* elf/dl-runtime.c (reloc_offset): Define.

	(reloc_index): Define.
	(_dl_fixup): Rename reloc_offset parameter to reloc_arg.
	(_dl_fixup_profile): Likewise.  Use reloc_index instead of
	computing index from reloc_offset.
	(_dl_call_pltexit): Likewise.
	* sysdeps/x86_64/dl-trampoline.S (_dl_runtime_resolve): Just pass
	the relocation index to _dl_fixup.
	(_dl_runtime_profile): Likewise for _dl_fixup_profile and
	_dl_call_pltexit.
	* sysdeps/x86_64/dl-runtime.c: New file.

ChangeLog

@@ -1,5 +1,17 @@
 2009-03-14  Ulrich Drepper  <drepper@redhat.com>
 
+	* elf/dl-runtime.c (reloc_offset): Define.
+	(reloc_index): Define.
+	(_dl_fixup): Rename reloc_offset parameter to reloc_arg.
+	(_dl_fixup_profile): Likewise.  Use reloc_index instead of
+	computing index from reloc_offset.
+	(_dl_call_pltexit): Likewise.
+	* sysdeps/x86_64/dl-trampoline.S (_dl_runtime_resolve): Just pass
+	the relocation index to _dl_fixup.
+	(_dl_runtime_profile): Likewise for _dl_fixup_profile and
+	_dl_call_pltexit.
+	* sysdeps/x86_64/dl-runtime.c: New file.
+
 	[BZ #9893]
 	* sysdeps/x86_64/dl-trampoline.S (_dl_runtime_profile): Fix
 	alignement of La_x86_64_regs.  Store xmm parameters.

elf/dl-runtime.c

@@ -1,5 +1,5 @@
 /* On-demand PLT fixup for shared objects.
-   Copyright (C) 1995-2006, 2007, 2008 Free Software Foundation, Inc.
+   Copyright (C) 1995-2006, 2007, 2008, 2009 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
 
    The GNU C Library is free software; you can redistribute it and/or
@@ -46,6 +46,12 @@
 # define ARCH_FIXUP_ATTRIBUTE
 #endif
 
+#ifndef reloc_offset
+# define reloc_offset reloc_arg
+# define reloc_index  reloc_arg / sizeof (PLTREL)
+#endif
+
+
 
 /* This function is called through a special trampoline from the PLT the
    first time each PLT entry is called.  We must perform the relocation
@@ -63,7 +69,7 @@ _dl_fixup (
 # endif
 	   /* GKM FIXME: Fix trampoline to pass bounds so we can do
 	      without the `__unbounded' qualifier.  */
-	   struct link_map *__unbounded l, ElfW(Word) reloc_offset)
+	   struct link_map *__unbounded l, ElfW(Word) reloc_arg)
 {
   const ElfW(Sym) *const symtab
     = (const void *) D_PTR (l, l_info[DT_SYMTAB]);
@@ -142,22 +148,20 @@ _dl_fixup (
 #endif
 
 #if !defined PROF && !defined ELF_MACHINE_NO_PLT && !__BOUNDED_POINTERS__
-
 DL_FIXUP_VALUE_TYPE
 __attribute ((noinline)) ARCH_FIXUP_ATTRIBUTE
 _dl_profile_fixup (
 #ifdef ELF_MACHINE_RUNTIME_FIXUP_ARGS
 		   ELF_MACHINE_RUNTIME_FIXUP_ARGS,
 #endif
-		   struct link_map *l, ElfW(Word) reloc_offset,
+		   struct link_map *l, ElfW(Word) reloc_arg,
 		   ElfW(Addr) retaddr, void *regs, long int *framesizep)
 {
   void (*mcount_fct) (ElfW(Addr), ElfW(Addr)) = INTUSE(_dl_mcount);
 
   /* This is the address in the array where we store the result of previous
      relocations.  */
-  struct reloc_result *reloc_result
-    = &l->l_reloc_result[reloc_offset / sizeof (PLTREL)];
+  struct reloc_result *reloc_result = &l->l_reloc_result[reloc_index];
   DL_FIXUP_VALUE_TYPE *resultp = &reloc_result->addr;
 
   DL_FIXUP_VALUE_TYPE value = *resultp;
@@ -415,16 +419,15 @@ _dl_profile_fixup (
 #include <stdio.h>
 void
 ARCH_FIXUP_ATTRIBUTE
-_dl_call_pltexit (struct link_map *l, ElfW(Word) reloc_offset,
+_dl_call_pltexit (struct link_map *l, ElfW(Word) reloc_arg,
 		  const void *inregs, void *outregs)
 {
 #ifdef SHARED
   /* This is the address in the array where we store the result of previous
      relocations.  */
   // XXX Maybe the bound information must be stored on the stack since
   // XXX with bind_not a new value could have been stored in the meantime.
-  struct reloc_result *reloc_result
-    = &l->l_reloc_result[reloc_offset / sizeof (PLTREL)];
+  struct reloc_result *reloc_result = &l->l_reloc_result[reloc_index];
   ElfW(Sym) *defsym = ((ElfW(Sym) *) D_PTR (reloc_result->bound,
 					    l_info[DT_SYMTAB])
 		       + reloc_result->boundndx);

sysdeps/x86_64/dl-runtime.c

@@ -0,0 +1,9 @@
+/* The ABI calls for the PLT stubs to pass the index of the relocation
+   and not its offset.  In _dl_profile_fixup and _dl_call_pltexit we
+   also use the index.  Therefore it is wasteful to compute the offset
+   in the trampoline just to reverse the operation immediately
+   afterwards.  */
+#define reloc_offset reloc_arg * sizeof (PLTREL)
+#define reloc_index  reloc_arg
+
+#include <elf/dl-runtime.c>

sysdeps/x86_64/dl-trampoline.S

@@ -35,10 +35,6 @@ _dl_runtime_resolve:
 	movq %r8, 40(%rsp)
 	movq %r9, 48(%rsp)
 	movq 64(%rsp), %rsi	# Copy args pushed by PLT in register.
-	movq %rsi, %r11		# Multiply by 24
-	addq %r11, %rsi
-	addq %r11, %rsi
-	shlq $3, %rsi
 	movq 56(%rsp), %rdi	# %rdi: link_map, %rsi: reloc_offset
 	call _dl_fixup		# Call resolver.
 	movq %rax, %r11		# Save return value
@@ -116,10 +112,6 @@ _dl_runtime_profile:
 	movq %rsp, %rcx		# La_x86_64_regs pointer to %rcx.
 	movq 48(%rbx), %rdx	# Load return address if needed.
 	movq 40(%rbx), %rsi	# Copy args pushed by PLT in register.
-	movq %rsi,%r11		# Multiply by 24.
-	addq %r11,%rsi
-	addq %r11,%rsi
-	shlq $3, %rsi
 	movq 32(%rbx), %rdi	# %rdi: link_map, %rsi: reloc_offset
 	leaq 16(%rbx), %r8
 	call _dl_profile_fixup	# Call resolver.
@@ -204,10 +196,6 @@ _dl_runtime_profile:
 
 	movq 24(%rbx), %rdx	# La_x86_64_regs argument to %rdx.
 	movq 40(%rbx), %rsi	# Copy args pushed by PLT in register.
-	movq %rsi,%r11		# Multiply by 24.
-	addq %r11,%rsi
-	addq %r11,%rsi
-	shlq $3, %rsi
         movq 32(%rbx), %rdi	# %rdi: link_map, %rsi: reloc_offset
 	call _dl_call_pltexit