diff --git a/.cvsignore b/.cvsignore
index fd72f71..04150eb 100644
--- a/.cvsignore
+++ b/.cvsignore
@@ -1 +1 @@
-gcc-4.1.2-20070821.tar.bz2
+gcc-4.1.2-20070925.tar.bz2
diff --git a/gcc41-artificial-attrib.patch b/gcc41-artificial-attrib.patch
new file mode 100644
index 0000000..c1b1dd8
--- /dev/null
+++ b/gcc41-artificial-attrib.patch
@@ -0,0 +1,4198 @@
+2007-09-23  Jakub Jelinek  <jakub@redhat.com>
+
+	* tree.h (block_nonartificial_location): New prototype.
+	* tree.c (block_nonartificial_location): New function.
+	* dwarf2out.c (gen_subprogram_die): Add DW_AT_artificial
+	if artificial attribute is present on abstract inline decl.
+	* c-common.c (handle_artificial_attribute): New function.
+	(c_common_attribute_table): Add artificial attribute.
+	* final.c (override_filename, override_linenum): New variables.
+	(final_scan_insn): For DBX_DEBUG or SDB_DEBUG, set override_filename
+	and override_linenum if inside of a block inlined from
+	__attribute__((__artificial__)) function.
+	(notice_source_line): Honor override_filename and override_linenum.
+	* doc/extend.texi: Document __attribute__((__artificial__)).
+	* config/i386/emmintrin.h: Add __artificial__ attribute to
+	all __always_inline__ functions.
+	* config/i386/mmintrin.h: Likewise.
+	* config/i386/mm3dnow.h: Likewise.
+	* config/i386/pmmintrin.h: Likewise.
+	* config/i386/ammintrin.h: Likewise.
+	* config/i386/xmmintrin.h: Likewise.
+
+--- gcc/doc/extend.texi.jj	2007-09-25 12:23:05.000000000 +0200
++++ gcc/doc/extend.texi	2007-09-25 15:04:15.000000000 +0200
+@@ -1589,7 +1589,8 @@ attributes are currently defined for fun
+ @code{section}, @code{constructor}, @code{destructor}, @code{used},
+ @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
+ @code{alias}, @code{warn_unused_result}, @code{nonnull},
+-@code{gnu_inline} and @code{externally_visible}.  Several other
++@code{gnu_inline}, @code{externally_visible} and @code{artificial}.
++Several other
+ attributes are defined for functions on particular target systems.  Other
+ attributes, including @code{section} are supported for variables declarations
+ (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
+@@ -1671,6 +1672,14 @@ In C++, this attribute does not depend o
+ but it still requires the @code{inline} keyword to enable its special
+ behavior.
+ 
++@cindex @code{artificial} function attribute
++@item artificial
++This attribute is useful for small inline wrappers which if possible
++should appear during debugging as a unit, depending on the debug
++info format it will either mean marking the function as artificial
++or using the caller location for all instructions within the inlined
++body.
++
+ @cindex @code{flatten} function attribute
+ @item flatten
+ Generally, inlining into a function is limited.  For a function marked with
+--- gcc/tree.c.jj	2007-04-03 13:18:26.000000000 +0200
++++ gcc/tree.c	2007-09-25 15:01:49.000000000 +0200
+@@ -7638,4 +7638,40 @@ empty_body_p (tree stmt)
+   return true;
+ }
+ 
++/* If BLOCK is inlined from an __attribute__((__artificial__))
++   routine, return pointer to location from where it has been
++   called.  */
++location_t *
++block_nonartificial_location (tree block)
++{
++  location_t *ret = NULL;
++
++  while (block && TREE_CODE (block) == BLOCK
++	 && BLOCK_ABSTRACT_ORIGIN (block))
++    {
++      tree ao = BLOCK_ABSTRACT_ORIGIN (block);
++
++      while (TREE_CODE (ao) == BLOCK && BLOCK_ABSTRACT_ORIGIN (ao))
++	ao = BLOCK_ABSTRACT_ORIGIN (ao);
++
++      if (TREE_CODE (ao) == FUNCTION_DECL)
++	{
++	  /* If AO is an artificial inline, point RET to the
++	     call site locus at which it has been inlined and continue
++	     the loop, in case AO's caller is also an artificial
++	     inline.  */
++	  if (DECL_DECLARED_INLINE_P (ao)
++	      && lookup_attribute ("artificial", DECL_ATTRIBUTES (ao)))
++	    ret = &BLOCK_SOURCE_LOCATION (block);
++	  else
++	    break;
++	}
++      else if (TREE_CODE (ao) != BLOCK)
++	break;
++
++      block = BLOCK_SUPERCONTEXT (block);
++    }
++  return ret;
++}
++
+ #include "gt-tree.h"
+--- gcc/tree.h.jj	2007-09-25 12:20:12.000000000 +0200
++++ gcc/tree.h	2007-09-25 15:02:37.000000000 +0200
+@@ -4270,6 +4270,8 @@ extern tree build_addr (tree, tree);
+ extern bool fields_compatible_p (tree, tree);
+ extern tree find_compatible_field (tree, tree);
+ 
++extern location_t *block_nonartificial_location (tree);
++
+ /* In function.c */
+ extern void expand_main_function (void);
+ extern void init_dummy_function_start (void);
+--- gcc/final.c.jj	2007-02-20 22:39:12.000000000 +0100
++++ gcc/final.c	2007-09-25 15:01:49.000000000 +0200
+@@ -141,6 +141,10 @@ static int high_function_linenum;
+ /* Filename of last NOTE.  */
+ static const char *last_filename;
+ 
++/* Override filename and line number.  */
++static const char *override_filename;
++static int override_linenum;
++
+ /* Whether to force emission of a line note before the next insn.  */
+ static bool force_source_line = false;
+   
+@@ -1822,6 +1826,18 @@ final_scan_insn (rtx insn, FILE *file, i
+ 	      /* Mark this block as output.  */
+ 	      TREE_ASM_WRITTEN (NOTE_BLOCK (insn)) = 1;
+ 	    }
++	  if (write_symbols == DBX_DEBUG
++	      || write_symbols == SDB_DEBUG)
++	    {
++	      location_t *locus_ptr
++		= block_nonartificial_location (NOTE_BLOCK (insn));
++
++	      if (locus_ptr != NULL)
++		{
++		  override_filename = LOCATION_FILE (*locus_ptr);
++		  override_linenum = LOCATION_LINE (*locus_ptr);
++		}
++	    }
+ 	  break;
+ 
+ 	case NOTE_INSN_BLOCK_END:
+@@ -1841,6 +1857,24 @@ final_scan_insn (rtx insn, FILE *file, i
+ 
+ 	      (*debug_hooks->end_block) (high_block_linenum, n);
+ 	    }
++	  if (write_symbols == DBX_DEBUG
++	      || write_symbols == SDB_DEBUG)
++	    {
++	      tree outer_block = BLOCK_SUPERCONTEXT (NOTE_BLOCK (insn));
++	      location_t *locus_ptr
++		= block_nonartificial_location (outer_block);
++
++	      if (locus_ptr != NULL)
++		{
++		  override_filename = LOCATION_FILE (*locus_ptr);
++		  override_linenum = LOCATION_LINE (*locus_ptr);
++		}
++	      else
++		{
++		  override_filename = NULL;
++		  override_linenum = 0;
++		}
++	    }
+ 	  break;
+ 
+ 	case NOTE_INSN_DELETED_LABEL:
+@@ -2521,8 +2555,19 @@ final_scan_insn (rtx insn, FILE *file, i
+ static bool
+ notice_source_line (rtx insn)
+ {
+-  const char *filename = insn_file (insn);
+-  int linenum = insn_line (insn);
++  const char *filename;
++  int linenum;
++
++  if (override_filename)
++    {
++      filename = override_filename;
++      linenum = override_linenum;
++    }
++  else
++    {
++      filename = insn_file (insn);
++      linenum = insn_line (insn);
++    }
+ 
+   if (filename
+       && (force_source_line
+--- gcc/dwarf2out.c.jj	2007-04-03 13:13:27.000000000 +0200
++++ gcc/dwarf2out.c	2007-09-25 15:01:49.000000000 +0200
+@@ -11667,6 +11667,10 @@ gen_subprogram_die (tree decl, dw_die_re
+             add_AT_unsigned (subr_die, DW_AT_inline, DW_INL_not_inlined);
+ 	}
+ 
++      if (DECL_DECLARED_INLINE_P (decl)
++	  && lookup_attribute ("artificial", DECL_ATTRIBUTES (decl)))
++	add_AT_flag (subr_die, DW_AT_artificial, 1);
++
+       equate_decl_number_to_die (decl, subr_die);
+     }
+   else if (!DECL_EXTERNAL (decl))
+--- gcc/c-common.c.jj	2007-03-27 15:40:15.000000000 +0200
++++ gcc/c-common.c	2007-09-25 15:01:49.000000000 +0200
+@@ -505,8 +505,8 @@ static tree handle_noreturn_attribute (t
+ static tree handle_noinline_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_always_inline_attribute (tree *, tree, tree, int,
+ 					    bool *);
+-static tree handle_gnu_inline_attribute (tree *, tree, tree, int,
+-					 bool *);
++static tree handle_gnu_inline_attribute (tree *, tree, tree, int, bool *);
++static tree handle_artificial_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_flatten_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_used_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_unused_attribute (tree *, tree, tree, int, bool *);
+@@ -577,6 +577,8 @@ const struct attribute_spec c_common_att
+ 			      handle_always_inline_attribute },
+   { "gnu_inline",             0, 0, true,  false, false,
+ 			      handle_gnu_inline_attribute },
++  { "artificial",             0, 0, true,  false, false,
++			      handle_artificial_attribute },
+   { "flatten",                0, 0, true,  false, false,
+                               handle_flatten_attribute },
+   { "used",                   0, 0, true,  false, false,
+@@ -4180,6 +4182,29 @@ handle_gnu_inline_attribute (tree *node,
+   return NULL_TREE;
+ }
+ 
++/* Handle an "artificial" attribute; arguments as in
++   struct attribute_spec.handler.  */
++
++static tree
++handle_artificial_attribute (tree *node, tree name,
++			     tree ARG_UNUSED (args),
++			     int ARG_UNUSED (flags),
++			     bool *no_add_attrs)
++{
++  if (TREE_CODE (*node) == FUNCTION_DECL && DECL_DECLARED_INLINE_P (*node))
++    {
++      /* Do nothing else, just set the attribute.  We'll get at
++	 it later with lookup_attribute.  */
++    }
++  else
++    {
++      warning (OPT_Wattributes, "%qE attribute ignored", name);
++      *no_add_attrs = true;
++    }
++
++  return NULL_TREE;
++}
++
+ /* Handle a "flatten" attribute; arguments as in
+    struct attribute_spec.handler.  */
+ 
+--- gcc/config/i386/mmintrin.h.jj	2007-03-27 15:40:18.000000000 +0200
++++ gcc/config/i386/mmintrin.h	2007-09-25 15:06:30.000000000 +0200
+@@ -43,26 +43,26 @@ typedef short __v4hi __attribute__ ((__v
+ typedef char __v8qi __attribute__ ((__vector_size__ (8)));
+ 
+ /* Empty the multimedia state.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_empty (void)
+ {
+   __builtin_ia32_emms ();
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _m_empty (void)
+ {
+   _mm_empty ();
+ }
+ 
+ /* Convert I to a __m64 object.  The integer is zero-extended to 64-bits.  */
+-static __inline __m64  __attribute__((__always_inline__))
++static __inline __m64  __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi32_si64 (int __i)
+ {
+   return (__m64) __builtin_ia32_vec_init_v2si (__i, 0);
+ }
+ 
+-static __inline __m64  __attribute__((__always_inline__))
++static __inline __m64  __attribute__((__always_inline__, __artificial__))
+ _m_from_int (int __i)
+ {
+   return _mm_cvtsi32_si64 (__i);
+@@ -72,26 +72,26 @@ _m_from_int (int __i)
+ /* Convert I to a __m64 object.  */
+ 
+ /* Intel intrinsic.  */
+-static __inline __m64  __attribute__((__always_inline__))
++static __inline __m64  __attribute__((__always_inline__, __artificial__))
+ _m_from_int64 (long long __i)
+ {
+   return (__m64) __i;
+ }
+ 
+-static __inline __m64  __attribute__((__always_inline__))
++static __inline __m64  __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64_m64 (long long __i)
+ {
+   return (__m64) __i;
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline __m64  __attribute__((__always_inline__))
++static __inline __m64  __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64x_si64 (long long __i)
+ {
+   return (__m64) __i;
+ }
+ 
+-static __inline __m64  __attribute__((__always_inline__))
++static __inline __m64  __attribute__((__always_inline__, __artificial__))
+ _mm_set_pi64x (long long __i)
+ {
+   return (__m64) __i;
+@@ -99,13 +99,13 @@ _mm_set_pi64x (long long __i)
+ #endif
+ 
+ /* Convert the lower 32 bits of the __m64 object into an integer.  */
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64_si32 (__m64 __i)
+ {
+   return __builtin_ia32_vec_ext_v2si ((__v2si)__i, 0);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _m_to_int (__m64 __i)
+ {
+   return _mm_cvtsi64_si32 (__i);
+@@ -115,20 +115,20 @@ _m_to_int (__m64 __i)
+ /* Convert the __m64 object to a 64bit integer.  */
+ 
+ /* Intel intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _m_to_int64 (__m64 __i)
+ {
+   return (long long)__i;
+ }
+ 
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvtm64_si64 (__m64 __i)
+ {
+   return (long long)__i;
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64_si64x (__m64 __i)
+ {
+   return (long long)__i;
+@@ -138,13 +138,13 @@ _mm_cvtsi64_si64x (__m64 __i)
+ /* Pack the four 16-bit values from M1 into the lower four 8-bit values of
+    the result, and the four 16-bit values from M2 into the upper four 8-bit
+    values of the result, all with signed saturation.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_packs_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_packsswb ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_packsswb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_packs_pi16 (__m1, __m2);
+@@ -153,13 +153,13 @@ _m_packsswb (__m64 __m1, __m64 __m2)
+ /* Pack the two 32-bit values from M1 in to the lower two 16-bit values of
+    the result, and the two 32-bit values from M2 into the upper two 16-bit
+    values of the result, all with signed saturation.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_packs_pi32 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_packssdw ((__v2si)__m1, (__v2si)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_packssdw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_packs_pi32 (__m1, __m2);
+@@ -168,13 +168,13 @@ _m_packssdw (__m64 __m1, __m64 __m2)
+ /* Pack the four 16-bit values from M1 into the lower four 8-bit values of
+    the result, and the four 16-bit values from M2 into the upper four 8-bit
+    values of the result, all with unsigned saturation.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_packs_pu16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_packuswb ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_packuswb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_packs_pu16 (__m1, __m2);
+@@ -182,13 +182,13 @@ _m_packuswb (__m64 __m1, __m64 __m2)
+ 
+ /* Interleave the four 8-bit values from the high half of M1 with the four
+    8-bit values from the high half of M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_pi8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_punpckhbw ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_punpckhbw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_unpackhi_pi8 (__m1, __m2);
+@@ -196,13 +196,13 @@ _m_punpckhbw (__m64 __m1, __m64 __m2)
+ 
+ /* Interleave the two 16-bit values from the high half of M1 with the two
+    16-bit values from the high half of M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_punpckhwd ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_punpckhwd (__m64 __m1, __m64 __m2)
+ {
+   return _mm_unpackhi_pi16 (__m1, __m2);
+@@ -210,13 +210,13 @@ _m_punpckhwd (__m64 __m1, __m64 __m2)
+ 
+ /* Interleave the 32-bit value from the high half of M1 with the 32-bit
+    value from the high half of M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_pi32 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_punpckhdq ((__v2si)__m1, (__v2si)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_punpckhdq (__m64 __m1, __m64 __m2)
+ {
+   return _mm_unpackhi_pi32 (__m1, __m2);
+@@ -224,13 +224,13 @@ _m_punpckhdq (__m64 __m1, __m64 __m2)
+ 
+ /* Interleave the four 8-bit values from the low half of M1 with the four
+    8-bit values from the low half of M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_pi8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_punpcklbw ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_punpcklbw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_unpacklo_pi8 (__m1, __m2);
+@@ -238,13 +238,13 @@ _m_punpcklbw (__m64 __m1, __m64 __m2)
+ 
+ /* Interleave the two 16-bit values from the low half of M1 with the two
+    16-bit values from the low half of M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_punpcklwd ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_punpcklwd (__m64 __m1, __m64 __m2)
+ {
+   return _mm_unpacklo_pi16 (__m1, __m2);
+@@ -252,59 +252,59 @@ _m_punpcklwd (__m64 __m1, __m64 __m2)
+ 
+ /* Interleave the 32-bit value from the low half of M1 with the 32-bit
+    value from the low half of M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_pi32 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_punpckldq ((__v2si)__m1, (__v2si)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_punpckldq (__m64 __m1, __m64 __m2)
+ {
+   return _mm_unpacklo_pi32 (__m1, __m2);
+ }
+ 
+ /* Add the 8-bit values in M1 to the 8-bit values in M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_add_pi8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_paddb ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_paddb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_add_pi8 (__m1, __m2);
+ }
+ 
+ /* Add the 16-bit values in M1 to the 16-bit values in M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_add_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_paddw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_paddw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_add_pi16 (__m1, __m2);
+ }
+ 
+ /* Add the 32-bit values in M1 to the 32-bit values in M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_add_pi32 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_paddd ((__v2si)__m1, (__v2si)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_paddd (__m64 __m1, __m64 __m2)
+ {
+   return _mm_add_pi32 (__m1, __m2);
+ }
+ 
+ /* Add the 64-bit values in M1 to the 64-bit values in M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_add_si64 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_paddq ((long long)__m1, (long long)__m2);
+@@ -312,13 +312,13 @@ _mm_add_si64 (__m64 __m1, __m64 __m2)
+ 
+ /* Add the 8-bit values in M1 to the 8-bit values in M2 using signed
+    saturated arithmetic.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_adds_pi8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_paddsb ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_paddsb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_adds_pi8 (__m1, __m2);
+@@ -326,13 +326,13 @@ _m_paddsb (__m64 __m1, __m64 __m2)
+ 
+ /* Add the 16-bit values in M1 to the 16-bit values in M2 using signed
+    saturated arithmetic.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_adds_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_paddsw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_paddsw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_adds_pi16 (__m1, __m2);
+@@ -340,13 +340,13 @@ _m_paddsw (__m64 __m1, __m64 __m2)
+ 
+ /* Add the 8-bit values in M1 to the 8-bit values in M2 using unsigned
+    saturated arithmetic.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_adds_pu8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_paddusb ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_paddusb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_adds_pu8 (__m1, __m2);
+@@ -354,59 +354,59 @@ _m_paddusb (__m64 __m1, __m64 __m2)
+ 
+ /* Add the 16-bit values in M1 to the 16-bit values in M2 using unsigned
+    saturated arithmetic.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_adds_pu16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_paddusw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_paddusw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_adds_pu16 (__m1, __m2);
+ }
+ 
+ /* Subtract the 8-bit values in M2 from the 8-bit values in M1.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sub_pi8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_psubb ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psubb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_sub_pi8 (__m1, __m2);
+ }
+ 
+ /* Subtract the 16-bit values in M2 from the 16-bit values in M1.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sub_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_psubw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psubw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_sub_pi16 (__m1, __m2);
+ }
+ 
+ /* Subtract the 32-bit values in M2 from the 32-bit values in M1.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sub_pi32 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_psubd ((__v2si)__m1, (__v2si)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psubd (__m64 __m1, __m64 __m2)
+ {
+   return _mm_sub_pi32 (__m1, __m2);
+ }
+ 
+ /* Add the 64-bit values in M1 to the 64-bit values in M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sub_si64 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_psubq ((long long)__m1, (long long)__m2);
+@@ -414,13 +414,13 @@ _mm_sub_si64 (__m64 __m1, __m64 __m2)
+ 
+ /* Subtract the 8-bit values in M2 from the 8-bit values in M1 using signed
+    saturating arithmetic.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_subs_pi8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_psubsb ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psubsb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_subs_pi8 (__m1, __m2);
+@@ -428,13 +428,13 @@ _m_psubsb (__m64 __m1, __m64 __m2)
+ 
+ /* Subtract the 16-bit values in M2 from the 16-bit values in M1 using
+    signed saturating arithmetic.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_subs_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_psubsw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psubsw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_subs_pi16 (__m1, __m2);
+@@ -442,13 +442,13 @@ _m_psubsw (__m64 __m1, __m64 __m2)
+ 
+ /* Subtract the 8-bit values in M2 from the 8-bit values in M1 using
+    unsigned saturating arithmetic.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_subs_pu8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_psubusb ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psubusb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_subs_pu8 (__m1, __m2);
+@@ -456,13 +456,13 @@ _m_psubusb (__m64 __m1, __m64 __m2)
+ 
+ /* Subtract the 16-bit values in M2 from the 16-bit values in M1 using
+    unsigned saturating arithmetic.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_subs_pu16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_psubusw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psubusw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_subs_pu16 (__m1, __m2);
+@@ -471,13 +471,13 @@ _m_psubusw (__m64 __m1, __m64 __m2)
+ /* Multiply four 16-bit values in M1 by four 16-bit values in M2 producing
+    four 32-bit intermediate results, which are then summed by pairs to
+    produce two 32-bit results.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_madd_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pmaddwd ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pmaddwd (__m64 __m1, __m64 __m2)
+ {
+   return _mm_madd_pi16 (__m1, __m2);
+@@ -485,13 +485,13 @@ _m_pmaddwd (__m64 __m1, __m64 __m2)
+ 
+ /* Multiply four signed 16-bit values in M1 by four signed 16-bit values in
+    M2 and produce the high 16 bits of the 32-bit results.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_mulhi_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pmulhw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pmulhw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_mulhi_pi16 (__m1, __m2);
+@@ -499,226 +499,226 @@ _m_pmulhw (__m64 __m1, __m64 __m2)
+ 
+ /* Multiply four 16-bit values in M1 by four 16-bit values in M2 and produce
+    the low 16 bits of the results.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_mullo_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pmullw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pmullw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_mullo_pi16 (__m1, __m2);
+ }
+ 
+ /* Shift four 16-bit values in M left by COUNT.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sll_pi16 (__m64 __m, __m64 __count)
+ {
+   return (__m64) __builtin_ia32_psllw ((__v4hi)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psllw (__m64 __m, __m64 __count)
+ {
+   return _mm_sll_pi16 (__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_slli_pi16 (__m64 __m, int __count)
+ {
+   return (__m64) __builtin_ia32_psllw ((__v4hi)__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psllwi (__m64 __m, int __count)
+ {
+   return _mm_slli_pi16 (__m, __count);
+ }
+ 
+ /* Shift two 32-bit values in M left by COUNT.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sll_pi32 (__m64 __m, __m64 __count)
+ {
+   return (__m64) __builtin_ia32_pslld ((__v2si)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pslld (__m64 __m, __m64 __count)
+ {
+   return _mm_sll_pi32 (__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_slli_pi32 (__m64 __m, int __count)
+ {
+   return (__m64) __builtin_ia32_pslld ((__v2si)__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pslldi (__m64 __m, int __count)
+ {
+   return _mm_slli_pi32 (__m, __count);
+ }
+ 
+ /* Shift the 64-bit value in M left by COUNT.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sll_si64 (__m64 __m, __m64 __count)
+ {
+   return (__m64) __builtin_ia32_psllq ((long long)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psllq (__m64 __m, __m64 __count)
+ {
+   return _mm_sll_si64 (__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_slli_si64 (__m64 __m, int __count)
+ {
+   return (__m64) __builtin_ia32_psllq ((long long)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psllqi (__m64 __m, int __count)
+ {
+   return _mm_slli_si64 (__m, __count);
+ }
+ 
+ /* Shift four 16-bit values in M right by COUNT; shift in the sign bit.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sra_pi16 (__m64 __m, __m64 __count)
+ {
+   return (__m64) __builtin_ia32_psraw ((__v4hi)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psraw (__m64 __m, __m64 __count)
+ {
+   return _mm_sra_pi16 (__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_srai_pi16 (__m64 __m, int __count)
+ {
+   return (__m64) __builtin_ia32_psraw ((__v4hi)__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psrawi (__m64 __m, int __count)
+ {
+   return _mm_srai_pi16 (__m, __count);
+ }
+ 
+ /* Shift two 32-bit values in M right by COUNT; shift in the sign bit.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sra_pi32 (__m64 __m, __m64 __count)
+ {
+   return (__m64) __builtin_ia32_psrad ((__v2si)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psrad (__m64 __m, __m64 __count)
+ {
+   return _mm_sra_pi32 (__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_srai_pi32 (__m64 __m, int __count)
+ {
+   return (__m64) __builtin_ia32_psrad ((__v2si)__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psradi (__m64 __m, int __count)
+ {
+   return _mm_srai_pi32 (__m, __count);
+ }
+ 
+ /* Shift four 16-bit values in M right by COUNT; shift in zeros.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_srl_pi16 (__m64 __m, __m64 __count)
+ {
+   return (__m64) __builtin_ia32_psrlw ((__v4hi)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psrlw (__m64 __m, __m64 __count)
+ {
+   return _mm_srl_pi16 (__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_srli_pi16 (__m64 __m, int __count)
+ {
+   return (__m64) __builtin_ia32_psrlw ((__v4hi)__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psrlwi (__m64 __m, int __count)
+ {
+   return _mm_srli_pi16 (__m, __count);
+ }
+ 
+ /* Shift two 32-bit values in M right by COUNT; shift in zeros.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_srl_pi32 (__m64 __m, __m64 __count)
+ {
+   return (__m64) __builtin_ia32_psrld ((__v2si)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psrld (__m64 __m, __m64 __count)
+ {
+   return _mm_srl_pi32 (__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_srli_pi32 (__m64 __m, int __count)
+ {
+   return (__m64) __builtin_ia32_psrld ((__v2si)__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psrldi (__m64 __m, int __count)
+ {
+   return _mm_srli_pi32 (__m, __count);
+ }
+ 
+ /* Shift the 64-bit value in M left by COUNT; shift in zeros.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_srl_si64 (__m64 __m, __m64 __count)
+ {
+   return (__m64) __builtin_ia32_psrlq ((long long)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psrlq (__m64 __m, __m64 __count)
+ {
+   return _mm_srl_si64 (__m, __count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_srli_si64 (__m64 __m, int __count)
+ {
+   return (__m64) __builtin_ia32_psrlq ((long long)__m, (long long)__count);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psrlqi (__m64 __m, int __count)
+ {
+   return _mm_srli_si64 (__m, __count);
+ }
+ 
+ /* Bit-wise AND the 64-bit values in M1 and M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_and_si64 (__m64 __m1, __m64 __m2)
+ {
+   return __builtin_ia32_pand (__m1, __m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pand (__m64 __m1, __m64 __m2)
+ {
+   return _mm_and_si64 (__m1, __m2);
+@@ -726,39 +726,39 @@ _m_pand (__m64 __m1, __m64 __m2)
+ 
+ /* Bit-wise complement the 64-bit value in M1 and bit-wise AND it with the
+    64-bit value in M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_andnot_si64 (__m64 __m1, __m64 __m2)
+ {
+   return __builtin_ia32_pandn (__m1, __m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pandn (__m64 __m1, __m64 __m2)
+ {
+   return _mm_andnot_si64 (__m1, __m2);
+ }
+ 
+ /* Bit-wise inclusive OR the 64-bit values in M1 and M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_or_si64 (__m64 __m1, __m64 __m2)
+ {
+   return __builtin_ia32_por (__m1, __m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_por (__m64 __m1, __m64 __m2)
+ {
+   return _mm_or_si64 (__m1, __m2);
+ }
+ 
+ /* Bit-wise exclusive OR the 64-bit values in M1 and M2.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_xor_si64 (__m64 __m1, __m64 __m2)
+ {
+   return __builtin_ia32_pxor (__m1, __m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pxor (__m64 __m1, __m64 __m2)
+ {
+   return _mm_xor_si64 (__m1, __m2);
+@@ -766,25 +766,25 @@ _m_pxor (__m64 __m1, __m64 __m2)
+ 
+ /* Compare eight 8-bit values.  The result of the comparison is 0xFF if the
+    test is true and zero if false.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_pi8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pcmpeqb ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pcmpeqb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_cmpeq_pi8 (__m1, __m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_pi8 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pcmpgtb ((__v8qi)__m1, (__v8qi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pcmpgtb (__m64 __m1, __m64 __m2)
+ {
+   return _mm_cmpgt_pi8 (__m1, __m2);
+@@ -792,25 +792,25 @@ _m_pcmpgtb (__m64 __m1, __m64 __m2)
+ 
+ /* Compare four 16-bit values.  The result of the comparison is 0xFFFF if
+    the test is true and zero if false.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pcmpeqw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pcmpeqw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_cmpeq_pi16 (__m1, __m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_pi16 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pcmpgtw ((__v4hi)__m1, (__v4hi)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pcmpgtw (__m64 __m1, __m64 __m2)
+ {
+   return _mm_cmpgt_pi16 (__m1, __m2);
+@@ -818,53 +818,53 @@ _m_pcmpgtw (__m64 __m1, __m64 __m2)
+ 
+ /* Compare two 32-bit values.  The result of the comparison is 0xFFFFFFFF if
+    the test is true and zero if false.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_pi32 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pcmpeqd ((__v2si)__m1, (__v2si)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pcmpeqd (__m64 __m1, __m64 __m2)
+ {
+   return _mm_cmpeq_pi32 (__m1, __m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_pi32 (__m64 __m1, __m64 __m2)
+ {
+   return (__m64) __builtin_ia32_pcmpgtd ((__v2si)__m1, (__v2si)__m2);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pcmpgtd (__m64 __m1, __m64 __m2)
+ {
+   return _mm_cmpgt_pi32 (__m1, __m2);
+ }
+ 
+ /* Creates a 64-bit zero.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_setzero_si64 (void)
+ {
+   return (__m64)0LL;
+ }
+ 
+ /* Creates a vector of two 32-bit values; I0 is least significant.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_set_pi32 (int __i1, int __i0)
+ {
+   return (__m64) __builtin_ia32_vec_init_v2si (__i0, __i1);
+ }
+ 
+ /* Creates a vector of four 16-bit values; W0 is least significant.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_set_pi16 (short __w3, short __w2, short __w1, short __w0)
+ {
+   return (__m64) __builtin_ia32_vec_init_v4hi (__w0, __w1, __w2, __w3);
+ }
+ 
+ /* Creates a vector of eight 8-bit values; B0 is least significant.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_set_pi8 (char __b7, char __b6, char __b5, char __b4,
+ 	     char __b3, char __b2, char __b1, char __b0)
+ {
+@@ -873,19 +873,19 @@ _mm_set_pi8 (char __b7, char __b6, char 
+ }
+ 
+ /* Similar, but with the arguments in reverse order.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_setr_pi32 (int __i0, int __i1)
+ {
+   return _mm_set_pi32 (__i1, __i0);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_setr_pi16 (short __w0, short __w1, short __w2, short __w3)
+ {
+   return _mm_set_pi16 (__w3, __w2, __w1, __w0);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_setr_pi8 (char __b0, char __b1, char __b2, char __b3,
+ 	      char __b4, char __b5, char __b6, char __b7)
+ {
+@@ -893,21 +893,21 @@ _mm_setr_pi8 (char __b0, char __b1, char
+ }
+ 
+ /* Creates a vector of two 32-bit values, both elements containing I.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_set1_pi32 (int __i)
+ {
+   return _mm_set_pi32 (__i, __i);
+ }
+ 
+ /* Creates a vector of four 16-bit values, all elements containing W.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_set1_pi16 (short __w)
+ {
+   return _mm_set_pi16 (__w, __w, __w, __w);
+ }
+ 
+ /* Creates a vector of eight 8-bit values, all elements containing B.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_set1_pi8 (char __b)
+ {
+   return _mm_set_pi8 (__b, __b, __b, __b, __b, __b, __b, __b);
+--- gcc/config/i386/pmmintrin.h.jj	2007-06-26 13:38:46.000000000 +0200
++++ gcc/config/i386/pmmintrin.h	2007-09-25 15:01:52.000000000 +0200
+@@ -48,79 +48,79 @@
+ #define _MM_GET_DENORMALS_ZERO_MODE() \
+   (_mm_getcsr() & _MM_DENORMALS_ZERO_MASK)
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_addsub_ps (__m128 __X, __m128 __Y)
+ {
+   return (__m128) __builtin_ia32_addsubps ((__v4sf)__X, (__v4sf)__Y);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_hadd_ps (__m128 __X, __m128 __Y)
+ {
+   return (__m128) __builtin_ia32_haddps ((__v4sf)__X, (__v4sf)__Y);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_hsub_ps (__m128 __X, __m128 __Y)
+ {
+   return (__m128) __builtin_ia32_hsubps ((__v4sf)__X, (__v4sf)__Y);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_movehdup_ps (__m128 __X)
+ {
+   return (__m128) __builtin_ia32_movshdup ((__v4sf)__X);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_moveldup_ps (__m128 __X)
+ {
+   return (__m128) __builtin_ia32_movsldup ((__v4sf)__X);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_addsub_pd (__m128d __X, __m128d __Y)
+ {
+   return (__m128d) __builtin_ia32_addsubpd ((__v2df)__X, (__v2df)__Y);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_hadd_pd (__m128d __X, __m128d __Y)
+ {
+   return (__m128d) __builtin_ia32_haddpd ((__v2df)__X, (__v2df)__Y);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_hsub_pd (__m128d __X, __m128d __Y)
+ {
+   return (__m128d) __builtin_ia32_hsubpd ((__v2df)__X, (__v2df)__Y);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_loaddup_pd (double const *__P)
+ {
+   return _mm_load1_pd (__P);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_movedup_pd (__m128d __X)
+ {
+   return _mm_shuffle_pd (__X, __X, _MM_SHUFFLE2 (0,0));
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_lddqu_si128 (__m128i const *__P)
+ {
+   return (__m128i) __builtin_ia32_lddqu ((char const *)__P);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_monitor (void const * __P, unsigned int __E, unsigned int __H)
+ {
+   __builtin_ia32_monitor (__P, __E, __H);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_mwait (unsigned int __E, unsigned int __H)
+ {
+   __builtin_ia32_mwait (__E, __H);
+--- gcc/config/i386/mm3dnow.h.jj	2007-02-20 22:38:59.000000000 +0100
++++ gcc/config/i386/mm3dnow.h	2007-09-25 15:01:52.000000000 +0200
+@@ -1,4 +1,4 @@
+-/* Copyright (C) 2004 Free Software Foundation, Inc.
