Re: [PATCH] PR ld/17973: Don't generate the fake FILE symbol when plugin is used

[Alan Modra <amodra at gmail dot com>]

LTO output objects have an STT_FILE symbol using the name of the file,
a temporary file.  This results in executables that can't be exactly
reproduced, so the file name needs to be dropped.  We don't want to
lose all file symbols when linking a mix of lto and non-lto objects as
a file symbol can be used to figure which source file generated a
given local symbol.  So lto output objects need to be marked.

I chose to mark lto output objects with a new bfd flag.  This flag is
also used to fix a bug in the link-once handling;  An object being
loaded after "loading_lto_outputs" is set might be one extracted from
an archive to satisfy new references from lto objects, not an lto
object itself.

The new flag is copied from archive to elements, and the same done
for no_export.  This fixes a bug in that --exclude-libs doesn't work
with thin archives.  I'm not completely happy with this part of the
patch and may revist this to avoid the hack in
_bfd_look_for_bfd_in_cache.

	PR ld/17973
include/
	* bfdlink.h (struct bfd_link_info): Delete loading_lto_outputs.
bfd/
	* bfd.c (struct bfd): Add lto_output.
	* linker.c (_bfd_handle_already_linked): Explicitly test for
	objects added by the lto plugin.
	* opncls.c (_bfd_new_bfd_contained_in): Copy lto_output and
	no_export flags from archive.
	* archive.c (open_nested_file): New function, setting lto_output
	and no_export, extracted from..
	(find_nested_archive): ..here.  Flip params.  Rename from
	_bfd_find_nested_archive.
	(_bfd_get_elt_at_filepos): Correct var typo.  Use open_nested_file.
	(_bfd_look_for_bfd_in_cache): Copy no_export.
	* elflink.c (elf_link_add_object_symbols): Remove now unnecessary
	my_archive->no_export test.
	(elf_link_input_bfd): Drop existing lto_output STT_FILE syms.
	Don't use the file name when adding lto_output STT_FILE sym.
	* bfd-in2.h: Regenerate.
ld/
	* ldlang.h (struct lang_input_statement_flags): Add lto_output.
	* ldlang.c (lang_process): Don't set loading_lto_outputs.
	* ldfile.c (ldfile_try_open_bfd): Transfer entry flags.lto_output
	to bfd.
	* plugin.c (add_input_file, add_input_library): Set flags.lto_output.

diff --git a/bfd/archive.c b/bfd/archive.c
index cc4c52f..3899d84 100644
--- a/bfd/archive.c
+++ b/bfd/archive.c
@@ -311,8 +311,12 @@ _bfd_look_for_bfd_in_cache (bfd *arch_bfd, file_ptr filepos)
       struct ar_cache *entry = (struct ar_cache *) htab_find (hash_table, &m);
       if (!entry)
 	return NULL;
-      else
-	return entry->arbfd;
+
+      /* Unfortunately this flag is set after checking that we have
+	 an archive, and checking for an archive means one element has
+	 sneaked into the cache.  */
+      entry->arbfd->no_export = arch_bfd->no_export;
+      return entry->arbfd;
     }
   else
     return NULL;
@@ -375,10 +379,27 @@ _bfd_add_bfd_to_archive_cache (bfd *arch_bfd, file_ptr filepos, bfd *new_elt)
 }
 
 static bfd *
-_bfd_find_nested_archive (bfd *arch_bfd, const char *filename)
+open_nested_file (const char *filename, bfd *archive)
 {
-  bfd *abfd;
   const char *target;
+  bfd *n_bfd;
+
+  target = NULL;
+  if (!archive->target_defaulted)
+    target = archive->xvec->name;
+  n_bfd = bfd_openr (filename, target);
+  if (n_bfd != NULL)
+    {
+      n_bfd->lto_output = archive->lto_output;
+      n_bfd->no_export = archive->no_export;
+    }
+  return n_bfd;
+}
+
+static bfd *
+find_nested_archive (const char *filename, bfd *arch_bfd)
+{
+  bfd *abfd;
 
