* emultempl/beos.em (sort_sections): Modify for 2001-08-03 change,
authorAlan Modra <amodra@gmail.com>
Sat, 4 Aug 2001 01:21:43 +0000 (01:21 +0000)
committerAlan Modra <amodra@gmail.com>
Sat, 4 Aug 2001 01:21:43 +0000 (01:21 +0000)
ie. iterate over wild_statement.section_list.
(gld${EMULATION_NAME}_place_orphan): Likewise.

ld/ChangeLog
ld/emultempl/beos.em

index efa02de..09fdc23 100644 (file)
@@ -1,3 +1,9 @@
+2001-08-04  Alan Modra  <amodra@bigpond.net.au>
+
+       * emultempl/beos.em (sort_sections): Modify for 2001-08-03 change,
+       ie. iterate over wild_statement.section_list.
+       (gld${EMULATION_NAME}_place_orphan): Likewise.
+
 2001-08-03  Stephane Carrez  <Stephane.Carrez@worldnet.fr>
 
        * scripttempl/elfm68hc12.sc (FINISH_CODE, FINISH_RELOC): New to handle
index 6c8bc6b..23ac5d4 100644 (file)
@@ -595,66 +595,73 @@ sort_sections (s)
       case lang_wild_statement_enum:
        {
          lang_statement_union_type **p = &s->wild_statement.children.head;
+         struct wildcard_list *sec;
 
-         /* Is this the .idata section?  */
-         if (s->wild_statement.section_name != NULL
-             && strncmp (s->wild_statement.section_name, ".idata", 6) == 0)
+         for (sec = s->wild_statement.section_list; sec; sec = sec->next)
            {
-             /* Sort the children.  We want to sort any objects in
-                 the same archive.  In order to handle the case of
-                 including a single archive multiple times, we sort
-                 all the children by archive name and then by object
-                 name.  After sorting them, we re-thread the pointer
-                 chain.  */
-
-             while (*p)
+             /* Is this the .idata section?  */
+             if (sec->spec.name != NULL
+                 && strncmp (sec->spec.name, ".idata", 6) == 0)
                {
-                 lang_statement_union_type *start = *p;
-                 if (start->header.type != lang_input_section_enum
-                     || !start->input_section.ifile->the_bfd->my_archive)
-                   p = &(start->header.next);
-                 else
+                 /* Sort the children.  We want to sort any objects in
+                    the same archive.  In order to handle the case of
+                    including a single archive multiple times, we sort
+                    all the children by archive name and then by object
+                    name.  After sorting them, we re-thread the pointer
+                    chain.  */
+
+                 while (*p)
                    {
-                     lang_statement_union_type *end;
-                     int count;
-
-                     for (end = start, count = 0;
-                          end && end->header.type == lang_input_section_enum;
-                          end = end->next)
-                       count++;
-
-                     p = sort_sections_1 (p, end, count, sort_by_file_name);
+                     lang_statement_union_type *start = *p;
+                     if (start->header.type != lang_input_section_enum
+                         || !start->input_section.ifile->the_bfd->my_archive)
+                       p = &(start->header.next);
+                     else
+                       {
+                         lang_statement_union_type *end;
+                         int count;
+
+                         for (end = start, count = 0;
+                              end && (end->header.type
+                                      == lang_input_section_enum);
+                              end = end->next)
+                           count++;
+
+                         p = sort_sections_1 (p, end, count,
+                                              sort_by_file_name);
+                       }
                    }
+                 break;
                }
-             break;
-           }
-
-         /* If this is a collection of grouped sections, sort them.
-            The linker script must explicitly mention "*(.foo\$)" or
-            "*(.foo\$*)".  Don't sort them if \$ is not the last
-            character (not sure if this is really useful, but it
-            allows explicitly mentioning some \$ sections and letting
-            the linker handle the rest).  */
-         if (s->wild_statement.section_name != NULL)
-           {
-             char *q = strchr (s->wild_statement.section_name, '\$');
 
-             if (q != NULL
-                 && (q[1] == '\0'
-                     || (q[1] == '*' && q[2] == '\0')))
+             /* If this is a collection of grouped sections, sort them.
+                The linker script must explicitly mention "*(.foo\$)" or
+                "*(.foo\$*)".  Don't sort them if \$ is not the last
+                character (not sure if this is really useful, but it
+                allows explicitly mentioning some \$ sections and letting
+                the linker handle the rest).  */
+             if (sec->spec.name != NULL)
                {
-                 lang_statement_union_type *end;
-                 int count;
+                 char *q = strchr (sec->spec.name, '\$');
 
-                 for (end = *p, count = 0; end; end = end->next)
+                 if (q != NULL
+                     && (q[1] == '\0'
+                         || (q[1] == '*' && q[2] == '\0')))
                    {
-                     if (end->header.type != lang_input_section_enum)
-                       abort ();
-                     count++;
+                     lang_statement_union_type *end;
+                     int count;
+
+                     for (end = *p, count = 0; end; end = end->next)
+                       {
+                         if (end->header.type != lang_input_section_enum)
+                           abort ();
+                         count++;
+                       }
+                     (void) sort_sections_1 (p, end, count,
+                                             sort_by_section_name);
                    }
-                 (void) sort_sections_1 (p, end, count, sort_by_section_name);
+                 break;
                }
-             break;
            }
        }
        break;
@@ -765,11 +772,16 @@ gld${EMULATION_NAME}_place_orphan (file, s)
   ps[0] = '\$';
   ps[1] = 0;
   for (l = os->children.head; l; l = l->next)
-    {
-      if (l->header.type == lang_wild_statement_enum
-         && strcmp (l->wild_statement.section_name, output_secname) == 0)
-       break;
-    }
+    if (l->header.type == lang_wild_statement_enum)
+      {
+       struct wildcard_list *sec;
+
+       for (sec = l->wild_statement.section_list; sec; sec = sec->next)
+         if (sec->spec.name && strcmp (sec->spec.name, output_secname) == 0)
+           break;
+       if (sec)
+         break;
+      }
   ps[0] = 0;
   if (l == NULL)
 #if 1
@@ -778,11 +790,20 @@ gld${EMULATION_NAME}_place_orphan (file, s)
         should one decide to not require *(.foo\$) to appear in the linker
         script.  */
     {
-      lang_wild_statement_type *new = new_stat (lang_wild_statement,
-                                               &os->children);
-      new->section_name = xmalloc (strlen (output_secname) + 2);
-      sprintf (new->section_name, "%s\$", output_secname);
+      lang_wild_statement_type *new;
+      struct wildcard_list *tmp;
+
+      tmp = (struct wildcard_list *) xmalloc (sizeof *tmp);
+      tmp->next = NULL;
+      tmp->spec.name = xmalloc (strlen (output_secname) + 2);
+      sprintf (tmp->spec.name, "%s\$", output_secname);
+      tmp->spec.exclude_name_list = NULL;
+      tmp->sorted = false;
+      new = new_stat (lang_wild_statement, &os->children);
       new->filename = NULL;
+      new->filenames_sorted = false;
+      new->section_list = tmp;
+      new->keep_sections = false;
       lang_list_init (&new->children);
       l = new;
     }