++/* Copyright (C) 2004, 2007 Free Software Foundation, Inc.
+ 
+    This file is part of GCC.
+ 
+--- gcc/config/i386/xmmintrin.h.jj	2007-06-26 13:38:46.000000000 +0200
++++ gcc/config/i386/xmmintrin.h	2007-09-25 15:06:47.000000000 +0200
+@@ -88,7 +88,7 @@ enum _mm_hint
+ #define _MM_FLUSH_ZERO_OFF    0x0000
+ 
+ /* Create a vector of zeros.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_setzero_ps (void)
+ {
+   return __extension__ (__m128){ 0.0f, 0.0f, 0.0f, 0.0f };
+@@ -98,55 +98,55 @@ _mm_setzero_ps (void)
+    floating-point) values of A and B; the upper three SPFP values are
+    passed through from A.  */
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_add_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_addss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_sub_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_subss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_mul_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_mulss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_div_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_divss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_sqrt_ss (__m128 __A)
+ {
+   return (__m128) __builtin_ia32_sqrtss ((__v4sf)__A);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_rcp_ss (__m128 __A)
+ {
+   return (__m128) __builtin_ia32_rcpss ((__v4sf)__A);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_rsqrt_ss (__m128 __A)
+ {
+   return (__m128) __builtin_ia32_rsqrtss ((__v4sf)__A);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_min_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_minss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_max_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_maxss ((__v4sf)__A, (__v4sf)__B);
+@@ -154,55 +154,55 @@ _mm_max_ss (__m128 __A, __m128 __B)
+ 
+ /* Perform the respective operation on the four SPFP values in A and B.  */
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_add_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_addps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_sub_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_subps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_mul_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_mulps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_div_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_divps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_sqrt_ps (__m128 __A)
+ {
+   return (__m128) __builtin_ia32_sqrtps ((__v4sf)__A);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_rcp_ps (__m128 __A)
+ {
+   return (__m128) __builtin_ia32_rcpps ((__v4sf)__A);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_rsqrt_ps (__m128 __A)
+ {
+   return (__m128) __builtin_ia32_rsqrtps ((__v4sf)__A);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_min_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_minps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_max_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_maxps ((__v4sf)__A, (__v4sf)__B);
+@@ -210,25 +210,25 @@ _mm_max_ps (__m128 __A, __m128 __B)
+ 
+ /* Perform logical bit-wise operations on 128-bit values.  */
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_and_ps (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_andps (__A, __B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_andnot_ps (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_andnps (__A, __B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_or_ps (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_orps (__A, __B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_xor_ps (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_xorps (__A, __B);
+@@ -238,25 +238,25 @@ _mm_xor_ps (__m128 __A, __m128 __B)
+    comparison is true, place a mask of all ones in the result, otherwise a
+    mask of zeros.  The upper three SPFP values are passed through from A.  */
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpeqss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmplt_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpltss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmple_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpless ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_movss ((__v4sf) __A,
+@@ -266,7 +266,7 @@ _mm_cmpgt_ss (__m128 __A, __m128 __B)
+ 								__A));
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpge_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_movss ((__v4sf) __A,
+@@ -276,25 +276,25 @@ _mm_cmpge_ss (__m128 __A, __m128 __B)
+ 								__A));
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpneq_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpneqss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnlt_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpnltss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnle_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpnless ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpngt_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_movss ((__v4sf) __A,
+@@ -304,7 +304,7 @@ _mm_cmpngt_ss (__m128 __A, __m128 __B)
+ 								 __A));
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnge_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_movss ((__v4sf) __A,
+@@ -314,13 +314,13 @@ _mm_cmpnge_ss (__m128 __A, __m128 __B)
+ 								 __A));
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpord_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpordss ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpunord_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpunordss ((__v4sf)__A, (__v4sf)__B);
+@@ -330,73 +330,73 @@ _mm_cmpunord_ss (__m128 __A, __m128 __B)
+    element, if the comparison is true, place a mask of all ones in the
+    result, otherwise a mask of zeros.  */
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpeqps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmplt_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpltps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmple_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpleps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpgtps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpge_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpgeps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpneq_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpneqps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnlt_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpnltps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnle_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpnleps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpngt_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpngtps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnge_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpngeps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpord_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpordps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cmpunord_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_cmpunordps ((__v4sf)__A, (__v4sf)__B);
+@@ -405,73 +405,73 @@ _mm_cmpunord_ps (__m128 __A, __m128 __B)
+ /* Compare the lower SPFP values of A and B and return 1 if true
+    and 0 if false.  */
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comieq_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_comieq ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comilt_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_comilt ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comile_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_comile ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comigt_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_comigt ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comige_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_comige ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comineq_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_comineq ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomieq_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_ucomieq ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomilt_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_ucomilt ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomile_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_ucomile ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomigt_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_ucomigt ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomige_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_ucomige ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomineq_ss (__m128 __A, __m128 __B)
+ {
+   return __builtin_ia32_ucomineq ((__v4sf)__A, (__v4sf)__B);
+@@ -479,13 +479,13 @@ _mm_ucomineq_ss (__m128 __A, __m128 __B)
+ 
+ /* Convert the lower SPFP value to a 32-bit integer according to the current
+    rounding mode.  */
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_cvtss_si32 (__m128 __A)
+ {
+   return __builtin_ia32_cvtss2si ((__v4sf) __A);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_cvt_ss2si (__m128 __A)
+ {
+   return _mm_cvtss_si32 (__A);
+@@ -496,14 +496,14 @@ _mm_cvt_ss2si (__m128 __A)
+    current rounding mode.  */
+ 
+ /* Intel intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvtss_si64 (__m128 __A)
+ {
+   return __builtin_ia32_cvtss2si64 ((__v4sf) __A);
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvtss_si64x (__m128 __A)
+ {
+   return __builtin_ia32_cvtss2si64 ((__v4sf) __A);
+@@ -512,26 +512,26 @@ _mm_cvtss_si64x (__m128 __A)
+ 
+ /* Convert the two lower SPFP values to 32-bit integers according to the
+    current rounding mode.  Return the integers in packed form.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtps_pi32 (__m128 __A)
+ {
+   return (__m64) __builtin_ia32_cvtps2pi ((__v4sf) __A);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cvt_ps2pi (__m128 __A)
+ {
+   return _mm_cvtps_pi32 (__A);
+ }
+ 
+ /* Truncate the lower SPFP value to a 32-bit integer.  */
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_cvttss_si32 (__m128 __A)
+ {
+   return __builtin_ia32_cvttss2si ((__v4sf) __A);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_cvtt_ss2si (__m128 __A)
+ {
+   return _mm_cvttss_si32 (__A);
+@@ -541,14 +541,14 @@ _mm_cvtt_ss2si (__m128 __A)
+ /* Truncate the lower SPFP value to a 32-bit integer.  */
+ 
+ /* Intel intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvttss_si64 (__m128 __A)
+ {
+   return __builtin_ia32_cvttss2si64 ((__v4sf) __A);
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvttss_si64x (__m128 __A)
+ {
+   return __builtin_ia32_cvttss2si64 ((__v4sf) __A);
+@@ -557,26 +557,26 @@ _mm_cvttss_si64x (__m128 __A)
+ 
+ /* Truncate the two lower SPFP values to 32-bit integers.  Return the
+    integers in packed form.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cvttps_pi32 (__m128 __A)
+ {
+   return (__m64) __builtin_ia32_cvttps2pi ((__v4sf) __A);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtt_ps2pi (__m128 __A)
+ {
+   return _mm_cvttps_pi32 (__A);
+ }
+ 
+ /* Convert B to a SPFP value and insert it as element zero in A.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi32_ss (__m128 __A, int __B)
+ {
+   return (__m128) __builtin_ia32_cvtsi2ss ((__v4sf) __A, __B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvt_si2ss (__m128 __A, int __B)
+ {
+   return _mm_cvtsi32_ss (__A, __B);
+@@ -586,14 +586,14 @@ _mm_cvt_si2ss (__m128 __A, int __B)
+ /* Convert B to a SPFP value and insert it as element zero in A.  */
+ 
+ /* Intel intrinsic.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64_ss (__m128 __A, long long __B)
+ {
+   return (__m128) __builtin_ia32_cvtsi642ss ((__v4sf) __A, __B);
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64x_ss (__m128 __A, long long __B)
+ {
+   return (__m128) __builtin_ia32_cvtsi642ss ((__v4sf) __A, __B);
+@@ -602,20 +602,20 @@ _mm_cvtsi64x_ss (__m128 __A, long long _
+ 
+ /* Convert the two 32-bit values in B to SPFP form and insert them
+    as the two lower elements in A.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpi32_ps (__m128 __A, __m64 __B)
+ {
+   return (__m128) __builtin_ia32_cvtpi2ps ((__v4sf) __A, (__v2si)__B);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvt_pi2ps (__m128 __A, __m64 __B)
+ {
+   return _mm_cvtpi32_ps (__A, __B);
+ }
+ 
+ /* Convert the four signed 16-bit values in A to SPFP form.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpi16_ps (__m64 __A)
+ {
+   __v4hi __sign;
+@@ -641,7 +641,7 @@ _mm_cvtpi16_ps (__m64 __A)
+ }
+ 
+ /* Convert the four unsigned 16-bit values in A to SPFP form.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpu16_ps (__m64 __A)
+ {
+   __v2si __hisi, __losi;
+@@ -661,7 +661,7 @@ _mm_cvtpu16_ps (__m64 __A)
+ }
+ 
+ /* Convert the low four signed 8-bit values in A to SPFP form.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpi8_ps (__m64 __A)
+ {
+   __v8qi __sign;
+@@ -678,7 +678,7 @@ _mm_cvtpi8_ps (__m64 __A)
+ }
+ 
+ /* Convert the low four unsigned 8-bit values in A to SPFP form.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpu8_ps(__m64 __A)
+ {
+   __A = (__m64) __builtin_ia32_punpcklbw ((__v8qi)__A, (__v8qi)0LL);
+@@ -686,7 +686,7 @@ _mm_cvtpu8_ps(__m64 __A)
+ }
+ 
+ /* Convert the four signed 32-bit values in A and B to SPFP form.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpi32x2_ps(__m64 __A, __m64 __B)
+ {
+   __v4sf __zero = (__v4sf) _mm_setzero_ps ();
+@@ -696,7 +696,7 @@ _mm_cvtpi32x2_ps(__m64 __A, __m64 __B)
+ }
+ 
+ /* Convert the four SPFP values in A to four signed 16-bit integers.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtps_pi16(__m128 __A)
+ {
+   __v4sf __hisf = (__v4sf)__A;
+@@ -707,7 +707,7 @@ _mm_cvtps_pi16(__m128 __A)
+ }
+ 
+ /* Convert the four SPFP values in A to four signed 8-bit integers.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtps_pi8(__m128 __A)
+ {
+   __v4hi __tmp = (__v4hi) _mm_cvtps_pi16 (__A);
+@@ -716,7 +716,7 @@ _mm_cvtps_pi8(__m128 __A)
+ 
+ /* Selects four specific SPFP values from A and B based on MASK.  */
+ #if 0
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_shuffle_ps (__m128 __A, __m128 __B, int __mask)
+ {
+   return (__m128) __builtin_ia32_shufps ((__v4sf)__A, (__v4sf)__B, __mask);
+@@ -728,14 +728,14 @@ _mm_shuffle_ps (__m128 __A, __m128 __B, 
+ 
+ 
+ /* Selects and interleaves the upper two SPFP values from A and B.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_unpckhps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+ /* Selects and interleaves the lower two SPFP values from A and B.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_unpcklps ((__v4sf)__A, (__v4sf)__B);
+@@ -743,28 +743,28 @@ _mm_unpacklo_ps (__m128 __A, __m128 __B)
+ 
+ /* Sets the upper two SPFP values with 64-bits of data loaded from P;
+    the lower two values are passed through from A.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_loadh_pi (__m128 __A, __m64 const *__P)
+ {
+   return (__m128) __builtin_ia32_loadhps ((__v4sf)__A, (__v2si *)__P);
+ }
+ 
+ /* Stores the upper two SPFP values of A into P.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storeh_pi (__m64 *__P, __m128 __A)
+ {
+   __builtin_ia32_storehps ((__v2si *)__P, (__v4sf)__A);
+ }
+ 
+ /* Moves the upper two values of B into the lower two values of A.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_movehl_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_movhlps ((__v4sf)__A, (__v4sf)__B);
+ }
+ 
+ /* Moves the lower two values of B into the upper two values of A.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_movelh_ps (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_movlhps ((__v4sf)__A, (__v4sf)__B);
+@@ -772,146 +772,146 @@ _mm_movelh_ps (__m128 __A, __m128 __B)
+ 
+ /* Sets the lower two SPFP values with 64-bits of data loaded from P;
+    the upper two values are passed through from A.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_loadl_pi (__m128 __A, __m64 const *__P)
+ {
+   return (__m128) __builtin_ia32_loadlps ((__v4sf)__A, (__v2si *)__P);
+ }
+ 
+ /* Stores the lower two SPFP values of A into P.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storel_pi (__m64 *__P, __m128 __A)
+ {
+   __builtin_ia32_storelps ((__v2si *)__P, (__v4sf)__A);
+ }
+ 
+ /* Creates a 4-bit mask from the most significant bits of the SPFP values.  */
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_movemask_ps (__m128 __A)
+ {
+   return __builtin_ia32_movmskps ((__v4sf)__A);
+ }
+ 
+ /* Return the contents of the control register.  */
+-static __inline unsigned int __attribute__((__always_inline__))
++static __inline unsigned int __attribute__((__always_inline__, __artificial__))
+ _mm_getcsr (void)
+ {
+   return __builtin_ia32_stmxcsr ();
+ }
+ 
+ /* Read exception bits from the control register.  */
+-static __inline unsigned int __attribute__((__always_inline__))
++static __inline unsigned int __attribute__((__always_inline__, __artificial__))
+ _MM_GET_EXCEPTION_STATE (void)
+ {
+   return _mm_getcsr() & _MM_EXCEPT_MASK;
+ }
+ 
+-static __inline unsigned int __attribute__((__always_inline__))
++static __inline unsigned int __attribute__((__always_inline__, __artificial__))
+ _MM_GET_EXCEPTION_MASK (void)
+ {
+   return _mm_getcsr() & _MM_MASK_MASK;
+ }
+ 
+-static __inline unsigned int __attribute__((__always_inline__))
++static __inline unsigned int __attribute__((__always_inline__, __artificial__))
+ _MM_GET_ROUNDING_MODE (void)
+ {
+   return _mm_getcsr() & _MM_ROUND_MASK;
+ }
+ 
+-static __inline unsigned int __attribute__((__always_inline__))
++static __inline unsigned int __attribute__((__always_inline__, __artificial__))
+ _MM_GET_FLUSH_ZERO_MODE (void)
+ {
+   return _mm_getcsr() & _MM_FLUSH_ZERO_MASK;
+ }
+ 
+ /* Set the control register to I.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_setcsr (unsigned int __I)
+ {
+   __builtin_ia32_ldmxcsr (__I);
+ }
+ 
+ /* Set exception bits in the control register.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _MM_SET_EXCEPTION_STATE(unsigned int __mask)
+ {
+   _mm_setcsr((_mm_getcsr() & ~_MM_EXCEPT_MASK) | __mask);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _MM_SET_EXCEPTION_MASK (unsigned int __mask)
+ {
+   _mm_setcsr((_mm_getcsr() & ~_MM_MASK_MASK) | __mask);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _MM_SET_ROUNDING_MODE (unsigned int __mode)
+ {
+   _mm_setcsr((_mm_getcsr() & ~_MM_ROUND_MASK) | __mode);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _MM_SET_FLUSH_ZERO_MODE (unsigned int __mode)
+ {
+   _mm_setcsr((_mm_getcsr() & ~_MM_FLUSH_ZERO_MASK) | __mode);
+ }
+ 
+ /* Create a vector with element 0 as F and the rest zero.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_set_ss (float __F)
+ {
+   return __extension__ (__m128)(__v4sf){ __F, 0, 0, 0 };
+ }
+ 
+ /* Create a vector with all four elements equal to F.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_set1_ps (float __F)
+ {
+   return __extension__ (__m128)(__v4sf){ __F, __F, __F, __F };
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_set_ps1 (float __F)
+ {
+   return _mm_set1_ps (__F);
+ }
+ 
+ /* Create a vector with element 0 as *P and the rest zero.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_load_ss (float const *__P)
+ {
+   return _mm_set_ss (*__P);
+ }
+ 
+ /* Create a vector with all four elements equal to *P.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_load1_ps (float const *__P)
+ {
+   return _mm_set1_ps (*__P);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_load_ps1 (float const *__P)
+ {
+   return _mm_load1_ps (__P);
+ }
+ 
+ /* Load four SPFP values from P.  The address must be 16-byte aligned.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_load_ps (float const *__P)
+ {
+   return (__m128) *(__v4sf *)__P;
+ }
+ 
+ /* Load four SPFP values from P.  The address need not be 16-byte aligned.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_loadu_ps (float const *__P)
+ {
+   return (__m128) __builtin_ia32_loadups (__P);
+ }
+ 
+ /* Load four SPFP values in reverse order.  The address must be aligned.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_loadr_ps (float const *__P)
+ {
+   __v4sf __tmp = *(__v4sf *)__P;
+@@ -919,48 +919,48 @@ _mm_loadr_ps (float const *__P)
+ }
+ 
+ /* Create the vector [Z Y X W].  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_set_ps (const float __Z, const float __Y, const float __X, const float __W)
+ {
+   return __extension__ (__m128)(__v4sf){ __W, __X, __Y, __Z };
+ }
+ 
+ /* Create the vector [W X Y Z].  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_setr_ps (float __Z, float __Y, float __X, float __W)
+ {
+   return __extension__ (__m128)(__v4sf){ __Z, __Y, __X, __W };
+ }
+ 
+ /* Stores the lower SPFP value.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store_ss (float *__P, __m128 __A)
+ {
+   *__P = __builtin_ia32_vec_ext_v4sf ((__v4sf)__A, 0);
+ }
+ 
+-static __inline float __attribute__((__always_inline__))
++static __inline float __attribute__((__always_inline__, __artificial__))
+ _mm_cvtss_f32 (__m128 __A)
+ {
+   return __builtin_ia32_vec_ext_v4sf ((__v4sf)__A, 0);
+ }
+ 
+ /* Store four SPFP values.  The address must be 16-byte aligned.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store_ps (float *__P, __m128 __A)
+ {
+   *(__v4sf *)__P = (__v4sf)__A;
+ }
+ 
+ /* Store four SPFP values.  The address need not be 16-byte aligned.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storeu_ps (float *__P, __m128 __A)
+ {
+   __builtin_ia32_storeups (__P, (__v4sf)__A);
+ }
+ 
+ /* Store the lower SPFP value across four words.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store1_ps (float *__P, __m128 __A)
+ {
+   __v4sf __va = (__v4sf)__A;
+@@ -968,14 +968,14 @@ _mm_store1_ps (float *__P, __m128 __A)
+   _mm_storeu_ps (__P, __tmp);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store_ps1 (float *__P, __m128 __A)
+ {
+   _mm_store1_ps (__P, __A);
+ }
+ 
+ /* Store four SPFP values in reverse order.  The address must be aligned.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storer_ps (float *__P, __m128 __A)
+ {
+   __v4sf __va = (__v4sf)__A;
+@@ -984,7 +984,7 @@ _mm_storer_ps (float *__P, __m128 __A)
+ }
+ 
+ /* Sets the low SPFP value of A from the low value of B.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_move_ss (__m128 __A, __m128 __B)
+ {
+   return (__m128) __builtin_ia32_movss ((__v4sf)__A, (__v4sf)__B);
+@@ -992,13 +992,13 @@ _mm_move_ss (__m128 __A, __m128 __B)
+ 
+ /* Extracts one of the four words of A.  The selector N must be immediate.  */
+ #if 0
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_extract_pi16 (__m64 const __A, int const __N)
+ {
+   return __builtin_ia32_vec_ext_v4hi ((__v4hi)__A, __N);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _m_pextrw (__m64 const __A, int const __N)
+ {
+   return _mm_extract_pi16 (__A, __N);
+@@ -1011,13 +1011,13 @@ _m_pextrw (__m64 const __A, int const __
+ /* Inserts word D into one of four words of A.  The selector N must be
+    immediate.  */
+ #if 0
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_insert_pi16 (__m64 const __A, int const __D, int const __N)
+ {
+   return (__m64) __builtin_ia32_vec_set_v4hi ((__v4hi)__A, __D, __N);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pinsrw (__m64 const __A, int const __D, int const __N)
+ {
+   return _mm_insert_pi16 (__A, __D, __N);
+@@ -1029,65 +1029,65 @@ _m_pinsrw (__m64 const __A, int const __
+ #endif
+ 
+ /* Compute the element-wise maximum of signed 16-bit values.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_max_pi16 (__m64 __A, __m64 __B)
+ {
+   return (__m64) __builtin_ia32_pmaxsw ((__v4hi)__A, (__v4hi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pmaxsw (__m64 __A, __m64 __B)
+ {
+   return _mm_max_pi16 (__A, __B);
+ }
+ 
+ /* Compute the element-wise maximum of unsigned 8-bit values.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_max_pu8 (__m64 __A, __m64 __B)
+ {
+   return (__m64) __builtin_ia32_pmaxub ((__v8qi)__A, (__v8qi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pmaxub (__m64 __A, __m64 __B)
+ {
+   return _mm_max_pu8 (__A, __B);
+ }
+ 
+ /* Compute the element-wise minimum of signed 16-bit values.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_min_pi16 (__m64 __A, __m64 __B)
+ {
+   return (__m64) __builtin_ia32_pminsw ((__v4hi)__A, (__v4hi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pminsw (__m64 __A, __m64 __B)
+ {
+   return _mm_min_pi16 (__A, __B);
+ }
+ 
+ /* Compute the element-wise minimum of unsigned 8-bit values.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_min_pu8 (__m64 __A, __m64 __B)
+ {
+   return (__m64) __builtin_ia32_pminub ((__v8qi)__A, (__v8qi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pminub (__m64 __A, __m64 __B)
+ {
+   return _mm_min_pu8 (__A, __B);
+ }
+ 
+ /* Create an 8-bit mask of the signs of 8-bit values.  */
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_movemask_pi8 (__m64 __A)
+ {
+   return __builtin_ia32_pmovmskb ((__v8qi)__A);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _m_pmovmskb (__m64 __A)
+ {
+   return _mm_movemask_pi8 (__A);
+@@ -1095,13 +1095,13 @@ _m_pmovmskb (__m64 __A)
+ 
+ /* Multiply four unsigned 16-bit values in A by four unsigned 16-bit values
+    in B and produce the high 16 bits of the 32-bit results.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_mulhi_pu16 (__m64 __A, __m64 __B)
+ {
+   return (__m64) __builtin_ia32_pmulhuw ((__v4hi)__A, (__v4hi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pmulhuw (__m64 __A, __m64 __B)
+ {
+   return _mm_mulhi_pu16 (__A, __B);
+@@ -1110,13 +1110,13 @@ _m_pmulhuw (__m64 __A, __m64 __B)
+ /* Return a combination of the four 16-bit values in A.  The selector
+    must be an immediate.  */
+ #if 0
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_shuffle_pi16 (__m64 __A, int __N)
+ {
+   return (__m64) __builtin_ia32_pshufw ((__v4hi)__A, __N);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pshufw (__m64 __A, int __N)
+ {
+   return _mm_shuffle_pi16 (__A, __N);
+@@ -1130,39 +1130,39 @@ _m_pshufw (__m64 __A, int __N)
+ /* Conditionally store byte elements of A into P.  The high bit of each
+    byte in the selector N determines whether the corresponding byte from
+    A is stored.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_maskmove_si64 (__m64 __A, __m64 __N, char *__P)
+ {
+   __builtin_ia32_maskmovq ((__v8qi)__A, (__v8qi)__N, __P);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _m_maskmovq (__m64 __A, __m64 __N, char *__P)
+ {
+   _mm_maskmove_si64 (__A, __N, __P);
+ }
+ 
+ /* Compute the rounded averages of the unsigned 8-bit values in A and B.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_avg_pu8 (__m64 __A, __m64 __B)
+ {
+   return (__m64) __builtin_ia32_pavgb ((__v8qi)__A, (__v8qi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pavgb (__m64 __A, __m64 __B)
+ {
+   return _mm_avg_pu8 (__A, __B);
+ }
+ 
+ /* Compute the rounded averages of the unsigned 16-bit values in A and B.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_avg_pu16 (__m64 __A, __m64 __B)
+ {
+   return (__m64) __builtin_ia32_pavgw ((__v4hi)__A, (__v4hi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_pavgw (__m64 __A, __m64 __B)
+ {
+   return _mm_avg_pu16 (__A, __B);
+@@ -1171,13 +1171,13 @@ _m_pavgw (__m64 __A, __m64 __B)
+ /* Compute the sum of the absolute differences of the unsigned 8-bit
+    values in A and B.  Return the value in the lower 16-bit word; the
+    upper words are cleared.  */
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_sad_pu8 (__m64 __A, __m64 __B)
+ {
+   return (__m64) __builtin_ia32_psadbw ((__v8qi)__A, (__v8qi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _m_psadbw (__m64 __A, __m64 __B)
+ {
+   return _mm_sad_pu8 (__A, __B);
+@@ -1186,7 +1186,7 @@ _m_psadbw (__m64 __A, __m64 __B)
+ /* Loads one cache line from address P to a location "closer" to the
+    processor.  The selector I specifies the type of prefetch operation.  */
+ #if 0
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_prefetch (void *__P, enum _mm_hint __I)
+ {
+   __builtin_prefetch (__P, 0, __I);
+@@ -1197,14 +1197,14 @@ _mm_prefetch (void *__P, enum _mm_hint _
+ #endif
+ 
+ /* Stores the data in A to the address P without polluting the caches.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_stream_pi (__m64 *__P, __m64 __A)
+ {
+   __builtin_ia32_movntq ((unsigned long long *)__P, (unsigned long long)__A);
+ }
+ 
+ /* Likewise.  The address must be 16-byte aligned.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_stream_ps (float *__P, __m128 __A)
+ {
+   __builtin_ia32_movntps (__P, (__v4sf)__A);
+@@ -1212,7 +1212,7 @@ _mm_stream_ps (float *__P, __m128 __A)
+ 
+ /* Guarantees that every preceding store is globally visible before
+    any subsequent store.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_sfence (void)
+ {
+   __builtin_ia32_sfence ();
+@@ -1221,7 +1221,7 @@ _mm_sfence (void)
+ /* The execution of the next instruction is delayed by an implementation
+    specific amount of time.  The instruction does not modify the
+    architectural state.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_pause (void)
+ {
+   __asm__ __volatile__ ("rep; nop" : : );
+--- gcc/config/i386/ammintrin.h.jj	2007-06-26 13:38:46.000000000 +0200
++++ gcc/config/i386/ammintrin.h	2007-09-25 15:05:56.000000000 +0200
+@@ -37,19 +37,19 @@
+ /* We need definitions from the SSE3, SSE2 and SSE header files*/
+ #include <pmmintrin.h>
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_stream_sd (double * __P, __m128d __Y)
+ {
+   __builtin_ia32_movntsd (__P, (__v2df) __Y);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_stream_ss (float * __P, __m128 __Y)
+ {
+   __builtin_ia32_movntss (__P, (__v4sf) __Y);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_extract_si64 (__m128i __X, __m128i __Y)
+ {
+   return (__m128i) __builtin_ia32_extrq ((__v2di) __X, (__v16qi) __Y);
+@@ -58,7 +58,7 @@ _mm_extract_si64 (__m128i __X, __m128i _
+ #define _mm_extracti_si64(X, I, L) \
+ ((__m128i) __builtin_ia32_extrqi ((__v2di)(X), I, L))
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_insert_si64 (__m128i __X,__m128i __Y)
+ {
+   return (__m128i) __builtin_ia32_insertq ((__v2di)__X, (__v2di)__Y);
+--- gcc/config/i386/emmintrin.h.jj	2007-06-26 13:38:46.000000000 +0200
++++ gcc/config/i386/emmintrin.h	2007-09-25 15:06:19.000000000 +0200
+@@ -54,89 +54,89 @@ typedef double __m128d __attribute__ ((_
+  (((fp1) << 1) | (fp0))
+ 
+ /* Create a vector with element 0 as F and the rest zero.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_set_sd (double __F)
+ {
+   return __extension__ (__m128d){ __F, 0 };
+ }
+ 
+ /* Create a vector with both elements equal to F.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_set1_pd (double __F)
+ {
+   return __extension__ (__m128d){ __F, __F };
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_set_pd1 (double __F)
+ {
+   return _mm_set1_pd (__F);
+ }
+ 
+ /* Create a vector with the lower value X and upper value W.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_set_pd (double __W, double __X)
+ {
+   return __extension__ (__m128d){ __X, __W };
+ }
+ 
+ /* Create a vector with the lower value W and upper value X.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_setr_pd (double __W, double __X)
+ {
+   return __extension__ (__m128d){ __W, __X };
+ }
+ 
+ /* Create a vector of zeros.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_setzero_pd (void)
+ {
+   return __extension__ (__m128d){ 0.0, 0.0 };
+ }
+ 
+ /* Sets the low DPFP value of A from the low value of B.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_move_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d) __builtin_ia32_movsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+ /* Load two DPFP values from P.  The address must be 16-byte aligned.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_load_pd (double const *__P)
+ {
+   return *(__m128d *)__P;
+ }
+ 
+ /* Load two DPFP values from P.  The address need not be 16-byte aligned.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_loadu_pd (double const *__P)
+ {
+   return __builtin_ia32_loadupd (__P);
+ }
+ 
+ /* Create a vector with all two elements equal to *P.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_load1_pd (double const *__P)
+ {
+   return _mm_set1_pd (*__P);
+ }
+ 
+ /* Create a vector with element 0 as *P and the rest zero.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_load_sd (double const *__P)
+ {
+   return _mm_set_sd (*__P);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_load_pd1 (double const *__P)
+ {
+   return _mm_load1_pd (__P);
+ }
+ 
+ /* Load two DPFP values in reverse order.  The address must be aligned.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_loadr_pd (double const *__P)
+ {
+   __m128d __tmp = _mm_load_pd (__P);
+@@ -144,40 +144,40 @@ _mm_loadr_pd (double const *__P)
+ }
+ 
+ /* Store two DPFP values.  The address must be 16-byte aligned.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store_pd (double *__P, __m128d __A)
+ {
+   *(__m128d *)__P = __A;
+ }
+ 
+ /* Store two DPFP values.  The address need not be 16-byte aligned.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storeu_pd (double *__P, __m128d __A)
+ {
+   __builtin_ia32_storeupd (__P, __A);
+ }
+ 
+ /* Stores the lower DPFP value.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store_sd (double *__P, __m128d __A)
+ {
+   *__P = __builtin_ia32_vec_ext_v2df (__A, 0);
+ }
+ 
+-static __inline double __attribute__((__always_inline__))
++static __inline double __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsd_f64 (__m128d __A)
+ {
+   return __builtin_ia32_vec_ext_v2df (__A, 0);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storel_pd (double *__P, __m128d __A)
+ {
+   _mm_store_sd (__P, __A);
+ }
+ 
+ /* Stores the upper DPFP value.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storeh_pd (double *__P, __m128d __A)
+ {
+   *__P = __builtin_ia32_vec_ext_v2df (__A, 1);
+@@ -185,26 +185,26 @@ _mm_storeh_pd (double *__P, __m128d __A)
+ 
+ /* Store the lower DPFP value across two words.
