([sys/resource.h])
 
 # Checks for header file macros
 # -----------------------------
@@ -876,6 +877,7 @@ if test "${enable_static}" = "yes"; then
 	fi
 fi
 
+AC_CHECK_FUNCS(GetCurrentThreadStackLimits)
 
 # Checks for missing prototypes
 # -----------------------------
@@ -890,6 +892,8 @@ if test "$have_ftruncate" = yes ; then
 	CHECK_PROTO(ftruncate, unistd.h)
 fi
 
+CHECK_PROTO(GetCurrentThreadStackLimits, windows.h)
+
 # Process library configuration options
 # -------------------------------------
 

main/e_msoft.h

@@ -29,6 +29,8 @@
 #define HAVE_FNMATCH_H 1
 #define HAVE_PUTENV 1
 #define TMPDIR "\\"
+#define HAVE_GETCURRENTTHREADSTACKLIMITS 1
+#define NEED_PROTO_GETCURRENTTHREADSTACKLIMITS 0 /* the function declared in windows.h  */
 
 int mkstemp (char *template_name);
 

main/entry.c

@@ -821,6 +821,18 @@ static char* getFullQualifiedScopeNameFromCorkQueue (const tagEntryInfo * inner_
 			/* Force break this while-loop. */
 			scope = NULL;
 		}
+
+#define SCOPE_DEPTH_LIMIT 32
+		if (scope
+			/* '* 2' considers the separators in the queue. */
+			&& stringListCount (queue) > (SCOPE_DEPTH_LIMIT * 2))
+		{
+			notice ("too deep scope (> %d): %s in %s:%lu",
+					SCOPE_DEPTH_LIMIT,
+					scope->name, scope->inputFileName, scope->lineNumber);
+			/* Force break this while-loop. */
+			scope = NULL;
+		}
 	}
 
 	n = vStringNew ();

main/main.c

@@ -61,6 +61,7 @@
 #include "parse_p.h"
 #include "read_p.h"
 #include "routines_p.h"
+#include "stackguard_p.h"
 #include "stats_p.h"
 #include "trace.h"
 #include "trashbox_p.h"
@@ -584,6 +585,9 @@ extern int ctags_cli_main (int argc CTAGS_ATTR_UNUSED, char **argv)
 	parseCmdlineOptions (args);
 	checkOptions ();
 
+	size_t stack_limit = stackGuardGetLimit ();
+	verbose ("Stack usage limit: %zu bytes (%.1f KiB)\n",
+			 stack_limit, stack_limit / 1024.0);
 	runMainLoop (args);
 
 	/*  Clean up.

main/options.c

@@ -40,6 +40,7 @@
 #include "writer_p.h"
 #include "trace.h"
 #include "flags_p.h"
+#include "stackguard_p.h"
 
 #ifdef HAVE_JANSSON
 #include <jansson.h>
@@ -498,7 +499,9 @@ static optionDescription LongOptionDescription [] = {
  {0,0,"       Don't make tags file but just print the guessed language name for"},
  {0,0,"       input file."},
  {1,0,"  --quiet[=(yes|no)]"},
- {0,0,"       Don't print NOTICE class messages [no]."},
+ {1,0,"       Don't print NOTICE class messages [no]."},
+ {1,0,"  --stack-limit=<bytes>"},
+ {1,0,"       Limit the stack usage while parsing, in bytes (default is system-dependent)."},
  {1,0,"  --totals[=(yes|no|extra)]"},
  {1,0,"       Print statistics about input and tag files [no]."},
  {1,0,"  --verbose[=(yes|no)]"},
@@ -566,6 +569,7 @@ static const char* const License2 =
 static struct Feature {
 	const char *name;
 	const char *description;
+	char *(*descriptionNew) (const char *template);
 } Features [] = {
 #ifdef _WIN32
 	{"win32", "TO BE WRITTEN"},
@@ -627,6 +631,14 @@ static struct Feature {
 	{"optscript", "can use the interpreter"},
 #ifdef HAVE_PCRE2
 	{"pcre2", "has pcre2 regex engine"},
+#endif
+#if HAVE_SYS_RESOURCE_H || (HAVE_GETCURRENTTHREADSTACKLIMITS \
+							&& !NEED_PROTO_GETCURRENTTHREADSTACKLIMITS)
+	{"dynamic-stack-guard", "sets the default stack limit dynamically",
+	 makeFeatureStackGuardDescription },
+#else
+	{"static-stack-guard", "sets the default stack limit statically",
+	 makeFeatureStackGuardDescription },
 #endif
 	{NULL,}
 };
@@ -1566,9 +1578,15 @@ static void processListFeaturesOption(const char *const option CTAGS_ATTR_UNUSED
 		if (strcmp (Features [i].name, "regex") != 0 || checkRegex ())
 		{
 			colprintLineAppendColumnCString (line, Features [i].name);
-			colprintLineAppendColumnCString (line, Features [i].description);
+			if (Features [i].descriptionNew)
+			{
+				char *desc = Features [i].descriptionNew(Features [i].description);
+				colprintLineAppendColumnCString (line, desc);
+				eFree (desc);
+			}
+			else
+				colprintLineAppendColumnCString (line, Features [i].description);
 		}
-
 	}
 
 	colprintTableSort (table, featureCompare);
@@ -2470,6 +2488,7 @@ static void processDescribeLanguage(const char *const option,
 	puts("Implementation specific status");
 	puts("-------------------------------------------------------");
 	printf("allow null tags: %s\n", doesLanguageAllowNullTag(language)? "yes": "no");
+	printf("stack guard: %s\n", doesParserSupportStackGuard (language)? "supported": "unsupported");
 
 	exit (0);
 
