* config.h.in (USE_MULTIARCH): Define.

	* configure.in: Handle --enable-multi-arch.
	* elf/dl-runtime.c (_dl_fixup): Handle STT_GNU_IFUNC.
	(_dl_fixup_profile): Likewise.
	* elf/do-lookup.c (dl_lookup_x): Likewise.
	* sysdeps/x86_64/dl-machine.h: Handle STT_GNU_IFUNC.
	* elf/elf.h (STT_GNU_IFUNC): Define.
	* include/libc-symbols.h (libc_ifunc): Define.
	* sysdeps/x86_64/cacheinfo.c: If USE_MULTIARCH is defined, use the
	framework in init-arch.h to get CPUID values.
	* sysdeps/x86_64/multiarch/Makefile: New file.
	* sysdeps/x86_64/multiarch/init-arch.c: New file.
	* sysdeps/x86_64/multiarch/init-arch.h: New file.
	* sysdeps/x86_64/multiarch/sched_cpucount.c: New file.

	* config.make.in (experimental-malloc): Define.
	* configure.in: Handle --enable-experimental-malloc.
	* malloc/Makefile: Handle experimental-malloc flag.
	* malloc/malloc.c: Implement PER_THREAD and ATOMIC_FASTBINS features.
	* malloc/arena.c: Likewise.
	* malloc/hooks.c: Likewise.
	* malloc/malloc.h: Define M_ARENA_TEST and M_ARENA_MAX.

20 files changed, 813 insertions, 75 deletions

diff --git a/ChangeLog b/ChangeLog
index 933e81cb4f..954285ff0f 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,28 @@
+2009-03-13  Ulrich Drepper  <drepper@redhat.com>
+
+	* config.h.in (USE_MULTIARCH): Define.
+	* configure.in: Handle --enable-multi-arch.
+	* elf/dl-runtime.c (_dl_fixup): Handle STT_GNU_IFUNC.
+	(_dl_fixup_profile): Likewise.
+	* elf/do-lookup.c (dl_lookup_x): Likewise.
+	* sysdeps/x86_64/dl-machine.h: Handle STT_GNU_IFUNC.
+	* elf/elf.h (STT_GNU_IFUNC): Define.
+	* include/libc-symbols.h (libc_ifunc): Define.
+	* sysdeps/x86_64/cacheinfo.c: If USE_MULTIARCH is defined, use the
+	framework in init-arch.h to get CPUID values.
+	* sysdeps/x86_64/multiarch/Makefile: New file.
+	* sysdeps/x86_64/multiarch/init-arch.c: New file.
+	* sysdeps/x86_64/multiarch/init-arch.h: New file.
+	* sysdeps/x86_64/multiarch/sched_cpucount.c: New file.
+
+	* config.make.in (experimental-malloc): Define.
+	* configure.in: Handle --enable-experimental-malloc.
+	* malloc/Makefile: Handle experimental-malloc flag.
+	* malloc/malloc.c: Implement PER_THREAD and ATOMIC_FASTBINS features.
+	* malloc/arena.c: Likewise.
+	* malloc/hooks.c: Likewise.
+	* malloc/malloc.h: Define M_ARENA_TEST and M_ARENA_MAX.
+
 2009-03-11  Ulrich Drepper  <drepper@redhat.com>
 
 	* sysdeps/x86_64/dl-machine.h (elf_machine_rela): Add branch
diff --git a/NEWS b/NEWS
index 2689db304e..55ba54469e 100644
--- a/NEWS
+++ b/NEWS
@@ -1,4 +1,4 @@
-GNU C Library NEWS -- history of user-visible changes.  2009-3-8
+GNU C Library NEWS -- history of user-visible changes.  2009-3-12
 Copyright (C) 1992-2008, 2009 Free Software Foundation, Inc.
 See the end for copying conditions.
 
@@ -7,7 +7,7 @@ using `glibc' in the "product" field.
 
 Version 2.10
 
-* New Linux interface: accept4
+* New Linux interfaces: accept4, fallocate, fallocate64.
 