+    The address must be 16-byte aligned.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store1_pd (double *__P, __m128d __A)
+ {
+   _mm_store_pd (__P, __builtin_ia32_shufpd (__A, __A, _MM_SHUFFLE2 (0,0)));
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store_pd1 (double *__P, __m128d __A)
+ {
+   _mm_store1_pd (__P, __A);
+ }
+ 
+ /* Store two DPFP values in reverse order.  The address must be aligned.  */
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storer_pd (double *__P, __m128d __A)
+ {
+   _mm_store_pd (__P, __builtin_ia32_shufpd (__A, __A, _MM_SHUFFLE2 (0,1)));
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi128_si32 (__m128i __A)
+ {
+   return __builtin_ia32_vec_ext_v4si ((__v4si)__A, 0);
+@@ -212,221 +212,221 @@ _mm_cvtsi128_si32 (__m128i __A)
+ 
+ #ifdef __x86_64__
+ /* Intel intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi128_si64 (__m128i __A)
+ {
+   return __builtin_ia32_vec_ext_v2di ((__v2di)__A, 0);
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi128_si64x (__m128i __A)
+ {
+   return __builtin_ia32_vec_ext_v2di ((__v2di)__A, 0);
+ }
+ #endif
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_add_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_addpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_add_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_addsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_sub_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_subpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_sub_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_subsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_mul_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_mulpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_mul_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_mulsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_div_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_divpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_div_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_divsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_sqrt_pd (__m128d __A)
+ {
+   return (__m128d)__builtin_ia32_sqrtpd ((__v2df)__A);
+ }
+ 
+ /* Return pair {sqrt (A[0), B[1]}.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_sqrt_sd (__m128d __A, __m128d __B)
+ {
+   __v2df __tmp = __builtin_ia32_movsd ((__v2df)__A, (__v2df)__B);
+   return (__m128d)__builtin_ia32_sqrtsd ((__v2df)__tmp);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_min_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_minpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_min_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_minsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_max_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_maxpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_max_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_maxsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_and_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_andpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_andnot_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_andnpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_or_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_orpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_xor_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_xorpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpeqpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmplt_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpltpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmple_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmplepd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpgtpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpge_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpgepd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpneq_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpneqpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnlt_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpnltpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnle_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpnlepd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpngt_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpngtpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnge_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpngepd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpord_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpordpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpunord_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpunordpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpeqsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmplt_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpltsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmple_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmplesd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d) __builtin_ia32_movsd ((__v2df) __A,
+@@ -436,7 +436,7 @@ _mm_cmpgt_sd (__m128d __A, __m128d __B)
+ 								 __A));
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpge_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d) __builtin_ia32_movsd ((__v2df) __A,
+@@ -446,25 +446,25 @@ _mm_cmpge_sd (__m128d __A, __m128d __B)
+ 								 __A));
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpneq_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpneqsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnlt_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpnltsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnle_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpnlesd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpngt_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d) __builtin_ia32_movsd ((__v2df) __A,
+@@ -474,7 +474,7 @@ _mm_cmpngt_sd (__m128d __A, __m128d __B)
+ 								  __A));
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpnge_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d) __builtin_ia32_movsd ((__v2df) __A,
+@@ -484,85 +484,85 @@ _mm_cmpnge_sd (__m128d __A, __m128d __B)
+ 								  __A));
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpord_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpordsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cmpunord_sd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_cmpunordsd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comieq_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_comisdeq ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comilt_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_comisdlt ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comile_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_comisdle ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comigt_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_comisdgt ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comige_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_comisdge ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_comineq_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_comisdneq ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomieq_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_ucomisdeq ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomilt_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_ucomisdlt ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomile_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_ucomisdle ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomigt_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_ucomisdgt ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomige_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_ucomisdge ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_ucomineq_sd (__m128d __A, __m128d __B)
+ {
+   return __builtin_ia32_ucomisdneq ((__v2df)__A, (__v2df)__B);
+@@ -570,25 +570,25 @@ _mm_ucomineq_sd (__m128d __A, __m128d __
+ 
+ /* Create a vector of Qi, where i is the element number.  */
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set_epi64x (long long __q1, long long __q0)
+ {
+   return __extension__ (__m128i)(__v2di){ __q0, __q1 };
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set_epi64 (__m64 __q1,  __m64 __q0)
+ {
+   return _mm_set_epi64x ((long long)__q1, (long long)__q0);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set_epi32 (int __q3, int __q2, int __q1, int __q0)
+ {
+   return __extension__ (__m128i)(__v4si){ __q0, __q1, __q2, __q3 };
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set_epi16 (short __q7, short __q6, short __q5, short __q4,
+ 	       short __q3, short __q2, short __q1, short __q0)
+ {
+@@ -596,7 +596,7 @@ _mm_set_epi16 (short __q7, short __q6, s
+     __q0, __q1, __q2, __q3, __q4, __q5, __q6, __q7 };
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set_epi8 (char __q15, char __q14, char __q13, char __q12,
+ 	      char __q11, char __q10, char __q09, char __q08,
+ 	      char __q07, char __q06, char __q05, char __q04,
+@@ -610,31 +610,31 @@ _mm_set_epi8 (char __q15, char __q14, ch
+ 
+ /* Set all of the elements of the vector to A.  */
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set1_epi64x (long long __A)
+ {
+   return _mm_set_epi64x (__A, __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set1_epi64 (__m64 __A)
+ {
+   return _mm_set_epi64 (__A, __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set1_epi32 (int __A)
+ {
+   return _mm_set_epi32 (__A, __A, __A, __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set1_epi16 (short __A)
+ {
+   return _mm_set_epi16 (__A, __A, __A, __A, __A, __A, __A, __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_set1_epi8 (char __A)
+ {
+   return _mm_set_epi8 (__A, __A, __A, __A, __A, __A, __A, __A,
+@@ -644,26 +644,26 @@ _mm_set1_epi8 (char __A)
+ /* Create a vector of Qi, where i is the element number.
+    The parameter order is reversed from the _mm_set_epi* functions.  */
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_setr_epi64 (__m64 __q0, __m64 __q1)
+ {
+   return _mm_set_epi64 (__q1, __q0);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_setr_epi32 (int __q0, int __q1, int __q2, int __q3)
+ {
+   return _mm_set_epi32 (__q3, __q2, __q1, __q0);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_setr_epi16 (short __q0, short __q1, short __q2, short __q3,
+ 	        short __q4, short __q5, short __q6, short __q7)
+ {
+   return _mm_set_epi16 (__q7, __q6, __q5, __q4, __q3, __q2, __q1, __q0);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_setr_epi8 (char __q00, char __q01, char __q02, char __q03,
+ 	       char __q04, char __q05, char __q06, char __q07,
+ 	       char __q08, char __q09, char __q10, char __q11,
+@@ -675,134 +675,134 @@ _mm_setr_epi8 (char __q00, char __q01, c
+ 
+ /* Create a vector with element 0 as *P and the rest zero.  */
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_load_si128 (__m128i const *__P)
+ {
+   return *__P;
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_loadu_si128 (__m128i const *__P)
+ {
+   return (__m128i) __builtin_ia32_loaddqu ((char const *)__P);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_loadl_epi64 (__m128i const *__P)
+ {
+   return _mm_set_epi64 ((__m64)0LL, *(__m64 *)__P);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_store_si128 (__m128i *__P, __m128i __B)
+ {
+   *__P = __B;
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storeu_si128 (__m128i *__P, __m128i __B)
+ {
+   __builtin_ia32_storedqu ((char *)__P, (__v16qi)__B);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_storel_epi64 (__m128i *__P, __m128i __B)
+ {
+   *(long long *)__P = __builtin_ia32_vec_ext_v2di ((__v2di)__B, 0);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_movepi64_pi64 (__m128i __B)
+ {
+   return (__m64) __builtin_ia32_vec_ext_v2di ((__v2di)__B, 0);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_movpi64_epi64 (__m64 __A)
+ {
+   return _mm_set_epi64 ((__m64)0LL, __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_move_epi64 (__m128i __A)
+ {
+   return _mm_set_epi64 ((__m64)0LL, _mm_movepi64_pi64 (__A));
+ }
+ 
+ /* Create a vector of zeros.  */
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_setzero_si128 (void)
+ {
+   return __extension__ (__m128i)(__v4si){ 0, 0, 0, 0 };
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cvtepi32_pd (__m128i __A)
+ {
+   return (__m128d)__builtin_ia32_cvtdq2pd ((__v4si) __A);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtepi32_ps (__m128i __A)
+ {
+   return (__m128)__builtin_ia32_cvtdq2ps ((__v4si) __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpd_epi32 (__m128d __A)
+ {
+   return (__m128i)__builtin_ia32_cvtpd2dq ((__v2df) __A);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpd_pi32 (__m128d __A)
+ {
+   return (__m64)__builtin_ia32_cvtpd2pi ((__v2df) __A);
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpd_ps (__m128d __A)
+ {
+   return (__m128)__builtin_ia32_cvtpd2ps ((__v2df) __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cvttpd_epi32 (__m128d __A)
+ {
+   return (__m128i)__builtin_ia32_cvttpd2dq ((__v2df) __A);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_cvttpd_pi32 (__m128d __A)
+ {
+   return (__m64)__builtin_ia32_cvttpd2pi ((__v2df) __A);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cvtpi32_pd (__m64 __A)
+ {
+   return (__m128d)__builtin_ia32_cvtpi2pd ((__v2si) __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cvtps_epi32 (__m128 __A)
+ {
+   return (__m128i)__builtin_ia32_cvtps2dq ((__v4sf) __A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cvttps_epi32 (__m128 __A)
+ {
+   return (__m128i)__builtin_ia32_cvttps2dq ((__v4sf) __A);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cvtps_pd (__m128 __A)
+ {
+   return (__m128d)__builtin_ia32_cvtps2pd ((__v4sf) __A);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsd_si32 (__m128d __A)
+ {
+   return __builtin_ia32_cvtsd2si ((__v2df) __A);
+@@ -810,21 +810,21 @@ _mm_cvtsd_si32 (__m128d __A)
+ 
+ #ifdef __x86_64__
+ /* Intel intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsd_si64 (__m128d __A)
+ {
+   return __builtin_ia32_cvtsd2si64 ((__v2df) __A);
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsd_si64x (__m128d __A)
+ {
+   return __builtin_ia32_cvtsd2si64 ((__v2df) __A);
+ }
+ #endif
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_cvttsd_si32 (__m128d __A)
+ {
+   return __builtin_ia32_cvttsd2si ((__v2df) __A);
+@@ -832,27 +832,27 @@ _mm_cvttsd_si32 (__m128d __A)
+ 
+ #ifdef __x86_64__
+ /* Intel intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvttsd_si64 (__m128d __A)
+ {
+   return __builtin_ia32_cvttsd2si64 ((__v2df) __A);
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline long long __attribute__((__always_inline__))
++static __inline long long __attribute__((__always_inline__, __artificial__))
+ _mm_cvttsd_si64x (__m128d __A)
+ {
+   return __builtin_ia32_cvttsd2si64 ((__v2df) __A);
+ }
+ #endif
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsd_ss (__m128 __A, __m128d __B)
+ {
+   return (__m128)__builtin_ia32_cvtsd2ss ((__v4sf) __A, (__v2df) __B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi32_sd (__m128d __A, int __B)
+ {
+   return (__m128d)__builtin_ia32_cvtsi2sd ((__v2df) __A, __B);
+@@ -860,21 +860,21 @@ _mm_cvtsi32_sd (__m128d __A, int __B)
+ 
+ #ifdef __x86_64__
+ /* Intel intrinsic.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64_sd (__m128d __A, long long __B)
+ {
+   return (__m128d)__builtin_ia32_cvtsi642sd ((__v2df) __A, __B);
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64x_sd (__m128d __A, long long __B)
+ {
+   return (__m128d)__builtin_ia32_cvtsi642sd ((__v2df) __A, __B);
+ }
+ #endif
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_cvtss_sd (__m128d __A, __m128 __B)
+ {
+   return (__m128d)__builtin_ia32_cvtss2sd ((__v2df) __A, (__v4sf)__B);
+@@ -882,242 +882,242 @@ _mm_cvtss_sd (__m128d __A, __m128 __B)
+ 
+ #define _mm_shuffle_pd(__A, __B, __C) ((__m128d)__builtin_ia32_shufpd ((__v2df)__A, (__v2df)__B, (__C)))
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_unpckhpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_pd (__m128d __A, __m128d __B)
+ {
+   return (__m128d)__builtin_ia32_unpcklpd ((__v2df)__A, (__v2df)__B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_loadh_pd (__m128d __A, double const *__B)
+ {
+   return (__m128d)__builtin_ia32_loadhpd ((__v2df)__A, __B);
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_loadl_pd (__m128d __A, double const *__B)
+ {
+   return (__m128d)__builtin_ia32_loadlpd ((__v2df)__A, __B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_movemask_pd (__m128d __A)
+ {
+   return __builtin_ia32_movmskpd ((__v2df)__A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_packs_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_packsswb128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_packs_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_packssdw128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_packus_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_packuswb128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_punpckhbw128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_punpckhwd128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_punpckhdq128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_unpackhi_epi64 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_punpckhqdq128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_punpcklbw128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_punpcklwd128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_punpckldq128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_unpacklo_epi64 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_punpcklqdq128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_add_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_paddb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_add_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_paddw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_add_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_paddd128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_add_epi64 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_paddq128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_adds_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_paddsb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_adds_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_paddsw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_adds_epu8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_paddusb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_adds_epu16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_paddusw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sub_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psubb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sub_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psubw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sub_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psubd128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sub_epi64 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psubq128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_subs_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psubsb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_subs_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psubsw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_subs_epu8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psubusb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_subs_epu16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psubusw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_madd_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pmaddwd128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_mulhi_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pmulhw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_mullo_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pmullw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m64 __attribute__((__always_inline__))
++static __inline __m64 __attribute__((__always_inline__, __artificial__))
+ _mm_mul_su32 (__m64 __A, __m64 __B)
+ {
+   return (__m64)__builtin_ia32_pmuludq ((__v2si)__A, (__v2si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_mul_epu32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pmuludq128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+ #if 0
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_slli_epi16 (__m128i __A, int __B)
+ {
+   return (__m128i)__builtin_ia32_psllwi128 ((__v8hi)__A, __B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_slli_epi32 (__m128i __A, int __B)
+ {
+   return (__m128i)__builtin_ia32_pslldi128 ((__v4si)__A, __B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_slli_epi64 (__m128i __A, int __B)
+ {
+   return (__m128i)__builtin_ia32_psllqi128 ((__v2di)__A, __B);
+@@ -1132,13 +1132,13 @@ _mm_slli_epi64 (__m128i __A, int __B)
+ #endif
+ 
+ #if 0
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srai_epi16 (__m128i __A, int __B)
+ {
+   return (__m128i)__builtin_ia32_psrawi128 ((__v8hi)__A, __B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srai_epi32 (__m128i __A, int __B)
+ {
+   return (__m128i)__builtin_ia32_psradi128 ((__v4si)__A, __B);
+@@ -1151,13 +1151,13 @@ _mm_srai_epi32 (__m128i __A, int __B)
+ #endif
+ 
+ #if 0
+-static __m128i __attribute__((__always_inline__))
++static __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srli_si128 (__m128i __A, int __B)
+ {
+   return ((__m128i)__builtin_ia32_psrldqi128 (__A, __B * 8));
+ }
+ 
+-static __m128i __attribute__((__always_inline__))
++static __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srli_si128 (__m128i __A, int __B)
+ {
+   return ((__m128i)__builtin_ia32_pslldqi128 (__A, __B * 8));
+@@ -1170,19 +1170,19 @@ _mm_srli_si128 (__m128i __A, int __B)
+ #endif
+ 
+ #if 0
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srli_epi16 (__m128i __A, int __B)
+ {
+   return (__m128i)__builtin_ia32_psrlwi128 ((__v8hi)__A, __B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srli_epi32 (__m128i __A, int __B)
+ {
+   return (__m128i)__builtin_ia32_psrldi128 ((__v4si)__A, __B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srli_epi64 (__m128i __A, int __B)
+ {
+   return (__m128i)__builtin_ia32_psrlqi128 ((__v2di)__A, __B);
+@@ -1196,140 +1196,140 @@ _mm_srli_epi64 (__m128i __A, int __B)
+   ((__m128i)__builtin_ia32_psrlqi128 ((__v4si)(__A), __B))
+ #endif
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sll_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psllw128((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sll_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pslld128((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sll_epi64 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psllq128((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sra_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psraw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sra_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psrad128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srl_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psrlw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srl_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psrld128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_srl_epi64 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psrlq128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_and_si128 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pand128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_andnot_si128 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pandn128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_or_si128 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_por128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_xor_si128 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pxor128 ((__v2di)__A, (__v2di)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpeqb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpeqw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmpeq_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpeqd128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmplt_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpgtb128 ((__v16qi)__B, (__v16qi)__A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmplt_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpgtw128 ((__v8hi)__B, (__v8hi)__A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmplt_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpgtd128 ((__v4si)__B, (__v4si)__A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_epi8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpgtb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpgtw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cmpgt_epi32 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pcmpgtd128 ((__v4si)__A, (__v4si)__B);
+ }
+ 
+ #if 0
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_extract_epi16 (__m128i const __A, int const __N)
+ {
+   return __builtin_ia32_vec_ext_v8hi ((__v8hi)__A, __N);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_insert_epi16 (__m128i const __A, int const __D, int const __N)
+ {
+   return (__m128i) __builtin_ia32_vec_set_v8hi ((__v8hi)__A, __D, __N);
+@@ -1341,37 +1341,37 @@ _mm_insert_epi16 (__m128i const __A, int
+   ((__m128i) __builtin_ia32_vec_set_v8hi ((__v8hi)(A), (D), (N)))
+ #endif
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_max_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pmaxsw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_max_epu8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pmaxub128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_min_epi16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pminsw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_min_epu8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pminub128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline int __attribute__((__always_inline__))
++static __inline int __attribute__((__always_inline__, __artificial__))
+ _mm_movemask_epi8 (__m128i __A)
+ {
+   return __builtin_ia32_pmovmskb128 ((__v16qi)__A);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_mulhi_epu16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pmulhuw128 ((__v8hi)__A, (__v8hi)__B);
+@@ -1381,67 +1381,67 @@ _mm_mulhi_epu16 (__m128i __A, __m128i __
+ #define _mm_shufflelo_epi16(__A, __B) ((__m128i)__builtin_ia32_pshuflw ((__v8hi)__A, __B))
+ #define _mm_shuffle_epi32(__A, __B) ((__m128i)__builtin_ia32_pshufd ((__v4si)__A, __B))
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_maskmoveu_si128 (__m128i __A, __m128i __B, char *__C)
+ {
+   __builtin_ia32_maskmovdqu ((__v16qi)__A, (__v16qi)__B, __C);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_avg_epu8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pavgb128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_avg_epu16 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_pavgw128 ((__v8hi)__A, (__v8hi)__B);
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_sad_epu8 (__m128i __A, __m128i __B)
+ {
+   return (__m128i)__builtin_ia32_psadbw128 ((__v16qi)__A, (__v16qi)__B);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_stream_si32 (int *__A, int __B)
+ {
+   __builtin_ia32_movnti (__A, __B);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_stream_si128 (__m128i *__A, __m128i __B)
+ {
+   __builtin_ia32_movntdq ((__v2di *)__A, (__v2di)__B);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_stream_pd (double *__A, __m128d __B)
+ {
+   __builtin_ia32_movntpd (__A, (__v2df)__B);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_clflush (void const *__A)
+ {
+   __builtin_ia32_clflush (__A);
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_lfence (void)
+ {
+   __builtin_ia32_lfence ();
+ }
+ 
+-static __inline void __attribute__((__always_inline__))
++static __inline void __attribute__((__always_inline__, __artificial__))
+ _mm_mfence (void)
+ {
+   __builtin_ia32_mfence ();
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi32_si128 (int __A)
+ {
+   return _mm_set_epi32 (0, 0, 0, __A);
+@@ -1449,14 +1449,14 @@ _mm_cvtsi32_si128 (int __A)
+ 
+ #ifdef __x86_64__
+ /* Intel intrinsic.  */
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64_si128 (long long __A)
+ {
+   return _mm_set_epi64x (0, __A);
+ }
+ 
+ /* Microsoft intrinsic.  */
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_cvtsi64x_si128 (long long __A)
+ {
+   return _mm_set_epi64x (0, __A);
+@@ -1465,37 +1465,37 @@ _mm_cvtsi64x_si128 (long long __A)
+ 
+ /* Casts between various SP, DP, INT vector types.  Note that these do no
+    conversion of values, they just change the type.  */
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_castpd_ps(__m128d __A)
+ {
+   return (__m128) __A;
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_castpd_si128(__m128d __A)
+ {
+   return (__m128i) __A;
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_castps_pd(__m128 __A)
+ {
+   return (__m128d) __A;
+ }
+ 
+-static __inline __m128i __attribute__((__always_inline__))
++static __inline __m128i __attribute__((__always_inline__, __artificial__))
+ _mm_castps_si128(__m128 __A)
+ {
+   return (__m128i) __A;
+ }
+ 
+-static __inline __m128 __attribute__((__always_inline__))
++static __inline __m128 __attribute__((__always_inline__, __artificial__))
+ _mm_castsi128_ps(__m128i __A)
+ {
+   return (__m128) __A;
+ }
+ 
+-static __inline __m128d __attribute__((__always_inline__))
++static __inline __m128d __attribute__((__always_inline__, __artificial__))
+ _mm_castsi128_pd(__m128i __A)
+ {
+   return (__m128d) __A;
diff --git a/gcc41-builtin-chk-anticipated.patch b/gcc41-builtin-chk-anticipated.patch
deleted file mode 100644
index 15b7846..0000000
--- a/gcc41-builtin-chk-anticipated.patch
+++ /dev/null
@@ -1,37 +0,0 @@
-2007-09-06  Jakub Jelinek  <jakub@redhat.com>
-
-	PR c++/33289
-	* decl.c (builtin_function_1): Set DECL_ANTICIPATED also
-	on __*_chk non-__builtin_* decls.
-
-	* g++.dg/eh/builtin4.C: New test.
-
---- gcc/cp/decl.c.jj	2007-08-31 14:28:19.000000000 +0200
-+++ gcc/cp/decl.c	2007-09-04 21:28:06.000000000 +0200
-@@ -3517,6 +3529,17 @@ builtin_function_1 (tree decl, tree cont
-      anticipated but not actually declared.  */
-   if (name[0] != '_' || name[1] != '_')
-     DECL_ANTICIPATED (decl) = 1;
-+  else if (strncmp (name + 2, "builtin_", strlen ("builtin_")) != 0)
-+    {
-+      size_t len = strlen (name);
-+
-+      /* Treat __*_chk fortification functions as anticipated as well,
-+	 unless they are __builtin_*.  */
-+      if (len > strlen ("___chk")
-+	  && memcmp (name + len - strlen ("_chk"),
-+		     "_chk", strlen ("_chk") + 1) == 0)
-+	DECL_ANTICIPATED (decl) = 1;
-+    }
- 
-   return decl;
- }
---- gcc/testsuite/g++.dg/eh/builtin4.C.jj	2007-09-04 22:06:06.000000000 +0200
-+++ gcc/testsuite/g++.dg/eh/builtin4.C	2007-09-04 22:07:03.000000000 +0200
-@@ -0,0 +1,6 @@
-+// PR c++/33289
-+// { dg-do compile }
-+
-+typedef __SIZE_TYPE__ size_t;
-+extern "C" int __sprintf_chk (char *__restrict, int, size_t, const char *, ...) throw ();
-+extern "C" int __sprintf_chk (char *__restrict, int, size_t, const char *, ...) throw ();
diff --git a/gcc41-builtin-throw.patch b/gcc41-builtin-throw.patch
deleted file mode 100644
index f9c4eaf..0000000
--- a/gcc41-builtin-throw.patch
+++ /dev/null
@@ -1,134 +0,0 @@
-2007-09-06  Jakub Jelinek  <jakub@redhat.com>
-
-	* builtin-attrs.def (ATTR_NONNULL_3): New.
-	(DEF_FORMAT_ATTRIBUTE): Use just ATTR_NONNULL_##FA instead of
-	ATTR_NOTHROW_NONNULL_##FA.
-	(DEF_FORMAT_ATTRIBUTE_NOTHROW, DEF_FORMAT_ATTRIBUTE_BOTH): New macros.
-	(ATTR_FORMAT_PRINTF_NOTHROW_2_0, ATTR_PRINTF_NOTHROW_2_3,
-	ATTR_FORMAT_PRINTF_NOTHROW_3_0, ATTR_FORMAT_PRINTF_NOTHROW_3_4,
-	ATTR_FORMAT_PRINTF_NOTHROW_4_0, ATTR_PRINTF_NOTHROW_4_5,
-	ATTR_FORMAT_PRINTF_NOTHROW_5_0, ATTR_FORMAT_PRINTF_NOTHROW_5_6,
-	ATTR_FORMAT_SCANF_NOTHROW_2_0, ATTR_FORMAT_SCANF_NOTHROW_2_3,
-	ATTR_FORMAT_STRFTIME_NOTHROW_3_0, ATTR_FORMAT_STRFMON_NOTHROW_3_4):
-	New.
-	(ATTR_FORMAT_PRINTF_4_0, ATTR_PRINTF_4_5, ATTR_FORMAT_PRINTF_5_0,
-	ATTR_FORMAT_PRINTF_5_6, ATTR_FORMAT_STRFTIME_3_0,
-	ATTR_FORMAT_NOTHROW_3_4): Remove.
-	* builtins.def (snprintf, sprintf, sscanf, vsnprintf, vsprintf,
-	vsscanf, strfmon, strftime, __snprintf_chk, __sprintf_chk,
-	__vsnprintf_chk, __vsprintf_chk): Use ATTR_FORMAT_*_NOTHROW_*
-	instead of ATTR_FORMAT_*_*.
-
---- gcc/builtin-attrs.def.jj	2007-08-13 15:11:18.000000000 +0200
-+++ gcc/builtin-attrs.def	2007-09-03 11:26:13.000000000 +0200
-@@ -118,6 +118,8 @@ DEF_ATTR_TREE_LIST (ATTR_NONNULL_LIST, A
- DEF_ATTR_TREE_LIST (ATTR_NONNULL_1, ATTR_NONNULL, ATTR_LIST_1, ATTR_NULL)
- /* Functions whose second parameter is a nonnull pointer.  */
- DEF_ATTR_TREE_LIST (ATTR_NONNULL_2, ATTR_NONNULL, ATTR_LIST_2, ATTR_NULL)
-+/* Functions whose third parameter is a nonnull pointer.  */
-+DEF_ATTR_TREE_LIST (ATTR_NONNULL_3, ATTR_NONNULL, ATTR_LIST_3, ATTR_NULL)
- /* Nothrow functions with the sentinel(1) attribute. */
- DEF_ATTR_TREE_LIST (ATTR_NOTHROW_SENTINEL_1, ATTR_SENTINEL, ATTR_LIST_1, \
- 			ATTR_NOTHROW_LIST)
-@@ -157,24 +159,38 @@ DEF_ATTR_TREE_LIST (ATTR_MALLOC_NOTHROW_
-   DEF_ATTR_TREE_LIST (ATTR_##TYPE##_##VALUES, ATTR_NULL,		 \
- 		      ATTR_##TYPE, ATTR_LIST_##VALUES)			 \
-   DEF_ATTR_TREE_LIST (ATTR_FORMAT_##TYPE##_##VALUES, ATTR_FORMAT,	 \
-+		      ATTR_##TYPE##_##VALUES, ATTR_NONNULL_##FA)
-+#define DEF_FORMAT_ATTRIBUTE_NOTHROW(TYPE, FA, VALUES)			 \
-+  DEF_ATTR_TREE_LIST (ATTR_##TYPE##_##VALUES, ATTR_NULL,		 \
-+		      ATTR_##TYPE, ATTR_LIST_##VALUES)			 \
-+  DEF_ATTR_TREE_LIST (ATTR_FORMAT_##TYPE##_NOTHROW_##VALUES, ATTR_FORMAT,\
-+		      ATTR_##TYPE##_##VALUES, ATTR_NOTHROW_NONNULL_##FA)
-+#define DEF_FORMAT_ATTRIBUTE_BOTH(TYPE, FA, VALUES)			 \
-+  DEF_ATTR_TREE_LIST (ATTR_##TYPE##_##VALUES, ATTR_NULL,		 \
-+		      ATTR_##TYPE, ATTR_LIST_##VALUES)			 \
-+  DEF_ATTR_TREE_LIST (ATTR_FORMAT_##TYPE##_##VALUES, ATTR_FORMAT,	 \
-+		      ATTR_##TYPE##_##VALUES, ATTR_NONNULL_##FA)	 \
-+  DEF_ATTR_TREE_LIST (ATTR_FORMAT_##TYPE##_NOTHROW_##VALUES, ATTR_FORMAT,\
- 		      ATTR_##TYPE##_##VALUES, ATTR_NOTHROW_NONNULL_##FA)
- DEF_FORMAT_ATTRIBUTE(PRINTF,1,1_0)
- DEF_FORMAT_ATTRIBUTE(PRINTF,1,1_2)
--DEF_FORMAT_ATTRIBUTE(PRINTF,2,2_0)
--DEF_FORMAT_ATTRIBUTE(PRINTF,2,2_3)
--DEF_FORMAT_ATTRIBUTE(PRINTF,3,3_0)
--DEF_FORMAT_ATTRIBUTE(PRINTF,3,3_4)
--DEF_FORMAT_ATTRIBUTE(PRINTF,4,4_0)
--DEF_FORMAT_ATTRIBUTE(PRINTF,4,4_5)
--DEF_FORMAT_ATTRIBUTE(PRINTF,5,5_0)
--DEF_FORMAT_ATTRIBUTE(PRINTF,5,5_6)
-+DEF_FORMAT_ATTRIBUTE_BOTH(PRINTF,2,2_0)
-+DEF_FORMAT_ATTRIBUTE_BOTH(PRINTF,2,2_3)
-+DEF_FORMAT_ATTRIBUTE_BOTH(PRINTF,3,3_0)
-+DEF_FORMAT_ATTRIBUTE_BOTH(PRINTF,3,3_4)
-+DEF_FORMAT_ATTRIBUTE_NOTHROW(PRINTF,4,4_0)
-+DEF_FORMAT_ATTRIBUTE_NOTHROW(PRINTF,4,4_5)
-+DEF_FORMAT_ATTRIBUTE_NOTHROW(PRINTF,5,5_0)
-+DEF_FORMAT_ATTRIBUTE_NOTHROW(PRINTF,5,5_6)
- DEF_FORMAT_ATTRIBUTE(SCANF,1,1_0)
- DEF_FORMAT_ATTRIBUTE(SCANF,1,1_2)
--DEF_FORMAT_ATTRIBUTE(SCANF,2,2_0)
--DEF_FORMAT_ATTRIBUTE(SCANF,2,2_3)
--DEF_FORMAT_ATTRIBUTE(STRFTIME,3,3_0)
--DEF_FORMAT_ATTRIBUTE(STRFMON,3,3_4)
-+DEF_FORMAT_ATTRIBUTE_BOTH(SCANF,2,2_0)
-+DEF_FORMAT_ATTRIBUTE_BOTH(SCANF,2,2_3)
-+DEF_FORMAT_ATTRIBUTE_NOTHROW(STRFTIME,3,3_0)
-+DEF_FORMAT_ATTRIBUTE_NOTHROW(STRFMON,3,3_4)
- #undef DEF_FORMAT_ATTRIBUTE
-+#undef DEF_FORMAT_ATTRIBUTE_NOTHROW
-+#undef DEF_FORMAT_ATTRIBUTE_BOTH
- 
- /* Construct a tree for a format_arg attribute.  */
- #define DEF_FORMAT_ARG_ATTRIBUTE(FA)					\
---- gcc/builtins.def.jj	2007-08-30 18:46:59.000000000 +0200
-+++ gcc/builtins.def	2007-09-03 10:52:18.000000000 +0200
-@@ -553,16 +553,16 @@ DEF_EXT_LIB_BUILTIN    (BUILT_IN_PUTCHAR
- DEF_LIB_BUILTIN        (BUILT_IN_PUTS, "puts", BT_FN_INT_CONST_STRING, ATTR_NONNULL_LIST)
- DEF_EXT_LIB_BUILTIN    (BUILT_IN_PUTS_UNLOCKED, "puts_unlocked", BT_FN_INT_CONST_STRING, ATTR_NONNULL_LIST)
- DEF_LIB_BUILTIN        (BUILT_IN_SCANF, "scanf", BT_FN_INT_CONST_STRING_VAR, ATTR_FORMAT_SCANF_1_2)
--DEF_C99_BUILTIN        (BUILT_IN_SNPRINTF, "snprintf", BT_FN_INT_STRING_SIZE_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_3_4)
--DEF_LIB_BUILTIN        (BUILT_IN_SPRINTF, "sprintf", BT_FN_INT_STRING_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_2_3)
--DEF_LIB_BUILTIN        (BUILT_IN_SSCANF, "sscanf", BT_FN_INT_CONST_STRING_CONST_STRING_VAR, ATTR_FORMAT_SCANF_2_3)
-+DEF_C99_BUILTIN        (BUILT_IN_SNPRINTF, "snprintf", BT_FN_INT_STRING_SIZE_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_NOTHROW_3_4)
-+DEF_LIB_BUILTIN        (BUILT_IN_SPRINTF, "sprintf", BT_FN_INT_STRING_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_NOTHROW_2_3)
-+DEF_LIB_BUILTIN        (BUILT_IN_SSCANF, "sscanf", BT_FN_INT_CONST_STRING_CONST_STRING_VAR, ATTR_FORMAT_SCANF_NOTHROW_2_3)
- DEF_LIB_BUILTIN        (BUILT_IN_VFPRINTF, "vfprintf", BT_FN_INT_FILEPTR_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_2_0)
- DEF_C99_BUILTIN        (BUILT_IN_VFSCANF, "vfscanf", BT_FN_INT_FILEPTR_CONST_STRING_VALIST_ARG, ATTR_FORMAT_SCANF_2_0)
- DEF_LIB_BUILTIN        (BUILT_IN_VPRINTF, "vprintf", BT_FN_INT_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_1_0)
- DEF_C99_BUILTIN        (BUILT_IN_VSCANF, "vscanf", BT_FN_INT_CONST_STRING_VALIST_ARG, ATTR_FORMAT_SCANF_1_0)
--DEF_C99_BUILTIN        (BUILT_IN_VSNPRINTF, "vsnprintf", BT_FN_INT_STRING_SIZE_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_3_0)
--DEF_LIB_BUILTIN        (BUILT_IN_VSPRINTF, "vsprintf", BT_FN_INT_STRING_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_2_0)
--DEF_C99_BUILTIN        (BUILT_IN_VSSCANF, "vsscanf", BT_FN_INT_CONST_STRING_CONST_STRING_VALIST_ARG, ATTR_FORMAT_SCANF_2_0)
-+DEF_C99_BUILTIN        (BUILT_IN_VSNPRINTF, "vsnprintf", BT_FN_INT_STRING_SIZE_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_NOTHROW_3_0)
-+DEF_LIB_BUILTIN        (BUILT_IN_VSPRINTF, "vsprintf", BT_FN_INT_STRING_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_NOTHROW_2_0)
-+DEF_C99_BUILTIN        (BUILT_IN_VSSCANF, "vsscanf", BT_FN_INT_CONST_STRING_CONST_STRING_VALIST_ARG, ATTR_FORMAT_SCANF_NOTHROW_2_0)
- 
- /* Category: ctype builtins.  */
- DEF_LIB_BUILTIN        (BUILT_IN_ISALNUM, "isalnum", BT_FN_INT_INT, ATTR_PURE_NOTHROW_LIST)
-@@ -693,8 +693,8 @@ DEF_GCC_BUILTIN        (BUILT_IN_RETURN_
- DEF_GCC_BUILTIN        (BUILT_IN_SAVEREGS, "saveregs", BT_FN_PTR_VAR, ATTR_NULL)
- DEF_GCC_BUILTIN        (BUILT_IN_SETJMP, "setjmp", BT_FN_INT_PTR, ATTR_NULL)
- DEF_GCC_BUILTIN        (BUILT_IN_STDARG_START, "stdarg_start", BT_FN_VOID_VALIST_REF_VAR, ATTR_NULL)
--DEF_EXT_LIB_BUILTIN    (BUILT_IN_STRFMON, "strfmon", BT_FN_SSIZE_STRING_SIZE_CONST_STRING_VAR, ATTR_FORMAT_STRFMON_3_4)
--DEF_LIB_BUILTIN        (BUILT_IN_STRFTIME, "strftime", BT_FN_SIZE_STRING_SIZE_CONST_STRING_CONST_PTR, ATTR_FORMAT_STRFTIME_3_0)
-+DEF_EXT_LIB_BUILTIN    (BUILT_IN_STRFMON, "strfmon", BT_FN_SSIZE_STRING_SIZE_CONST_STRING_VAR, ATTR_FORMAT_STRFMON_NOTHROW_3_4)
-+DEF_LIB_BUILTIN        (BUILT_IN_STRFTIME, "strftime", BT_FN_SIZE_STRING_SIZE_CONST_STRING_CONST_PTR, ATTR_FORMAT_STRFTIME_NOTHROW_3_0)
- DEF_GCC_BUILTIN        (BUILT_IN_TRAP, "trap", BT_FN_VOID, ATTR_NORETURN_NOTHROW_LIST)
- DEF_GCC_BUILTIN        (BUILT_IN_UNWIND_INIT, "unwind_init", BT_FN_VOID, ATTR_NULL)
- DEF_GCC_BUILTIN        (BUILT_IN_UPDATE_SETJMP_BUF, "update_setjmp_buf", BT_FN_VOID_PTR_INT, ATTR_NULL)
-@@ -730,10 +730,10 @@ DEF_EXT_LIB_BUILTIN    (BUILT_IN_STRCAT_
- DEF_EXT_LIB_BUILTIN    (BUILT_IN_STRCPY_CHK, "__strcpy_chk", BT_FN_STRING_STRING_CONST_STRING_SIZE, ATTR_NOTHROW_NONNULL)
- DEF_EXT_LIB_BUILTIN    (BUILT_IN_STRNCAT_CHK, "__strncat_chk", BT_FN_STRING_STRING_CONST_STRING_SIZE_SIZE, ATTR_NOTHROW_NONNULL)
- DEF_EXT_LIB_BUILTIN    (BUILT_IN_STRNCPY_CHK, "__strncpy_chk", BT_FN_STRING_STRING_CONST_STRING_SIZE_SIZE, ATTR_NOTHROW_NONNULL)
--DEF_EXT_LIB_BUILTIN    (BUILT_IN_SNPRINTF_CHK, "__snprintf_chk", BT_FN_INT_STRING_SIZE_INT_SIZE_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_5_6)
--DEF_EXT_LIB_BUILTIN    (BUILT_IN_SPRINTF_CHK, "__sprintf_chk", BT_FN_INT_STRING_INT_SIZE_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_4_5)
--DEF_EXT_LIB_BUILTIN    (BUILT_IN_VSNPRINTF_CHK, "__vsnprintf_chk", BT_FN_INT_STRING_SIZE_INT_SIZE_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_5_0)
--DEF_EXT_LIB_BUILTIN    (BUILT_IN_VSPRINTF_CHK, "__vsprintf_chk", BT_FN_INT_STRING_INT_SIZE_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_4_0)
-+DEF_EXT_LIB_BUILTIN    (BUILT_IN_SNPRINTF_CHK, "__snprintf_chk", BT_FN_INT_STRING_SIZE_INT_SIZE_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_NOTHROW_5_6)
-+DEF_EXT_LIB_BUILTIN    (BUILT_IN_SPRINTF_CHK, "__sprintf_chk", BT_FN_INT_STRING_INT_SIZE_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_NOTHROW_4_5)
-+DEF_EXT_LIB_BUILTIN    (BUILT_IN_VSNPRINTF_CHK, "__vsnprintf_chk", BT_FN_INT_STRING_SIZE_INT_SIZE_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_NOTHROW_5_0)
-+DEF_EXT_LIB_BUILTIN    (BUILT_IN_VSPRINTF_CHK, "__vsprintf_chk", BT_FN_INT_STRING_INT_SIZE_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_NOTHROW_4_0)
- DEF_EXT_LIB_BUILTIN    (BUILT_IN_FPRINTF_CHK, "__fprintf_chk", BT_FN_INT_FILEPTR_INT_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_3_4)
- DEF_EXT_LIB_BUILTIN    (BUILT_IN_PRINTF_CHK, "__printf_chk", BT_FN_INT_INT_CONST_STRING_VAR, ATTR_FORMAT_PRINTF_2_3)
- DEF_EXT_LIB_BUILTIN    (BUILT_IN_VFPRINTF_CHK, "__vfprintf_chk", BT_FN_INT_FILEPTR_INT_CONST_STRING_VALIST_ARG, ATTR_FORMAT_PRINTF_3_0)
diff --git a/gcc41-builtin-va-arg-pack-len.patch b/gcc41-builtin-va-arg-pack-len.patch
deleted file mode 100644
index 672cecd..0000000
--- a/gcc41-builtin-va-arg-pack-len.patch
+++ /dev/null
@@ -1,368 +0,0 @@
-2007-09-11  Jakub Jelinek  <jakub@redhat.com>
-
-	* builtins.def (BUILT_IN_VA_ARG_PACK_LEN): New builtin.
