Re: Add section caches to coff_data_type

[Alan Modra <amodra@gmail.com>]

On Thu, May 18, 2023 at 12:24:27PM +0930, Alan Modra wrote:
> NULL.  Well, yes, the hash table isn't initialised for the output bfd.
> mkobject_hook is the wrong place to do that.

One other reason why it's the wrong place to call htab_create is that
any memory malloc'd in an object_p function should result in the
object_p returning a bfd_cleanup function that frees the memory.
bfd_alloc memory is fine in object_p, as alloc'd memory handled by
bfd_check_format_matches.

Another thing, section target_index is renumbered in
coff_compute_section_file_positions and _bfd_xcoff_bfd_final_link.  I
don't know that there is currently any way that the output bfd
section_by_target_index could be populated before this point but
clear them out so no one need worry about it.

	* coffcode.h (coff_compute_section_file_positions): Clear
	section_by_target_index hash table when changing target_index.
	(_bfd_xcoff_bfd_final_link): Likewise.

diff --git a/bfd/coffcode.h b/bfd/coffcode.h
index 777515e82e1..2d429828822 100644
--- a/bfd/coffcode.h
+++ b/bfd/coffcode.h
@@ -3067,6 +3067,9 @@ coff_compute_section_file_positions (bfd * abfd)
       sofar += bfd_coff_scnhsz (abfd);
 #endif
 
+  if (coff_data (abfd)->section_by_target_index)
+    htab_empty (coff_data (abfd)->section_by_target_index);
+
 #ifdef COFF_IMAGE_WITH_PE
   {
     /* PE requires the sections to be in memory order when listed in
diff --git a/bfd/xcofflink.c b/bfd/xcofflink.c
index b57ed21afb9..a27721301dd 100644
--- a/bfd/xcofflink.c
+++ b/bfd/xcofflink.c
@@ -7076,6 +7076,8 @@ _bfd_xcoff_bfd_final_link (bfd *abfd, struct bfd_link_info *info)
 
 	  /* Reset the section indices after inserting the new
 	     sections.  */
+	  if (xcoff_data (abfd)->coff.section_by_target_index)
+	    htab_empty (xcoff_data (abfd)->coff.section_by_target_index);
 	  indx = 0;
 	  for (o = abfd->sections; o != NULL; o = o->next)
 	    {

-- 
Alan Modra
Australia Development Lab, IBM