[xml] ctxt->token patches

["Peter Jacobi" <pj walter-graphtek com>]

Hi Daniel, All,

Find attached patches for removing ctxt->token
usage in parser.c and parserInternals.c

As a previous patch successfully eliminated
all non-zero assignments to ctxt->token, in theory 
all uses of ctxt->token can be removed.

But at some places, removal of this
code make performance worse (must
be cache line alignment, cache preload
or static branch prediction).

So I tried to find the removals which 
enhance performance and now get a
performance gain of 7% for api.xml
on slow x86 (200Mhz PPro).

Regards,
Peter Jacobi


*** baseline\parserinternals.c	Thu Mar 21 08:31:14 2002
--- parserinternals.c	Fri Jun 28 10:23:45 2002
***************
*** 1260,1269 ****
      if (ctxt->instate == XML_PARSER_EOF)
  	return(0);
  
-     if (ctxt->token != 0) {
- 	*len = 0;
- 	return(ctxt->token);
-     }	
      if ((*ctxt->input->cur >= 0x20) && (*ctxt->input->cur <= 0x7F)) {
  	    *len = 1;
  	    return((int) *ctxt->input->cur);
--- 1260,1265 ----

*** baseline\parser.c	Sat Jun 15 09:32:20 2002
--- parser.c	Fri Jun 28 12:48:06 2002
***************
*** 268,275 ****
   *   GROW, SHRINK  handling of input buffers
   */
  
! #define RAW (ctxt->token ? -1 : (*ctxt->input->cur))
! #define CUR (ctxt->token ? ctxt->token : (*ctxt->input->cur))
  #define NXT(val) ctxt->input->cur[(val)]
  #define CUR_PTR ctxt->input->cur
  
--- 268,275 ----
   *   GROW, SHRINK  handling of input buffers
   */
  
! #define RAW (*ctxt->input->cur)
! #define CUR (*ctxt->input->cur)
  #define NXT(val) ctxt->input->cur[(val)]
  #define CUR_PTR ctxt->input->cur
  
***************
*** 2366,2394 ****
      while (((NXT(0) != limit) && /* checked */
  	   (c != '<')) || (ctxt->token != 0)) {
  	if (c == 0) break;
! 	if (ctxt->token == '&') {
! 	    if (ctxt->replaceEntities) {
! 		if (len > buf_size - 10) {
! 		    growBuffer(buf);
! 		}
! 		buf[len++] = '&';
! 	    } else {
! 		/*
! 		 * The reparsing will be done in xmlStringGetNodeList()
! 		 * called by the attribute() function in SAX.c
! 		 */
! 		static xmlChar buffer[6] = "&#38;";
! 
! 		if (len > buf_size - 10) {
! 		    growBuffer(buf);
! 		}
! 		current = &buffer[0];
! 		while (*current != 0) { /* non input consuming */
! 		    buf[len++] = *current++;
! 		}
! 		ctxt->token = 0;
! 	    }
! 	} else if (c == '&') {
  	    if (NXT(1) == '#') {
  		int val = xmlParseCharRef(ctxt);
  		if (val == '&') {
--- 2366,2372 ----
      while (((NXT(0) != limit) && /* checked */
  	   (c != '<')) || (ctxt->token != 0)) {
  	if (c == 0) break;
! 	if (c == '&') {
  	    if (NXT(1) == '#') {
  		int val = xmlParseCharRef(ctxt);
  		if (val == '&') {
***************
*** 6884,6894 ****
  void
  xmlParseContent(xmlParserCtxtPtr ctxt) {
      GROW;
!     while (((RAW != 0) || (ctxt->token != 0)) &&
  	   ((RAW != '<') || (NXT(1) != '/'))) {
  	const xmlChar *test = CUR_PTR;
  	int cons = ctxt->input->consumed;
- 	int tok = ctxt->token;
  	const xmlChar *cur = ctxt->input->cur;
  
  	/*
--- 6862,6871 ----
  void
  xmlParseContent(xmlParserCtxtPtr ctxt) {
      GROW;
!     while ((RAW != 0) &&
  	   ((RAW != '<') || (NXT(1) != '/'))) {
  	const xmlChar *test = CUR_PTR;
  	int cons = ctxt->input->consumed;
  	const xmlChar *cur = ctxt->input->cur;
  
  	/*
***************
*** 6955,6962 ****
  	    xmlPopInput(ctxt);
  	SHRINK;
  
! 	if ((cons == ctxt->input->consumed) && (test == CUR_PTR) &&
! 	    (tok == ctxt->token)) {
  	    ctxt->errNo = XML_ERR_INTERNAL_ERROR;
  	    if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL))
  	        ctxt->sax->error(ctxt->userData,
--- 6932,6938 ----
  	    xmlPopInput(ctxt);
  	SHRINK;
  
! 	if ((cons == ctxt->input->consumed) && (test == CUR_PTR)) {
  	    ctxt->errNo = XML_ERR_INTERNAL_ERROR;
  	    if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL))
  	        ctxt->sax->error(ctxt->userData,

Attachment: ctxt-token-patch-1.zip
Description: Zip archive