fix pesky memory leak finally

Highlight.cpp

@@ -170,6 +170,10 @@ int32_t Highlight::set ( SafeBuf *sb ,
 	//m_bufLen = bufLen;
 	//m_bufPtr = buf;
 	m_sb = sb;
+
+	// label it
+	m_sb->setLabel ("highw");
+
 	// save room for terminating \0
 	//m_bufEnd = m_buf + m_bufLen - 1;
 

Mem.cpp

@@ -530,6 +530,11 @@ void Mem::addMem ( void *mem , int32_t size , const char *note , char isnew ) {
 
 	//validate();
 
+	// if ( note && note[0] == 'S' && note[1] == 'a' &&
+	//      note[2] == 'f' && size == 13371521 )
+	// 	log("mem: got mystery safebuf");
+
+
         //m_memtablesize = 0;//DMEMTABLESIZE;
 	  // 4G/x = 600*1024 -> x = 4000000000.0/(600*1024) = 6510
 	// crap, g_hostdb.init() is called inmain.cpp before

Msg2.cpp

@@ -413,6 +413,8 @@ bool Msg2::getLists ( ) {
 		// mem. we should also report the size of each termlist
 		// in bytes in the query info header.
 		//int32_t minRecSizes = DEFAULT_POSDB_READSIZE;
+		// MDW TODO fix this later we go oom too easily for queries
+		// like 'www.disney.nl'
 		int32_t minRecSizes = -1;
 
 		// start up the read. thread will wait in thread queue to 

Msg3a.cpp

@@ -380,7 +380,9 @@ bool Msg3a::gotCacheReply ( ) {
 		rs = DEFAULT_POSDB_READSIZE;//90000000; // 90MB!
 		// it is better to go oom then leave users scratching their
 		// heads as to why some results are not being returned.
-		rs = -1;
+		// no, because we are going out of mem for queries like
+		// 'www.disney.nl' etc.
+		//rs = -1;
 		// if section stats, limit to 1MB
 		//if ( m_r->m_getSectionStats ) rs = 1000000;
 		// get the jth query term

Pages.cpp

@@ -3519,6 +3519,7 @@ bool printApiForPage ( SafeBuf *sb , int32_t PAGENUM , CollectionRec *cr ) {
 		if ( pageNum != PAGENUM ) continue;
 
 		SafeBuf tmp;
+		tmp.setLabel("apisb");
 		char diff = 0;
 		bool printVal = false;
 		if ( parm->m_type != TYPE_CMD &&

Query.cpp

@@ -28,6 +28,7 @@ void Query::constructor ( ) {
 	//m_bmap      = NULL;
 	m_bitScores = NULL;
 	m_qwords      = NULL;
+	m_numWords = 0;
 	//m_expressions = NULL;
 	m_qwordsAllocSize      = 0;
 	//m_expressionsAllocSize = 0;
@@ -68,6 +69,11 @@ void Query::reset ( ) {
 		qt->m_facetIndexBuf.purge();
 	}
 
+	for ( int32_t i = 0 ; i < m_numWords ; i++ ) {
+		QueryWord *qw = &m_qwords[i];
+		qw->destructor();
+	}
+
 	m_sb.purge();
 	m_osb.purge();
 	m_docIdRestriction = 0LL;
@@ -86,6 +92,7 @@ void Query::reset ( ) {
 	//if ( m_bitScores && m_bitScoresSize ) //  != m_bsbuf )
 	//	mfree ( m_bitScores , m_bitScoresSize , "Query2" );
 	//m_bmap = NULL;
+
 	m_bitScores = NULL;
 	//m_bmapSize      = 0;
 	m_bitScoresSize = 0;
@@ -1402,6 +1409,7 @@ bool Query::setQTerms ( Words &words , Phrases &phrases ) {
 		// sanity
 		if ( naids > MAX_SYNS ) { char *xx=NULL;*xx=0; }
 		// now make the buffer to hold them for us
+		qw->m_synWordBuf.setLabel("qswbuf");
 		qw->m_synWordBuf.safeMemcpy ( &syn.m_synWordBuf );
 		// get the term for this word
 		QueryTerm *origTerm = qw->m_queryWordTerm;
@@ -2057,6 +2065,9 @@ bool Query::setQWords ( char boolFlag ,
 			return log("query: Could not allocate mem for query.");
 		m_qwordsAllocSize = need;
 	}
+	// reset safebuf in there
+	for ( int32_t i = 0 ; i < m_numWords ; i++ )
+		m_qwords[i].constructor();
 
