Update.

1998-04-30 16:45  Ulrich Drepper  <drepper@cygnus.com>

	* inet/ether_aton.c: Including netinet/if_ether.h is not necessary.

	* locale/Makefile (distribute): Add programs/repertoire.h.
	(localedef-modules): Add repertoire.
	(CPPFLAGS): Define REPERTOIREMAP_PATH.
	* locale/programs/repertoire.c: New file.
	* locale/programs/repertoire.h: New file.
	* locale/programs/charmap.c: Starting fixing character set handling
	to handle multi-byte encodings.
	* locale/programs/charset.c: Likewise.
	* locale/programs/charset.h: Likewise.
	* locale/programs/ld-collate.c: Likewise.
	* locale/programs/ld-ctype.c: Likewise.
	* locale/programs/linereader.c: Likewise.
	* locale/programs/localedef.c: Likewise.
	* locale/programs/locfile-kw.gperf: Likewise.
	* locale/programs/locfile-kw.h: Likewise.
	* locale/programs/locfile-token.h: Likewise.
	* locale/programs/locfile.h: Likewise.
	* locale/programs/stringtrans.c: Likewise.

1998-04-18  Philip Blundell  <Philip.Blundell@pobox.com>

	* sysdeps/arm/memset.S: Fix off by one error.

	* sysdeps/unix/sysv/linux/arm/sysdep.h (PSEUDO): On error, call
	__syscall_error rather than syscall_error directly.

1998-04-17  Philip Blundell  <Philip.Blundell@pobox.com>

	* sysdeps/unix/sysv/linux/arm/mmap.S: New file; implementation of
	mmap() syscall for ARM.

	* sysdeps/unix/arm/start.c: New file; startup code for ARM a.out
	binaries.

14 files changed, 565 insertions, 171 deletions

diff --git a/locale/programs/charmap.c b/locale/programs/charmap.c
index 0cd62fbfd4..7114a237a0 100644
--- a/locale/programs/charmap.c
+++ b/locale/programs/charmap.c
@@ -33,6 +33,8 @@
 #include "error.h"
 #include "linereader.h"
 #include "charset.h"
+#include "locfile.h"
+#include "repertoire.h"
 
 
 /* Uncomment following line for production version.  */
@@ -209,6 +211,8 @@ parse_charmap (const char *filename)
   memset (result, '\0', sizeof (struct charset_t));
   /* The default DEFAULT_WIDTH is 1.  */
   result->width_default = 1;
+  /* Let the user overwrite the repertoire map we use.  */
+  result->repertoiremap = repertoiremap;
 
 #define obstack_chunk_alloc malloc
 #define obstack_chunk_free free
@@ -265,6 +269,17 @@ parse_charmap (const char *filename)
 
 	      lr_ignore_rest (cmfile, 1);
 
+	      /* Read the repertoire map now.  */
+	      if (result->repertoiremap == NULL)
+		/* This is fatal.  */
+		error (4, 0, _("no repertoire map specified: cannot proceed"));
+
+	      result->repertoire = repertoire_read (result->repertoiremap);
+	      if (result->repertoire == NULL)
+		/* This is also fatal.  */
+		error (4, errno, _("cannot read repertoire map `%s'"),
+		       result->repertoiremap);
+
 	      state = 2;
 	      continue;
 	    }
@@ -273,7 +288,7 @@ parse_charmap (const char *filename)
 	      && nowtok != tok_mb_cur_min && nowtok != tok_escape_char
 	      && nowtok != tok_comment_char && nowtok != tok_g0esc
 	      && nowtok != tok_g1esc && nowtok != tok_g2esc
-	      && nowtok != tok_g3esc)
+	      && nowtok != tok_g3esc && nowtok != tok_repertoiremap)
 	    {
 	      lr_error (cmfile, _("syntax error in prolog: %s"),
 			_("illegal definition"));
@@ -305,6 +320,18 @@ parse_charmap (const char *filename)
 	      lr_ignore_rest (cmfile, 1);
 	      continue;
 
+	    case tok_repertoiremap:
+	      if (arg->tok != tok_ident)
+		goto badarg;
+
+	      if (result->repertoiremap == NULL)
+		result->repertoiremap = obstack_copy0 (&result->mem_pool,
+						       arg->val.str.start,
+						       arg->val.str.len);
+
+	      lr_ignore_rest (cmfile, 1);
+	      continue;
+
 	    case tok_mb_cur_max:
 	    case tok_mb_cur_min:
 	      if (arg->tok != tok_number)
@@ -437,14 +464,14 @@ argument to <%s> must be a single character"),
 	      continue;
 	    }
 