@@ -2683,6 +2702,22 @@ static void processPseudoTags (const char *const option CTAGS_ATTR_UNUSED,
 	vStringDelete (str);
 }
 
+static void processStackLimit (
+		const char *const option, const char *const parameter)
+{
+	if (parameter == NULL || parameter[0] == '\0')
+		error (FATAL, "A positive number or 0 is needed after --%s option", option);
+
+	unsigned long limit = 0;
+	if (!strToULong(parameter, 0, &limit))
+		error (FATAL, "Invalid stack limit: %s", parameter);
+	if (limit > SIZE_MAX)
+		error (FATAL, "Too large limit: %s (> %lu)",
+			   parameter, SIZE_MAX);
+
+	stackGuardSetLimit ((size_t)limit);
+}
+
 static void processSortOption (
 		const char *const option, const char *const parameter)
 {
@@ -2989,10 +3024,10 @@ static void processMaxRecursionDepthOption (const char *const option, const char
 static void processPatternLengthLimit(const char *const option, const char *const parameter)
 {
 	if (parameter == NULL || parameter[0] == '\0')
-		error (FATAL, "A parameter is needed after \"%s\" option", option);
+		error (FATAL, "A positive integer is needed after --%s option", option);
 
 	if (!strToUInt(parameter, 0, &Option.patternLengthLimit))
-		error (FATAL, "-%s: Invalid pattern length limit", option);
+		error (FATAL, "Invalid pattern length limit: %s", parameter);
 }
 
 extern bool ptagMakePatternLengthLimit (ptagDesc *pdesc, langType language CTAGS_ATTR_UNUSED,
@@ -3083,6 +3118,7 @@ static parametricOption ParametricOptions [] = {
 	{ "output-format",          processOutputFormat,            true,   STAGE_ANY },
 	{ "pattern-length-limit",   processPatternLengthLimit,      true,   STAGE_ANY },
 	{ "pseudo-tags",            processPseudoTags,              false,  STAGE_ANY },
+	{ "stack-limit",            processStackLimit,              true,   STAGE_ANY },
 	{ "sort",                   processSortOption,              true,   STAGE_ANY },
 	{ "tag-relative",           processTagRelative,             true,   STAGE_ANY },
 	{ "totals",                 processTotals,                  true,   STAGE_ANY },

main/parse.c

@@ -41,6 +41,7 @@
 #include "rexprcode_p.h"
 #include "routines.h"
 #include "routines_p.h"
+#include "stackguard_p.h"
 #include "stats_p.h"
 #include "subparser.h"
 #include "subparser_p.h"
@@ -4377,6 +4378,18 @@ extern bool processFielddefOption (const char *const option, const char *const p
 *   File parsing
 */
 
+static void condPrepareStackGuard (parserDefinition *const lang)
+{
+	if (lang->discardInput)
+		stackGuardPrepare (lang->discardInput);
+}
+
+static void condReleaseStackGuard (parserDefinition *const lang)
+{
+	if (lang->discardInput)
+		stackGuardRelease ();
+}
+
 static rescanReason createTagsForFile (const langType language,
 					   const unsigned int passCount)
 {
@@ -4389,11 +4402,15 @@ static rescanReason createTagsForFile (const langType language,
 
 	notifyInputStart ();
 
+	condPrepareStackGuard (lang);
+
 	if (lang->parser != NULL)
 		lang->parser ();
 	else if (lang->parser2 != NULL)
 		rescan = lang->parser2 (passCount);
 
+	condReleaseStackGuard (lang);
+
 	notifyInputEnd ();
 
 	return rescan;
@@ -4751,6 +4768,14 @@ extern bool doesParserRequireMemoryStream (const langType language)
 	return false;
 }
 
+extern bool doesParserSupportStackGuard (const langType language)
+{
+	Assert (0 <= language  &&  language < (int) LanguageCount);
+	parserDefinition *const lang = LanguageTable [language].def;
+
+	return !!lang->discardInput;
+}
+
 extern bool parseFile (const char *const fileName)
 {
 	TRACE_ENTER_TEXT("Parsing file %s",fileName);

main/parse.h

@@ -42,6 +42,9 @@ typedef void (*parserInitialize) (langType language);
 typedef void (*initStatistics) (langType language);
 typedef void (*printStatistics) (langType langType);
 
+/* Used in the stack guard. */
+typedef void (*discardInputFn) (void *);
+
 /* Per language finalizer is called anytime when ctags exits.
    (Exceptions are a kind of options are given when invoked. Here
    options are: --version, --help, --list-*, and so on.)
@@ -158,6 +161,8 @@ struct sParserDefinition {
 	initStatistics initStats;
 	printStatistics printStats;
 
+	discardInputFn discardInput;
+
 	/* used internally */
 	langType id;		    /* id assigned to language */
 	unsigned int traced:1;
@@ -224,4 +229,7 @@ extern vString *anonGenerateNewFull (const char *prefix, langType lang, int kind
 #define anonGenerateNew(P,K) anonGenerateNewFull((P), LANG_AUTO, (K))
 extern void anonHashString (const char *filename, char buf[9]);
 
+/* Return false if unsafe. */
+extern bool stackGuardCheck (void *dataForDiscardFunc);
+
 #endif  /* CTAGS_MAIN_PARSE_H */

main/parse_p.h

@@ -155,6 +155,7 @@ extern void freeEncodingResources (void);
 #endif
 
 extern bool doesLanguageHaveForeignDependency (const langType language, const langType foreign_lang);
+extern bool doesParserSupportStackGuard (const langType language);
 
 /* Regex interface */
 extern bool processLanguageRegexOption (langType language, enum regexParserType regptype, const char *const parameter);