 * Correct declarations of string function when used in C++ code.  This
   could lead to compile error for invalid C++ code.
diff --git a/config.h.in b/config.h.in
index b5abb10187..8dbc224a7d 100644
--- a/config.h.in
+++ b/config.h.in
@@ -189,6 +189,9 @@
 /* Define if __stack_chk_guard canary should be randomized at program startup.  */
 #undef ENABLE_STACKGUARD_RANDOMIZE
 
+/* Define if multi-arch DSOs should be generated.  */
+#undef USE_MULTIARCH
+
 /*
  */
 
diff --git a/config.make.in b/config.make.in
index aa73466713..6da6362f24 100644
--- a/config.make.in
+++ b/config.make.in
@@ -70,6 +70,7 @@ versioning = @VERSIONING@
 oldest-abi = @oldest_abi@
 no-whole-archive = @no_whole_archive@
 exceptions = @exceptions@
+multi-arch = @multi_arch@
 
 mach-interface-list = @mach_interface_list@
 
@@ -78,6 +79,8 @@ have-ksh = @libc_cv_have_ksh@
 
 sizeof-long-double = @sizeof_long_double@
 
+experimental-malloc = @experimental_malloc@
+
 # Configuration options.
 build-static = @static@
 build-shared = @shared@
diff --git a/configure b/configure
index c5c6dc4789..69af8525be 100755
--- a/configure
+++ b/configure
@@ -660,6 +660,8 @@ oldest_abi
 bindnow
 force_install
 all_warnings
+multi_arch
+experimental_malloc
 build
 build_cpu
 build_vendor
@@ -1380,6 +1382,10 @@ Optional Features:
   --enable-kernel=VERSION compile for compatibility with kernel not older than
                           VERSION
   --enable-all-warnings   enable all useful warnings gcc can issue
+  --enable-multi-arch     enable single DSO with optimizations for multiple
+                          architectures
+  --enable-experimental-malloc
+                          enable experimental malloc features
 
 Optional Packages:
   --with-PACKAGE[=ARG]    use PACKAGE [ARG=yes]
@@ -2173,6 +2179,29 @@ fi
 
 
 
+# Check whether --enable-multi-arch was given.
+if test "${enable_multi_arch+set}" = set; then
+  enableval=$enable_multi_arch; multi_arch=$enableval
+else
+  multi_arch=no
+fi
+
+if test x"$multi_arch" = xyes; then
+  cat >>confdefs.h <<\_ACEOF
+#define USE_MULTIARCH 1
+_ACEOF
+
+  multi_arch_d=/multiarch
+fi
+
+
+# Check whether --enable-experimental-malloc was given.
+if test "${enable_experimental_malloc+set}" = set; then
+  enableval=$enable_experimental_malloc; experimental_malloc=$enableval
+fi
+
+
+
 # Make sure we can run config.sub.
 $SHELL "$ac_aux_dir/config.sub" sun4 >/dev/null 2>&1 ||
   { { echo "$as_me:$LINENO: error: cannot run $SHELL $ac_aux_dir/config.sub" >&5
@@ -2627,7 +2656,7 @@ for b in $base ''; do
       test "$v" = / && continue
       for o in /$ostry ''; do
 	test "$o" = / && continue
-	for m in $mach ''; do
+	for m in $multi_arch_d $mach ''; do
 	  for d in $add_ons_pfx ''; do
 	    for a in $add_ons_sfx ''; do
 	      if test -n "$m0$m0sub$b$v$o$m$msub"; then
@@ -5684,6 +5713,37 @@ _ACEOF
 