-	  if (nowtok == tok_charcode)
-	    /* Write char value in table.  */
-	    charset_new_char (cmfile, result, now->val.charcode.nbytes,
-			      now->val.charcode.val, from_name, to_name);
+	  if (now->val.charcode.nbytes < result->mb_cur_min)
+	    lr_error (cmfile, _("too few bytes in character encoding"));
+	  else if (now->val.charcode.nbytes > result->mb_cur_max)
+	    lr_error (cmfile, _("too many bytes in character encoding"));
 	  else
-	    /* Determine ISO 10646 value and write into table.  */
-	    charset_new_unicode (cmfile, result, now->val.charcode.nbytes,
-				 now->val.charcode.val, from_name, to_name);
+	    charset_new_char (cmfile, &result->char_table,
+			      now->val.charcode.nbytes,
+			      now->val.charcode.val, from_name, to_name);
 
 	  /* Ignore trailing comment silently.  */
 	  lr_ignore_rest (cmfile, 0);
@@ -466,8 +493,7 @@ argument to <%s> must be a single character"),
 	      continue;
 	    }
 
-	  /* If the previous line was not completely correct free the
-	     used memory.  */
+	  /* Copy the to-name in a safe place.  */
 	  to_name = (char *) obstack_copy0 (&result->mem_pool,
 					    cmfile->token.val.str.start,
 					    cmfile->token.val.str.len);
@@ -694,7 +720,7 @@ new_width (struct linereader *cmfile, struct charset_t *result,
 {
   unsigned int from_val, to_val;
 
-  from_val = charset_find_value (result, from, strlen (from));
+  from_val = charset_find_value (&result->char_table, from, strlen (from));
   if ((wchar_t) from_val == ILLEGAL_CHAR_VALUE)
     {
       lr_error (cmfile, _("unknown character `%s'"), from);
@@ -705,7 +731,7 @@ new_width (struct linereader *cmfile, struct charset_t *result,
     to_val = from_val;
   else
     {
-      to_val = charset_find_value (result, to, strlen (to));
+      to_val = charset_find_value (&result->char_table, to, strlen (to));
       if ((wchar_t) to_val == ILLEGAL_CHAR_VALUE)
 	{
 	  lr_error (cmfile, _("unknown character `%s'"), to);
diff --git a/locale/programs/charset.c b/locale/programs/charset.c
index fdacf25328..767fafb93a 100644
--- a/locale/programs/charset.c
+++ b/locale/programs/charset.c
@@ -1,4 +1,4 @@
-/* Copyright (C) 1996, 1997 Free Software Foundation, Inc.
+/* Copyright (C) 1996, 1997, 1998 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Contributed by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1996.
 
@@ -23,6 +23,8 @@
 
 #include <alloca.h>
 #include <ctype.h>
+#include <errno.h>
+#include <limits.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -31,49 +33,24 @@
 #include "charset.h"
 
 
-static void
-insert_char (struct linereader *lr, struct charset_t *cs, int bytes,
-	     unsigned int value, const char *from, const char *to);
-
-
-void
-charset_new_char (struct linereader *lr, struct charset_t *cs, int bytes,
-		  unsigned int value, const char *from, const char *to)
-{
-  if (bytes < cs->mb_cur_min)
-    lr_error (lr, _("too few bytes in character encoding"));
-  else if (bytes > cs->mb_cur_max)
-    lr_error (lr, _("too many bytes in character encoding"));
-  else
-    insert_char (lr, cs, bytes, value, from, to);
-}
-
-
-void
-charset_new_unicode (struct linereader *lr, struct charset_t *cs, int bytes,
-		     unsigned int value, const char *from, const char *to)
-{
-  /* For now: perhaps <Uxxxx> support will be removed again... */
-  insert_char (lr, cs, bytes, value, from, to);
-}
-
-
 unsigned int
-charset_find_value (const struct charset_t *cs, const char *name, size_t len)
+charset_find_value (const hash_table *ht, const char *name, size_t len)
 {
   void *result;
 