   /* PR 15140: Don't allow a nested archive pointing to itself.  */
   if (filename_cmp (filename, arch_bfd->filename) == 0)
@@ -394,10 +415,7 @@ _bfd_find_nested_archive (bfd *arch_bfd, const char *filename)
       if (filename_cmp (filename, abfd->filename) == 0)
 	return abfd;
     }
-  target = NULL;
-  if (!arch_bfd->target_defaulted)
-    target = arch_bfd->xvec->name;
-  abfd = bfd_openr (filename, target);
+  abfd = open_nested_file (filename, arch_bfd);
   if (abfd)
     {
       abfd->archive_next = arch_bfd->nested_archives;
@@ -626,12 +644,12 @@ bfd *
 _bfd_get_elt_at_filepos (bfd *archive, file_ptr filepos)
 {
   struct areltdata *new_areldata;
-  bfd *n_nfd;
+  bfd *n_bfd;
   char *filename;
 
-  n_nfd = _bfd_look_for_bfd_in_cache (archive, filepos);
-  if (n_nfd)
-    return n_nfd;
+  n_bfd = _bfd_look_for_bfd_in_cache (archive, filepos);
+  if (n_bfd)
+    return n_bfd;
 
   if (0 > bfd_seek (archive, filepos, SEEK_SET))
     return NULL;
@@ -643,8 +661,6 @@ _bfd_get_elt_at_filepos (bfd *archive, file_ptr filepos)
 
   if (bfd_is_thin_archive (archive))
     {
-      const char *target;
-
       /* This is a proxy entry for an external file.  */
       if (! IS_ABSOLUTE_PATH (filename))
 	{
@@ -660,7 +676,7 @@ _bfd_get_elt_at_filepos (bfd *archive, file_ptr filepos)
 	{
 	  /* This proxy entry refers to an element of a nested archive.
 	     Locate the member of that archive and return a bfd for it.  */
-	  bfd *ext_arch = _bfd_find_nested_archive (archive, filename);
+	  bfd *ext_arch = find_nested_archive (filename, archive);
 
 	  if (ext_arch == NULL
 	      || ! bfd_check_format (ext_arch, bfd_archive))
@@ -668,57 +684,55 @@ _bfd_get_elt_at_filepos (bfd *archive, file_ptr filepos)
 	      free (new_areldata);
 	      return NULL;
 	    }
-	  n_nfd = _bfd_get_elt_at_filepos (ext_arch, new_areldata->origin);
-	  if (n_nfd == NULL)
+	  n_bfd = _bfd_get_elt_at_filepos (ext_arch, new_areldata->origin);
+	  if (n_bfd == NULL)
 	    {
 	      free (new_areldata);
 	      return NULL;
 	    }
-	  n_nfd->proxy_origin = bfd_tell (archive);
-	  return n_nfd;
+	  n_bfd->proxy_origin = bfd_tell (archive);
+	  return n_bfd;
 	}
+
       /* It's not an element of a nested archive;
 	 open the external file as a bfd.  */
-      target = NULL;
-      if (!archive->target_defaulted)
-	target = archive->xvec->name;
-      n_nfd = bfd_openr (filename, target);
-      if (n_nfd == NULL)
+      n_bfd = open_nested_file (filename, archive);
+      if (n_bfd == NULL)
 	bfd_set_error (bfd_error_malformed_archive);
     }
   else
     {
-      n_nfd = _bfd_create_empty_archive_element_shell (archive);
+      n_bfd = _bfd_create_empty_archive_element_shell (archive);
     }
 
-  if (n_nfd == NULL)
+  if (n_bfd == NULL)
     {
       free (new_areldata);
       return NULL;
     }
 
-  n_nfd->proxy_origin = bfd_tell (archive);
+  n_bfd->proxy_origin = bfd_tell (archive);
 
   if (bfd_is_thin_archive (archive))
     {
-      n_nfd->origin = 0;
+      n_bfd->origin = 0;
     }
   else
     {
-      n_nfd->origin = n_nfd->proxy_origin;
-      n_nfd->filename = xstrdup (filename);
+      n_bfd->origin = n_bfd->proxy_origin;
+      n_bfd->filename = xstrdup (filename);
     }
 
-  n_nfd->arelt_data = new_areldata;
+  n_bfd->arelt_data = new_areldata;
 
   /* Copy BFD_COMPRESS and BFD_DECOMPRESS flags.  */
-  n_nfd->flags |= archive->flags & (BFD_COMPRESS | BFD_DECOMPRESS);
+  n_bfd->flags |= archive->flags & (BFD_COMPRESS | BFD_DECOMPRESS);
 