-	* builtins.c (expand_builtin) <case BUILT_IN_VA_ARG_PACK_LEN>: Issue
-	error if __builtin_va_arg_pack_len () wasn't optimized out during
-	inlining.
-	* tree-inline.c (copy_bb): Replace __builtin_va_arg_pack_len ()
-	with the number of inline's anonymous arguments.
-	* doc/extend.texi: Document __builtin_va_arg_pack_len ().
-
-	* gcc.dg/va-arg-pack-len-1.c: New test.
-	* g++.dg/va-arg-pack-len-1.C: New test.
-
---- gcc/builtins.def.jj	2007-09-06 10:19:24.000000000 +0200
-+++ gcc/builtins.def	2007-09-14 21:07:41.000000000 +0200
-@@ -666,6 +666,7 @@ DEF_GCC_BUILTIN        (BUILT_IN_VA_COPY
- DEF_GCC_BUILTIN        (BUILT_IN_VA_END, "va_end", BT_FN_VOID_VALIST_REF, ATTR_NULL)
- DEF_GCC_BUILTIN        (BUILT_IN_VA_START, "va_start", BT_FN_VOID_VALIST_REF_VAR, ATTR_NULL)
- DEF_GCC_BUILTIN        (BUILT_IN_VA_ARG_PACK, "va_arg_pack", BT_FN_INT, ATTR_PURE_NOTHROW_LIST)
-+DEF_GCC_BUILTIN        (BUILT_IN_VA_ARG_PACK_LEN, "va_arg_pack_len", BT_FN_INT, ATTR_PURE_NOTHROW_LIST)
- DEF_EXT_LIB_BUILTIN    (BUILT_IN__EXIT, "_exit", BT_FN_VOID_INT, ATTR_NORETURN_NOTHROW_LIST)
- DEF_C99_BUILTIN        (BUILT_IN__EXIT2, "_Exit", BT_FN_VOID_INT, ATTR_NORETURN_NOTHROW_LIST)
- 
---- gcc/builtins.c.jj	2007-09-06 11:04:56.000000000 +0200
-+++ gcc/builtins.c	2007-09-14 21:07:41.000000000 +0200
-@@ -5985,6 +5985,12 @@ expand_builtin (tree exp, rtx target, rt
-       error ("invalid use of %<__builtin_va_arg_pack ()%>");
-       return const0_rtx;
- 
-+    case BUILT_IN_VA_ARG_PACK_LEN:
-+      /* All valid uses of __builtin_va_arg_pack_len () are removed during
-+	 inlining.  */
-+      error ("invalid use of %<__builtin_va_arg_pack_len ()%>");
-+      return const0_rtx;
-+
-       /* Return the address of the first anonymous stack arg.  */
-     case BUILT_IN_NEXT_ARG:
-       if (fold_builtin_next_arg (arglist))
---- gcc/tree-inline.c.jj	2007-09-06 12:35:57.000000000 +0200
-+++ gcc/tree-inline.c	2007-09-14 21:22:48.000000000 +0200
-@@ -741,6 +741,37 @@ copy_bb (copy_body_data *id, basic_block
- 	      *a = copy_list (arglist);
- 	      CALL_EXPR_VA_ARG_PACK (call) = 0;
- 	    }
-+	  else if (call
-+		   && id->call_expr
-+		   && (decl = get_callee_fndecl (call))
-+		   && DECL_BUILT_IN_CLASS (decl) == BUILT_IN_NORMAL
-+		   && DECL_FUNCTION_CODE (decl)
-+		      == BUILT_IN_VA_ARG_PACK_LEN)
-+	    {
-+	      /* __builtin_va_arg_pack_len () should be replaced by
-+		 the number of anonymous arguments.  */
-+	      int nargs;
-+	      tree count, *call_ptr, p, a;
-+
-+	      a = TREE_OPERAND (id->call_expr, 1);
-+	      for (p = DECL_ARGUMENTS (id->src_fn); p; p = TREE_CHAIN (p))
-+		a = TREE_CHAIN (a);
-+
-+              for (nargs = 0; a; a = TREE_CHAIN (a))
-+		nargs++;
-+
-+	      count = build_int_cst (integer_type_node, nargs);
-+	      call_ptr = &stmt;
-+	      if (TREE_CODE (*call_ptr) == MODIFY_EXPR)
-+		call_ptr = &TREE_OPERAND (*call_ptr, 1);
-+	      if (TREE_CODE (*call_ptr) == WITH_SIZE_EXPR)
-+		call_ptr = &TREE_OPERAND (*call_ptr, 0);
-+	      gcc_assert (*call_ptr == call && call_ptr != &stmt);
-+	      *call_ptr = count;
-+	      update_stmt (stmt);
-+	      call = NULL_TREE;
-+	    }
-+
- 	  /* We're duplicating a CALL_EXPR.  Find any corresponding
- 	     callgraph edges and update or duplicate them.  */
- 	  if (call && (decl = get_callee_fndecl (call)))
---- gcc/doc/extend.texi.jj	2007-09-06 10:19:25.000000000 +0200
-+++ gcc/doc/extend.texi	2007-09-14 21:07:41.000000000 +0200
-@@ -578,6 +578,41 @@ myprintf (FILE *f, const char *format, .
- @end smallexample
- @end deftypefn
- 
-+@deftypefn {Built-in Function} __builtin_va_arg_pack_len ()
-+This built-in function returns the number of anonymous arguments of
-+an inline function.  It can be used only in inline functions which
-+will be always inlined, never compiled as a separate function, such
-+as those using @code{__attribute__ ((__always_inline__))} or
-+@code{__attribute__ ((__gnu_inline__))} extern inline functions.
-+For example following will do link or runtime checking of open
-+arguments for optimized code:
-+@smallexample
-+#ifdef __OPTIMIZE__
-+extern inline __attribute__((__gnu_inline__)) int
-+myopen (const char *path, int oflag, ...)
-+@{
-+  if (__builtin_va_arg_pack_len () > 1)
-+    warn_open_too_many_arguments ();
-+
-+  if (__builtin_constant_p (oflag))
-+    @{
-+      if ((oflag & O_CREAT) != 0 && __builtin_va_arg_pack_len () < 1)
-+        @{
-+          warn_open_missing_mode ();
-+          return __open_2 (path, oflag);
-+        @}
-+      return open (path, oflag, __builtin_va_arg_pack ());
-+    @}
-+    
-+  if (__builtin_va_arg_pack_len () < 1)
-+    return __open_2 (path, oflag);
-+
-+  return open (path, oflag, __builtin_va_arg_pack ());
-+@}
-+#endif
-+@end smallexample
-+@end deftypefn
-+
- @node Typeof
- @section Referring to a Type with @code{typeof}
- @findex typeof
---- gcc/testsuite/gcc.dg/va-arg-pack-len-1.c.jj	2007-09-14 21:07:41.000000000 +0200
-+++ gcc/testsuite/gcc.dg/va-arg-pack-len-1.c	2007-09-14 21:07:41.000000000 +0200
-@@ -0,0 +1,120 @@
-+/* { dg-do run } */
-+/* { dg-options "-O2" } */
-+
-+#include <stdarg.h>
-+
-+extern int warn_open_missing_mode (void);
-+extern int warn_open_too_many_arguments (void);
-+extern void abort (void);
-+
-+char expected_char;
-+
-+__attribute__((noinline)) int
-+myopen2 (const char *path, int oflag)
-+{
-+  if (expected_char++ != path[0] || path[1] != '\0')
-+    abort ();
-+  switch (path[0])
-+    {
-+    case 'f':
-+      if (oflag != 0x2)
-+	abort ();
-+      break;
-+    case 'g':
-+      if (oflag != 0x43)
-+	abort ();
-+      /* In real __open_2 this would terminate the program:
-+	 open with O_CREAT without third argument.  */
-+      return -6;
-+    default:
-+      abort ();
-+    }
-+  return 0;
-+}
-+
-+__attribute__((noinline)) int
-+myopenva (const char *path, int oflag, ...)
-+{
-+  int mode = 0;
-+  va_list ap;
-+  if ((oflag & 0x40) != 0)
-+    {
-+      va_start (ap, oflag);
-+      mode = va_arg (ap, int);
-+      va_end (ap);
-+    }
-+  if (expected_char++ != path[0] || path[1] != '\0')
-+    abort ();
-+  switch (path[0])
-+    {
-+    case 'a':
-+      if (oflag != 0x43 || mode != 0644)
-+	abort ();
-+      break;
-+    case 'b':
-+      if (oflag != 0x3)
-+	abort ();
-+      break;
-+    case 'c':
-+      if (oflag != 0x2)
-+	abort ();
-+      break;
-+    case 'd':
-+      if (oflag != 0x43 || mode != 0600)
-+	abort ();
-+      break;
-+    case 'e':
-+      if (oflag != 0x3)
-+	abort ();
-+      break;
-+    default:
-+      abort ();
-+    }
-+  return 0;
-+}
-+
-+extern inline __attribute__((always_inline, gnu_inline)) int
-+myopen (const char *path, int oflag, ...)
-+{
-+  if (__builtin_va_arg_pack_len () > 1)
-+    warn_open_too_many_arguments ();
-+
-+  if (__builtin_constant_p (oflag))
-+    {
-+      if ((oflag & 0x40) != 0 && __builtin_va_arg_pack_len () < 1)
-+	{
-+	  warn_open_missing_mode ();
-+	  return myopen2 (path, oflag);
-+	}
-+      return myopenva (path, oflag, __builtin_va_arg_pack ());
-+    }
-+
-+  if (__builtin_va_arg_pack_len () < 1)
-+    return myopen2 (path, oflag);
-+
-+  return myopenva (path, oflag, __builtin_va_arg_pack ());
-+}
-+
-+volatile int l0;
-+
-+int
-+main (void)
-+{
-+  expected_char = 'a';
-+  if (myopen ("a", 0x43, 0644))
-+    abort ();
-+  if (myopen ("b", 0x3, 0755))
-+    abort ();
-+  if (myopen ("c", 0x2))
-+    abort ();
-+  if (myopen ("d", l0 + 0x43, 0600))
-+    abort ();
-+  if (myopen ("e", l0 + 0x3, 0700))
-+    abort ();
-+  if (myopen ("f", l0 + 0x2))
-+    abort ();
-+  /* Invalid use of myopen, but only detectable at runtime.  */
-+  if (myopen ("g", l0 + 0x43) != -6)
-+    abort ();
-+  return 0;
-+}
---- gcc/testsuite/g++.dg/ext/va-arg-pack-len-1.C.jj	2007-09-14 21:07:41.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/va-arg-pack-len-1.C	2007-09-14 21:07:41.000000000 +0200
-@@ -0,0 +1,120 @@
-+// { dg-do run }
-+// { dg-options "-O2" }
-+
-+#include <stdarg.h>
-+
-+extern "C" int warn_open_missing_mode (void);
-+extern "C" int warn_open_too_many_arguments (void);
-+extern "C" void abort (void);
-+
-+char expected_char;
-+
-+__attribute__((noinline)) int
-+myopen2 (const char *path, int oflag)
-+{
-+  if (expected_char++ != path[0] || path[1] != '\0')
-+    abort ();
-+  switch (path[0])
-+    {
-+    case 'f':
-+      if (oflag != 0x2)
-+	abort ();
-+      break;
-+    case 'g':
-+      if (oflag != 0x43)
-+	abort ();
-+      // In real __open_2 this would terminate the program:
-+      // open with O_CREAT without third argument.
-+      return -6;
-+    default:
-+      abort ();
-+    }
-+  return 0;
-+}
-+
-+__attribute__((noinline)) int
-+myopenva (const char *path, int oflag, ...)
-+{
-+  int mode = 0;
-+  va_list ap;
-+  if ((oflag & 0x40) != 0)
-+    {
-+      va_start (ap, oflag);
-+      mode = va_arg (ap, int);
-+      va_end (ap);
-+    }
-+  if (expected_char++ != path[0] || path[1] != '\0')
-+    abort ();
-+  switch (path[0])
-+    {
-+    case 'a':
-+      if (oflag != 0x43 || mode != 0644)
-+	abort ();
-+      break;
-+    case 'b':
-+      if (oflag != 0x3)
-+	abort ();
-+      break;
-+    case 'c':
-+      if (oflag != 0x2)
-+	abort ();
-+      break;
-+    case 'd':
-+      if (oflag != 0x43 || mode != 0600)
-+	abort ();
-+      break;
-+    case 'e':
-+      if (oflag != 0x3)
-+	abort ();
-+      break;
-+    default:
-+      abort ();
-+    }
-+  return 0;
-+}
-+
-+extern inline __attribute__((always_inline, gnu_inline)) int
-+myopen (const char *path, int oflag, ...)
-+{
-+  if (__builtin_va_arg_pack_len () > 1)
-+    warn_open_too_many_arguments ();
-+
-+  if (__builtin_constant_p (oflag))
-+    {
-+      if ((oflag & 0x40) != 0 && __builtin_va_arg_pack_len () < 1)
-+	{
-+	  warn_open_missing_mode ();
-+	  return myopen2 (path, oflag);
-+	}
-+      return myopenva (path, oflag, __builtin_va_arg_pack ());
-+    }
-+
-+  if (__builtin_va_arg_pack_len () < 1)
-+    return myopen2 (path, oflag);
-+
-+  return myopenva (path, oflag, __builtin_va_arg_pack ());
-+}
-+
-+volatile int l0;
-+
-+int
-+main (void)
-+{
-+  expected_char = 'a';
-+  if (myopen ("a", 0x43, 0644))
-+    abort ();
-+  if (myopen ("b", 0x3, 0755))
-+    abort ();
-+  if (myopen ("c", 0x2))
-+    abort ();
-+  if (myopen ("d", l0 + 0x43, 0600))
-+    abort ();
-+  if (myopen ("e", l0 + 0x3, 0700))
-+    abort ();
-+  if (myopen ("f", l0 + 0x2))
-+    abort ();
-+  // Invalid use of myopen, but only detectable at runtime.
-+  if (myopen ("g", l0 + 0x43) != -6)
-+    abort ();
-+  return 0;
-+}
diff --git a/gcc41-builtin-va-arg-pack.patch b/gcc41-builtin-va-arg-pack.patch
deleted file mode 100644
index 177e94b..0000000
--- a/gcc41-builtin-va-arg-pack.patch
+++ /dev/null
@@ -1,722 +0,0 @@
-2007-09-07  Jakub Jelinek  <jakub@redhat.com>
-
-	* tree.c (cp_cannot_inline_tree_fn): Don't mark varargs_function_p
-	as uninlinable.
-
-	* g++.dg/ext/va-arg-pack-1.C: New test.
-	* g++.dg/ext/va-arg-pack-2.C: New test.
-
-2007-09-05  Jakub Jelinek  <jakub@redhat.com>
-
-	* builtins.def (BUILT_IN_VA_ARG_PACK): New built-in.
-	* tree.h (CALL_EXPR_VA_ARG_PACK): Define.
-	* tree-inline.h (copy_body_data): Add call_expr field.
-	* tree-inline.c (expand_call_inline): Initialize call_expr.
-	(copy_bb): Append anonymous inline fn arguments to arguments
-	when inlining a CALL_EXPR_VA_ARG_PACK call.
-	* builtins.c (expand_builtin): Issue an error if
-	BUILT_IN_VA_ARG_PACK is seen during expand.
-	(fold_builtin_1): Don't fold calls with
-	__builtin_va_arg_pack () call as last argument.
-	* gimplify.c (gimplify_call_expr): If last argument to a vararg
-	function is __builtin_va_arg_pack (), decrease number of call
-	arguments and instead set CALL_EXPR_VA_ARG_PACK on the CALL_EXPR.
-	* fold-const.c (fold): Don't fold CALL_EXPRs with
-	CALL_EXPR_VA_ARG_PACK bit set.
-	* expr.c (expand_expr_real_1): Issue an error if
-	CALL_EXPR_VA_ARG_PACK CALL_EXPR is seen during expand.
-	* tree-pretty-print.c (dump_generic_node): Handle printing
-	CALL_EXPR_VA_ARG_PACK bit on CALL_EXPRs.
-	* doc/extend.texi (__builtin_va_arg_pack): Document.
-
-	* gcc.c-torture/execute/va-arg-pack-1.c: New test.
-	* gcc.dg/va-arg-pack-1.c: New test.
-	* gcc.dg/va-arg-pack-1a.c: New test.
-
---- gcc/builtins.def.jj	2007-09-06 10:19:09.000000000 +0200
-+++ gcc/builtins.def	2007-09-06 10:19:24.000000000 +0200
-@@ -665,6 +665,7 @@ DEF_GCC_BUILTIN        (BUILT_IN_UPDATE_
- DEF_GCC_BUILTIN        (BUILT_IN_VA_COPY, "va_copy", BT_FN_VOID_VALIST_REF_VALIST_ARG, ATTR_NULL)
- DEF_GCC_BUILTIN        (BUILT_IN_VA_END, "va_end", BT_FN_VOID_VALIST_REF, ATTR_NULL)
- DEF_GCC_BUILTIN        (BUILT_IN_VA_START, "va_start", BT_FN_VOID_VALIST_REF_VAR, ATTR_NULL)
-+DEF_GCC_BUILTIN        (BUILT_IN_VA_ARG_PACK, "va_arg_pack", BT_FN_INT, ATTR_PURE_NOTHROW_LIST)
- DEF_EXT_LIB_BUILTIN    (BUILT_IN__EXIT, "_exit", BT_FN_VOID_INT, ATTR_NORETURN_NOTHROW_LIST)
- DEF_C99_BUILTIN        (BUILT_IN__EXIT2, "_Exit", BT_FN_VOID_INT, ATTR_NORETURN_NOTHROW_LIST)
- 
---- gcc/tree.h.jj	2007-02-20 22:39:12.000000000 +0100
-+++ gcc/tree.h	2007-09-06 11:50:56.000000000 +0200
-@@ -411,6 +411,8 @@ struct tree_common GTY(())
-            VAR_DECL or FUNCTION_DECL or IDENTIFIER_NODE
-        ASM_VOLATILE_P in
-            ASM_EXPR
-+       CALL_EXPR_VA_ARG_PACK in
-+	  CALL_EXPR
-        TYPE_CACHED_VALUES_P in
-           ..._TYPE
-        SAVE_EXPR_RESOLVED_P in
-@@ -1066,6 +1068,11 @@ extern void omp_clause_range_check_faile
- #define SAVE_EXPR_RESOLVED_P(NODE) \
-   (TREE_CHECK (NODE, SAVE_EXPR)->common.public_flag)
- 
-+/* Set on a CALL_EXPR if this stdarg call should be passed the argument
-+   pack.  */
-+#define CALL_EXPR_VA_ARG_PACK(NODE) \
-+  (CALL_EXPR_CHECK(NODE)->common.public_flag)
-+
- /* In any expression, decl, or constant, nonzero means it has side effects or
-    reevaluation of the whole expression could produce a different value.
-    This is set if any subexpression is a function call, a side effect or a
---- gcc/tree-inline.h.jj	2007-02-20 22:39:13.000000000 +0100
-+++ gcc/tree-inline.h	2007-09-06 10:19:24.000000000 +0200
-@@ -57,6 +57,10 @@ typedef struct copy_body_data
-   /* Current BLOCK.  */
-   tree block;
- 
-+  /* CALL_EXPR if va arg parameter packs should be expanded or NULL
-+     is not.  */
-+  tree call_expr;
-+
-   /* Exception region the inlined call lie in.  */
-   int eh_region;
-   /* Take region number in the function being copied, add this value and
---- gcc/tree-inline.c.jj	2007-04-26 09:41:49.000000000 +0200
-+++ gcc/tree-inline.c	2007-09-06 12:35:57.000000000 +0200
-@@ -725,6 +725,22 @@ copy_bb (copy_body_data *id, basic_block
- 
-           bsi_insert_after (&copy_bsi, stmt, BSI_NEW_STMT);
- 	  call = get_call_expr_in (stmt);
-+	  if (call && CALL_EXPR_VA_ARG_PACK (call) && id->call_expr)
-+	    {
-+	      tree arglist, *a, p;
-+	      TREE_OPERAND (call, 1) = copy_list (TREE_OPERAND (call, 1));
-+
-+	      for (a = &TREE_OPERAND (call, 1); *a; a = &TREE_CHAIN (*a))
-+		;
-+
-+	      p = DECL_ARGUMENTS (id->src_fn);
-+	      for (arglist = TREE_OPERAND (id->call_expr, 1);
-+		   p; p = TREE_CHAIN (p), arglist = TREE_CHAIN (arglist))
-+		;
-+
-+	      *a = copy_list (arglist);
-+	      CALL_EXPR_VA_ARG_PACK (call) = 0;
-+	    }
- 	  /* We're duplicating a CALL_EXPR.  Find any corresponding
- 	     callgraph edges and update or duplicate them.  */
- 	  if (call && (decl = get_callee_fndecl (call)))
-@@ -2085,6 +2101,7 @@ expand_call_inline (basic_block bb, tree
-   /* Record the function we are about to inline.  */
-   id->src_fn = fn;
-   id->src_node = cg_edge->callee;
-+  id->call_expr = t;
- 
-   initialize_inlined_parameters (id, args, TREE_OPERAND (t, 2), fn, bb);
- 
---- gcc/builtins.c.jj	2007-08-31 09:47:46.000000000 +0200
-+++ gcc/builtins.c	2007-09-06 11:04:56.000000000 +0200
-@@ -5979,6 +5979,12 @@ expand_builtin (tree exp, rtx target, rt
-     case BUILT_IN_ARGS_INFO:
-       return expand_builtin_args_info (arglist);
- 
-+    case BUILT_IN_VA_ARG_PACK:
-+      /* All valid uses of __builtin_va_arg_pack () are removed during
-+	 inlining.  */
-+      error ("invalid use of %<__builtin_va_arg_pack ()%>");
-+      return const0_rtx;
-+
-       /* Return the address of the first anonymous stack arg.  */
-     case BUILT_IN_NEXT_ARG:
-       if (fold_builtin_next_arg (arglist))
-@@ -9017,7 +9023,27 @@ fold_builtin_1 (tree fndecl, tree arglis
- {
-   tree type = TREE_TYPE (TREE_TYPE (fndecl));
-   enum built_in_function fcode;
-+  tree a;
-+
-+  if (arglist)
-+    {
-+      for (a = arglist; TREE_CHAIN (a); a = TREE_CHAIN (a))
-+	;
- 
-+      /* Before gimplification CALL_EXPR_VA_ARG_PACK is not set, but
-+	 instead last argument is __builtin_va_arg_pack ().  Defer folding
-+	 even in that case, until arguments are finalized.  */
-+      if (TREE_CODE (TREE_VALUE (a)) == CALL_EXPR)
-+	{
-+	  tree fndecl2 = get_callee_fndecl (TREE_VALUE (a));
-+	  if (fndecl2
-+	      && TREE_CODE (fndecl2) == FUNCTION_DECL
-+	      && DECL_BUILT_IN_CLASS (fndecl2) == BUILT_IN_NORMAL
-+	      && DECL_FUNCTION_CODE (fndecl2) == BUILT_IN_VA_ARG_PACK)
-+	    return NULL_TREE;
-+	}
-+    }
-+
-   if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_MD)
-     return targetm.fold_builtin (fndecl, arglist, ignore);
- 
---- gcc/gimplify.c.jj	2007-08-31 09:47:46.000000000 +0200
-+++ gcc/gimplify.c	2007-09-06 11:42:01.000000000 +0200
-@@ -2018,6 +2018,55 @@ gimplify_call_expr (tree *expr_p, tree *
-   ret = gimplify_expr (&TREE_OPERAND (*expr_p, 0), pre_p, NULL,
- 		       is_gimple_call_addr, fb_rvalue);
- 
-+  if (ret != GS_ERROR && TREE_OPERAND (*expr_p, 1))
-+    {
-+      int nargs;
-+
-+      for (nargs = 1, arglist = TREE_OPERAND (*expr_p, 1);
-+	   TREE_CHAIN (arglist);
-+	   arglist = TREE_CHAIN (arglist))
-+	nargs++;
-+
-+      if (TREE_CODE (TREE_VALUE (arglist)) == CALL_EXPR)
-+	{
-+	  tree last_arg = TREE_VALUE (arglist);
-+	  tree last_arg_fndecl = get_callee_fndecl (last_arg);
-+
-+	  if (last_arg_fndecl
-+	      && TREE_CODE (last_arg_fndecl) == FUNCTION_DECL
-+	      && DECL_BUILT_IN_CLASS (last_arg_fndecl) == BUILT_IN_NORMAL
-+	      && DECL_FUNCTION_CODE (last_arg_fndecl) == BUILT_IN_VA_ARG_PACK)
-+	    {
-+	      tree p = NULL_TREE, *aptr;
-+	      int i;
-+
-+	      if (decl && DECL_ARGUMENTS (decl))
-+		p = DECL_ARGUMENTS (decl);
-+	      else if (decl)
-+		p = TYPE_ARG_TYPES (TREE_TYPE (decl));
-+	      else if (POINTER_TYPE_P (TREE_TYPE (TREE_OPERAND (*expr_p, 0))))
-+		p = TYPE_ARG_TYPES (TREE_TYPE (TREE_TYPE
-+						(TREE_OPERAND (*expr_p, 0))));
-+	      for (i = 0; p; p = TREE_CHAIN (p))
-+		i++;
-+
-+	      if (i < nargs)
-+		{
-+		  TREE_OPERAND (*expr_p, 1)
-+		    = copy_list (TREE_OPERAND (*expr_p, 1));
-+
-+		  for (aptr = &TREE_OPERAND (*expr_p, 1);
-+		       TREE_CHAIN (*aptr);
-+		       aptr = &TREE_CHAIN (*aptr))
-+		    ;
-+
-+		  *aptr = NULL_TREE;
-+		  CALL_EXPR_VA_ARG_PACK (*expr_p) = 1;
-+		}
-+	    }
-+	}
-+    }
-+
-   if (PUSH_ARGS_REVERSED)
-     TREE_OPERAND (*expr_p, 1) = nreverse (TREE_OPERAND (*expr_p, 1));
-   for (arglist = TREE_OPERAND (*expr_p, 1); arglist;
-@@ -2037,7 +2086,7 @@ gimplify_call_expr (tree *expr_p, tree *
-   if (ret != GS_ERROR)
-     {
-       decl = get_callee_fndecl (*expr_p);
--      if (decl && DECL_BUILT_IN (decl))
-+      if (decl && DECL_BUILT_IN (decl) && !CALL_EXPR_VA_ARG_PACK (*expr_p))
- 	{
- 	  tree arglist = TREE_OPERAND (*expr_p, 1);
- 	  tree new = fold_builtin (decl, arglist, !want_value);
---- gcc/fold-const.c.jj	2007-08-31 09:47:39.000000000 +0200
-+++ gcc/fold-const.c	2007-09-06 11:40:55.000000000 +0200
-@@ -10409,6 +10409,8 @@ fold (tree expr)
- 	  op0 = TREE_OPERAND (t, 0);
- 	  op1 = TREE_OPERAND (t, 1);
- 	  op2 = TREE_OPERAND (t, 2);
-+	  if (code == CALL_EXPR && CALL_EXPR_VA_ARG_PACK (t))
-+	    return expr;
- 	  tem = fold_ternary (code, type, op0, op1, op2);
- 	  return tem ? tem : expr;
- 	default:
---- gcc/expr.c.jj	2007-04-24 22:58:53.000000000 +0200
-+++ gcc/expr.c	2007-09-06 10:19:24.000000000 +0200
-@@ -7475,6 +7475,10 @@ expand_expr_real_1 (tree exp, rtx target
-       return expand_expr (OBJ_TYPE_REF_EXPR (exp), target, tmode, modifier);
- 
-     case CALL_EXPR:
-+      /* All valid uses of __builtin_va_arg_pack () are removed during
-+	 inlining.  */
-+      if (CALL_EXPR_VA_ARG_PACK (exp))
-+	error ("invalid use of %<__builtin_va_arg_pack ()%>");
-       /* Check for a built-in function.  */
-       if (TREE_CODE (TREE_OPERAND (exp, 0)) == ADDR_EXPR
- 	  && (TREE_CODE (TREE_OPERAND (TREE_OPERAND (exp, 0), 0))
---- gcc/doc/extend.texi.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/doc/extend.texi	2007-09-06 10:19:25.000000000 +0200
-@@ -552,6 +552,32 @@ the containing function.  You should spe
- returned by @code{__builtin_apply}.
- @end deftypefn
- 
-+@deftypefn {Built-in Function} __builtin_va_arg_pack ()
-+This built-in function represents all anonymous arguments of an inline
-+function.  It can be used only in inline functions which will be always
-+inlined, never compiled as a separate function, such as those using
-+@code{__attribute__ ((__always_inline__))} or
-+@code{__attribute__ ((__gnu_inline__))} extern inline functions.