-  if (find_entry ((hash_table *) &cs->char_table, name, len, &result) < 0)
+  if (find_entry ((hash_table *) ht, name, len, &result) < 0)
     return ILLEGAL_CHAR_VALUE;
 
   return (unsigned int) ((unsigned long int) result);
 }
 
 
-static void
-insert_char (struct linereader *lr, struct charset_t *cs, int bytes,
-	     unsigned int value, const char *from, const char *to)
+void
+charset_new_char (struct linereader *lr, hash_table *ht, int bytes,
+		  unsigned int value, const char *from, const char *to)
 {
+  char *from_end;
+  char *to_end;
   const char *cp;
   char *buf;
   int prefix_len, len1, len2;
@@ -81,7 +58,7 @@ insert_char (struct linereader *lr, struct charset_t *cs, int bytes,
 
   if (to == NULL)
     {
-      if (insert_entry (&cs->char_table, from, strlen (from),
+      if (insert_entry (ht, from, strlen (from),
 			(void *) (unsigned long int) value)
 	  < 0)
 	lr_error (lr, _("duplicate character name `%s'"), from);
@@ -111,8 +88,16 @@ insert_char (struct linereader *lr, struct charset_t *cs, int bytes,
   if (cp == &from[len1 - 1] || strncmp (from, to, prefix_len) != 0)
     goto illegal_range;
 
-  from_nr = strtoul (&from[prefix_len], NULL, 10);
-  to_nr = strtoul (&to[prefix_len], NULL, 10);
+  errno = 0;
+  from_nr = strtoul (&from[prefix_len], &from_end, 10);
+  if (*from_end != '\0' || (from_nr == ULONG_MAX && errno == ERANGE)
+      || ((to_nr = strtoul (&to[prefix_len], &to_end, 10)) == ULONG_MAX
+	  && errno == ERANGE)
+      || *to_end != '\0')
+    {
+      lr_error (lr, _("<%s> and <%s> are illegal names for range"));
+      return;
+    }
 
   if (from_nr > to_nr)
     {
@@ -127,7 +112,7 @@ insert_char (struct linereader *lr, struct charset_t *cs, int bytes,
     {
       sprintf (&buf[prefix_len], "%0*d", len1 - prefix_len, cnt);
 
-      if (insert_entry (&cs->char_table, buf, len1,
+      if (insert_entry (ht, buf, len1,
 			(void *) (unsigned long int) (value + (cnt - from_nr)))
 	  < 0)
 	lr_error (lr, _("duplicate character name `%s'"), buf);
diff --git a/locale/programs/charset.h b/locale/programs/charset.h
index 82c4ef0c57..db93f16306 100644
--- a/locale/programs/charset.h
+++ b/locale/programs/charset.h
@@ -1,4 +1,4 @@
-/* Copyright (C) 1996, 1997 Free Software Foundation, Inc.
+/* Copyright (C) 1996, 1997, 1998 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Contributed by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1996.
 
@@ -22,6 +22,7 @@
 
 #include <obstack.h>
 
+#include "repertoire.h"
 #include "simple-hash.h"
 #include "linereader.h"
 
@@ -36,6 +37,9 @@ struct width_rule
 
 struct charset_t
 {
+  const char *repertoiremap;
+  struct repertoire_t *repertoire;
+
   const char *code_set_name;
   int mb_cur_min;
   int mb_cur_max;
@@ -63,14 +67,11 @@ extern int be_quiet;
 struct charset_t *charmap_read (const char *filename);
 
 /* Prototypes for function to insert new character.  */
-void charset_new_char (struct linereader *lr, struct charset_t *cs, int bytes,
+void charset_new_char (struct linereader *lr, hash_table *ht, int bytes,
 		       unsigned int value, const char *from, const char *to);
 
-void charset_new_unicode (struct linereader *lr, struct charset_t *cs,
-			  int bytes, unsigned int value, const char *from,
-			  const char *to);
-
-unsigned int charset_find_value (const struct charset_t *__cs,
-				 const char *__name, size_t __len);
+/* Return the value stored under the given key in the hashing table.  */
+unsigned int charset_find_value (const hash_table *ht,
+				 const char *name, size_t len);
 
 #endif /* charset.h */
diff --git a/locale/programs/ld-collate.c b/locale/programs/ld-collate.c
index 57b97767af..a92ff1154a 100644
--- a/locale/programs/ld-collate.c
+++ b/locale/programs/ld-collate.c
@@ -212,7 +212,7 @@ collate_finish (struct localedef_t *locale, struct charset_t *charset)
       void *ptmp;
       unsigned int value = 0;
 
-      wch = charset_find_value (charset, patch->token, toklen);
+      wch = charset_find_value (&charset->char_table, patch->token, toklen);
       if (wch != ILLEGAL_CHAR_VALUE)
 	{
 	  element_t *runp;
@@ -1054,7 +1054,8 @@ collate_element_to (struct linereader *lr, struct localedef_t *locale,
       collate->combine_token = NULL;
     }
 