 	// is all alpha chars in query in upper case? caps lock on?
 	bool allUpper = true;
@@ -5655,3 +5666,11 @@ int64_t Query::getQueryHash() {
 	}
 	return qh;
 }
+
+void QueryWord::constructor () {
+	m_synWordBuf.constructor();
+}
+
+void QueryWord::destructor () {
+	m_synWordBuf.purge();
+}

Query.h

@@ -274,6 +274,9 @@ class QueryWord {
 			if ( is_wspace_utf8 ( p ) ) return true;
 		return false;
 	};
+	void constructor ();
+	void destructor ();
+
 	//UCScript wordScript() { 
 	//	UChar*foo;
 	//	return ucGetScript(utf16Decode((UChar*)(m_word),&foo));

SafeBuf.cpp

@@ -522,7 +522,8 @@ int32_t SafeBuf::safeSave (char *filename ) {
 }
 
 
-int32_t SafeBuf::fillFromFile(char *dir,char *filename) {
+int32_t SafeBuf::fillFromFile(char *dir,char *filename,char *label) {
+	m_label = label;
 	char buf[1024];
 	if ( dir ) snprintf(buf,1024,"%s/%s",dir,filename);
 	else       snprintf(buf,1024,"%s",filename);

SafeBuf.h

@@ -73,8 +73,9 @@ public:
 	int32_t safeSave (char *filename );
 
 	int32_t  fillFromFile(char *filename);
-	int32_t  fillFromFile(char *dir,char *filename);
-	int32_t  load(char *dir,char *fname) { return fillFromFile(dir,fname);};
+	int32_t  fillFromFile(char *dir,char *filename, char *label=NULL);
+	int32_t  load(char *dir,char *fname,char *label = NULL) { 
+		return fillFromFile(dir,fname,label);};
 	int32_t  load(char *fname) { return fillFromFile(fname);};
 
 	void filterTags();

SearchInput.cpp

@@ -825,6 +825,9 @@ bool SearchInput::setQueryBuffers ( HttpRequest *hr ) {
 			m_sbuf2.safeStrcpy(" AND ");
 		}
 	}
+	m_sbuf1.setLabel("sisbuf1");
+	m_sbuf2.setLabel("sisbuf2");
+	m_sbuf3.setLabel("sisbuf3");
 	// append the natural query
 	if ( m_query && m_query[0] ) {
 		//if ( p  > pstart  ) *p++  = ' ';

Synonyms.cpp

@@ -12,6 +12,7 @@
 #include "Wiktionary.h"
 
 Synonyms::Synonyms() {
+	m_synWordBuf.setLabel("syswbuf");
 }
 
 Synonyms::~Synonyms() {

Tagdb.cpp

@@ -5049,8 +5049,8 @@ bool Tagdb::loadMinSiteInlinksBuffer2 ( ) {
 	// use 4 bytes for the first 130,000 entries or so to hold
 	// # of site inlinks. then we only need 1 byte since the remaining
 	// 25M are <256 sitenuminlinksunqiecblocks
-	m_siteBuf1.load(g_hostdb.m_dir,"sitelinks1.dat");
-	m_siteBuf2.load(g_hostdb.m_dir,"sitelinks2.dat");
+	m_siteBuf1.load(g_hostdb.m_dir,"sitelinks1.dat","stelnks1");
+	m_siteBuf2.load(g_hostdb.m_dir,"sitelinks2.dat","stelnks2");
 
 	m_siteBuf1.setLabel("sitelnks");
 	m_siteBuf2.setLabel("sitelnks");