-+It must be only passed as last argument to some other function
-+with variable arguments.  This is useful for writing small wrapper
-+inlines for variable argument functions, when using preprocessor
-+macros is undesirable.  For example:
-+@smallexample
-+extern int myprintf (FILE *f, const char *format, ...);
-+extern inline __attribute__ ((__gnu_inline__)) int
-+myprintf (FILE *f, const char *format, ...)
-+@{
-+  int r = fprintf (f, "myprintf: ");
-+  if (r < 0)
-+    return r;
-+  int s = fprintf (f, format, __builtin_va_arg_pack ());
-+  if (s < 0)
-+    return s;
-+  return r + s;
-+@}
-+@end smallexample
-+@end deftypefn
-+
- @node Typeof
- @section Referring to a Type with @code{typeof}
- @findex typeof
---- gcc/tree-pretty-print.c.jj	2007-02-20 22:39:12.000000000 +0100
-+++ gcc/tree-pretty-print.c	2007-09-06 10:24:51.000000000 +0200
-@@ -1128,6 +1128,15 @@ dump_generic_node (pretty_printer *buffe
-       op1 = TREE_OPERAND (node, 1);
-       if (op1)
- 	dump_generic_node (buffer, op1, spc, flags, false);
-+      if (CALL_EXPR_VA_ARG_PACK (node))
-+	{
-+	  if (op1)
-+	    {
-+	      pp_character (buffer, ',');
-+	      pp_space (buffer);
-+	    }
-+	  pp_string (buffer, "__builtin_va_arg_pack ()");
-+	}
-       pp_character (buffer, ')');
- 
-       op1 = TREE_OPERAND (node, 2);
---- gcc/cp/tree.c.jj	2007-03-12 08:28:01.000000000 +0100
-+++ gcc/cp/tree.c	2007-09-07 09:03:07.000000000 +0200
-@@ -2078,12 +2078,6 @@ cp_cannot_inline_tree_fn (tree* fnp)
-       return 1;
-     }
- 
--  if (varargs_function_p (fn))
--    {
--      DECL_UNINLINABLE (fn) = 1;
--      return 1;
--    }
--
-   if (! function_attribute_inlinable_p (fn))
-     {
-       DECL_UNINLINABLE (fn) = 1;
---- gcc/testsuite/gcc.c-torture/execute/va-arg-pack-1.c.jj	2007-09-06 10:19:25.000000000 +0200
-+++ gcc/testsuite/gcc.c-torture/execute/va-arg-pack-1.c	2007-09-06 10:19:25.000000000 +0200
-@@ -0,0 +1,143 @@
-+/* __builtin_va_arg_pack () builtin tests.  */
-+
-+#include <stdarg.h>
-+
-+extern void abort (void);
-+
-+int v1 = 8;
-+long int v2 = 3;
-+void *v3 = (void *) &v2;
-+struct A { char c[16]; } v4 = { "foo" };
-+long double v5 = 40;
-+char seen[20];
-+int cnt;
-+
-+__attribute__ ((noinline)) int
-+foo1 (int x, int y, ...)
-+{
-+  int i;
-+  long int l;
-+  void *v;
-+  struct A a;
-+  long double ld;
-+  va_list ap;
-+
-+  va_start (ap, y);
-+  if (x < 0 || x >= 20 || seen[x])
-+    abort ();
-+  seen[x] = ++cnt;
-+  if (y != 6)
-+    abort ();
-+  i = va_arg (ap, int);
-+  if (i != 5)
-+    abort ();
-+  switch (x)
-+    {
-+    case 0:
-+      i = va_arg (ap, int);
-+      if (i != 9 || v1 != 9)
-+	abort ();
-+      a = va_arg (ap, struct A);
-+      if (__builtin_memcmp (a.c, v4.c, sizeof (a.c)) != 0)
-+	abort ();
-+      v = (void *) va_arg (ap, struct A *);
-+      if (v != (void *) &v4)
-+	abort ();
-+      l = va_arg (ap, long int);
-+      if (l != 3 || v2 != 4)
-+	abort ();
-+      break;
-+    case 1:
-+      ld = va_arg (ap, long double);
-+      if (ld != 41 || v5 != ld)
-+	abort ();
-+      i = va_arg (ap, int);
-+      if (i != 8)
-+	abort ();
-+      v = va_arg (ap, void *);
-+      if (v != &v2)
-+	abort ();
-+      break;
-+    case 2:
-+      break;
-+    default:
-+      abort ();
-+    }
-+  va_end (ap);
-+  return x;
-+}
-+
-+__attribute__ ((noinline)) int
-+foo2 (int x, int y, ...)
-+{
-+  long long int ll;
-+  void *v;
-+  struct A a, b;
-+  long double ld;
-+  va_list ap;
-+
-+  va_start (ap, y);
-+  if (x < 0 || x >= 20 || seen[x])
-+    abort ();
-+  seen[x] = ++cnt | 64;
-+  if (y != 10)
-+    abort ();
-+  switch (x)
-+    {
-+    case 11:
-+      break;
-+    case 12:
-+      ld = va_arg (ap, long double);
-+      if (ld != 41 || v5 != 40)
-+	abort ();
-+      a = va_arg (ap, struct A);
-+      if (__builtin_memcmp (a.c, v4.c, sizeof (a.c)) != 0)
-+	abort ();
-+      b = va_arg (ap, struct A);
-+      if (__builtin_memcmp (b.c, v4.c, sizeof (b.c)) != 0)
-+	abort ();
-+      v = va_arg (ap, void *);
-+      if (v != &v2)
-+	abort ();
-+      ll = va_arg (ap, long long int);
-+      if (ll != 16LL)
-+	abort ();
-+      break;
-+    case 2:
-+      break;
-+    default:
-+      abort ();
-+    }
-+  va_end (ap);
-+  return x + 8;
-+}
-+
-+__attribute__ ((noinline)) int
-+foo3 (void)
-+{
-+  return 6;
-+}
-+
-+extern inline __attribute__ ((always_inline, gnu_inline)) int
-+bar (int x, ...)
-+{
-+  if (x < 10)
-+    return foo1 (x, foo3 (), 5, __builtin_va_arg_pack ());
-+  return foo2 (x, foo3 () + 4, __builtin_va_arg_pack ());
-+}
-+
-+int
-+main (void)
-+{
-+  if (bar (0, ++v1, v4, &v4, v2++) != 0)
-+    abort ();
-+  if (bar (1, ++v5, 8, v3) != 1)
-+    abort ();
-+  if (bar (2) != 2)
-+    abort ();
-+  if (bar (v1 + 2) != 19)
-+    abort ();
-+  if (bar (v1 + 3, v5--, v4, v4, v3, 16LL) != 20)
-+    abort ();
-+  return 0;
-+}
---- gcc/testsuite/gcc.dg/va-arg-pack-1.c.jj	2007-09-06 10:19:25.000000000 +0200
-+++ gcc/testsuite/gcc.dg/va-arg-pack-1.c	2007-09-06 12:52:00.000000000 +0200
-@@ -0,0 +1,46 @@
-+/* { dg-do compile } */
-+/* { dg-options "-O2" } */
-+
-+int bar (int, const char *, int, ...);
-+int baz (int, const char *, long int);
-+
-+extern inline __attribute__((always_inline)) int
-+f2 (int y, ...)
-+{
-+  return bar (y, "", __builtin_va_arg_pack ());		/* { dg-error "invalid use of" } */
-+}
-+
-+extern inline __attribute__((always_inline)) int
-+f3 (int y, ...)
-+{
-+  return bar (y, "", 5, __builtin_va_arg_pack ());
-+}
-+
-+extern inline __attribute__((always_inline)) int
-+f4 (int y, ...)
-+{
-+  return bar (y, "", 4, __builtin_va_arg_pack (), 6);	/* { dg-error "invalid use of" } */
-+}
-+
-+extern inline __attribute__((always_inline)) int
-+f5 (int y, ...)
-+{
-+  return baz (y, "", __builtin_va_arg_pack ());		/* { dg-error "invalid use of" } */
-+}
-+
-+extern inline __attribute__((always_inline)) int
-+f6 (int y, ...)
-+{
-+  return __builtin_va_arg_pack ();			/* { dg-error "invalid use of" } */
-+}
-+
-+int
-+test (void)
-+{
-+  int a = f2 (5, "a", 6);
-+  a += f3 (6, "ab", 17LL);
-+  a += f4 (7, 1, 2, 3);
-+  a += f5 (8, 7L);
-+  a += f6 (9);
-+  return a;
-+}
---- gcc/testsuite/gcc.dg/va-arg-pack-1a.c.jj	2007-09-06 12:51:51.000000000 +0200
-+++ gcc/testsuite/gcc.dg/va-arg-pack-1a.c	2007-09-06 12:52:09.000000000 +0200
-@@ -0,0 +1,11 @@
-+/* { dg-do compile } */
-+/* { dg-options "-O2" } */
-+
-+int bar (int, const char *, int, ...);
-+int baz (int, const char *, long int);
-+
-+int
-+f1 (int x, ...)
-+{
-+  return bar (5, "", 6, __builtin_va_arg_pack ());	/* { dg-error "invalid use of" } */
-+}
---- gcc/testsuite/g++.dg/ext/va-arg-pack-1.C.jj	2007-09-07 08:52:55.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/va-arg-pack-1.C	2007-09-07 08:53:45.000000000 +0200
-@@ -0,0 +1,145 @@
-+// __builtin_va_arg_pack () builtin tests.
-+// { dg-do run }
-+// { dg-options "-O2" }
-+
-+#include <stdarg.h>
-+
-+extern "C" void abort (void);
-+
-+int v1 = 8;
-+long int v2 = 3;
-+void *v3 = (void *) &v2;
-+struct A { char c[16]; } v4 = { "foo" };
-+long double v5 = 40;
-+char seen[20];
-+int cnt;
-+
-+__attribute__ ((noinline)) int
-+foo1 (int x, int y, ...)
-+{
-+  int i;
-+  long int l;
-+  void *v;
-+  struct A a;
-+  long double ld;
-+  va_list ap;
-+
-+  va_start (ap, y);
-+  if (x < 0 || x >= 20 || seen[x])
-+    abort ();
-+  seen[x] = ++cnt;
-+  if (y != 6)
-+    abort ();
-+  i = va_arg (ap, int);
-+  if (i != 5)
-+    abort ();
-+  switch (x)
-+    {
-+    case 0:
-+      i = va_arg (ap, int);
-+      if (i != 9 || v1 != 9)
-+	abort ();
-+      a = va_arg (ap, struct A);
-+      if (__builtin_memcmp (a.c, v4.c, sizeof (a.c)) != 0)
-+	abort ();
-+      v = (void *) va_arg (ap, struct A *);
-+      if (v != (void *) &v4)
-+	abort ();
-+      l = va_arg (ap, long int);
-+      if (l != 3 || v2 != 4)
-+	abort ();
-+      break;
-+    case 1:
-+      ld = va_arg (ap, long double);
-+      if (ld != 41 || v5 != ld)
-+	abort ();
-+      i = va_arg (ap, int);
-+      if (i != 8)
-+	abort ();
-+      v = va_arg (ap, void *);
-+      if (v != &v2)
-+	abort ();
-+      break;
-+    case 2:
-+      break;
-+    default:
-+      abort ();
-+    }
-+  va_end (ap);
-+  return x;
-+}
-+
-+__attribute__ ((noinline)) int
-+foo2 (int x, int y, ...)
-+{
-+  long long int ll;
-+  void *v;
-+  struct A a, b;
-+  long double ld;
-+  va_list ap;
-+
-+  va_start (ap, y);
-+  if (x < 0 || x >= 20 || seen[x])
-+    abort ();
-+  seen[x] = ++cnt | 64;
-+  if (y != 10)
-+    abort ();
-+  switch (x)
-+    {
-+    case 11:
-+      break;
-+    case 12:
-+      ld = va_arg (ap, long double);
-+      if (ld != 41 || v5 != 40)
-+	abort ();
-+      a = va_arg (ap, struct A);
-+      if (__builtin_memcmp (a.c, v4.c, sizeof (a.c)) != 0)
-+	abort ();
-+      b = va_arg (ap, struct A);
-+      if (__builtin_memcmp (b.c, v4.c, sizeof (b.c)) != 0)
-+	abort ();
-+      v = va_arg (ap, void *);
-+      if (v != &v2)
-+	abort ();
-+      ll = va_arg (ap, long long int);
-+      if (ll != 16LL)
-+	abort ();
-+      break;
-+    case 2:
-+      break;
-+    default:
-+      abort ();
-+    }
-+  va_end (ap);
-+  return x + 8;
-+}
-+
-+__attribute__ ((noinline)) int
-+foo3 (void)
-+{
-+  return 6;
-+}
-+
-+extern inline __attribute__ ((always_inline, gnu_inline)) int
-+bar (int x, ...)
-+{
-+  if (x < 10)
-+    return foo1 (x, foo3 (), 5, __builtin_va_arg_pack ());
-+  return foo2 (x, foo3 () + 4, __builtin_va_arg_pack ());
-+}
-+
-+int
-+main (void)
-+{
-+  if (bar (0, ++v1, v4, &v4, v2++) != 0)
-+    abort ();
-+  if (bar (1, ++v5, 8, v3) != 1)
-+    abort ();
-+  if (bar (2) != 2)
-+    abort ();
-+  if (bar (v1 + 2) != 19)
-+    abort ();
-+  if (bar (v1 + 3, v5--, v4, v4, v3, 16LL) != 20)
-+    abort ();
-+  return 0;
-+}
---- gcc/testsuite/g++.dg/ext/va-arg-pack-2.C.jj	2007-09-07 08:53:23.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/va-arg-pack-2.C	2007-09-07 08:54:25.000000000 +0200
-@@ -0,0 +1,46 @@
-+// { dg-do compile }
-+// { dg-options "-O2" }
-+
-+int bar (int, const char *, int, ...);
-+int baz (int, const char *, long int);
-+
-+extern inline __attribute__((always_inline)) int
-+f2 (int y, ...)
-+{
-+  return bar (y, "", __builtin_va_arg_pack ());		/* { dg-error "invalid use of" } */
-+}
-+
-+extern inline __attribute__((always_inline)) int
-+f3 (int y, ...)
-+{
-+  return bar (y, "", 5, __builtin_va_arg_pack ());
-+}
-+
-+extern inline __attribute__((always_inline)) int
-+f4 (int y, ...)
-+{
-+  return bar (y, "", 4, __builtin_va_arg_pack (), 6);	/* { dg-error "invalid use of" } */
-+}
-+
-+extern inline __attribute__((always_inline)) int
-+f5 (int y, ...)
-+{
-+  return baz (y, "", __builtin_va_arg_pack ());		/* { dg-error "invalid use of" } */
-+}
-+
-+extern inline __attribute__((always_inline)) int
-+f6 (int y, ...)
-+{
-+  return __builtin_va_arg_pack ();			/* { dg-error "invalid use of" } */
-+}
-+
-+int
-+test (void)
-+{
-+  int a = f2 (5, "a", 6);
-+  a += f3 (6, "ab", 17LL);
-+  a += f4 (7, 1, 2, 3);
-+  a += f5 (8, 7L);
-+  a += f6 (9);
-+  return a;
-+}
diff --git a/gcc41-c++-gnu_inline.patch b/gcc41-c++-gnu_inline.patch
deleted file mode 100644
index 5175a3d..0000000
--- a/gcc41-c++-gnu_inline.patch
+++ /dev/null
@@ -1,608 +0,0 @@
-2007-08-31  Jakub Jelinek  <jakub@redhat.com>
-
-	* cp-tree.h (cp_disregard_inline_limits): New prototype.
-	* decl.c (cp_disregard_inline_limits): New function.
-	* cp-objcp-common.h
-	(LANG_HOOKS_TREE_INLINING_DISREGARD_INLINE_LIMITS): Define.
-
-2007-08-30  Jakub Jelinek  <jakub@redhat.com>
-
-	* gcc.dg/inline-24.c: New test.
-	* g++.dg/opt/inline11.C: New test.
-
-2007-08-27  Alexandre Oliva  <aoliva@redhat.com>
-
-	* doc/extend.texi (gnu_inline funtion attribute): Document C++
-	behavior.
-
-	* decl.c (GNU_INLINE_P): New.
-	(duplicate_decls): Handle gnu_inline.  Merge attributes and
-	some flags in overriding definitions.
-	(redeclaration_error_message): Handle gnu_inline.
-	(start_preparsed_function): Likewise.
-
-	* g++.dg/ext/gnu-inline-common.h: New.
-	* g++.dg/ext/gnu-inline-global-reject.C: New.
-	* g++.dg/ext/gnu-inline-global.C: New.
-	* g++.dg/ext/gnu-inline-namespace.C: New.
-	* g++.dg/ext/gnu-inline-anon-namespace.C: New.
-	* g++.dg/ext/gnu-inline-class.C: New.
-	* g++.dg/ext/gnu-inline-class-static.C: New.
-	* g++.dg/ext/gnu-inline-template-class.C: New.
-	* g++.dg/ext/gnu-inline-template-func.C: New.
-
---- gcc/cp/cp-tree.h.jj	2007-08-31 09:47:43.000000000 +0200
-+++ gcc/cp/cp-tree.h	2007-08-31 14:09:08.000000000 +0200
-@@ -4417,6 +4417,7 @@ extern linkage_kind decl_linkage		(tree)
- extern tree cp_walk_subtrees (tree*, int*, walk_tree_fn,
- 			      void*, struct pointer_set_t*);
- extern int cp_cannot_inline_tree_fn		(tree*);
-+extern int cp_disregard_inline_limits		(tree);
- extern tree cp_add_pending_fn_decls		(void*,tree);
- extern int cp_auto_var_in_fn_p			(tree,tree);
- extern tree fold_if_not_in_template		(tree);
---- gcc/cp/decl.c.jj	2007-08-31 09:47:43.000000000 +0200
-+++ gcc/cp/decl.c	2007-08-31 14:27:12.000000000 +0200
-@@ -1097,6 +1097,10 @@ check_redeclaration_exception_specificat
-     }
- }
- 
-+#define GNU_INLINE_P(fn) (DECL_DECLARED_INLINE_P (fn)			\
-+			  && lookup_attribute ("gnu_inline",		\
-+					       DECL_ATTRIBUTES (fn)))
-+
- /* If NEWDECL is a redeclaration of OLDDECL, merge the declarations.
-    If the redeclaration is invalid, a diagnostic is issued, and the
-    error_mark_node is returned.  Otherwise, OLDDECL is returned.
-@@ -1626,19 +1630,45 @@ duplicate_decls (tree newdecl, tree oldd
- 	= chainon (DECL_TEMPLATE_SPECIALIZATIONS (olddecl),
- 		   DECL_TEMPLATE_SPECIALIZATIONS (newdecl));
- 
-+      DECL_ATTRIBUTES (old_result)
-+	= (*targetm.merge_decl_attributes) (old_result, new_result);
-+
-       if (DECL_FUNCTION_TEMPLATE_P (newdecl))
- 	{
--	  DECL_INLINE (old_result)
--	    |= DECL_INLINE (new_result);
--	  DECL_DECLARED_INLINE_P (old_result)
--	    |= DECL_DECLARED_INLINE_P (new_result);
--	  check_redeclaration_exception_specification (newdecl, olddecl);
-+	  if (GNU_INLINE_P (old_result) != GNU_INLINE_P (new_result)
-+	      && DECL_INITIAL (new_result))
-+	    {
-+	      if (DECL_INITIAL (old_result))
-+		{
-+		  DECL_INLINE (old_result) = 0;
-+		  DECL_UNINLINABLE (old_result) = 1;
-+		}
-+	      else
-+		{
-+		  DECL_INLINE (old_result) = DECL_INLINE (new_result);
-+		  DECL_UNINLINABLE (old_result) = DECL_UNINLINABLE (new_result);
-+		}
-+	      DECL_EXTERNAL (old_result) = DECL_EXTERNAL (new_result);
-+	      DECL_NOT_REALLY_EXTERN (old_result)
-+		= DECL_NOT_REALLY_EXTERN (new_result);
-+	      DECL_INTERFACE_KNOWN (old_result)
-+		= DECL_INTERFACE_KNOWN (new_result);
-+	      DECL_DECLARED_INLINE_P (old_result)
-+		= DECL_DECLARED_INLINE_P (new_result);
-+	    }
-+	  else
-+	    {
-+	      DECL_INLINE (old_result)
-+		|= DECL_INLINE (new_result);
-+	      DECL_DECLARED_INLINE_P (old_result)
-+		|= DECL_DECLARED_INLINE_P (new_result);
-+	      check_redeclaration_exception_specification (newdecl, olddecl);
-+	    }
- 	}
- 
-       /* If the new declaration is a definition, update the file and
- 	 line information on the declaration.  */
--      if (DECL_INITIAL (old_result) == NULL_TREE
--	  && DECL_INITIAL (new_result) != NULL_TREE)
-+      if (DECL_INITIAL (new_result) != NULL_TREE)
- 	{
- 	  DECL_SOURCE_LOCATION (olddecl)
- 	    = DECL_SOURCE_LOCATION (old_result)
-@@ -1795,9 +1825,29 @@ duplicate_decls (tree newdecl, tree oldd
-   new_template = NULL_TREE;
-   if (DECL_LANG_SPECIFIC (newdecl) && DECL_LANG_SPECIFIC (olddecl))
-     {
--      DECL_INTERFACE_KNOWN (newdecl) |= DECL_INTERFACE_KNOWN (olddecl);
--      DECL_NOT_REALLY_EXTERN (newdecl) |= DECL_NOT_REALLY_EXTERN (olddecl);
--      DECL_COMDAT (newdecl) |= DECL_COMDAT (olddecl);
-+      bool old_decl_gnu_inline;
-+
-+      if ((DECL_INTERFACE_KNOWN (olddecl)
-+	   && TREE_CODE (olddecl) == FUNCTION_DECL)
-+	  || (TREE_CODE (olddecl) == TEMPLATE_DECL
-+	      && TREE_CODE (DECL_TEMPLATE_RESULT (olddecl)) == FUNCTION_DECL))
-+	{
-+	  tree fn = olddecl;
-+
-+	  if (TREE_CODE (fn) == TEMPLATE_DECL)
-+	    fn = DECL_TEMPLATE_RESULT (olddecl);
-+
-+	  old_decl_gnu_inline = GNU_INLINE_P (fn) && DECL_INITIAL (fn);
-+	}
-+      else
-+	old_decl_gnu_inline = false;
-+
-+      if (!old_decl_gnu_inline)
-+	{
-+	  DECL_INTERFACE_KNOWN (newdecl) |= DECL_INTERFACE_KNOWN (olddecl);
-+	  DECL_NOT_REALLY_EXTERN (newdecl) |= DECL_NOT_REALLY_EXTERN (olddecl);
-+	  DECL_COMDAT (newdecl) |= DECL_COMDAT (olddecl);
-+	}
-       DECL_TEMPLATE_INSTANTIATED (newdecl)
- 	|= DECL_TEMPLATE_INSTANTIATED (olddecl);
- 
-@@ -1871,6 +1921,13 @@ duplicate_decls (tree newdecl, tree oldd
- 	  /* [temp.expl.spec/14] We don't inline explicit specialization
- 	     just because the primary template says so.  */
- 	}
-+      else if (new_defines_function && DECL_INITIAL (olddecl))
-+	{
-+	  /* C++ is always in in unit-at-a-time mode, so we never
-+	     inline re-defined extern inline functions.  */
-+	  DECL_INLINE (newdecl) = 0;
-+	  DECL_UNINLINABLE (newdecl) = 1;
-+	}
-       else
- 	{
- 	  if (DECL_PENDING_INLINE_INFO (newdecl) == 0)
-@@ -2113,9 +2170,25 @@ redeclaration_error_message (tree newdec
- 	{
- 	  if (DECL_NAME (olddecl) == NULL_TREE)
- 	    return "%q#D not declared in class";
--	  else
-+	  else if (!GNU_INLINE_P (olddecl)
-+		   || GNU_INLINE_P (newdecl))
- 	    return "redefinition of %q#D";
- 	}
-+
-+      if (DECL_DECLARED_INLINE_P (olddecl) && DECL_DECLARED_INLINE_P (newdecl))
-+	{
-+	  bool olda = GNU_INLINE_P (olddecl);
-+	  bool newa = GNU_INLINE_P (newdecl);
-+
-+	  if (olda != newa)
-+	    {
-+	      if (newa)
-+		return "%q+D redeclared inline with %<gnu_inline%> attribute";
-+	      else
-+		return "%q+D redeclared inline without %<gnu_inline%> attribute";
-+	    }
-+	}
-+
-       return 0;
-     }
-   else if (TREE_CODE (newdecl) == TEMPLATE_DECL)
-@@ -2141,9 +2214,24 @@ redeclaration_error_message (tree newdec
-       ot = DECL_TEMPLATE_RESULT (olddecl);
-       if (DECL_TEMPLATE_INFO (ot))
- 	ot = DECL_TEMPLATE_RESULT (template_for_substitution (ot));
--      if (DECL_INITIAL (nt) && DECL_INITIAL (ot))
-+      if (DECL_INITIAL (nt) && DECL_INITIAL (ot)
-+	  && (!GNU_INLINE_P (ot) || GNU_INLINE_P (nt)))
- 	return "redefinition of %q#D";
- 
-+      if (DECL_DECLARED_INLINE_P (ot) && DECL_DECLARED_INLINE_P (nt))
-+	{
-+	  bool olda = GNU_INLINE_P (ot);
-+	  bool newa = GNU_INLINE_P (nt);
-+
-+	  if (olda != newa)
-+	    {
-+	      if (newa)
-+		return "%q+D redeclared inline with %<gnu_inline%> attribute";
-+	      else
-+		return "%q+D redeclared inline without %<gnu_inline%> attribute";
-+	    }
-+	}
-+
-       return NULL;
-     }
-   else if (TREE_CODE (newdecl) == VAR_DECL
-@@ -10447,6 +10535,14 @@ start_preparsed_function (tree decl1, tr
-       && lookup_attribute ("noinline", attrs))
-     warning (0, "inline function %q+D given attribute noinline", decl1);
- 
-+  /* Handle gnu_inline attribute.  */
-+  if (GNU_INLINE_P (decl1))
-+    {
-+      DECL_EXTERNAL (decl1) = 1;
-+      DECL_NOT_REALLY_EXTERN (decl1) = 0;
-+      DECL_INTERFACE_KNOWN (decl1) = 1;
-+    }
-+
-   if (DECL_MAYBE_IN_CHARGE_CONSTRUCTOR_P (decl1))
-     /* This is a constructor, we must ensure that any default args
-        introduced by this definition are propagated to the clones
-@@ -10702,8 +10798,9 @@ start_preparsed_function (tree decl1, tr
-   else
-     {
-       /* This is a definition, not a reference.
--	 So clear DECL_EXTERNAL.  */
--      DECL_EXTERNAL (decl1) = 0;
-+	 So clear DECL_EXTERNAL, unless this is a GNU extern inline.  */
-+      if (!GNU_INLINE_P (decl1))
-+	DECL_EXTERNAL (decl1) = 0;
- 
-       if ((DECL_DECLARED_INLINE_P (decl1)
- 	   || DECL_TEMPLATE_INSTANTIATION (decl1))
-@@ -11698,4 +11795,17 @@ cxx_comdat_group (tree decl)
-   return IDENTIFIER_POINTER (name);
- }
- 
-+/* We want to inline __gnu_inline__ functions even if this would
-+   violate inlining limits.  Some glibc and linux constructs depend on
-+   such functions always being inlined when optimizing.  */
-+
-+int
-+cp_disregard_inline_limits (tree fn)
-+{
-+  if (lookup_attribute ("always_inline", DECL_ATTRIBUTES (fn)) != NULL)
-+    return 1;
-+
-+  return (!flag_really_no_inline && GNU_INLINE_P (fn));
-+}
-+
- #include "gt-cp-decl.h"
---- gcc/cp/cp-objcp-common.h.jj	2007-02-20 22:37:34.000000000 +0100
-+++ gcc/cp/cp-objcp-common.h	2007-08-31 13:58:32.000000000 +0200
-@@ -107,6 +107,9 @@ extern tree objcp_tsubst_copy_and_build 
- #undef LANG_HOOKS_TREE_INLINING_CANNOT_INLINE_TREE_FN
- #define LANG_HOOKS_TREE_INLINING_CANNOT_INLINE_TREE_FN \
-   cp_cannot_inline_tree_fn
-+#undef LANG_HOOKS_TREE_INLINING_DISREGARD_INLINE_LIMITS
-+#define LANG_HOOKS_TREE_INLINING_DISREGARD_INLINE_LIMITS \
-+  cp_disregard_inline_limits
- #undef LANG_HOOKS_TREE_INLINING_ADD_PENDING_FN_DECLS
- #define LANG_HOOKS_TREE_INLINING_ADD_PENDING_FN_DECLS \
-   cp_add_pending_fn_decls
---- gcc/doc/extend.texi.jj	2007-08-31 09:47:40.000000000 +0200
-+++ gcc/doc/extend.texi	2007-08-31 13:00:28.000000000 +0200
-@@ -1587,8 +1587,8 @@ refer to the single copy in the library.
- definitions of the functions need not be precisely the same, although
- if they do not have the same effect your program may behave oddly.
- 
--If the function is neither @code{extern} nor @code{static}, then the
--function is compiled as a standalone function, as well as being
-+In C, if the function is neither @code{extern} nor @code{static}, then
-+the function is compiled as a standalone function, as well as being
- inlined where possible.
- 
- This is how GCC traditionally handled functions declared
-@@ -1606,6 +1606,10 @@ assume that it is always present, whethe
- In versions prior to 4.3, the only effect of explicitly including it is
- to disable warnings about using inline functions in C99 mode.
- 
-+In C++, this attribute does not depend on @code{extern} in any way,
-+but it still requires the @code{inline} keyword to enable its special
-+behavior.
-+
- @cindex @code{flatten} function attribute
- @item flatten
- Generally, inlining into a function is limited.  For a function marked with
---- gcc/testsuite/gcc.dg/inline-24.c.jj	2007-08-31 14:11:45.000000000 +0200
-+++ gcc/testsuite/gcc.dg/inline-24.c	2007-08-30 18:43:55.000000000 +0200
-@@ -0,0 +1,28 @@
-+/* Verify that gnu_inline inlines disregard inlining limits.  */
-+/* { dg-do link } */
-+/* { dg-options "-O2" } */
-+
-+extern int foo (int);
-+extern int baz (int);
-+
-+extern inline __attribute__((gnu_inline))
-+int foo (int x)
-+{
-+  int i;
-+  if (!__builtin_constant_p (x))
-+    {
-+#define B(n) baz (1##n) + baz (2##n) + baz (3##n) \
-+	     + baz (4##n) + baz (5##n) + baz (6##n)
-+#define C(n) B(1##n) + B(2##n) + B(3##n) + B(4##n) + B(5##n) + B(6##n)
-+#define D(n) C(1##n) + C(2##n) + C(3##n) + C(4##n) + C(5##n) + C(6##n)
-+      return D(0) + D(1) + D(2) + D(3) + D(4)
-+	     + D(5) + D(6) + D(7) + D(8) + D(9);
-+    }
-+  return 0;
-+}
-+
-+int
-+main (void)
-+{
-+  return foo (0);
-+}
---- gcc/testsuite/g++.dg/opt/inline11.C.jj	2007-08-31 14:12:05.000000000 +0200
-+++ gcc/testsuite/g++.dg/opt/inline11.C	2007-08-30 18:43:55.000000000 +0200
-@@ -0,0 +1,28 @@
-+/* Verify that gnu_inline inlines disregard inlining limits.  */
-+/* { dg-do link } */
-+/* { dg-options "-O2" } */
-+
-+extern int foo (int);
-+extern int baz (int);
-+
-+extern inline __attribute__((gnu_inline))
-+int foo (int x)
-+{
-+  int i;
-+  if (!__builtin_constant_p (x))
-+    {
-+#define B(n) baz (1##n) + baz (2##n) + baz (3##n) \
-+	     + baz (4##n) + baz (5##n) + baz (6##n)
-+#define C(n) B(1##n) + B(2##n) + B(3##n) + B(4##n) + B(5##n) + B(6##n)
-+#define D(n) C(1##n) + C(2##n) + C(3##n) + C(4##n) + C(5##n) + C(6##n)
-+      return D(0) + D(1) + D(2) + D(3) + D(4)
-+	     + D(5) + D(6) + D(7) + D(8) + D(9);
-+    }
-+  return 0;
-+}
-+
-+int
-+main (void)
-+{
-+  return foo (0);
-+}
---- gcc/testsuite/g++.dg/ext/gnu-inline-global-reject.C.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-global-reject.C	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,55 @@
-+/* Test __attribute__((gnu_inline)).
-+
-+   Check that we reject various forms of duplicate definitions.