-  value = charset_find_value (charset, code->val.str.start, code->val.str.len);
+  value = charset_find_value (&charset->char_table, code->val.str.start,
+			      code->val.str.len);
   if ((wchar_t) value != ILLEGAL_CHAR_VALUE)
     {
       lr_error (lr, _("symbol for multicharacter collating element "
@@ -1181,7 +1182,8 @@ collate_symbol (struct linereader *lr, struct localedef_t *locale,
   wchar_t value;
   void *not_used;
 
-  value = charset_find_value (charset, code->val.str.start, code->val.str.len);
+  value = charset_find_value (&charset->char_table, code->val.str.start,
+			      code->val.str.len);
   if (value != ILLEGAL_CHAR_VALUE)
     {
       lr_error (lr, _("symbol for multicharacter collating element "
@@ -1268,7 +1270,7 @@ collate_order_elem (struct linereader *lr, struct localedef_t *locale,
     {
     case tok_bsymbol:
       /* We have a string to find in one of the three hashing tables.  */
-      value = charset_find_value (charset, code->val.str.start,
+      value = charset_find_value (&charset->char_table, code->val.str.start,
 				  code->val.str.len);
       if (value != ILLEGAL_CHAR_VALUE)
 	{
@@ -1533,7 +1535,8 @@ collate_weight_bsymbol (struct linereader *lr, struct localedef_t *locale,
 
   assert (code->tok == tok_bsymbol);
 
-  value = charset_find_value (charset, code->val.str.start, code->val.str.len);
+  value = charset_find_value (&charset->char_table, code->val.str.start,
+			      code->val.str.len);
   if (value != ILLEGAL_CHAR_VALUE)
     {
       element_t *runp;
@@ -1706,7 +1709,8 @@ collate_simple_weight (struct linereader *lr, struct localedef_t *locale,
 		    return -1;
 		  }
 
-		wch = charset_find_value (charset, startp, putp - startp);
+		wch = charset_find_value (&charset->char_table, startp,
+					  putp - startp);
 		if (wch != ILLEGAL_CHAR_VALUE)
 		  {
 		    element_t *pelem;
diff --git a/locale/programs/ld-ctype.c b/locale/programs/ld-ctype.c
index f2f32cc214..2f9a9a2caf 100644
--- a/locale/programs/ld-ctype.c
+++ b/locale/programs/ld-ctype.c
@@ -308,7 +308,7 @@ character %s'%s' in class `%s' must not be in class `%s'"),
     }
 
   /* ... and now test <SP> as a special case.  */
-  space_value = charset_find_value (charset, "SP", 2);
+  space_value = charset_find_value (&charset->char_table, "SP", 2);
   if ((wchar_t) space_value == ILLEGAL_CHAR_VALUE)
     {
       if (!be_quiet)
@@ -634,7 +634,8 @@ ctype_class_from (struct linereader *lr, struct localedef_t *locale,
   struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype;
   unsigned int value;
 
-  value = charset_find_value (charset, code->val.str.start, code->val.str.len);
+  value = charset_find_value (&charset->char_table, code->val.str.start,
+			      code->val.str.len);
 
   ctype->last_class_char = value;
 
@@ -656,7 +657,8 @@ ctype_class_to (struct linereader *lr, struct localedef_t *locale,
   struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype;
   unsigned int value, cnt;
 
-  value = charset_find_value (charset, code->val.str.start, code->val.str.len);
+  value = charset_find_value (&charset->char_table, code->val.str.start,
+			      code->val.str.len);
 
   /* In the LC_CTYPE category it is no error when a character is
      not found.  This has to be ignored silently.  */
@@ -750,7 +752,8 @@ ctype_map_from (struct linereader *lr, struct localedef_t *locale,
   struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype;
   unsigned int value;
 
-  value = charset_find_value (charset, code->val.str.start, code->val.str.len);
+  value = charset_find_value (&charset->char_table, code->val.str.start,
+			      code->val.str.len);
 
   if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
     /* In the LC_CTYPE category it is no error when a character is
@@ -770,7 +773,8 @@ ctype_map_to (struct linereader *lr, struct localedef_t *locale,
   struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype;
   unsigned int value;
 
-  value = charset_find_value (charset, code->val.str.start, code->val.str.len);
+  value = charset_find_value (&charset->char_table, code->val.str.start,
+			      code->val.str.len);
 
   if ((wchar_t) ctype->from_map_char == ILLEGAL_CHAR_VALUE
       || (wchar_t) value == ILLEGAL_CHAR_VALUE)
@@ -948,7 +952,7 @@ set_class_defaults (struct locale_ctype_t *ctype, struct charset_t *charset)
 	  unsigned int value;
 	  tmp[0] = ch;
 