 fi
 
+# For the multi-arch option we need support in the assembler.
+if test "$multi_arch" = yes; then
+  if test "x$libc_cv_asm_type_prefix" != xno; then
+{ echo "$as_me:$LINENO: checking for assembler gnu_indirect_function symbol type support" >&5
+echo $ECHO_N "checking for assembler gnu_indirect_function symbol type support... $ECHO_C" >&6; }
+if test "${libc_cv_asm_gnu_indirect_function+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  cat > conftest.s <<EOF
+.type foo,%gnu_indirect_function
+EOF
+if ${CC-cc} -c $ASFLAGS conftest.s 1>&5 2>&5;
+then
+  libc_cv_asm_gnu_indirect_function=yes
+else
+  libc_cv_asm_gnu_indirect_function=no
+fi
+rm -f conftest*
+fi
+{ echo "$as_me:$LINENO: result: $libc_cv_asm_gnu_indirect_function" >&5
+echo "${ECHO_T}$libc_cv_asm_gnu_indirect_function" >&6; }
+  else
+    libc_cv_asm_gnu_indirect_function=no
+  fi
+  if test x"$libc_cv_asm_gnu_indirect_function" != xyes; then
+    { { echo "$as_me:$LINENO: error: --enable-multi-arch support requires assembler and linker support" >&5
+echo "$as_me: error: --enable-multi-arch support requires assembler and linker support" >&2;}
+   { (exit 1); exit 1; }; }
+  fi
+fi
+
 { echo "$as_me:$LINENO: checking for .symver assembler directive" >&5
 echo $ECHO_N "checking for .symver assembler directive... $ECHO_C" >&6; }
 if test "${libc_cv_asm_symver_directive+set}" = set; then
@@ -9184,6 +9244,8 @@ oldest_abi!$oldest_abi$ac_delim
 bindnow!$bindnow$ac_delim
 force_install!$force_install$ac_delim
 all_warnings!$all_warnings$ac_delim
+multi_arch!$multi_arch$ac_delim
+experimental_malloc!$experimental_malloc$ac_delim
 build!$build$ac_delim
 build_cpu!$build_cpu$ac_delim
 build_vendor!$build_vendor$ac_delim
@@ -9235,8 +9297,6 @@ libc_cv_have_bash2!$libc_cv_have_bash2$ac_delim
 KSH!$KSH$ac_delim
 libc_cv_have_ksh!$libc_cv_have_ksh$ac_delim
 AWK!$AWK$ac_delim
-PERL!$PERL$ac_delim
-INSTALL_INFO!$INSTALL_INFO$ac_delim
 _ACEOF
 
   if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 97; then
@@ -9278,6 +9338,8 @@ _ACEOF
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   cat >conf$$subs.sed <<_ACEOF
+PERL!$PERL$ac_delim
+INSTALL_INFO!$INSTALL_INFO$ac_delim
 BISON!$BISON$ac_delim
 VERSIONING!$VERSIONING$ac_delim
 libc_cv_cc_with_libunwind!$libc_cv_cc_with_libunwind$ac_delim
@@ -9334,7 +9396,7 @@ LIBOBJS!$LIBOBJS$ac_delim
 LTLIBOBJS!$LTLIBOBJS$ac_delim
 _ACEOF
 
-  if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 54; then
+  if test `sed -n "s/.*$ac_delim\$/X/p" conf$$subs.sed | grep -c X` = 56; then
     break
   elif $ac_last_try; then
     { { echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
diff --git a/configure.in b/configure.in
index d7b22f3a3b..4015722d5a 100644
--- a/configure.in
+++ b/configure.in
@@ -254,6 +254,24 @@ AC_ARG_ENABLE([all-warnings],
 	      [])
 AC_SUBST(all_warnings)
 
+AC_ARG_ENABLE([multi-arch],
+	      AC_HELP_STRING([--enable-multi-arch],
+			     [enable single DSO with optimizations for multiple architectures]),
+	      [multi_arch=$enableval],
+	      [multi_arch=no])
+if test x"$multi_arch" = xyes; then
+  AC_DEFINE(USE_MULTIARCH)
+  multi_arch_d=/multiarch
+fi
+AC_SUBST(multi_arch)
+
+AC_ARG_ENABLE([experimental-malloc],
+	      AC_HELP_STRING([--enable-experimental-malloc],
+			     [enable experimental malloc features]),
+	      [experimental_malloc=$enableval],
+	      [])
+AC_SUBST(experimental_malloc)
+
 AC_CANONICAL_HOST
 
