public inbox for binutils@sourceware.org
 help / color / mirror / Atom feed
From: "H. J. Lu" <hjl@lucon.org>
To: binutils@sources.redhat.com
Cc: matz@suse.de
Subject: Re: PATCH: Cache the result of _bfd_elf_check_kept_section
Date: Tue, 25 Apr 2006 16:11:00 -0000	[thread overview]
Message-ID: <20060425022744.GA24749@lucon.org> (raw)
In-Reply-To: <20060424231056.GA23563@lucon.org>

On Mon, Apr 24, 2006 at 04:10:56PM -0700, H. J. Lu wrote:
> On Mon, Apr 24, 2006 at 10:52:53AM -0700, H. J. Lu wrote:
> > This patch addresses the link speed issue by caching the result of
> > _bfd_elf_check_kept_section.
> >  
> 
> Apparently, I didn't include the correct patch. Here is the right one.
> 

Here is a patch without adding a new field to ELF section.


H.J.
----
2006-04-24  Michael Matz  <matz@suse.de>
	    H.J. Lu  <hongjiu.lu@intel.com>

	* elf.c (_bfd_elf_new_section_hook): Initialize kept_section to
	(asection *) -1L.
	(assign_section_numbers): Check kept_section before calling
	_bfd_elf_check_kept_section.
	* elflink.c (elf_link_input_bfd): Likewise.

	* elflink.c (elf_link_add_object_symbols): Also check if
	kept_section isn't (asection *) -1L before using it.
	(elf_link_input_bfd): Likewise.
	(match_group_member): Correctly iterate group members.
	(_bfd_elf_check_kept_section): Cache the result in
	elf_kept_section.

--- bfd/elf.c.kept	2006-04-24 19:09:01.000000000 -0700
+++ bfd/elf.c	2006-04-24 19:16:14.000000000 -0700
@@ -2481,6 +2481,8 @@ _bfd_elf_new_section_hook (bfd *abfd, as
       if (sdata == NULL)
 	return FALSE;
       sec->used_by_bfd = sdata;
+      /* Initialized it with -1L so that it can be cached.  */
+      sec->kept_section = (asection *) -1L;
     }
 
   /* Indicate whether or not this section should use RELA relocations.  */
@@ -3131,7 +3133,11 @@ assign_section_numbers (bfd *abfd, struc
 			 s, s->owner);
 		      /* Point to the kept section if it has the same
 			 size as the discarded one.  */
-		      kept = _bfd_elf_check_kept_section (s);
+		      if (s->kept_section != NULL
+			  && s->kept_section != (asection *) -1L)
+			kept = _bfd_elf_check_kept_section (s);
+		      else
+			kept = NULL;
 		      if (kept == NULL)
 			{
 			  bfd_set_error (bfd_error_bad_value);
--- bfd/elflink.c.kept	2006-04-24 15:36:18.000000000 -0700
+++ bfd/elflink.c	2006-04-24 19:22:22.000000000 -0700
@@ -3587,7 +3587,8 @@ elf_link_add_object_symbols (bfd *abfd, 
 	  sec = bfd_section_from_elf_index (abfd, isym->st_shndx);
 	  if (sec == NULL)
 	    sec = bfd_abs_section_ptr;
-	  else if (sec->kept_section)
+	  else if (sec->kept_section
+		   && sec->kept_section != (asection *) -1L)
 	    {
 	      /* Symbols from discarded section are undefined, and have
 		 default visibility.  */
@@ -6782,6 +6783,7 @@ match_group_member (asection *sec, asect
       if (bfd_elf_match_symbols_in_sections (s, sec))
 	return s;
 
+      s = elf_next_in_group (s);
       if (s == first)
 	break;
     }
@@ -6806,6 +6808,13 @@ _bfd_elf_check_kept_section (asection *s
       if (kept != NULL && sec->size != kept->size)
 	kept = NULL;
     }
+
+  /* kept_section is initialized to -1L.  */
+  if (kept == (asection *) -1L)
+    abort ();
+
+  sec->kept_section = kept;
+
   return kept;
 }
 
@@ -7157,7 +7166,9 @@ elf_link_input_bfd (struct elf_final_lin
 			 is that we warn in non-debug sections, and
 			 debug sections tend to come after other
 			 sections.  */
-		      if (action & PRETEND)
+		      if ((action & PRETEND)
+			  && sec->kept_section
+			  && sec->kept_section != (asection *) -1L)
 			{
 			  asection *kept;
 
@@ -7334,7 +7345,8 @@ elf_link_input_bfd (struct elf_final_lin
 			     case of discarded link-once and discarded
 			     SEC_MERGE sections, use the kept section.  */
 			  if (bfd_is_abs_section (osec)
-			      && sec->kept_section != NULL
+			      && sec->kept_section
+			      && sec->kept_section != (asection *) -1L 
 			      && sec->kept_section->output_section != NULL)
 			    {
 			      osec = sec->kept_section->output_section;

  reply	other threads:[~2006-04-25  2:27 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-04-24 20:12 PATCH: PR ld/2342: linkonce debug is broken H. J. Lu
2006-04-25 15:47 ` PATCH: Cache the result of _bfd_elf_check_kept_section H. J. Lu
2006-04-25 16:11   ` H. J. Lu [this message]
2006-04-25 16:37     ` H. J. Lu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20060425022744.GA24749@lucon.org \
    --to=hjl@lucon.org \
    --cc=binutils@sources.redhat.com \
    --cc=matz@suse.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).