-  if (_bfd_add_bfd_to_archive_cache (archive, filepos, n_nfd))
-    return n_nfd;
+  if (_bfd_add_bfd_to_archive_cache (archive, filepos, n_bfd))
+    return n_bfd;
 
   free (new_areldata);
-  n_nfd->arelt_data = NULL;
+  n_bfd->arelt_data = NULL;
   return NULL;
 }
 
diff --git a/bfd/bfd-in2.h b/bfd/bfd-in2.h
index 678eaed..d02fc02 100644
--- a/bfd/bfd-in2.h
+++ b/bfd/bfd-in2.h
@@ -6438,6 +6438,9 @@ struct bfd
   /* If this is an input for a compiler plug-in library.  */
   ENUM_BITFIELD (bfd_plugin_format) plugin_format : 2;
 
+  /* Set if this is a plugin output file.  */
+  unsigned int lto_output : 1;
+
   /* Set to dummy BFD created when claimed by a compiler plug-in
      library.  */
   bfd *plugin_dummy_bfd;
diff --git a/bfd/bfd.c b/bfd/bfd.c
index 69c6bde..5ae5eca 100644
--- a/bfd/bfd.c
+++ b/bfd/bfd.c
@@ -210,6 +210,9 @@ CODE_FRAGMENT
 .  {* If this is an input for a compiler plug-in library.  *}
 .  ENUM_BITFIELD (bfd_plugin_format) plugin_format : 2;
 .
+.  {* Set if this is a plugin output file.  *}
+.  unsigned int lto_output : 1;
+.
 .  {* Set to dummy BFD created when claimed by a compiler plug-in
 .     library.  *}
 .  bfd *plugin_dummy_bfd;
diff --git a/bfd/elflink.c b/bfd/elflink.c
index 16d9f20..bf28885 100644
--- a/bfd/elflink.c
+++ b/bfd/elflink.c
@@ -4085,8 +4085,7 @@ error_free_dyn:
 	     requested we not re-export it, then mark it as hidden.  */
 	  if (definition
 	      && !dynamic
-	      && (abfd->no_export
-		  || (abfd->my_archive && abfd->my_archive->no_export))
+	      && abfd->no_export
 	      && ELF_ST_VISIBILITY (isym->st_other) != STV_INTERNAL)
 	    isym->st_other = (STV_HIDDEN
 			      | (isym->st_other & ~ELF_ST_VISIBILITY (-1)));
@@ -9520,6 +9519,10 @@ elf_link_input_bfd (struct elf_final_link_info *flinfo, bfd *input_bfd)
 
       if (ELF_ST_TYPE (isym->st_info) == STT_FILE)
 	{
+	  if (input_bfd->lto_output)
+	    /* -flto puts a temp file name here.  This means builds
+	       are not reproducible.  Discard the symbol.  */
+	    continue;
 	  have_file_sym = TRUE;
 	  flinfo->filesym_count += 1;
 	}
@@ -9536,8 +9539,10 @@ elf_link_input_bfd (struct elf_final_link_info *flinfo, bfd *input_bfd)
 	  memset (&osym, 0, sizeof (osym));
 	  osym.st_info = ELF_ST_INFO (STB_LOCAL, STT_FILE);
 	  osym.st_shndx = SHN_ABS;
-	  if (!elf_link_output_sym (flinfo, input_bfd->filename, &osym,
-				    bfd_abs_section_ptr, NULL))
+	  if (!elf_link_output_sym (flinfo,
+				    (input_bfd->lto_output ? NULL
+				     : input_bfd->filename),
+				    &osym, bfd_abs_section_ptr, NULL))
 	    return FALSE;
 	}
 
diff --git a/bfd/linker.c b/bfd/linker.c
index c1a2e82..dec6d1d 100644
--- a/bfd/linker.c
+++ b/bfd/linker.c
@@ -2896,7 +2896,7 @@ _bfd_handle_already_linked (asection *sec,
 	 files over IR because the first pass may contain a
 	 mix of LTO and normal objects and we must keep the
 	 first match, be it IR or real.  */
-      if (info->loading_lto_outputs
+      if (sec->owner->lto_output
 	  && (l->sec->owner->flags & BFD_PLUGIN) != 0)
 	{
 	  l->sec = sec;
diff --git a/bfd/opncls.c b/bfd/opncls.c
index 1320ee1..f0f2e64 100644
--- a/bfd/opncls.c
+++ b/bfd/opncls.c
@@ -109,6 +109,8 @@ _bfd_new_bfd_contained_in (bfd *obfd)
   nbfd->my_archive = obfd;
   nbfd->direction = read_direction;
   nbfd->target_defaulted = obfd->target_defaulted;
+  nbfd->lto_output = obfd->lto_output;
+  nbfd->no_export = obfd->no_export;
   return nbfd;
 }
 