 # The way shlib-versions is used to generate soversions.mk uses a
@@ -608,7 +626,7 @@ for b in $base ''; do
       test "$v" = / && continue
       for o in /$ostry ''; do
 	test "$o" = / && continue
-	for m in $mach ''; do
+	for m in $multi_arch_d $mach ''; do
 	  for d in $add_ons_pfx ''; do
 	    for a in $add_ons_sfx ''; do
 	      if test -n "$m0$m0sub$b$v$o$m$msub"; then
@@ -1157,6 +1175,29 @@ if test "x$libc_cv_asm_type_prefix" != xno; then
   AC_DEFINE_UNQUOTED(ASM_TYPE_DIRECTIVE_PREFIX, ${libc_cv_asm_type_prefix})
 fi
 
+# For the multi-arch option we need support in the assembler.
+if test "$multi_arch" = yes; then
+  if test "x$libc_cv_asm_type_prefix" != xno; then
+AC_CACHE_CHECK([for assembler gnu_indirect_function symbol type support],
+               libc_cv_asm_gnu_indirect_function, [dnl
+cat > conftest.s <<EOF
+.type foo,%gnu_indirect_function
+EOF
+if ${CC-cc} -c $ASFLAGS conftest.s 1>&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD;
+then
+  libc_cv_asm_gnu_indirect_function=yes
+else
+  libc_cv_asm_gnu_indirect_function=no
+fi
+rm -f conftest*])
+  else
+    libc_cv_asm_gnu_indirect_function=no
+  fi
+  if test x"$libc_cv_asm_gnu_indirect_function" != xyes; then
+    AC_MSG_ERROR([--enable-multi-arch support requires assembler and linker support])
+  fi
+fi
+
 AC_CACHE_CHECK(for .symver assembler directive, libc_cv_asm_symver_directive,
 [cat > conftest.s <<EOF
 ${libc_cv_dot_text}
diff --git a/elf/dl-runtime.c b/elf/dl-runtime.c
index 968e293409..962f47de64 100644
--- a/elf/dl-runtime.c
+++ b/elf/dl-runtime.c
@@ -1,5 +1,5 @@
 /* On-demand PLT fixup for shared objects.
-   Copyright (C) 1995-2006, 2007 Free Software Foundation, Inc.
+   Copyright (C) 1995-2006, 2007, 2008 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
@@ -130,6 +130,9 @@ _dl_fixup (
   /* And now perhaps the relocation addend.  */
   value = elf_machine_plt_value (l, reloc, value);
 
+  if (__builtin_expect (ELFW(ST_TYPE) (sym->st_info) == STT_GNU_IFUNC, 0))
+    value = ((DL_FIXUP_VALUE_TYPE (*) (void)) value) ();
+
   /* Finally, fix up the plt itself.  */
   if (__builtin_expect (GLRO(dl_bind_not), 0))
     return value;
@@ -215,12 +218,21 @@ _dl_profile_fixup (
 				       defsym != NULL
 				       ? LOOKUP_VALUE_ADDRESS (result)
 					 + defsym->st_value : 0);
+
+	  if (__builtin_expect (ELFW(ST_TYPE) (defsym->st_info)
+				== STT_GNU_IFUNC, 0))
+	    value = ((DL_FIXUP_VALUE_TYPE (*) (void)) value) ();
 	}
       else
 	{
 	  /* We already found the symbol.  The module (and therefore its load
 	     address) is also known.  */
 	  value = DL_FIXUP_MAKE_VALUE (l, l->l_addr + refsym->st_value);
+
+	  if (__builtin_expect (ELFW(ST_TYPE) (refsym->st_info)
+				== STT_GNU_IFUNC, 0))
+	    value = ((DL_FIXUP_VALUE_TYPE (*) (void)) value) ();
+
 	  result = l;
 	}
       /* And now perhaps the relocation addend.  */
diff --git a/elf/do-lookup.h b/elf/do-lookup.h
index ebb9ed5b47..41e5fc137c 100644
--- a/elf/do-lookup.h
+++ b/elf/do-lookup.h
@@ -1,5 +1,5 @@
 /* Look up a symbol in the loaded objects.
-   Copyright (C) 1995-2004, 2005, 2006, 2007 Free Software Foundation, Inc.
+   Copyright (C) 1995-2007, 2008 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
@@ -88,10 +88,12 @@ do_lookup_x (const char *undef_name, uint_fast32_t new_hash,
 