-	  value = charset_find_value (charset, tmp, 1);
+	  value = charset_find_value (&charset->char_table, tmp, 1);
 	  if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	    {
 	      if (!be_quiet)
@@ -1013,7 +1017,7 @@ character `%s' not defined while needed as default value"),
     {
       unsigned int value;
 
-      value = charset_find_value (charset, "space", 5);
+      value = charset_find_value (&charset->char_table, "space", 5);
       if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1024,7 +1028,7 @@ character `%s' not defined while needed as default value"),
       else
 	ELEM (ctype, class_collection, , value) |= BIT (tok_space);
 
-      value = charset_find_value (charset, "form-feed", 9);
+      value = charset_find_value (&charset->char_table, "form-feed", 9);
       if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1035,7 +1039,7 @@ character `%s' not defined while needed as default value"),
       else
 	ELEM (ctype, class_collection, , value) |= BIT (tok_space);
 
-      value = charset_find_value (charset, "newline", 7);
+      value = charset_find_value (&charset->char_table, "newline", 7);
       if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1046,7 +1050,7 @@ character `%s' not defined while needed as default value"),
       else
 	ELEM (ctype, class_collection, , value) |= BIT (tok_space);
 
-      value = charset_find_value (charset, "carriage-return", 15);
+      value = charset_find_value (&charset->char_table, "carriage-return", 15);
       if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1057,7 +1061,7 @@ character `%s' not defined while needed as default value"),
       else
 	ELEM (ctype, class_collection, , value) |= BIT (tok_space);
 
-      value = charset_find_value (charset, "tab", 3);
+      value = charset_find_value (&charset->char_table, "tab", 3);
       if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1068,7 +1072,7 @@ character `%s' not defined while needed as default value"),
       else
 	ELEM (ctype, class_collection, , value) |= BIT (tok_space);
 
-      value = charset_find_value (charset, "vertical-tab", 12);
+      value = charset_find_value (&charset->char_table, "vertical-tab", 12);
       if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1097,7 +1101,7 @@ character `%s' not defined while needed as default value"),
    {
       unsigned int value;
 
-      value = charset_find_value (charset, "space", 5);
+      value = charset_find_value (&charset->char_table, "space", 5);
       if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1108,7 +1112,7 @@ character `%s' not defined while needed as default value"),
       else
 	ELEM (ctype, class_collection, , value) |= BIT (tok_blank);
 
-      value = charset_find_value (charset, "tab", 3);
+      value = charset_find_value (&charset->char_table, "tab", 3);
       if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1149,7 +1153,7 @@ character `%s' not defined while needed as default value"),
 	if ((ctype->class_collection[cnt] & mask) != 0)
 	  ctype->class_collection[cnt] |= BIT (tok_print);
 
-      space = charset_find_value (charset, "space", 5);
+      space = charset_find_value (&charset->char_table, "space", 5);
       if (space == ILLEGAL_CHAR_VALUE)
 	{
 	  if (!be_quiet)
@@ -1178,7 +1182,7 @@ character `%s' not defined while needed as default value"),
 
 	  tmp[1] = (char) ch;
 
-	  value_from = charset_find_value (charset, &tmp[1], 1);
+	  value_from = charset_find_value (&charset->char_table, &tmp[1], 1);
 	  if ((wchar_t) value_from == ILLEGAL_CHAR_VALUE)
 	    {
 	      if (!be_quiet)
@@ -1190,7 +1194,7 @@ character `%s' not defined while needed as default value"),
 
 	  /* This conversion is implementation defined.  */
 	  tmp[1] = (char) (ch + ('A' - 'a'));
-	  value_to = charset_find_value (charset, &tmp[1], 1);
+	  value_to = charset_find_value (&charset->char_table, &tmp[1], 1);
 	  if ((wchar_t) value_to == ILLEGAL_CHAR_VALUE)
 	    {
 	      if (!be_quiet)
diff --git a/locale/programs/linereader.c b/locale/programs/linereader.c
index 6692164b60..4406e1ab72 100644
--- a/locale/programs/linereader.c
+++ b/locale/programs/linereader.c
@@ -524,7 +524,7 @@ get_string (struct linereader *lr, const struct charset_t *charset)
 
 	if (lr->translate_strings)
 	  {
-	    value = charset_find_value (charset, &buf[startidx],
+	    value = charset_find_value (&charset->char_table, &buf[startidx],
 					bufact - startidx);
 	    if ((wchar_t) value == ILLEGAL_CHAR_VALUE)
 	      illegal_string = 1;