-+*/
-+
-+/* { dg-do compile } */
-+
-+#include "gnu-inline-common.h"
-+
-+#undef fn
-+#define fn pfx(func_decl_inline_before)
-+decl(inline, fn) // { dg-error "previous" "" }
-+gnuindef(fn, 0) // { dg-error "redeclared" "" }
-+
-+#undef fn
-+#define fn pfx(func_decl_inline_after)
-+gnuindef(fn, 0) // { dg-error "previous" "" }
-+decl(inline, fn) // { dg-error "redeclared" "" }
-+
-+#undef fn
-+#define fn pfx(func_def_gnuin_redef)
-+gnuindef(fn, 0) // { dg-error "previous" "" }
-+gnuindef(fn, 1) // { dg-error "redefinition" "" }
-+
-+#undef fn
-+#define fn pfx(func_def_inline_redef)
-+def(inline, fn, 0) // { dg-error "previous" "" }
-+def(inline, fn, 1) // { dg-error "redefinition" "" }
-+
-+#undef fn
-+#define fn pfx(func_def_inline_after)
-+gnuindef(fn, 0) // { dg-error "previous" "" }
-+def(inline, fn, 1) // { dg-error "redeclare" "" }
-+
-+#undef fn
-+#define fn pfx(func_def_inline_before)
-+def(inline, fn, 0) // { dg-error "previous" "" }
-+gnuindef(fn, 1) // { dg-error "redefinition" "" }
-+
-+#undef fn
-+#define fn pfx(func_def_before)
-+def(, fn, 0) // { dg-error "previous" "" }
-+gnuindef(fn, 1) // { dg-error "redefinition" "" }
-+
-+#undef fn
-+#define fn pfx(func_decl_static_inline_before)
-+decl(static inline, fn) // { dg-error "previous" "" }
-+gnuindef(fn, 0) // { dg-error "redeclared" "" }
-+
-+#undef fn
-+#define fn pfx(func_def_static_inline_after)
-+decl(static, fn)
-+gnuindef(fn, 0) // { dg-error "previous" "" }
-+decl(static, fn)
-+def(static inline, fn, 1) // { dg-error "redeclare" "" }
---- gcc/testsuite/g++.dg/ext/gnu-inline-template-func.C.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-template-func.C	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,17 @@
-+/* { dg-do compile } */
-+/* { dg-options "-O" } */ // such that static functions are optimized out
-+/* { dg-final { scan-assembler "func1" } } */
-+/* { dg-final { scan-assembler "func2" } } */
-+/* { dg-final { scan-assembler-not "func3" } } */
-+/* { dg-final { scan-assembler "func4" } } */
-+/* { dg-final { scan-assembler-not "func5" } } */
-+
-+#define defpfx template <typename T>
-+
-+#include "gnu-inline-global.C"
-+
-+template int func1<int>(void);
-+template int func2<int>(void);
-+template int func3<int>(void);
-+template int func4<int>(void);
-+template int func5<int>(void);
---- gcc/testsuite/g++.dg/ext/gnu-inline-common.h.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-common.h	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,24 @@
-+#ifndef gnu
-+# define gnu_inline __attribute__((gnu_inline)) inline
-+#endif
-+
-+#define declspec(spec, name) spec int name (void)
-+#ifdef IN_CLASS
-+# define decl(spec, name)
-+#else
-+# define decl(spec, name) defpfx declspec(spec, name);
-+#endif
-+#define def(spec, name, ret) defpfx declspec(spec, name) { return ret; }
-+#define gnuindef(name, ret) def(gnu_inline, name, ret)
-+
-+#ifndef pfx
-+# ifdef IN_CLASS
-+#  define pfx(x) IN_CLASS::x
-+# else
-+#  define pfx(x) x
-+# endif
-+#endif
-+
-+#ifndef defpfx
-+# define defpfx
-+#endif
---- gcc/testsuite/g++.dg/ext/gnu-inline-anon-namespace.C.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-anon-namespace.C	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,11 @@
-+/* { dg-do compile } */
-+/* { dg-options "-O" } */ // such that static functions are optimized out
-+/* { dg-final { scan-assembler-not "func1" } } */
-+/* { dg-final { scan-assembler-not "func2" } } */
-+/* { dg-final { scan-assembler-not "func3" } } */
-+/* { dg-final { scan-assembler-not "func4" } } */
-+/* { dg-final { scan-assembler-not "func5" } } */
-+
-+namespace {
-+#include "gnu-inline-global.C"
-+}
---- gcc/testsuite/g++.dg/ext/gnu-inline-class-static.C.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-class-static.C	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,20 @@
-+/* { dg-do compile } */
-+/* { dg-options "-O" } */ // such that static functions are optimized out
-+/* { dg-final { scan-assembler "func1" } } */
-+/* { dg-final { scan-assembler "func2" } } */
-+/* { dg-final { scan-assembler-not "func3" } } */
-+/* { dg-final { scan-assembler "func4" } } */
-+/* { dg-final { scan-assembler "func5" } } */
-+
-+#undef IN_CLASS
-+#define IN_CLASS gnu_test_static
-+
-+struct IN_CLASS {
-+  static int func1(void);
-+  static int func2(void);
-+  static int func3(void);
-+  static int func4(void);
-+  static int func5(void);
-+};
-+
-+#include "gnu-inline-global.C"
---- gcc/testsuite/g++.dg/ext/gnu-inline-global.C.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-global.C	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,50 @@
-+/* Test __attribute__((gnu_inline)).
-+
-+   Check that __attribute__((gnu_inline)) has no effect, in the
-+   absence of extern and/or inline.
-+
-+   Check that we don't get out-of-line definitions for extern inline
-+   gnu_inline functions, regardless of declarations or definitions.
-+
-+   Check that such functions can be overridden by out-of-line
-+   definitions.
-+
-+ */
-+
-+/* { dg-do compile } */
-+/* { dg-options "-O" } */ // such that static functions are optimized out
-+/* { dg-final { scan-assembler "func1" } } */
-+/* { dg-final { scan-assembler "func2" } } */
-+/* { dg-final { scan-assembler-not "func3" } } */
-+/* { dg-final { scan-assembler "func4" } } */
-+/* { dg-final { scan-assembler-not "func5" } } */
-+
-+#include "gnu-inline-common.h"
-+
-+#undef fn
-+#define fn pfx(func1) // must be emitted out-of-line
-+gnuindef(fn, 0)
-+def(, fn, 2)
-+
-+#undef fn
-+#define fn pfx(func2) // must be emitted out-of-line
-+decl(extern, fn)
-+gnuindef(fn, 0)
-+def(, fn, 2)
-+
-+#undef fn
-+#define fn pfx(func3) // must not be emitted
-+decl(extern, fn)
-+gnuindef(fn, 0)
-+
-+#undef fn
-+#define fn pfx(func4) // must be emitted out-of-line
-+decl(extern, fn)
-+gnuindef(fn, 0)
-+def(, fn, 1)
-+
-+#undef fn
-+#define fn pfx(func5) // must NOT be emitted, because it's static and unused
-+decl(static, fn)
-+gnuindef(fn, 0)
-+def(, fn, 1)
---- gcc/testsuite/g++.dg/ext/gnu-inline-template-class.C.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-template-class.C	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,22 @@
-+/* { dg-do compile } */
-+/* { dg-options "-O" } */ // such that static functions are optimized out
-+/* { dg-final { scan-assembler "func1" } } */
-+/* { dg-final { scan-assembler "func2" } } */
-+/* { dg-final { scan-assembler-not "func3" } } */
-+/* { dg-final { scan-assembler "func4" } } */
-+/* { dg-final { scan-assembler "func5" } } */
-+
-+template <typename T> struct gnu_test {
-+  int func1(void);
-+  int func2(void);
-+  int func3(void);
-+  int func4(void);
-+  int func5(void);
-+};
-+
-+#define defpfx template <typename T>
-+#define IN_CLASS gnu_test<T>
-+
-+#include "gnu-inline-global.C"
-+
-+template struct gnu_test<int>;
---- gcc/testsuite/g++.dg/ext/gnu-inline-namespace.C.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-namespace.C	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,11 @@
-+/* { dg-do compile } */
-+/* { dg-options "-O" } */ // such that static functions are optimized out
-+/* { dg-final { scan-assembler "func1" } } */
-+/* { dg-final { scan-assembler "func2" } } */
-+/* { dg-final { scan-assembler-not "func3" } } */
-+/* { dg-final { scan-assembler "func4" } } */
-+/* { dg-final { scan-assembler-not "func5" } } */
-+
-+namespace gnu_test {
-+#include "gnu-inline-global.C"
-+}
---- gcc/testsuite/g++.dg/ext/gnu-inline-class.C.jj	2007-08-31 13:00:28.000000000 +0200
-+++ gcc/testsuite/g++.dg/ext/gnu-inline-class.C	2007-08-31 13:00:28.000000000 +0200
-@@ -0,0 +1,19 @@
-+/* { dg-do compile } */
-+/* { dg-options "-O" } */ // such that static functions are optimized out
-+/* { dg-final { scan-assembler "func1" } } */
-+/* { dg-final { scan-assembler "func2" } } */
-+/* { dg-final { scan-assembler-not "func3" } } */
-+/* { dg-final { scan-assembler "func4" } } */
-+/* { dg-final { scan-assembler "func5" } } */
-+
-+#define IN_CLASS gnu_test
-+
-+struct IN_CLASS {
-+  int func1(void);
-+  int func2(void);
-+  int func3(void);
-+  int func4(void);
-+  int func5(void);
-+};
-+
-+#include "gnu-inline-global.C"
diff --git a/gcc41-error-attrib.patch b/gcc41-error-attrib.patch
new file mode 100644
index 0000000..a306821
--- /dev/null
+++ b/gcc41-error-attrib.patch
@@ -0,0 +1,295 @@
+2007-09-23  Jakub Jelinek  <jakub@redhat.com>
+
+	* expr.c (expand_expr_real_1) <case CALL_EXPR>: Use get_callee_fndecl
+	instead of checking CALL_EXPR_FN directly to test for builtins.
+	If error or warning attributes are present, print
+	error resp. warning.
+	* c-common.c (handle_error_attribute): New function.
+	(c_common_attribute_table): Add error and warning
+	attributes.
+	* doc/extend.texi: Document error and warning attributes.
+
+	* gcc.dg/va-arg-pack-len-1.c: Use error and warning
+	attributes.
+	* gcc.dg/va-arg-pack-len-2.c: New test.
+	* g++.dg/ext/va-arg-pack-len-1.C: Use error and warning
+	attributes.
+	* g++.dg/ext/va-arg-pack-len-2.C: New test.
+
+--- gcc/doc/extend.texi.jj	2007-09-25 15:04:15.000000000 +0200
++++ gcc/doc/extend.texi	2007-09-25 15:32:22.000000000 +0200
+@@ -1589,8 +1589,8 @@ attributes are currently defined for fun
+ @code{section}, @code{constructor}, @code{destructor}, @code{used},
+ @code{unused}, @code{deprecated}, @code{weak}, @code{malloc},
+ @code{alias}, @code{warn_unused_result}, @code{nonnull},
+-@code{gnu_inline}, @code{externally_visible} and @code{artificial}.
+-Several other
++@code{gnu_inline}, @code{externally_visible}, @code{artificial},
++@code{error} and @code{warning}.  Several other
+ attributes are defined for functions on particular target systems.  Other
+ attributes, including @code{section} are supported for variables declarations
+ (@pxref{Variable Attributes}) and for types (@pxref{Type Attributes}).
+@@ -1688,6 +1688,30 @@ Whether the function itself is considere
+ the current inlining parameters.  The @code{flatten} attribute only works
+ reliably in unit-at-a-time mode.
+ 
++@item error ("@var{message}")
++@cindex @code{error} function attribute
++If this attribute is used on a function declaration and a call to such a function
++is not eliminated through dead code elimination or other optimizations, an error
++which will include @var{message} will be diagnosed.  This is useful
++for compile time checking, especially together with @code{__builtin_constant_p}
++and inline functions where checking the inline function arguments is not
++possible through @code{extern char [(condition) ? 1 : -1];} tricks.
++While it is possible to leave the function undefined and thus invoke
++a link failure, when using this attribute the problem will be diagnosed
++earlier and with exact location of the call even in presence of inline
++functions or when not emitting debugging information.
++
++@item warning ("@var{message}")
++@cindex @code{warning} function attribute
++If this attribute is used on a function declaration and a call to such a function
++is not eliminated through dead code elimination or other optimizations, a warning
++which will include @var{message} will be diagnosed.  This is useful
++for compile time checking, especially together with @code{__builtin_constant_p}
++and inline functions.  While it is possible to define the function with
++a message in @code{.gnu.warning*} section, when using this attribute the problem
++will be diagnosed earlier and with exact location of the call even in presence
++of inline functions or when not emitting debugging information.
++
+ @item cdecl
+ @cindex functions that do pop the argument stack on the 386
+ @opindex mrtd
+--- gcc/expr.c.jj	2007-09-25 14:58:40.000000000 +0200
++++ gcc/expr.c	2007-09-25 15:19:15.000000000 +0200
+@@ -7513,19 +7513,31 @@ expand_expr_real_1 (tree exp, rtx target
+ 	 inlining.  */
+       if (CALL_EXPR_VA_ARG_PACK (exp))
+ 	error ("invalid use of %<__builtin_va_arg_pack ()%>");
+-      /* Check for a built-in function.  */
+-      if (TREE_CODE (TREE_OPERAND (exp, 0)) == ADDR_EXPR
+-	  && (TREE_CODE (TREE_OPERAND (TREE_OPERAND (exp, 0), 0))
+-	      == FUNCTION_DECL)
+-	  && DECL_BUILT_IN (TREE_OPERAND (TREE_OPERAND (exp, 0), 0)))
+-	{
+-	  if (DECL_BUILT_IN_CLASS (TREE_OPERAND (TREE_OPERAND (exp, 0), 0))
+-	      == BUILT_IN_FRONTEND)
+-	    return lang_hooks.expand_expr (exp, original_target,
+-					   tmode, modifier,
+-					   alt_rtl);
+-	  else
+-	    return expand_builtin (exp, target, subtarget, tmode, ignore);
++      {
++	tree fndecl = get_callee_fndecl (exp), attr;
++
++	if (fndecl
++	    && (attr = lookup_attribute ("error",
++					 DECL_ATTRIBUTES (fndecl))) != NULL)
++	  error ("call to %qs declared with attribute error: %s",
++		 lang_hooks.decl_printable_name (fndecl, 1),
++		 TREE_STRING_POINTER (TREE_VALUE (TREE_VALUE (attr))));
++	if (fndecl
++	    && (attr = lookup_attribute ("warning",
++					 DECL_ATTRIBUTES (fndecl))) != NULL)
++	  warning (0, "call to %qs declared with attribute warning: %s",
++		   lang_hooks.decl_printable_name (fndecl, 1),
++		   TREE_STRING_POINTER (TREE_VALUE (TREE_VALUE (attr))));
++
++	/* Check for a built-in function.  */
++	if (fndecl && DECL_BUILT_IN (fndecl))
++	  {
++	    if (DECL_BUILT_IN_CLASS (fndecl) == BUILT_IN_FRONTEND)
++	      return lang_hooks.expand_expr (exp, original_target,
++					     tmode, modifier, alt_rtl);
++	    else
++	      return expand_builtin (exp, target, subtarget, tmode, ignore);
++	  }
+ 	}
+ 
+       return expand_call (exp, target, ignore);
+--- gcc/c-common.c.jj	2007-09-25 15:01:49.000000000 +0200
++++ gcc/c-common.c	2007-09-25 15:24:34.000000000 +0200
+@@ -508,6 +508,7 @@ static tree handle_always_inline_attribu
+ static tree handle_gnu_inline_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_artificial_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_flatten_attribute (tree *, tree, tree, int, bool *);
++static tree handle_error_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_used_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_unused_attribute (tree *, tree, tree, int, bool *);
+ static tree handle_externally_visible_attribute (tree *, tree, tree, int,
+@@ -641,6 +642,10 @@ const struct attribute_spec c_common_att
+ 			      handle_warn_unused_result_attribute },
+   { "sentinel",               0, 1, false, true, true,
+ 			      handle_sentinel_attribute },
++  { "warning",		      1, 1, true,  false, false,
++			      handle_error_attribute },
++  { "error",		      1, 1, true,  false, false,
++			      handle_error_attribute },
+   { NULL,                     0, 0, false, false, false, NULL }
+ };
+ 
+@@ -4226,6 +4231,26 @@ handle_flatten_attribute (tree *node, tr
+   return NULL_TREE;
+ }
+ 
++/* Handle a "warning" or "error" attribute; arguments as in
++   struct attribute_spec.handler.  */
++
++static tree
++handle_error_attribute (tree *node, tree name, tree args,
++			int ARG_UNUSED (flags), bool *no_add_attrs)
++{
++  if (TREE_CODE (*node) == FUNCTION_DECL
++      || TREE_CODE (TREE_VALUE (args)) == STRING_CST)
++    /* Do nothing else, just set the attribute.  We'll get at
++       it later with lookup_attribute.  */
++    ;
++  else
++    {
++      warning (OPT_Wattributes, "%qE attribute ignored", name);
++      *no_add_attrs = true;
++    }
++
++  return NULL_TREE;
++}
+ 
+ /* Handle a "used" attribute; arguments as in
+    struct attribute_spec.handler.  */
+--- gcc/testsuite/gcc.dg/va-arg-pack-len-1.c.jj	2007-09-25 12:23:05.000000000 +0200
++++ gcc/testsuite/gcc.dg/va-arg-pack-len-1.c	2007-09-25 15:16:03.000000000 +0200
+@@ -3,8 +3,10 @@
+ 
+ #include <stdarg.h>
+ 
+-extern int warn_open_missing_mode (void);
+-extern int warn_open_too_many_arguments (void);
++extern int error_open_missing_mode (void)
++  __attribute__((__error__ ("open with O_CREAT needs 3 arguments, only 2 were given")));
++extern int warn_open_too_many_arguments (void)
++  __attribute__((__warning__ ("open called with more than 3 arguments")));
+ extern void abort (void);
+ 
+ char expected_char;
+@@ -83,7 +85,7 @@ myopen (const char *path, int oflag, ...
+     {
+       if ((oflag & 0x40) != 0 && __builtin_va_arg_pack_len () < 1)
+ 	{
+-	  warn_open_missing_mode ();
++	  error_open_missing_mode ();
+ 	  return myopen2 (path, oflag);
+ 	}
+       return myopenva (path, oflag, __builtin_va_arg_pack ());
+--- gcc/testsuite/gcc.dg/va-arg-pack-len-2.c.jj	2007-09-25 15:16:03.000000000 +0200
++++ gcc/testsuite/gcc.dg/va-arg-pack-len-2.c	2007-09-25 15:16:03.000000000 +0200
+@@ -0,0 +1,42 @@
++/* { dg-do compile } */
++/* { dg-options "-O2" } */
++
++#include <stdarg.h>
++
++extern int error_open_missing_mode (void)
++  __attribute__((__error__ ("open with O_CREAT needs 3 arguments, only 2 were given")));
++extern int warn_open_too_many_arguments (void)
++  __attribute__((__warning__ ("open called with more than 3 arguments")));
++
++extern int myopen2 (const char *path, int oflag);
++extern int myopenva (const char *path, int oflag, ...);
++
++extern inline __attribute__((always_inline, gnu_inline)) int
++myopen (const char *path, int oflag, ...)
++{
++  if (__builtin_va_arg_pack_len () > 1)
++    warn_open_too_many_arguments ();	/* { dg-warning "called with more than 3" } */
++
++  if (__builtin_constant_p (oflag))
++    {
++      if ((oflag & 0x40) != 0 && __builtin_va_arg_pack_len () < 1)
++	{
++	  error_open_missing_mode ();	/* { dg-error "needs 3 arguments, only 2 were given" } */
++	  return myopen2 (path, oflag);
++	}
++      return myopenva (path, oflag, __builtin_va_arg_pack ());
++    }
++
++  if (__builtin_va_arg_pack_len () < 1)
++    return myopen2 (path, oflag);
++
++  return myopenva (path, oflag, __builtin_va_arg_pack ());
++}
++
++int
++main (void)
++{
++  myopen ("h", 0x43);
++  myopen ("i", 0x43, 0644, 0655);
++  return 0;
++}
+--- gcc/testsuite/g++.dg/ext/va-arg-pack-len-1.C.jj	2007-09-25 12:23:05.000000000 +0200
++++ gcc/testsuite/g++.dg/ext/va-arg-pack-len-1.C	2007-09-25 15:16:03.000000000 +0200
+@@ -3,8 +3,10 @@
+ 
+ #include <stdarg.h>
+ 
+-extern "C" int warn_open_missing_mode (void);
+-extern "C" int warn_open_too_many_arguments (void);
++extern "C" int error_open_missing_mode (void)
++  __attribute__((__error__ ("open with O_CREAT needs 3 arguments, only 2 were given")));
++extern "C" int warn_open_too_many_arguments (void)
++  __attribute__((__warning__ ("open called with more than 3 arguments")));
+ extern "C" void abort (void);
+ 
+ char expected_char;
+@@ -83,7 +85,7 @@ myopen (const char *path, int oflag, ...
+     {
+       if ((oflag & 0x40) != 0 && __builtin_va_arg_pack_len () < 1)
+ 	{
+-	  warn_open_missing_mode ();
++	  error_open_missing_mode ();
+ 	  return myopen2 (path, oflag);
+ 	}
+       return myopenva (path, oflag, __builtin_va_arg_pack ());
+--- gcc/testsuite/g++.dg/ext/va-arg-pack-len-2.C.jj	2007-09-25 15:16:03.000000000 +0200
++++ gcc/testsuite/g++.dg/ext/va-arg-pack-len-2.C	2007-09-25 15:16:03.000000000 +0200
+@@ -0,0 +1,42 @@
++// { dg-do compile }
++// { dg-options "-O2" }
++
++#include <stdarg.h>
++
++extern int error_open_missing_mode (void)
++  __attribute__((__error__ ("open with O_CREAT needs 3 arguments, only 2 were given")));
++extern int warn_open_too_many_arguments (void)
++  __attribute__((__warning__ ("open called with more than 3 arguments")));
++
++extern int myopen2 (const char *path, int oflag);
++extern int myopenva (const char *path, int oflag, ...);
++
++extern inline __attribute__((always_inline, gnu_inline)) int
++myopen (const char *path, int oflag, ...)
++{
++  if (__builtin_va_arg_pack_len () > 1)
++    warn_open_too_many_arguments ();	// { dg-warning "called with more than 3" }
++
++  if (__builtin_constant_p (oflag))
++    {
++      if ((oflag & 0x40) != 0 && __builtin_va_arg_pack_len () < 1)
++	{
++	  error_open_missing_mode ();	// { dg-error "needs 3 arguments, only 2 were given" }
++	  return myopen2 (path, oflag);
++	}
++      return myopenva (path, oflag, __builtin_va_arg_pack ());
++    }
++
++  if (__builtin_va_arg_pack_len () < 1)
++    return myopen2 (path, oflag);
++
++  return myopenva (path, oflag, __builtin_va_arg_pack ());
++}
++
++int
++main (void)
++{
++  myopen ("h", 0x43);
++  myopen ("i", 0x43, 0644, 0655);
++  return 0;
++}
diff --git a/gcc41-jdwp.patch b/gcc41-jdwp.patch
new file mode 100644
index 0000000..35e4b7b
--- /dev/null
+++ b/gcc41-jdwp.patch
@@ -0,0 +1,242 @@
+2007-09-24  Keith Seitz  <keiths@redhat.com>
+
+	* include/jvm.h (struct natThread): Add new field 'frame'.
+	* include/java-interp.h (_Jv_Frame): Use _Jv_ThreadStackPeek,
+	_Jv_ThreadStackPop, and _Jv_ThreadStackPush instead of
+	java.lang.Thread.frame.
+	(~_Jv_Frame): Use _Jv_ThreadStackPop.
+	* java/lang/natThread.cc (_Jv_ThreadStackPeek): New function.
+	(_Jv_ThreadStackPush): New function.
+	(_Jv_ThreadStackPop): New function.
+	* java/lang/Thread.java (frame): Remove field to restore
+	C++ ABI compatibility.
+	* gnu/classpath/jdwp/natVMVirtualMachine.cc (getFrames): Use
+	_Jv_ThreadStackPeek.
+	(getFrame): Likewise.
+	* gnu/classpath/jdwp/natVMFrame.cc (getFrameDepth): Likewise.
+	* jvmti.cc (getLocalFrame): Likewise.
+	(_Jv_JVMTI_GetFrameCount): Likewise.
+	(_Jv_JVMTI_GetThreadState): Likewise.
+	(_Jv_JVMTI_GetStackTrace): Likewise.
+	* interpret.cc (_Jv_ReportJVMTIExceptionThrow): Likewise.
+	* headers.txt (java/lang/Thread.h): Prepend declarations
+	for _Jv_ThreadStackPeek, _Jv_ThreadStackPush, and _Jv_ThreadStackPop.
+	Add as friend functions.
+	* jni.cc (_Jv_JNIMethod::call): Push a frame onto the stack when
+	calling a JNI method.
+
+--- libjava/interpret.cc	(revision 128603)
++++ libjava/interpret.cc	(working copy)
+@@ -1709,7 +1709,7 @@
+ _Jv_ReportJVMTIExceptionThrow (jthrowable ex)
+ {
+   jthread thread = ::java::lang::Thread::currentThread ();
+-  _Jv_Frame *frame = reinterpret_cast<_Jv_Frame *> (thread->frame);
++  _Jv_Frame *frame = _Jv_ThreadStackPeek (thread);
+   jmethodID throw_meth = frame->self->get_method ();
+   jlocation throw_loc = -1;
+   if (frame->frame_type == frame_interpreter)
+--- libjava/include/java-interp.h	(revision 128603)
++++ libjava/include/java-interp.h	(working copy)
+@@ -391,14 +391,14 @@
+   {
+     self = s;
+     frame_type = type;
+-    next = (_Jv_Frame *) thr->frame;
+-    thr->frame = (gnu::gcj::RawData *) this;
++    next = _Jv_ThreadStackPeek (thr);
++    _Jv_ThreadStackPush (thr, this);
+     thread = thr;
+   }
+ 
+   ~_Jv_Frame ()
+   {
+-    thread->frame = (gnu::gcj::RawData *) next;
++    _Jv_ThreadStackPop (thread);
+   }
+ 
+   int depth ()
+--- libjava/include/jvm.h	(revision 128603)
++++ libjava/include/jvm.h	(working copy)
+@@ -34,6 +34,8 @@
+ 
+ #include <sysdep/locks.h>
+ 
++class _Jv_Frame;
++
+ /* Macro for possible unused arguments.  */
+ #define MAYBE_UNUSED __attribute__((__unused__))
+ 
+@@ -767,6 +769,12 @@
+ 
+   // Each thread has its own JNI object.
+   _Jv_JNIEnv *jni_env;
++
++  // Describes the topmost frame in the thread's composite
++  // (interp + JNI) stack. Added here to maintain C++ ABI
++  // compatibility with previous versions. Newer versions
++  // of gcj put this in java/lang/Thread.java.
++  _Jv_Frame *frame;
+ };
+ 
+ #endif /* __JAVA_JVM_H__ */
+--- libjava/jni.cc	(revision 128603)
++++ libjava/jni.cc	(working copy)
+@@ -2343,6 +2343,10 @@
+   // Copy over passed-in arguments.
+   memcpy (&real_args[offset], args, _this->args_raw_size);
+ 
++  // Add a frame to the composite (interpreted + JNI) call stack
++  java::lang::Thread *thread = java::lang::Thread::currentThread ();
++  _Jv_NativeFrame nat_frame (_this, thread);
++
+   // The actual call to the JNI function.
+ #if FFI_NATIVE_RAW_API
+   ffi_raw_call (&_this->jni_cif, (void (*)()) _this->function,
+--- libjava/jvmti.cc	(revision 128603)
++++ libjava/jvmti.cc	(working copy)
+@@ -228,7 +228,7 @@
+   THREAD_CHECK_VALID (thread);
+   THREAD_CHECK_IS_ALIVE (thread);
+   
+-  _Jv_Frame *frame = reinterpret_cast<_Jv_Frame *> (thread->frame);
++  _Jv_Frame *frame = _Jv_ThreadStackPeek (thread);
+   
+   for (int i = 0; i < depth; i++)
+     {    
+@@ -516,7 +516,7 @@
+   THREAD_CHECK_VALID (thread);
+   THREAD_CHECK_IS_ALIVE (thread);
+    
+-  _Jv_Frame *frame = reinterpret_cast<_Jv_Frame *> (thread->frame);
++  _Jv_Frame *frame = _Jv_ThreadStackPeek (thread);
+   (*frame_count) = frame->depth ();
+   return JVMTI_ERROR_NONE;
+ }
+@@ -543,7 +543,7 @@
+       if (thread->isInterrupted ())
+ 	state |= JVMTI_THREAD_STATE_INTERRUPTED;
+ 
+-      _Jv_Frame *frame = reinterpret_cast<_Jv_Frame *> (thread->frame);
++      _Jv_Frame *frame = _Jv_ThreadStackPeek (thread);
+       if (frame != NULL && frame->frame_type == frame_native)
+ 	state |= JVMTI_THREAD_STATE_IN_NATIVE;
+ 
+@@ -1273,7 +1273,7 @@
+   ILLEGAL_ARGUMENT (start_depth >= (*frame_count));
+   ILLEGAL_ARGUMENT (start_depth < (-(*frame_count)));
+   
+-  _Jv_Frame *frame = reinterpret_cast<_Jv_Frame *> (thread->frame);
++  _Jv_Frame *frame = _Jv_ThreadStackPeek (thread);
+ 
+   // If start_depth is negative use this to determine at what depth to start
+   // the trace by adding it to the length of the call stack.  This allows the
+--- libjava/headers.txt	(revision 128603)
++++ libjava/headers.txt	(working copy)
+@@ -6,6 +6,7 @@
+ 
+ class java/lang/Thread
+ prepend class _Jv_JNIEnv;
++prepend class _Jv_Frame;
+ prepend #define _JV_NOT_OWNER 1
+ prepend #define _JV_INTERRUPTED 2
+ prepend _Jv_JNIEnv * _Jv_GetCurrentJNIEnv ();
+@@ -17,6 +18,9 @@
+ prepend jint _Jv_DetachCurrentThread ();
+ prepend struct _Jv_Thread_t;
+ prepend _Jv_Thread_t* _Jv_ThreadGetData (java::lang::Thread* thread);
++prepend _Jv_Frame *_Jv_ThreadStackPeek (java::lang::Thread *thread);
++prepend void _Jv_ThreadStackPush (java::lang::Thread *thread, _Jv_Frame *frame);
++prepend void _Jv_ThreadStackPop (java::lang::Thread *thread);
+ friend _Jv_JNIEnv * ::_Jv_GetCurrentJNIEnv ();
+ friend void ::_Jv_SetCurrentJNIEnv (_Jv_JNIEnv *env);
+ friend void ::_Jv_ThreadRun (java::lang::Thread* thread);
+@@ -24,6 +28,9 @@
+ friend java::lang::Thread* ::_Jv_AttachCurrentThread (jstring name, java::lang::ThreadGroup* group);
+ friend java::lang::Thread* ::_Jv_AttachCurrentThreadAsDaemon (jstring name, java::lang::ThreadGroup* group);
+ friend jint (::_Jv_DetachCurrentThread) ();
++friend _Jv_Frame *::_Jv_ThreadStackPeek (java::lang::Thread *thread);
++friend void ::_Jv_ThreadStackPush (java::lang::Thread *thread, _Jv_Frame *frame);
++friend void ::_Jv_ThreadStackPop (java::lang::Thread *thread);
+ 
+ class java/lang/String
+ prepend jchar* _Jv_GetStringChars (jstring str);
+--- libjava/gnu/classpath/jdwp/natVMFrame.cc	(revision 128603)
++++ libjava/gnu/classpath/jdwp/natVMFrame.cc	(working copy)
+@@ -181,7 +181,7 @@
+ getFrameDepth (_Jv_Frame *frame)
+ {
+   jint depth = 0;
+-  _Jv_Frame *top_frame = (_Jv_Frame *) frame->thread->frame;
++  _Jv_Frame *top_frame = _Jv_ThreadStackPeek (frame->thread);
+   jint num_frames = VMVirtualMachine::getFrameCount (frame->thread);
+   
+   while (frame != top_frame)
+--- libjava/gnu/classpath/jdwp/natVMVirtualMachine.cc	(revision 128603)
++++ libjava/gnu/classpath/jdwp/natVMVirtualMachine.cc	(working copy)
+@@ -553,7 +553,7 @@
+      
+   frame_list = new ::java::util::ArrayList (num_frames);
+   
+-  _Jv_Frame *vm_frame = reinterpret_cast<_Jv_Frame *> (thread->frame);
++  _Jv_Frame *vm_frame = _Jv_ThreadStackPeek (thread);
+   
+   // Take start frames off the top of the stack
+   while (vm_frame != NULL && start > 0)
+@@ -584,7 +584,7 @@
+ {
+   using namespace gnu::classpath::jdwp::exception;
+   
+-  _Jv_Frame *vm_frame = (_Jv_Frame *) thread->frame;
++  _Jv_Frame *vm_frame = _Jv_ThreadStackPeek (thread);
+   jint depth = 0;
+   _Jv_Frame *frame = reinterpret_cast<_Jv_Frame *> (frameID); 
+   
+--- libjava/java/lang/Thread.java	(revision 128603)
++++ libjava/java/lang/Thread.java	(working copy)
+@@ -186,9 +186,6 @@
+   // This describes the top-most interpreter frame for this thread.
+   RawData interp_frame;
+   
+-  // This describes the top most frame in the composite (interp + JNI) stack
+-  RawData frame;
+-
+   // Current state.
+   volatile int state;
+ 
+--- libjava/java/lang/natThread.cc	(revision 128603)
++++ libjava/java/lang/natThread.cc	(working copy)
+@@ -15,6 +15,7 @@
+ #include <gcj/cni.h>
+ #include <jvm.h>
+ #include <java-threads.h>
++#include <java-interp.h>
+ 
+ #include <gnu/gcj/RawDataManaged.h>
+ #include <java/lang/Thread.h>
+@@ -525,3 +526,25 @@
+ 
+   return 0;
+ }
++
++_Jv_Frame *
++_Jv_ThreadStackPeek (java::lang::Thread *thread)
++{
++  struct natThread *nt = (natThread *) thread->data;
++  return nt->frame;
++}
++
++void
++_Jv_ThreadStackPush (java::lang::Thread *thread, _Jv_Frame *frame)
++{
++  struct natThread *nt = (natThread *) thread->data;
++  nt->frame = frame;
++}
++
++void
++_Jv_ThreadStackPop (java::lang::Thread *thread)
++{
++  struct natThread *nt = (natThread *) thread->data;
++  _Jv_Frame *next = nt->frame->next;
++  nt->frame = next;
++}
diff --git a/gcc41-makeinfo-ver.patch b/gcc41-makeinfo-ver.patch
deleted file mode 100644
index 8c4047c..0000000
--- a/gcc41-makeinfo-ver.patch
+++ /dev/null
@@ -1,54 +0,0 @@
-2007-09-22  Jakub Jelinek  <jakub@redhat.com>
-
-	* configure.ac (MAKEINFO): Handle makeinfo version 4.10 and above.
-	* configure: Regenerated.
-
-2007-09-15  Alan Modra  <amodra@bigpond.net.au>
-
-	* configure.in: Correct makeinfo version check.
-	* configure: Regenerate.