 	if (__builtin_expect (ELFW(ST_TYPE) (sym->st_info) > STT_FUNC
 			      && ELFW(ST_TYPE) (sym->st_info) != STT_COMMON
-			      && ELFW(ST_TYPE) (sym->st_info) != STT_TLS, 0))
-	  /* Ignore all but STT_NOTYPE, STT_OBJECT, STT_FUNC, and STT_COMMON
-	     entries (and STT_TLS if TLS is supported) since these
-	     are no code/data definitions.  */
+			      && ELFW(ST_TYPE) (sym->st_info) != STT_TLS
+			      && ELFW(ST_TYPE) (sym->st_info) != STT_GNU_IFUNC,
+			      0))
+	  /* Ignore all but STT_NOTYPE, STT_OBJECT, STT_FUNC, STT_COMMON,
+	     STT_TLS, and STT_GNU_IFUNC since these are no code/data
+	     definitions.  */
 	  return NULL;
 
 	if (sym != ref && strcmp (strtab + sym->st_name, undef_name))
diff --git a/elf/elf.h b/elf/elf.h
index 2792820c35..cd74d510ee 100644
--- a/elf/elf.h
+++ b/elf/elf.h
@@ -459,6 +459,7 @@ typedef struct
 #define STT_TLS		6		/* Symbol is thread-local data object*/
 #define	STT_NUM		7		/* Number of defined types.  */
 #define STT_LOOS	10		/* Start of OS-specific */
+#define STT_GNU_IFUNC	10		/* Symbol is indirect code object */
 #define STT_HIOS	12		/* End of OS-specific */
 #define STT_LOPROC	13		/* Start of processor-specific */
 #define STT_HIPROC	15		/* End of processor-specific */
diff --git a/include/libc-symbols.h b/include/libc-symbols.h
index a2faeafb32..d53bcb9b77 100644
--- a/include/libc-symbols.h
+++ b/include/libc-symbols.h
@@ -1,7 +1,6 @@
 /* Support macros for making weak and strong aliases for symbols,
    and for using symbol sets and linker warnings with GNU ld.
-   Copyright (C) 1995-1998,2000-2003,2004,2005,2006
-	Free Software Foundation, Inc.
+   Copyright (C) 1995-1998, 2000-2006, 2008 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
@@ -835,4 +834,15 @@ for linking")
 # define compat_data_section .section ".data.compat", "aw";
 #endif
 
+/* Marker used for indirection function symbols.  */
+#define libc_ifunc(name, expr)						\
+  extern void *name##_ifunc (void) __asm__ (#name);			\
+  void *name##_ifunc (void)						\
+  {									\
+    INIT_ARCH ();							\
+    __typeof (name) *res = expr;					\
+    return res;								\
+  }									\
+  __asm__ (".type " #name ", %gnu_indirect_function");
+
 #endif /* libc-symbols.h */
diff --git a/malloc/Makefile b/malloc/Makefile
index 22b14eac77..1099335fff 100644
--- a/malloc/Makefile
+++ b/malloc/Makefile
@@ -1,4 +1,4 @@
-# Copyright (C) 1991-1999, 2000, 2001, 2002, 2003, 2005, 2006, 2007
+# Copyright (C) 1991-1999, 2000, 2001, 2002, 2003, 2005, 2006, 2007, 2009
 # Free Software Foundation, Inc.
 # This file is part of the GNU C Library.
 