diff --git a/include/bfdlink.h b/include/bfdlink.h
index add4971..6a02a3c 100644
--- a/include/bfdlink.h
+++ b/include/bfdlink.h
@@ -301,9 +301,6 @@ struct bfd_link_info
   /* TRUE if the LTO plugin is active.  */
   unsigned int lto_plugin_active: 1;
 
-  /* TRUE if we are loading LTO outputs.  */
-  unsigned int loading_lto_outputs: 1;
-
   /* TRUE if global symbols in discarded sections should be stripped.  */
   unsigned int strip_discarded: 1;
 
diff --git a/ld/ldfile.c b/ld/ldfile.c
index e19f55e..21bdbf4 100644
--- a/ld/ldfile.c
+++ b/ld/ldfile.c
@@ -142,6 +142,11 @@ ldfile_try_open_bfd (const char *attempt,
   /* Linker needs to decompress sections.  */
   entry->the_bfd->flags |= BFD_DECOMPRESS;
 
+#ifdef ENABLE_PLUGINS
+  if (entry->flags.lto_output)
+    entry->the_bfd->lto_output = 1;
+#endif
+
   /* If we are searching for this file, see if the architecture is
      compatible with the output file.  If it isn't, keep searching.
      If we can't open the file as an object file, stop the search
diff --git a/ld/ldlang.c b/ld/ldlang.c
index 5c4581f..ce333b4 100644
--- a/ld/ldlang.c
+++ b/ld/ldlang.c
@@ -6643,7 +6643,6 @@ lang_process (void)
 	einfo (_("%P%F: %s: plugin reported error after all symbols read\n"),
 	       plugin_error_plugin ());
       /* Open any newly added files, updating the file chains.  */
-      link_info.loading_lto_outputs = TRUE;
       open_input_bfds (*added.tail, OPEN_BFD_NORMAL);
       /* Restore the global list pointer now they have all been added.  */
       lang_list_remove_tail (stat_ptr, &added);
diff --git a/ld/ldlang.h b/ld/ldlang.h
index 69d21a7..7cb7610 100644
--- a/ld/ldlang.h
+++ b/ld/ldlang.h
@@ -283,6 +283,9 @@ struct lang_input_statement_flags
 
   /* Set if the file was claimed from an archive.  */
   unsigned int claim_archive : 1;
+
+  /* Set if added by the lto plugin add_input_file callback.  */
+  unsigned int lto_output : 1;
 #endif /* ENABLE_PLUGINS */
 
   /* Head of list of pushed flags.  */
diff --git a/ld/plugin.c b/ld/plugin.c
index 4fee305..31527ae 100644
--- a/ld/plugin.c
+++ b/ld/plugin.c
@@ -772,10 +772,14 @@ get_symbols_v2 (const void *handle, int nsyms, struct ld_plugin_symbol *syms)
 static enum ld_plugin_status
 add_input_file (const char *pathname)
 {
+  lang_input_statement_type *is;
+
   ASSERT (called_plugin);
-  if (!lang_add_input_file (xstrdup (pathname), lang_input_file_is_file_enum,
-			    NULL))
+  is = lang_add_input_file (xstrdup (pathname), lang_input_file_is_file_enum,
+			    NULL);
+  if (!is)
     return LDPS_ERR;
+  is->flags.lto_output = 1;
   return LDPS_OK;
 }
 
@@ -783,10 +787,14 @@ add_input_file (const char *pathname)
 static enum ld_plugin_status
 add_input_library (const char *pathname)
 {
+  lang_input_statement_type *is;
+
   ASSERT (called_plugin);
-  if (!lang_add_input_file (xstrdup (pathname), lang_input_file_is_l_enum,
-			    NULL))
+  is = lang_add_input_file (xstrdup (pathname), lang_input_file_is_l_enum,
+			    NULL);
+  if (!is)
     return LDPS_ERR;
+  is->flags.lto_output = 1;
   return LDPS_OK;
 }
 

-- 
Alan Modra
Australia Development Lab, IBM