-
---- configure.in.jj	2007-02-20 16:41:55.000000000 -0500
-+++ configure.in	2007-09-22 09:36:04.000000000 -0400
-@@ -2159,7 +2159,7 @@ changequote(,)
-     # For an installed makeinfo, we require it to be from texinfo 4.2 or
-     # higher, else we use the "missing" dummy.
-     if ${MAKEINFO} --version \
--       | egrep 'texinfo[^0-9]*([1-3][0-9]|4\.[2-9]|[5-9])' >/dev/null 2>&1; then
-+       | egrep 'texinfo[^0-9]*(4\.([2-9]|[1-9][0-9])|[5-9]|[1-9][0-9])' >/dev/null 2>&1; then
-       :
-     else
-       MAKEINFO="$MISSING makeinfo"
---- configure.jj	2007-02-20 16:41:55.000000000 -0500
-+++ configure	2007-09-22 09:36:25.000000000 -0400
-@@ -3574,7 +3574,7 @@ case " $build_configdirs " in
-     # For an installed makeinfo, we require it to be from texinfo 4.2 or
-     # higher, else we use the "missing" dummy.
-     if ${MAKEINFO} --version \
--       | egrep 'texinfo[^0-9]*([1-3][0-9]|4\.[2-9]|[5-9])' >/dev/null 2>&1; then
-+       | egrep 'texinfo[^0-9]*(4\.([2-9]|[1-9][0-9])|[5-9]|[1-9][0-9])' >/dev/null 2>&1; then
-       :
-     else
-       MAKEINFO="$MISSING makeinfo"
---- gcc/configure.ac.jj	2007-02-20 16:39:12.000000000 -0500
-+++ gcc/configure.ac	2007-09-22 08:38:07.000000000 -0400
-@@ -833,7 +833,7 @@ MISSING="${CONFIG_SHELL-/bin/sh} $srcdir
- # that we can use it.
- gcc_AC_CHECK_PROG_VER(MAKEINFO, makeinfo, --version,
-   [GNU texinfo.* \([0-9][0-9.]*\)],
--  [4.[2-9]*])
-+  [4.[2-9]*|4.[1-9][0-9]*|[5-9]*|[1-9][0-9]*])
- if test $gcc_cv_prog_makeinfo_modern = no; then
-   MAKEINFO="$MISSING makeinfo"
-   AC_MSG_WARN([
---- gcc/configure.jj	2007-02-20 16:39:12.000000000 -0500
-+++ gcc/configure	2007-09-22 08:38:30.000000000 -0400
-@@ -7496,7 +7496,7 @@ else
-   echo "configure:7496: version of makeinfo is $ac_prog_version" >&5
-   case $ac_prog_version in
-     '')     gcc_cv_prog_makeinfo_modern=no;;
--    4.[2-9]*)
-+    4.[2-9]*|4.[1-9][0-9]*|[5-9]*|[1-9][0-9]*)
-             gcc_cv_prog_makeinfo_modern=yes;;
-     *)      gcc_cv_prog_makeinfo_modern=no;;
-   esac
diff --git a/gcc41-ppc-sync-qihi.patch b/gcc41-ppc-sync-qihi.patch
deleted file mode 100644
index cf73e8a..0000000
--- a/gcc41-ppc-sync-qihi.patch
+++ /dev/null
@@ -1,82 +0,0 @@
-2007-08-29  Jakub Jelinek  <jakub@redhat.com>
-
-	* config/rs6000/rs6000.c (rs6000_emit_sync): For QI or HI mode
-	used_m, even if it is 32-bit aligned, adjust used_m MEM to have
-	SImode and update m.  Don't run gen_lowpart_common on arbitrary
-	memory address, force it to register first.
-
-	* gcc.dg/sync-2.c (AI_ALIGN): Define if not defined.
-	(AI): Add AI_ALIGN.
-	* gcc.dg/sync-3.c: New test.
-
---- gcc/config/rs6000/rs6000.c.jj	2007-08-27 14:56:36.000000000 +0200
-+++ gcc/config/rs6000/rs6000.c	2007-08-29 22:15:41.000000000 +0200
-@@ -12757,14 +12757,16 @@ rs6000_emit_sync (enum rtx_code code, en
- 	    ishift = GET_MODE_BITSIZE (SImode) - GET_MODE_BITSIZE (mode);
- 
- 	  shift = GEN_INT (ishift);
-+	  used_m = change_address (used_m, SImode, 0);
- 	}
-       else
- 	{
- 	  rtx addrSI, aligned_addr;
- 	  int shift_mask = mode == QImode ? 0x18 : 0x10;
- 
--	  addrSI = force_reg (SImode, gen_lowpart_common (SImode,
--							  XEXP (used_m, 0)));
-+	  addrSI = gen_lowpart_common (SImode,
-+				       force_reg (Pmode, XEXP (used_m, 0)));
-+	  addrSI = force_reg (SImode, addrSI);
- 	  shift = gen_reg_rtx (SImode);
- 
- 	  emit_insn (gen_rlwinm (shift, addrSI, GEN_INT (3),
-@@ -12777,14 +12779,14 @@ rs6000_emit_sync (enum rtx_code code, en
- 				       1, OPTAB_LIB_WIDEN);
- 	  used_m = change_address (used_m, SImode, aligned_addr);
- 	  set_mem_align (used_m, 32);
--	  /* It's safe to keep the old alias set of USED_M, because
--	     the operation is atomic and only affects the original
--	     USED_M.  */
--	  if (GET_CODE (m) == NOT)
--	    m = gen_rtx_NOT (SImode, used_m);
--	  else
--	    m = used_m;
- 	}
-+      /* It's safe to keep the old alias set of USED_M, because
-+	 the operation is atomic and only affects the original
-+	 USED_M.  */
-+      if (GET_CODE (m) == NOT)
-+	m = gen_rtx_NOT (SImode, used_m);
-+      else
-+	m = used_m;
- 
-       if (GET_CODE (op) == NOT)
- 	{
---- gcc/testsuite/gcc.dg/sync-2.c.jj	2006-10-05 00:25:32.000000000 +0200
-+++ gcc/testsuite/gcc.dg/sync-2.c	2007-08-29 21:49:47.000000000 +0200
-@@ -8,7 +8,11 @@
- extern void abort (void);
- extern void *memcpy (void *, const void *, __SIZE_TYPE__);
- 
--static char AI[18];
-+#ifndef AI_ALIGN
-+#define AI_ALIGN
-+#endif
-+
-+static char AI[18] AI_ALIGN;
- static char init_qi[18] = { 3,5,7,9,0,0,0,0,-1,0,0,0,0,0,-1,0,0,0 };
- static char test_qi[18] = { 3,5,7,9,1,4,22,-12,7,8,9,7,1,-12,7,8,9,7 };
- 
---- gcc/testsuite/gcc.dg/sync-3.c.jj	2007-08-29 21:49:59.000000000 +0200
-+++ gcc/testsuite/gcc.dg/sync-3.c	2007-08-29 22:00:31.000000000 +0200
-@@ -0,0 +1,10 @@
-+/* { dg-do run } */
-+/* { dg-require-effective-target sync_char_short } */
-+/* { dg-options "-O2" } */
-+/* { dg-options "-march=i486 -O2" { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
-+/* { dg-options "-mcpu=v9 -O2" { target sparc*-*-* } } */
-+
-+/* Test functionality of the intrinsics for 'short' and 'char'.  */
-+
-+#define AI_ALIGN __attribute__((__aligned__ (4)))
-+#include "sync-2.c"
diff --git a/gcc41-ppc-tramp.patch b/gcc41-ppc-tramp.patch
deleted file mode 100644
index 145b628..0000000
--- a/gcc41-ppc-tramp.patch
+++ /dev/null
@@ -1,34 +0,0 @@
-2007-08-20  Jakub Jelinek  <jakub@redhat.com>
-
-	* config/rs6000/tramp.asm: Include config.h.
-	Check __PIC__ or __pic__ macro instead of SHARED.
-
---- gcc/config/rs6000/tramp.asm.jj	2006-10-05 00:28:33.000000000 +0200
-+++ gcc/config/rs6000/tramp.asm	2007-08-20 23:20:52.000000000 +0200
-@@ -1,6 +1,6 @@
- /*  Special support for trampolines
-  *
-- *   Copyright (C) 1996, 1997, 2000 Free Software Foundation, Inc.
-+ *   Copyright (C) 1996, 1997, 2000, 2007 Free Software Foundation, Inc.
-  *   Written By Michael Meissner
-  * 
-  * This file is free software; you can redistribute it and/or modify it
-@@ -37,7 +37,8 @@
- 
- 	.file	"tramp.asm"
- 	.section ".text"
--	#include "ppc-asm.h"
-+#include "ppc-asm.h"
-+#include "config.h"
- 
- #ifndef __powerpc64__
- 	.type	trampoline_initial,@object
-@@ -105,7 +106,7 @@ FUNC_START(__trampoline_setup)
- 	blr
- 
- .Labort:
--#if defined SHARED && defined HAVE_AS_REL16
-+#if (defined __PIC__ || defined __pic__) && defined HAVE_AS_REL16
- 	bcl	20,31,1f
- 1:	mflr	r30
- 	addis	r30,r30,_GLOBAL_OFFSET_TABLE_-1b@ha
diff --git a/gcc41-pr20880.patch b/gcc41-pr20880.patch
new file mode 100644
index 0000000..be72a81
--- /dev/null
+++ b/gcc41-pr20880.patch
@@ -0,0 +1,59 @@
+2006-11-24  Paul Thomas  <pault@gcc.gnu.org>
+
+	PR fortran/20880
+	* parse.c (parse_interface): Error if procedure name is that of
+	encompassing scope.
+
+	* gfortran.dg/interface_3a.f90: New test.
+
+--- gcc/fortran/parse.c	(revision 119172)
++++ gcc/fortran/parse.c	(revision 119173)
+@@ -1694,6 +1694,7 @@ parse_interface (void)
+   gfc_interface_info save;
+   gfc_state_data s1, s2;
+   gfc_statement st;
++  locus proc_locus;
+ 
+   accept_statement (ST_INTERFACE);
+ 
+@@ -1781,6 +1782,7 @@ loop:
+   accept_statement (st);
+   prog_unit = gfc_new_block;
+   prog_unit->formal_ns = gfc_current_ns;
++  proc_locus = gfc_current_locus;
+ 
+ decl:
+   /* Read data declaration statements.  */
+@@ -1796,8 +1798,15 @@ decl:
+ 
+   current_interface = save;
+   gfc_add_interface (prog_unit);
+-
+   pop_state ();
++
++  if (current_interface.ns
++	&& current_interface.ns->proc_name
++	&& strcmp (current_interface.ns->proc_name->name,
++		   prog_unit->name) == 0)
++    gfc_error ("INTERFACE procedure '%s' at %L has the same name as the "
++	       "enclosing procedure", prog_unit->name, &proc_locus);
++
+   goto loop;
+ 
+ done:
+--- gcc/testsuite/gfortran.dg/interface_3a.f90
++++ gcc/testsuite/gfortran.dg/interface_3a.f90
+@@ -0,0 +1,13 @@
++! { dg-do compile }
++! Contributed by Joost VandeVondele  <jv244@cam.ac.uk>
++!
++! This was found whilst investigating => segfault
++subroutine thy_sub (a)
++  interface 
++    subroutine thy_sub (a) ! { dg-error "enclosing procedure" }
++      real a
++    end subroutine
++  end interface
++  real a
++  print *, a
++end subroutine
diff --git a/gcc41-pr22244.patch b/gcc41-pr22244.patch
deleted file mode 100644
index b7e037e..0000000
--- a/gcc41-pr22244.patch
+++ /dev/null
@@ -1,63 +0,0 @@
-2007-08-14  Jakub Jelinek  <jakub@redhat.com>
-
-	PR fortran/22244
-	* Make-lang.in (fortran/trans-types.o): Depend on $(FLAGS_H).
-	* trans-types.c: Include flags.h.
-	(gfc_get_nodesc_array_type): Add TYPE_DECL TYPE_NAME with
-	correct bounds and dimensions for packed arrays.
-
---- gcc/fortran/Make-lang.in	(revision 127395)
-+++ gcc/fortran/Make-lang.in	(working copy)
-@@ -292,7 +292,7 @@ fortran/trans-decl.o: $(GFORTRAN_TRANS_D
-   cgraph.h $(TARGET_H) function.h $(FLAGS_H) $(RTL_H) tree-gimple.h \
-   tree-dump.h
- fortran/trans-types.o: $(GFORTRAN_TRANS_DEPS) gt-fortran-trans-types.h \
--  real.h toplev.h $(TARGET_H)
-+  real.h toplev.h $(TARGET_H) $(FLAGS_H)
- fortran/trans-const.o: $(GFORTRAN_TRANS_DEPS)
- fortran/trans-expr.o: $(GFORTRAN_TRANS_DEPS) fortran/dependency.h
- fortran/trans-stmt.o: $(GFORTRAN_TRANS_DEPS) fortran/dependency.h
---- gcc/fortran/trans-types.c	(revision 127395)
-+++ gcc/fortran/trans-types.c	(working copy)
-@@ -35,6 +35,7 @@ Software Foundation, 51 Franklin Street,
- #include "trans-types.h"
- #include "trans-const.h"
- #include "real.h"
-+#include "flags.h"
- 
- 
- #if (GFC_MAX_DIMENSIONS < 10)
-@@ -1005,7 +1006,7 @@ gfc_get_nodesc_array_type (tree etype, g
-     {
-       /* Fill in the stride and bound components of the type.  */
-       if (known_stride)
--	tmp =  gfc_conv_mpz_to_tree (stride, gfc_index_integer_kind);
-+	tmp = gfc_conv_mpz_to_tree (stride, gfc_index_integer_kind);
-       else
-         tmp = NULL_TREE;
-       GFC_TYPE_ARRAY_STRIDE (type, n) = tmp;
-@@ -1103,6 +1104,24 @@ gfc_get_nodesc_array_type (tree etype, g
-   mpz_clear (stride);
-   mpz_clear (delta);
- 
-+  /* In debug info represent packed arrays as multi-dimensional
-+     if they have rank > 1 and with proper bounds, instead of flat
-+     arrays.  */
-+  if (known_stride && write_symbols != NO_DEBUG)
-+    {
-+      tree gtype = etype, rtype, type_decl;
-+
-+      for (n = as->rank - 1; n >= 0; n--)
-+	{
-+	  rtype = build_range_type (gfc_array_index_type,
-+				    GFC_TYPE_ARRAY_LBOUND (type, n),
-+				    GFC_TYPE_ARRAY_UBOUND (type, n));
-+	  gtype = build_array_type (gtype, rtype);
-+	}
-+      TYPE_NAME (type) = type_decl = build_decl (TYPE_DECL, NULL, gtype);
-+      DECL_ORIGINAL_TYPE (type_decl) = gtype;
-+    }
-+
-   if (packed < 3 || !known_stride)
-     {
-       /* For dummy arrays and automatic (heap allocated) arrays we
diff --git a/gcc41-pr27954.patch b/gcc41-pr27954.patch
deleted file mode 100644
index db11c3a..0000000
--- a/gcc41-pr27954.patch
+++ /dev/null
@@ -1,84 +0,0 @@
-2006-10-27  Jerry DeLisle  <jvdelisle@gcc.gnu.org>
-
-	PR fortran/27954
-	* decl.c (gfc_free_data_all): New function to free all data structures
-	after errors in DATA statements and declarations.
-	(top_var_list): Use new function.(top_val_list): Use new function.
-	(gfc_match_data_decl): Use new function.
-
-	PR libgfortran/27954
-	* gfortran.dg/error_recovery_2.f90: New test.
-
---- gcc/fortran/decl.c	(revision 118083)
-+++ gcc/fortran/decl.c	(revision 118086)
-@@ -128,6 +128,21 @@ gfc_free_data (gfc_data * p)
- }
- 
- 
-+/* Free all data in a namespace.  */
-+static void
-+gfc_free_data_all (gfc_namespace * ns)
-+{
-+  gfc_data *d;
-+
-+  for (;ns->data;)
-+    {
-+      d = ns->data->next;
-+      gfc_free (ns->data);
-+      ns->data = d;
-+    }
-+}
-+
-+
- static match var_element (gfc_data_variable *);
- 
- /* Match a list of variables terminated by an iterator and a right
-@@ -262,6 +277,7 @@ top_var_list (gfc_data * d)
- 
- syntax:
-   gfc_syntax_error (ST_DATA);
-+  gfc_free_data_all (gfc_current_ns);
-   return MATCH_ERROR;
- }
- 
-@@ -374,6 +390,7 @@ top_val_list (gfc_data * data)
- 
- syntax:
-   gfc_syntax_error (ST_DATA);
-+  gfc_free_data_all (gfc_current_ns);
-   return MATCH_ERROR;
- }
- 
-@@ -2368,6 +2385,8 @@ ok:
-   gfc_error ("Syntax error in data declaration at %C");
-   m = MATCH_ERROR;
- 
-+  gfc_free_data_all (gfc_current_ns);
-+
- cleanup:
-   gfc_free_array_spec (current_as);
-   current_as = NULL;
---- gcc/testsuite/gfortran.dg/error_recovery_2.f90	(revision 0)
-+++ gcc/testsuite/gfortran.dg/error_recovery_2.f90	(revision 118086)
-@@ -0,0 +1,21 @@
-+! { dg-do compile }
-+! PR27954 Internal compiler error on bad statements
-+! Derived from test case submitted in PR.
-+subroutine bad1
-+  character*20 :: y, x 00  ! { dg-error "Syntax error" }
-+  data  y /'abcdef'/, x /'jbnhjk'/ pp  ! { dg-error "Syntax error" }
-+end subroutine bad1
-+
-+subroutine bad2
-+  character*20 :: y, x 00  ! { dg-error "Syntax error" }
-+  data  y /'abcdef'/, x /'jbnhjk'/ pp  ! { dg-error "Syntax error" }
-+  print *, "basket case."
-+end subroutine bad2
-+
-+subroutine bad3
-+  implicit none
-+  character*20 :: y, x 00  ! { dg-error "Syntax error" }
-+  data  y /'abcdef'/, x /'jbnhjk'/ pp  ! { dg-error "Syntax error" }
-+  print *, "basket case that segfaults without patch."
-+end subroutine bad3
-+
diff --git a/gcc41-pr28755.patch b/gcc41-pr28755.patch
index bef8ce8..d6bb90a 100644
--- a/gcc41-pr28755.patch
+++ b/gcc41-pr28755.patch
@@ -1,44 +1,210 @@
-2006-08-17  Jakub Jelinek  <jakub@redhat.com>
+2007-09-23  Jakub Jelinek  <jakub@redhat.com>
 
 	PR middle-end/28755
-	* expr.c (expand_expr_real_1) <case ARRAY_REF>: Make sure
-	the const array field optimization doesn't create an extra constant
-	MEM.
+	* expr.c (expand_constructor): New function.
+	(expand_expr_real_1) <case CONSTRUCTOR>: Call it.
+	(expand_expr_real_1) <case ARRAY_REF>: Call it if VALUE is
+	CONSTRUCTOR.
 
 	* gcc.dg/pr28755.c: New test.
 
---- gcc/expr.c.jj	2006-08-09 17:54:03.000000000 +0200
-+++ gcc/expr.c	2006-08-17 13:00:33.000000000 +0200
-@@ -7105,13 +7105,25 @@ expand_expr_real_1 (tree exp, rtx target
+--- gcc/expr.c	(revision 128684)
++++ gcc/expr.c	(revision 128685)
+@@ -6435,6 +6435,89 @@ expand_expr_addr_expr (tree exp, rtx tar
+   return result;
+ }
+ 
++/* Generate code for computing CONSTRUCTOR EXP.
++   An rtx for the computed value is returned.  If AVOID_TEMP_MEM
++   is TRUE, instead of creating a temporary variable in memory
++   NULL is returned and the caller needs to handle it differently.  */
++
++static rtx
++expand_constructor (tree exp, rtx target, enum expand_modifier modifier,
++		    bool avoid_temp_mem)
++{
++  tree type = TREE_TYPE (exp);
++  enum machine_mode mode = TYPE_MODE (type);
++
++  /* Try to avoid creating a temporary at all.  This is possible
++     if all of the initializer is zero.
++     FIXME: try to handle all [0..255] initializers we can handle
++     with memset.  */
++  if (TREE_STATIC (exp)
++      && !TREE_ADDRESSABLE (exp)
++      && target != 0 && mode == BLKmode
++      && all_zeros_p (exp))
++    {
++      clear_storage (target, expr_size (exp), BLOCK_OP_NORMAL);
++      return target;
++    }
++
++  /* All elts simple constants => refer to a constant in memory.  But
++     if this is a non-BLKmode mode, let it store a field at a time
++     since that should make a CONST_INT or CONST_DOUBLE when we
++     fold.  Likewise, if we have a target we can use, it is best to
++     store directly into the target unless the type is large enough
++     that memcpy will be used.  If we are making an initializer and
++     all operands are constant, put it in memory as well.
++
++     FIXME: Avoid trying to fill vector constructors piece-meal.
++     Output them with output_constant_def below unless we're sure
++     they're zeros.  This should go away when vector initializers
++     are treated like VECTOR_CST instead of arrays.  */
++  if ((TREE_STATIC (exp)
++       && ((mode == BLKmode
++	    && ! (target != 0 && safe_from_p (target, exp, 1)))
++		  || TREE_ADDRESSABLE (exp)
++		  || (host_integerp (TYPE_SIZE_UNIT (type), 1)
++		      && (! MOVE_BY_PIECES_P
++				     (tree_low_cst (TYPE_SIZE_UNIT (type), 1),
++				      TYPE_ALIGN (type)))
++		      && ! mostly_zeros_p (exp))))
++      || ((modifier == EXPAND_INITIALIZER || modifier == EXPAND_CONST_ADDRESS)
++	  && TREE_CONSTANT (exp)))
++    {
++      rtx constructor;
++
++      if (avoid_temp_mem)
++	return NULL_RTX;
++
++      constructor = output_constant_def (exp, 1);
++
++      if (modifier != EXPAND_CONST_ADDRESS
++	  && modifier != EXPAND_INITIALIZER
++	  && modifier != EXPAND_SUM)
++	constructor = validize_mem (constructor);
++
++      return constructor;
++    }
++
++  /* Handle calls that pass values in multiple non-contiguous
++     locations.  The Irix 6 ABI has examples of this.  */
++  if (target == 0 || ! safe_from_p (target, exp, 1)
++      || GET_CODE (target) == PARALLEL || modifier == EXPAND_STACK_PARM)
++    {
++      if (avoid_temp_mem)
++	return NULL_RTX;
++
++      target
++	= assign_temp (build_qualified_type (type, (TYPE_QUALS (type)
++						    | (TREE_READONLY (exp)
++						       * TYPE_QUAL_CONST))),
++		       0, TREE_ADDRESSABLE (exp), 1);
++    }
++
++  store_constructor (exp, target, 0, int_expr_size (exp));
++  return target;
++}
++
+ 
+ /* expand_expr: generate code for computing expression EXP.
+    An rtx for the computed value is returned.  The value is never null.
+@@ -6899,71 +6982,7 @@ expand_expr_real_1 (tree exp, rtx target
+ 	  return const0_rtx;
+ 	}
+ 
+-      /* Try to avoid creating a temporary at all.  This is possible
+-	 if all of the initializer is zero.
+-	 FIXME: try to handle all [0..255] initializers we can handle
+-	 with memset.  */
+-      else if (TREE_STATIC (exp)
+-	       && !TREE_ADDRESSABLE (exp)
+-	       && target != 0 && mode == BLKmode
+-	       && all_zeros_p (exp))
+-	{
+-	  clear_storage (target, expr_size (exp), BLOCK_OP_NORMAL);
+-	  return target;
+-	}
+-
+-      /* All elts simple constants => refer to a constant in memory.  But
+-	 if this is a non-BLKmode mode, let it store a field at a time
+-	 since that should make a CONST_INT or CONST_DOUBLE when we
+-	 fold.  Likewise, if we have a target we can use, it is best to
+-	 store directly into the target unless the type is large enough
+-	 that memcpy will be used.  If we are making an initializer and
+-	 all operands are constant, put it in memory as well.
+-
+-	FIXME: Avoid trying to fill vector constructors piece-meal.
+-	Output them with output_constant_def below unless we're sure
+-	they're zeros.  This should go away when vector initializers
+-	are treated like VECTOR_CST instead of arrays.
+-      */
+-      else if ((TREE_STATIC (exp)
+-		&& ((mode == BLKmode
+-		     && ! (target != 0 && safe_from_p (target, exp, 1)))
+-		    || TREE_ADDRESSABLE (exp)
+-		    || (host_integerp (TYPE_SIZE_UNIT (type), 1)
+-			&& (! MOVE_BY_PIECES_P
+-			    (tree_low_cst (TYPE_SIZE_UNIT (type), 1),
+-			     TYPE_ALIGN (type)))
+-			&& ! mostly_zeros_p (exp))))
+-	       || ((modifier == EXPAND_INITIALIZER
+-		    || modifier == EXPAND_CONST_ADDRESS)
+-		   && TREE_CONSTANT (exp)))
+-	{
+-	  rtx constructor = output_constant_def (exp, 1);
+-
+-	  if (modifier != EXPAND_CONST_ADDRESS
+-	      && modifier != EXPAND_INITIALIZER
+-	      && modifier != EXPAND_SUM)
+-	    constructor = validize_mem (constructor);
+-
+-	  return constructor;
+-	}
+-      else
+-	{
+-	  /* Handle calls that pass values in multiple non-contiguous
+-	     locations.  The Irix 6 ABI has examples of this.  */
+-	  if (target == 0 || ! safe_from_p (target, exp, 1)
+-	      || GET_CODE (target) == PARALLEL
+-	      || modifier == EXPAND_STACK_PARM)
+-	    target
+-	      = assign_temp (build_qualified_type (type,
+-						   (TYPE_QUALS (type)
+-						    | (TREE_READONLY (exp)
+-						       * TYPE_QUAL_CONST))),
+-			     0, TREE_ADDRESSABLE (exp), 1);
+-
+-	  store_constructor (exp, target, 0, int_expr_size (exp));
+-	  return target;
+-	}
++      return expand_constructor (exp, target, modifier, false);
+ 
+     case MISALIGNED_INDIRECT_REF:
+     case ALIGN_INDIRECT_REF:
+@@ -7105,10 +7124,25 @@ expand_expr_real_1 (tree exp, rtx target
  					      field, value)
  		      if (tree_int_cst_equal (field, index))
  			{
 -			  if (!TREE_SIDE_EFFECTS (value))
+-			    return expand_expr (fold (value), target, tmode,
+-						modifier);
+-			  break;
 +			  if (TREE_SIDE_EFFECTS (value))
 +			    break;
 +
-+			  if (TREE_CODE (value) != CONSTRUCTOR)
- 			    return expand_expr (fold (value), target, tmode,
- 						modifier);
++			  if (TREE_CODE (value) == CONSTRUCTOR)
++			    {
++			      /* If VALUE is a CONSTRUCTOR, this
++				 optimization is only useful if
++				 this doesn't store the CONSTRUCTOR
++				 into memory.  If it does, it is more
++				 efficient to just load the data from
++				 the array directly.  */
++			      rtx ret = expand_constructor (value, target,
++							    modifier, true);
++			      if (ret == NULL_RTX)
++				break;
++			    }
 +
-+			  /* For CONSTRUCTOR this optimization is not always
-+			     a win - if expand_expr creates a temporary
-+			     constant, we just waste unnecessarily rodata
-+			     space.  */
-+			  temp = expand_expr (value, target, tmode, modifier);
-+			  if (temp == target
-+			      || (temp && GET_CODE (temp) != MEM))
-+			    return temp;
- 			  break;
++			  return expand_expr (fold (value), target, tmode,
++					      modifier);
  			}
  		  }
--		else if(TREE_CODE (init) == STRING_CST)
-+		else if (TREE_CODE (init) == STRING_CST)
- 		  {
- 		    tree index1 = index;
- 		    tree low_bound = array_ref_low_bound (exp);
---- gcc/testsuite/gcc.dg/pr28755.c.jj	2006-08-17 12:59:57.000000000 +0200
-+++ gcc/testsuite/gcc.dg/pr28755.c	2006-08-17 12:59:57.000000000 +0200
+ 		else if(TREE_CODE (init) == STRING_CST)
+--- gcc/testsuite/gcc.dg/pr28755.c	(revision 0)
++++ gcc/testsuite/gcc.dg/pr28755.c	(revision 128685)
 @@ -0,0 +1,22 @@
 +/* PR middle-end/28755 */
 +/* { dg-do compile } */
diff --git a/gcc41-pr32678.patch b/gcc41-pr32678.patch
deleted file mode 100644
index 61692bb..0000000
--- a/gcc41-pr32678.patch
+++ /dev/null
@@ -1,63 +0,0 @@
-2007-07-21  Jerry DeLisle  <jvdelisle@gcc.gnu.org>
-
-	PR libgfortran/32678
-	* io/transfer.c (formatted_transfer_scalar): Fix off by one error in
-	calculation of pos and skips. Don't allow pending_spaces to go
-	negative.
-
-	PR fortran/32678
-	* gfortran.dg/fmt_t_5.f90: New test.
-
---- libgfortran/io/transfer.c	(revision 126821)
-+++ libgfortran/io/transfer.c	(revision 126823)
-@@ -893,9 +893,9 @@ formatted_transfer_scalar (st_parameter_
- 	case FMT_TR:
- 	  consume_data_flag = 0 ;
- 
--	  pos = bytes_used + f->u.n + dtp->u.p.skips;
--	  dtp->u.p.skips = f->u.n + dtp->u.p.skips;
--	  dtp->u.p.pending_spaces = pos - dtp->u.p.max_pos;
-+	  dtp->u.p.skips += f->u.n;
-+	  pos = bytes_used + dtp->u.p.skips - 1;
-+	  dtp->u.p.pending_spaces = pos - dtp->u.p.max_pos + 1;
- 
- 	  /* Writes occur just before the switch on f->format, above, so
- 	     that trailing blanks are suppressed, unless we are doing a
-@@ -922,8 +922,6 @@ formatted_transfer_scalar (st_parameter_
- 	      if (bytes_used == 0)
- 		{
- 		  dtp->u.p.pending_spaces -= f->u.n;
--		  dtp->u.p.pending_spaces = dtp->u.p.pending_spaces < 0 ? 0
--					    : dtp->u.p.pending_spaces;
- 		  dtp->u.p.skips -= f->u.n;
- 		  dtp->u.p.skips = dtp->u.p.skips < 0 ? 0 : dtp->u.p.skips;
- 		}
-@@ -945,6 +943,8 @@ formatted_transfer_scalar (st_parameter_
- 	  dtp->u.p.skips = dtp->u.p.skips + pos - bytes_used;
- 	  dtp->u.p.pending_spaces = dtp->u.p.pending_spaces
- 				    + pos - dtp->u.p.max_pos;
-+	  dtp->u.p.pending_spaces = dtp->u.p.pending_spaces < 0
-+				    ? 0 : dtp->u.p.pending_spaces;
- 
- 	  if (dtp->u.p.skips == 0)
- 	    break;
---- gcc/testsuite/gfortran.dg/fmt_t_5.f90	(revision 126821)
-+++ gcc/testsuite/gfortran.dg/fmt_t_5.f90	(revision 126823)
-@@ -0,0 +1,17 @@
-+! { dg-do run }
-+! PR32678 GFortan works incorrectly when writing with FORMAT Tx
-+! Before patch, NULLs were inserted in output.
-+! Test case from reporter enhanced to detect this problem.
-+      character(25) :: output
-+      character(1)  :: c
-+      output = ""
-+      open (unit=10, file="pr32678testfile", status="replace")
-+      write (10,10) '12','a','b'
-+      close (10, status="keep")
-+      open (unit=10, file="pr32678testfile")
-+      read(10,20) output(1:21)
-+      if (output(1:21).ne."ab                  x") call abort
-+      close (10, status="delete")
-+ 10   format (a2,t1,a1,t2,a1,t20,' x')
-+ 20   format (a21)
-+      end
diff --git a/gcc41-pr32694.patch b/gcc41-pr32694.patch
new file mode 100644
index 0000000..4b172e4
--- /dev/null
+++ b/gcc41-pr32694.patch
@@ -0,0 +1,46 @@
+2007-09-25  Jakub Jelinek  <jakub@redhat.com>
+
+	PR tree-optimization/32694
+	2006-11-08  Roger Sayle  <roger@eyesopen.com>
+	* tree-ssa-propagate.c (set_rhs): Verify tcc_comparison the same way
+	as tcc_binary.
+
+	* gcc.c-torture/compile/20070925-1.c: New test.
+
+--- gcc/tree-ssa-propagate.c.jj	2007-09-23 19:43:36.000000000 +0200
++++ gcc/tree-ssa-propagate.c	2007-09-25 09:30:50.000000000 +0200
+@@ -570,7 +570,8 @@ set_rhs (tree *stmt_p, tree expr)
+   ssa_op_iter iter;
+ 
+   /* Verify the constant folded result is valid gimple.  */
+-  if (TREE_CODE_CLASS (code) == tcc_binary)
++  if (TREE_CODE_CLASS (code) == tcc_binary
++      || TREE_CODE_CLASS (code) == tcc_comparison)
+     {
+       if (!is_gimple_val (TREE_OPERAND (expr, 0))
+ 	  || !is_gimple_val (TREE_OPERAND (expr, 1)))
+--- gcc/testsuite/gcc.c-torture/compile/20070925-1.c.jj	2007-09-25 09:28:37.000000000 +0200
++++ gcc/testsuite/gcc.c-torture/compile/20070925-1.c	2007-09-25 09:29:41.000000000 +0200
+@@ -0,0 +1,22 @@
++/* PR tree-optimization/32694 */
++
++typedef signed long long int WordS64;
++typedef unsigned long long int Word64;
++
++int
++foo (Word64 * p)
++{
++  while (1)
++    {
++      WordS64 c = 0x1llu;
++      WordS64 x = *p;
++      if (c >= 0)
++	{
++	  if (x > (WordS64) 0x7FFFFFFFFFFFFFFFll - c)
++	    return 6;
++	}
++      else if (x < (WordS64) 0x8000000000000000ll - c)
++	return 7;
++      p++;
++    }
++}
diff --git a/gcc41-pr32912.patch b/gcc41-pr32912.patch
deleted file mode 100644
index 67ce830..0000000
--- a/gcc41-pr32912.patch
+++ /dev/null
@@ -1,138 +0,0 @@
-2007-08-20  Jakub Jelinek  <jakub@redhat.com>
-
-	PR middle-end/32912
-	* fold-const.c (fold_binary): Only optimize X | ~X and X ^ ~X for
-	integral types.