@@ -124,6 +124,9 @@ endif
 
 tst-mcheck-ENV = MALLOC_CHECK_=3
 
+ifeq ($(experimental-malloc),yes)
+CPPFLAGS-malloc.c += -DPER_THREAD -DATOMIC_FASTBINS
+endif
 # Uncomment this for test releases.  For public releases it is too expensive.
 #CPPFLAGS-malloc.o += -DMALLOC_DEBUG=1
 
diff --git a/malloc/arena.c b/malloc/arena.c
index cc03dc4a5b..f280d38811 100644
--- a/malloc/arena.c
+++ b/malloc/arena.c
@@ -1,5 +1,5 @@
 /* Malloc implementation for multiple threads without lock contention.
-   Copyright (C) 2001,2002,2003,2004,2005,2006,2007
+   Copyright (C) 2001,2002,2003,2004,2005,2006,2007,2009
    Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Contributed by Wolfram Gloger <wg@malloc.de>, 2001.
@@ -78,6 +78,10 @@ extern int sanity_check_heap_info_alignment[(sizeof (heap_info)
 
 static tsd_key_t arena_key;
 static mutex_t list_lock;
+#ifdef PER_THREAD
+static size_t narenas;
+static mstate free_list;
+#endif
 
 #if THREAD_STATS
 static int stat_n_heaps;
@@ -105,13 +109,30 @@ int __malloc_initialized = -1;
    in the new arena. */
 
 #define arena_get(ptr, size) do { \
+  arena_lookup(ptr); \
+  arena_lock(ptr, size); \
+} while(0)
+
+#define arena_lookup(ptr) do { \
   Void_t *vptr = NULL; \
   ptr = (mstate)tsd_getspecific(arena_key, vptr); \
+} while(0)
+
+#ifdef PER_THREAD
+#define arena_lock(ptr, size) do { \
+  if(ptr) \
+    (void)mutex_lock(&ptr->mutex); \
+  else \
+    ptr = arena_get2(ptr, (size)); \
+} while(0)
+#else
+#define arena_lock(ptr, size) do { \
   if(ptr && !mutex_trylock(&ptr->mutex)) { \
     THREAD_STAT(++(ptr->stat_lock_direct)); \
   } else \
     ptr = arena_get2(ptr, (size)); \
 } while(0)
+#endif
 
 /* find the heap and corresponding arena for a given ptr */
 
@@ -219,6 +240,11 @@ free_atfork(Void_t* mem, const Void_t *caller)
   }
 #endif
 
+#ifdef ATOMIC_FASTBINS
+  ar_ptr = arena_for_chunk(p);
+  tsd_getspecific(arena_key, vptr);
+  _int_free(ar_ptr, p, vptr == ATFORK_ARENA_PTR);
+#else
   ar_ptr = arena_for_chunk(p);
   tsd_getspecific(arena_key, vptr);
   if(vptr != ATFORK_ARENA_PTR)
@@ -226,6 +252,7 @@ free_atfork(Void_t* mem, const Void_t *caller)
   _int_free(ar_ptr, p);
   if(vptr != ATFORK_ARENA_PTR)
     (void)mutex_unlock(&ar_ptr->mutex);
+#endif
 }
 
 
@@ -312,8 +339,17 @@ ptmalloc_unlock_all2 (void)
   __malloc_hook = save_malloc_hook;
   __free_hook = save_free_hook;
 #endif
+#ifdef PER_THREAD
+  free_list = NULL;
+#endif
   for(ar_ptr = &main_arena;;) {
     mutex_init(&ar_ptr->mutex);
+#ifdef PER_THREAD
+    if (ar_ptr != save_arena) {
+      ar_ptr->next_free = free_list;
+      free_list = ar_ptr;
+    }
+#endif
     ar_ptr = ar_ptr->next;
     if(ar_ptr == &main_arena) break;
   }
@@ -377,6 +413,11 @@ ptmalloc_init_minimal (void)
   mp_.mmap_threshold = DEFAULT_MMAP_THRESHOLD;
   mp_.trim_threshold = DEFAULT_TRIM_THRESHOLD;
   mp_.pagesize       = malloc_getpagesize;
+#ifdef PER_THREAD
+# define NARENAS_FROM_NCORES(n) ((n) * (sizeof(long) == 4 ? 2 : 8))
+  mp_.arena_test     = NARENAS_FROM_NCORES (1);
+  narenas = 1;
+#endif
 }
 
 
@@ -529,9 +570,25 @@ ptmalloc_init (void)
 		}
 	      break;
 	    case 9:
-	      if (! secure &&am