This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: orphan section logic
> iterates over sections that match the name..
Doh! Of course it does. New patch attached. But is this the
functionality we want? Or is it a user error to assume an input
section matches an output section just because the names match?
Index: emultempl/elf32.em
===================================================================
RCS file: /cvs/src/src/ld/emultempl/elf32.em,v
retrieving revision 1.207
diff -p -U3 -r1.207 elf32.em
--- emultempl/elf32.em 11 Dec 2009 13:42:15 -0000 1.207
+++ emultempl/elf32.em 12 Jan 2010 03:42:13 -0000
@@ -1782,6 +1782,7 @@ gld${EMULATION_NAME}_place_orphan (asect
struct orphan_save *place;
lang_output_section_statement_type *after;
lang_output_section_statement_type *os;
+ lang_output_section_statement_type *match_by_name = NULL;
int isdyn = 0;
int iself = s->owner->xvec->flavour == bfd_target_elf_flavour;
unsigned int sh_type = iself ? elf_section_type (s) : SHT_NULL;
@@ -1837,8 +1838,21 @@ gld${EMULATION_NAME}_place_orphan (asect
lang_add_section (&os->children, s, os);
return os;
}
+
+ /* Save unused output sections in case we can match them
+ against orphans later. */
+ if (os->bfd_section == NULL)
+ match_by_name = os;
}
+ /* If we didn't match an active output section, see if we matched an
+ unused one and use that. */
+ if (match_by_name)
+ {
+ lang_add_section (&match_by_name->children, s, match_by_name);
+ return match_by_name;
+ }
+
if (!orphan_init_done)
{
lang_output_section_statement_type *lookup;