-
-	* gcc.dg/pr32912-1.c: New test.
-	* gcc.dg/pr32912-2.c: New test.
-
---- gcc/fold-const.c.jj	2007-08-13 15:11:18.000000000 +0200
-+++ gcc/fold-const.c	2007-08-20 15:49:05.000000000 +0200
-@@ -8079,6 +8079,7 @@ fold_binary (enum tree_code code, tree t
- 
-       /* ~X | X is -1.  */
-       if (TREE_CODE (arg0) == BIT_NOT_EXPR
-+	  && INTEGRAL_TYPE_P (TREE_TYPE (arg1))
- 	  && operand_equal_p (TREE_OPERAND (arg0, 0), arg1, 0))
- 	{
- 	  t1 = build_int_cst (type, -1);
-@@ -8088,6 +8089,7 @@ fold_binary (enum tree_code code, tree t
- 
-       /* X | ~X is -1.  */
-       if (TREE_CODE (arg1) == BIT_NOT_EXPR
-+	  && INTEGRAL_TYPE_P (TREE_TYPE (arg0))
- 	  && operand_equal_p (arg0, TREE_OPERAND (arg1, 0), 0))
- 	{
- 	  t1 = build_int_cst (type, -1);
-@@ -8175,6 +8177,7 @@ fold_binary (enum tree_code code, tree t
- 
-       /* ~X ^ X is -1.  */
-       if (TREE_CODE (arg0) == BIT_NOT_EXPR
-+	  && INTEGRAL_TYPE_P (TREE_TYPE (arg1))
- 	  && operand_equal_p (TREE_OPERAND (arg0, 0), arg1, 0))
- 	{
- 	  t1 = build_int_cst (type, -1);
-@@ -8184,6 +8187,7 @@ fold_binary (enum tree_code code, tree t
- 
-       /* X ^ ~X is -1.  */
-       if (TREE_CODE (arg1) == BIT_NOT_EXPR
-+	  && INTEGRAL_TYPE_P (TREE_TYPE (arg0))
- 	  && operand_equal_p (arg0, TREE_OPERAND (arg1, 0), 0))
- 	{
- 	  t1 = build_int_cst (type, -1);
---- gcc/testsuite/gcc.dg/pr32912-1.c.jj	2007-08-20 14:43:05.000000000 +0200
-+++ gcc/testsuite/gcc.dg/pr32912-1.c	2007-08-20 14:43:23.000000000 +0200
-@@ -0,0 +1,44 @@
-+/* PR middle-end/32912 */
-+/* { dg-do run } */
-+/* { dg-options "-O2 -w" } */
-+
-+extern void abort (void);
-+
-+typedef int __m128i __attribute__ ((__vector_size__ (16)));
-+
-+__m128i a, b, c, d, e, f;
-+
-+void
-+foo (__m128i x)
-+{
-+  a = x ^ ~x;
-+  b = ~x ^ x;
-+  c = x | ~x;
-+  d = ~x | x;
-+  e = x & ~x;
-+  f = ~x & x;
-+}
-+
-+int
-+main (void)
-+{
-+  union { __m128i v; int i[sizeof (__m128i) / sizeof (int)]; } u;
-+  int i;
-+
-+  for (i = 0; i < sizeof (u.i) / sizeof (u.i[0]); i++)
-+    u.i[i] = i * 49 - 36;
-+  foo (u.v);
-+#define check(x, val) \
-+  u.v = (x); \
-+  for (i = 0; i < sizeof (u.i) / sizeof (u.i[0]); i++) \
-+    if (u.i[i] != (val)) \
-+      abort ()
-+
-+  check (a, ~0);
-+  check (b, ~0);
-+  check (c, ~0);
-+  check (d, ~0);
-+  check (e, 0);
-+  check (f, 0);
-+  return 0;
-+}
---- gcc/testsuite/gcc.dg/pr32912-2.c.jj	2007-08-20 15:58:47.000000000 +0200
-+++ gcc/testsuite/gcc.dg/pr32912-2.c	2007-08-20 15:55:32.000000000 +0200
-@@ -0,0 +1,45 @@
-+/* { dg-do run } */
-+/* { dg-options "-O2 -w" } */
-+
-+extern void abort (void);
-+
-+typedef int __m128i __attribute__ ((__vector_size__ (16)));
-+
-+__m128i a, b, c, d, e, f;
-+
-+__m128i
-+foo (void)
-+{
-+  __m128i x = { 0x11111111, 0x22222222, 0x44444444 };
-+  return x;
-+}
-+
-+__m128i
-+bar (void)
-+{
-+  __m128i x = { 0x11111111, 0x22222222, 0x44444444 };
-+  return ~x;
-+}
-+
-+int
-+main (void)
-+{
-+  union { __m128i v; int i[sizeof (__m128i) / sizeof (int)]; } u, v;
-+  int i;
-+
-+  u.v = foo ();
-+  v.v = bar ();
-+  for (i = 0; i < sizeof (u.i) / sizeof (u.i[0]); i++)
-+    {
-+      if (u.i[i] != ~v.i[i])
-+	abort ();
-+      if (i < 3)
-+	{
-+	  if (u.i[i] != (0x11111111 << i))
-+	    abort ();
-+	}
-+      else if (u.i[i])
-+	abort ();
-+    }
-+  return 0;
-+}
diff --git a/gcc41-pr33423.patch b/gcc41-pr33423.patch
deleted file mode 100644
index 46cfe29..0000000
--- a/gcc41-pr33423.patch
+++ /dev/null
@@ -1,61 +0,0 @@
-2007-09-15  Jakub Jelinek  <jakub@redhat.com>
-
-	PR middle-end/33423
-	* builtins.c (expand_builtin_memory_chk): Handle COMPOUND_EXPRs
-	returned by build_call_expr.
-
-	* gcc.c-torture/compile/20070915-1.c: New test.
-
---- gcc/builtins.c.jj	2007-09-14 21:07:41.000000000 +0200
-+++ gcc/builtins.c	2007-09-16 22:31:40.000000000 +0200
-@@ -10383,6 +10383,13 @@ expand_builtin_memory_chk (tree exp, rtx
- 	return 0;
- 
-       fn = build_function_call_expr (fn, arglist);
-+      STRIP_TYPE_NOPS (fn);
-+      while (TREE_CODE (fn) == COMPOUND_EXPR)
-+	{
-+	  expand_expr (TREE_OPERAND (fn, 0), const0_rtx, VOIDmode,
-+		       EXPAND_NORMAL);
-+	  fn = TREE_OPERAND (fn, 1);
-+	}
-       if (TREE_CODE (fn) == CALL_EXPR)
- 	CALL_EXPR_TAILCALL (fn) = CALL_EXPR_TAILCALL (exp);
-       return expand_expr (fn, target, mode, EXPAND_NORMAL);
-@@ -10432,6 +10439,13 @@ expand_builtin_memory_chk (tree exp, rtx
- 	      if (!fn)
- 		return 0;
- 	      fn = build_function_call_expr (fn, arglist);
-+	      STRIP_TYPE_NOPS (fn);
-+	      while (TREE_CODE (fn) == COMPOUND_EXPR)
-+		{
-+		  expand_expr (TREE_OPERAND (fn, 0), const0_rtx, VOIDmode,
-+			       EXPAND_NORMAL);
-+		  fn = TREE_OPERAND (fn, 1);
-+		}
- 	      if (TREE_CODE (fn) == CALL_EXPR)
- 		CALL_EXPR_TAILCALL (fn) = CALL_EXPR_TAILCALL (exp);
- 	      return expand_expr (fn, target, mode, EXPAND_NORMAL);
---- gcc/testsuite/gcc.c-torture/compile/20070915-1.c.jj	2007-09-16 22:29:37.000000000 +0200
-+++ gcc/testsuite/gcc.c-torture/compile/20070915-1.c	2007-09-16 22:29:37.000000000 +0200
-@@ -0,0 +1,20 @@
-+/* PR middle-end/33423 */
-+
-+static struct
-+{
-+  char buf[15];
-+} u2;
-+
-+void
-+test6 (void)
-+{
-+  int len;
-+  char *p;
-+
-+  for (len = 0; len < 2; len++)
-+    {
-+      p = __builtin___memset_chk (u2.buf, '\0', len, 15);
-+      if (p != u2.buf)
-+	return;
-+    }
-+}
diff --git a/gcc41-rh253102.patch b/gcc41-rh253102.patch
deleted file mode 100644
index d4cad4f..0000000
--- a/gcc41-rh253102.patch
+++ /dev/null
@@ -1,89 +0,0 @@
-2007-08-17  Jakub Jelinek  <jakub@redhat.com>
-
-	* decl.c (variable_decl): Don't share charlen structs if
-	length == NULL.
-	* trans-decl.c (create_function_arglist): Assert
-	f->sym->ts.cl->backend_decl is NULL instead of unsharing
-	charlen struct here.
-
-	* gfortran.dg/assumed_charlen_sharing.f90: New test.
-
---- gcc/fortran/decl.c.jj	2007-02-20 22:38:20.000000000 +0100
-+++ gcc/fortran/decl.c	2007-08-21 20:50:33.000000000 +0200
-@@ -1086,10 +1086,11 @@ variable_decl (int elem)
- 	  break;
- 
- 	/* Non-constant lengths need to be copied after the first
--	   element.  */
-+	   element.  Also copy assumed lengths.  */
- 	case MATCH_NO:
--	  if (elem > 1 && current_ts.cl->length
--		&& current_ts.cl->length->expr_type != EXPR_CONSTANT)
-+	  if (elem > 1
-+	      && (current_ts.cl->length == NULL
-+		  || current_ts.cl->length->expr_type != EXPR_CONSTANT))
- 	    {
- 	      cl = gfc_get_charlen ();
- 	      cl->next = gfc_current_ns->cl_list;
---- gcc/fortran/trans-decl.c.jj	2007-03-12 08:28:13.000000000 +0100
-+++ gcc/fortran/trans-decl.c	2007-08-21 20:50:33.000000000 +0200
-@@ -1417,25 +1417,8 @@ create_function_arglist (gfc_symbol * sy
- 	  if (!f->sym->ts.cl->length)
- 	    {
- 	      TREE_USED (length) = 1;
--	      if (!f->sym->ts.cl->backend_decl)
--		f->sym->ts.cl->backend_decl = length;
--	      else
--		{
--		  /* there is already another variable using this
--		     gfc_charlen node, build a new one for this variable
--		     and chain it into the list of gfc_charlens.
--		     This happens for e.g. in the case
--		     CHARACTER(*)::c1,c2
--		     since CHARACTER declarations on the same line share
--		     the same gfc_charlen node.  */
--		  gfc_charlen *cl;
--	      
--		  cl = gfc_get_charlen ();
--		  cl->backend_decl = length;
--		  cl->next = f->sym->ts.cl->next;
--		  f->sym->ts.cl->next = cl;
--		  f->sym->ts.cl = cl;
--		}
-+	      gcc_assert (!f->sym->ts.cl->backend_decl);
-+	      f->sym->ts.cl->backend_decl = length;
- 	    }
- 
- 	  hidden_typelist = TREE_CHAIN (hidden_typelist);
---- gcc/testsuite/gfortran.dg/assumed_charlen_sharing.f90.jj	2007-08-21 08:29:57.000000000 +0200
-+++ gcc/testsuite/gfortran.dg/assumed_charlen_sharing.f90	2007-08-21 08:29:57.000000000 +0200
-@@ -0,0 +1,29 @@
-+! This testcase was miscompiled, because ts.cl
-+! in function bar was initially shared between both
-+! dummy arguments.  Although it was later unshared,
-+! all expressions which copied ts.cl from bar2
-+! before that used incorrectly bar1's length
-+! instead of bar2.
-+! { dg-do run }
-+
-+subroutine foo (foo1, foo2)
-+  implicit none
-+  integer, intent(in) :: foo2
-+  character(*), intent(in) :: foo1(foo2)
-+end subroutine foo
-+
-+subroutine bar (bar1, bar2)
-+  implicit none
-+  character(*), intent(in) :: bar1, bar2
-+
-+  call foo ((/ bar2 /), 1)
-+end subroutine bar
-+
-+program test
-+  character(80) :: str1
-+  character(5) :: str2
-+
-+  str1 = 'String'
-+  str2 = 'Strng'
-+  call bar (str2, str1)
-+end program test
diff --git a/gcc41-scanf-fmt-check.patch b/gcc41-scanf-fmt-check.patch
deleted file mode 100644
index 4143e38..0000000
--- a/gcc41-scanf-fmt-check.patch
+++ /dev/null
@@ -1,433 +0,0 @@
-2007-09-17  Jakub Jelinek  <jakub@redhat.com>
-
-	* c-format.h (format_kind_info): Add alloc_char field.
-	* c-format.c (scanf_flag_specs): Add 'm'.
-	(scanf_flag_pairs): Add 'a', 'm' pair.
-	(scan_char_table): Allow 'm' modifier for c, s, [, C and S.
-	(format_types_orig): Add alloc_char fields.
-	(check_format_info_main): Rename aflag to alloc_flag.
-	Handle fki->alloc_char. modifier after width and before length
-	modifiers.  Move FMT_FLAG_SCANF_A_KLUDGE handling before
-	length modifiers as well.
-	* config/sol2-c.c (solaris_format_types): Add alloc_char field.
-
-	* gcc.dg/format/c90-scanf-5.c: New test.
-	* gcc.dg/format/c99-scanf-4.c: New test.
-	* gcc.dg/format/ext-7.c: New test.
-	* gcc.dg/format/ext-8.c: New test.
-
---- gcc/c-format.h.jj	2007-02-20 22:39:12.000000000 +0100
-+++ gcc/c-format.h	2007-09-17 22:44:17.000000000 +0200
-@@ -234,6 +234,8 @@ typedef struct
-      specifiers, but is used to check for bad combinations such as length
-      modifier with assignment suppression in scanf.  */
-   int length_code_char;
-+  /* Assignment-allocation flag character ('m' in scanf), otherwise 0.  */
-+  int alloc_char;
-   /* Pointer to type of argument expected if '*' is used for a width,
-      or NULL if '*' not used for widths.  */
-   tree *width_type;
---- gcc/c-format.c.jj	2007-02-20 22:39:12.000000000 +0100
-+++ gcc/c-format.c	2007-09-17 22:51:08.000000000 +0200
-@@ -423,6 +423,7 @@ static const format_flag_spec scanf_flag
- {
-   { '*',  0, 0, N_("assignment suppression"), N_("the assignment suppression scanf feature"), STD_C89 },
-   { 'a',  0, 0, N_("'a' flag"),               N_("the 'a' scanf flag"),                       STD_EXT },
-+  { 'm',  0, 0, N_("'m' flag"),               N_("the 'm' scanf flag"),                       STD_EXT },
-   { 'w',  0, 0, N_("field width"),            N_("field width in scanf format"),              STD_C89 },
-   { 'L',  0, 0, N_("length modifier"),        N_("length modifier in scanf format"),          STD_C89 },
-   { '\'', 0, 0, N_("''' flag"),               N_("the ''' scanf flag"),                       STD_EXT },
-@@ -434,6 +435,7 @@ static const format_flag_spec scanf_flag
- static const format_flag_pair scanf_flag_pairs[] =
- {
-   { '*', 'L', 0, 0 },
-+  { 'a', 'm', 0, 0 },
-   { 0, 0, 0, 0 }
- };
- 
-@@ -648,16 +650,16 @@ static const format_char_info scan_char_
-   { "u",     1, STD_C89, { T89_UI,  T99_UC,  T89_US,  T89_UL,  T9L_ULL, TEX_ULL, T99_ST,  T99_UPD, T99_UIM }, "*w'I", "W",   NULL },
-   { "oxX",   1, STD_C89, { T89_UI,  T99_UC,  T89_US,  T89_UL,  T9L_ULL, TEX_ULL, T99_ST,  T99_UPD, T99_UIM }, "*w",   "W",   NULL },
-   { "efgEG", 1, STD_C89, { T89_F,   BADLEN,  BADLEN,  T89_D,   BADLEN,  T89_LD,  BADLEN,  BADLEN,  BADLEN  }, "*w'",  "W",   NULL },
--  { "c",     1, STD_C89, { T89_C,   BADLEN,  BADLEN,  T94_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*w",   "cW",  NULL },
--  { "s",     1, STD_C89, { T89_C,   BADLEN,  BADLEN,  T94_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*aw",  "cW",  NULL },
--  { "[",     1, STD_C89, { T89_C,   BADLEN,  BADLEN,  T94_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*aw",  "cW[", NULL },
-+  { "c",     1, STD_C89, { T89_C,   BADLEN,  BADLEN,  T94_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*mw",  "cW",  NULL },
-+  { "s",     1, STD_C89, { T89_C,   BADLEN,  BADLEN,  T94_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*amw", "cW",  NULL },
-+  { "[",     1, STD_C89, { T89_C,   BADLEN,  BADLEN,  T94_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*amw", "cW[", NULL },
-   { "p",     2, STD_C89, { T89_V,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*w",   "W",   NULL },
-   { "n",     1, STD_C89, { T89_I,   T99_SC,  T89_S,   T89_L,   T9L_LL,  BADLEN,  T99_SST, T99_PD,  T99_IM  }, "",     "W",   NULL },
-   /* C99 conversion specifiers.  */
-   { "FaA",   1, STD_C99, { T99_F,   BADLEN,  BADLEN,  T99_D,   BADLEN,  T99_LD,  BADLEN,  BADLEN,  BADLEN  }, "*w'",  "W",   NULL },
-   /* X/Open conversion specifiers.  */
--  { "C",     1, STD_EXT, { TEX_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*w",   "W",   NULL },
--  { "S",     1, STD_EXT, { TEX_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*aw",  "W",   NULL },
-+  { "C",     1, STD_EXT, { TEX_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*mw",  "W",   NULL },
-+  { "S",     1, STD_EXT, { TEX_W,   BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN,  BADLEN  }, "*amw", "W",   NULL },
-   { NULL, 0, 0, NOLENGTHS, NULL, NULL, NULL }
- };
- 
-@@ -700,59 +702,59 @@ static const format_kind_info format_typ
-   { "printf",   printf_length_specs,  print_char_table, " +#0-'I", NULL, 
-     printf_flag_specs, printf_flag_pairs,
-     FMT_FLAG_ARG_CONVERT|FMT_FLAG_DOLLAR_MULTIPLE|FMT_FLAG_USE_DOLLAR|FMT_FLAG_EMPTY_PREC_OK,
--    'w', 0, 'p', 0, 'L',
-+    'w', 0, 'p', 0, 'L', 0,
-     &integer_type_node, &integer_type_node
-   },
-   { "asm_fprintf",   asm_fprintf_length_specs,  asm_fprintf_char_table, " +#0-", NULL, 
-     asm_fprintf_flag_specs, asm_fprintf_flag_pairs,
-     FMT_FLAG_ARG_CONVERT|FMT_FLAG_EMPTY_PREC_OK,
--    'w', 0, 'p', 0, 'L',
-+    'w', 0, 'p', 0, 'L', 0,
-     NULL, NULL
-   },
-   { "gcc_diag",   gcc_diag_length_specs,  gcc_diag_char_table, "q+", NULL, 
-     gcc_diag_flag_specs, gcc_diag_flag_pairs,
-     FMT_FLAG_ARG_CONVERT,
--    0, 0, 'p', 0, 'L',
-+    0, 0, 'p', 0, 'L', 0,
-     NULL, &integer_type_node
-   },
-   { "gcc_tdiag",   gcc_tdiag_length_specs,  gcc_tdiag_char_table, "q+", NULL, 
-     gcc_tdiag_flag_specs, gcc_tdiag_flag_pairs,
-     FMT_FLAG_ARG_CONVERT,
--    0, 0, 'p', 0, 'L',
-+    0, 0, 'p', 0, 'L', 0,
-     NULL, &integer_type_node
-   },
-   { "gcc_cdiag",   gcc_cdiag_length_specs,  gcc_cdiag_char_table, "q+", NULL, 
-     gcc_cdiag_flag_specs, gcc_cdiag_flag_pairs,
-     FMT_FLAG_ARG_CONVERT,
--    0, 0, 'p', 0, 'L',
-+    0, 0, 'p', 0, 'L', 0,
-     NULL, &integer_type_node
-   },
-   { "gcc_cxxdiag",   gcc_cxxdiag_length_specs,  gcc_cxxdiag_char_table, "q+#", NULL, 
-     gcc_cxxdiag_flag_specs, gcc_cxxdiag_flag_pairs,
-     FMT_FLAG_ARG_CONVERT,
--    0, 0, 'p', 0, 'L',
-+    0, 0, 'p', 0, 'L', 0,
-     NULL, &integer_type_node
-   },
-   { "gcc_gfc", NULL, gcc_gfc_char_table, "", NULL, 
-     NULL, gcc_gfc_flag_pairs,
-     FMT_FLAG_ARG_CONVERT,
--    0, 0, 0, 0, 0,
-+    0, 0, 0, 0, 0, 0,
-     NULL, NULL
-   },
-   { "scanf",    scanf_length_specs,   scan_char_table,  "*'I", NULL, 
-     scanf_flag_specs, scanf_flag_pairs,
-     FMT_FLAG_ARG_CONVERT|FMT_FLAG_SCANF_A_KLUDGE|FMT_FLAG_USE_DOLLAR|FMT_FLAG_ZERO_WIDTH_BAD|FMT_FLAG_DOLLAR_GAP_POINTER_OK,
--    'w', 0, 0, '*', 'L',
-+    'w', 0, 0, '*', 'L', 'm',
-     NULL, NULL
-   },
-   { "strftime", NULL,                 time_char_table,  "_-0^#", "EO",
-     strftime_flag_specs, strftime_flag_pairs,
--    FMT_FLAG_FANCY_PERCENT_OK, 'w', 0, 0, 0, 0,
-+    FMT_FLAG_FANCY_PERCENT_OK, 'w', 0, 0, 0, 0, 0,
-     NULL, NULL
-   },
-   { "strfmon",  strfmon_length_specs, monetary_char_table, "=^+(!-", NULL, 
-     strfmon_flag_specs, strfmon_flag_pairs,
--    FMT_FLAG_ARG_CONVERT, 'w', '#', 'p', 0, 'L',
-+    FMT_FLAG_ARG_CONVERT, 'w', '#', 'p', 0, 'L', 0,
-     NULL, NULL
-   }
- };
-@@ -1462,7 +1464,7 @@ check_format_info_main (format_check_res
-       const format_length_info *fli = NULL;
-       const format_char_info *fci = NULL;
-       char flag_chars[256];
--      int aflag = 0;
-+      int alloc_flag = 0;
-       const char *format_start = format_chars;
-       if (*format_chars == 0)
- 	{
-@@ -1721,6 +1723,31 @@ check_format_info_main (format_check_res
- 	    }
- 	}
- 
-+      if (fki->alloc_char && fki->alloc_char == *format_chars)
-+	{
-+	  i = strlen (flag_chars);
-+	  flag_chars[i++] = fki->alloc_char;
-+	  flag_chars[i] = 0;
-+	  format_chars++;
-+	}
-+
-+      /* Handle the scanf allocation kludge.  */
-+      if (fki->flags & (int) FMT_FLAG_SCANF_A_KLUDGE)
-+	{
-+	  if (*format_chars == 'a' && !flag_isoc99)
-+	    {
-+	      if (format_chars[1] == 's' || format_chars[1] == 'S'
-+		  || format_chars[1] == '[')
-+		{
-+		  /* 'a' is used as a flag.  */
-+		  i = strlen (flag_chars);
-+		  flag_chars[i++] = 'a';
-+		  flag_chars[i] = 0;
-+		  format_chars++;
-+		}
-+	    }
-+	}
-+
-       /* Read any length modifier, if this kind of format has them.  */
-       fli = fki->length_char_specs;
-       length_chars = NULL;
-@@ -1783,23 +1810,6 @@ check_format_info_main (format_check_res
- 	    }
- 	}
- 
--      /* Handle the scanf allocation kludge.  */
--      if (fki->flags & (int) FMT_FLAG_SCANF_A_KLUDGE)
--	{
--	  if (*format_chars == 'a' && !flag_isoc99)
--	    {
--	      if (format_chars[1] == 's' || format_chars[1] == 'S'
--		  || format_chars[1] == '[')
--		{
--		  /* 'a' is used as a flag.  */
--		  i = strlen (flag_chars);
--		  flag_chars[i++] = 'a';
--		  flag_chars[i] = 0;
--		  format_chars++;
--		}
--	    }
--	}
--
-       format_char = *format_chars;
-       if (format_char == 0
- 	  || (!(fki->flags & (int) FMT_FLAG_FANCY_PERCENT_OK)
-@@ -1872,7 +1882,9 @@ check_format_info_main (format_check_res
- 
-       if ((fki->flags & (int) FMT_FLAG_SCANF_A_KLUDGE)
- 	  && strchr (flag_chars, 'a') != 0)
--	aflag = 1;
-+	alloc_flag = 1;
-+      if (fki->alloc_char && strchr (flag_chars, fki->alloc_char) != 0)
-+	alloc_flag = 1;
- 
-       if (fki->suppression_char
- 	  && strchr (flag_chars, fki->suppression_char) != 0)
-@@ -2044,13 +2056,13 @@ check_format_info_main (format_check_res
- 
- 	      wanted_type_ptr->wanted_type = wanted_type;
- 	      wanted_type_ptr->wanted_type_name = wanted_type_name;
--	      wanted_type_ptr->pointer_count = fci->pointer_count + aflag;
-+	      wanted_type_ptr->pointer_count = fci->pointer_count + alloc_flag;
- 	      wanted_type_ptr->char_lenient_flag = 0;
- 	      if (strchr (fci->flags2, 'c') != 0)
- 		wanted_type_ptr->char_lenient_flag = 1;
- 	      wanted_type_ptr->writing_in_flag = 0;
- 	      wanted_type_ptr->reading_from_flag = 0;
--	      if (aflag)
-+	      if (alloc_flag)
- 		wanted_type_ptr->writing_in_flag = 1;
- 	      else
- 		{
---- gcc/config/sol2-c.c.jj	2007-02-20 22:39:11.000000000 +0100
-+++ gcc/config/sol2-c.c	2007-09-17 22:44:17.000000000 +0200
-@@ -74,7 +74,7 @@ const format_kind_info solaris_format_ty
-   { "cmn_err",  cmn_err_length_specs,  cmn_err_char_table, "", NULL,
-     cmn_err_flag_specs, cmn_err_flag_pairs,
-     FMT_FLAG_ARG_CONVERT|FMT_FLAG_EMPTY_PREC_OK,
--    'w', 0, 0, 0, 'L',
-+    'w', 0, 0, 0, 'L', 0,
-     &integer_type_node, &integer_type_node
-   }
- };
---- gcc/testsuite/gcc.dg/format/c90-scanf-5.c.jj	2007-09-17 22:44:17.000000000 +0200
-+++ gcc/testsuite/gcc.dg/format/c90-scanf-5.c	2007-09-17 22:44:17.000000000 +0200
-@@ -0,0 +1,19 @@
-+/* Test for scanf formats.  Formats using extensions to the standard
-+   should be rejected in strict pedantic mode.
-+*/
-+/* { dg-do compile } */
-+/* { dg-options "-std=iso9899:1990 -pedantic -Wformat" } */
-+
-+#include "format.h"
-+
-+void
-+foo (char **sp, wchar_t **lsp)
-+{
-+  /* m assignment-allocation modifier, recognized in both C90
-+     and C99 modes, is a POSIX and ISO/IEC WDTR 24731-2 extension.  */
-+  scanf ("%ms", sp); /* { dg-warning "C" "%ms" } */
-+  scanf ("%mS", lsp); /* { dg-warning "C" "%mS" } */
-+  scanf ("%mls", lsp); /* { dg-warning "C" "%mls" } */
-+  scanf ("%m[bcd]", sp); /* { dg-warning "C" "%m[]" } */
-+  scanf ("%ml[bcd]", lsp); /* { dg-warning "C" "%ml[]" } */
-+}
---- gcc/testsuite/gcc.dg/format/c99-scanf-4.c.jj	2007-09-17 22:44:17.000000000 +0200
-+++ gcc/testsuite/gcc.dg/format/c99-scanf-4.c	2007-09-17 22:44:17.000000000 +0200
-@@ -0,0 +1,19 @@
-+/* Test for scanf formats.  Formats using extensions to the standard
-+   should be rejected in strict pedantic mode.
-+*/
-+/* { dg-do compile } */
-+/* { dg-options "-std=iso9899:1999 -pedantic -Wformat" } */
-+
-+#include "format.h"
-+
-+void
-+foo (char **sp, wchar_t **lsp)
-+{
-+  /* m assignment-allocation modifier, recognized in both C90
-+     and C99 modes, is a POSIX and ISO/IEC WDTR 24731-2 extension.  */
-+  scanf ("%ms", sp); /* { dg-warning "C" "%ms" } */
-+  scanf ("%mS", lsp); /* { dg-warning "C" "%mS" } */
-+  scanf ("%mls", lsp); /* { dg-warning "C" "%mls" } */
-+  scanf ("%m[bcd]", sp); /* { dg-warning "C" "%m[]" } */
-+  scanf ("%ml[bcd]", lsp); /* { dg-warning "C" "%ml[]" } */
-+}
---- gcc/testsuite/gcc.dg/format/ext-7.c.jj	2007-09-17 22:44:17.000000000 +0200
-+++ gcc/testsuite/gcc.dg/format/ext-7.c	2007-09-17 22:44:17.000000000 +0200
-@@ -0,0 +1,85 @@
-+/* Test for scanf formats.  %a and %m extensions.  */
-+/* { dg-do compile } */
-+/* { dg-options "-std=gnu89 -Wformat" } */
-+
-+#include "format.h"
-+
-+void
-+foo (char **sp, wchar_t **lsp, int *ip, float *fp, void **pp, double *dp)
-+{
-+  /* %a formats for allocation, only recognized in C90 mode, are a
-+     GNU extension.  Followed by other characters, %a is not treated
-+     specially.
-+  */
-+  scanf ("%as", sp);
-+  scanf ("%aS", lsp);
-+  scanf ("%las", dp);
-+  scanf ("%la", lsp);	/* { dg-warning "but argument 2 has type" } */
-+  scanf ("%las", lsp);	/* { dg-warning "but argument 2 has type" } */
-+  scanf ("%a[bcd]", sp);
-+  scanf ("%la[bcd]", dp);
-+  scanf ("%*as");
-+  scanf ("%*aS");
-+  scanf ("%*las");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*a[bcd]");
-+  scanf ("%*la[bcd]");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%10as", sp);
-+  scanf ("%5aS", lsp);
-+  scanf ("%9las", dp);
-+  scanf ("%25a[bcd]", sp);
-+  scanf ("%48la[bcd]", dp);
-+  scanf ("%*10as");
-+  scanf ("%*5aS");
-+  scanf ("%*9las");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*25a[bcd]");
-+  scanf ("%*48la[bcd]"); /* { dg-warning "assignment suppression and length modifier" } */
-+
-+  /* m assignment-allocation modifier, recognized in both C90
-+     and C99 modes, is a POSIX and ISO/IEC WDTR 24731-2 extension.  */
-+  scanf ("%ms", sp);
-+  scanf ("%mS", lsp);
-+  scanf ("%mls", lsp);
-+  scanf ("%m[bcd]", sp);
-+  scanf ("%ml[bcd]", lsp);
-+  scanf ("%mc", sp);
-+  scanf ("%mlc", lsp);
-+  scanf ("%mC", lsp);
-+  scanf ("%*ms");
-+  scanf ("%*mS");
-+  scanf ("%*mls");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*m[bcd]");
-+  scanf ("%*ml[bcd]");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*mc");
-+  scanf ("%*mlc");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*mC");
-+  scanf ("%10ms", sp);
-+  scanf ("%5mS", lsp);
-+  scanf ("%9mls", lsp);
-+  scanf ("%25m[bcd]", sp);
-+  scanf ("%41ml[bcd]", lsp);
-+  scanf ("%131mc", sp);
-+  scanf ("%27mlc", lsp);
-+  scanf ("%2mC", lsp);
-+  scanf ("%*10ms");
-+  scanf ("%*5mS");
-+  scanf ("%*9mls");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*25m[bcd]");
-+  scanf ("%*41ml[bcd]"); /* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*131mc");
-+  scanf ("%*27mlc");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*2mC");
-+
-+  scanf ("%md", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%mi", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%mo", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%mu", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%mx", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%me", fp);	/* { dg-warning "flag used with" } */
-+  scanf ("%mf", fp);	/* { dg-warning "flag used with" } */
-+  scanf ("%mg", fp);	/* { dg-warning "flag used with" } */
-+  scanf ("%mp", pp);	/* { dg-warning "flag used with" } */
-+
-+  scanf ("%mas", sp);	/* { dg-warning "flag together" } */
-+  scanf ("%maS", lsp);	/* { dg-warning "flag together" } */
-+  scanf ("%ma[bcd]", sp);	/* { dg-warning "flag together" } */
-+}
---- gcc/testsuite/gcc.dg/format/ext-8.c.jj	2007-09-17 22:44:17.000000000 +0200
-+++ gcc/testsuite/gcc.dg/format/ext-8.c	2007-09-17 22:44:17.000000000 +0200
-@@ -0,0 +1,56 @@
-+/* Test for scanf formats.  %m extensions.  */
-+/* { dg-do compile } */
-+/* { dg-options "-std=gnu99 -Wformat" } */
-+
-+#include "format.h"
-+
-+void
-+foo (char **sp, wchar_t **lsp, int *ip, float *fp, void **pp)
-+{
-+  /* m assignment-allocation modifier, recognized in both C90
-+     and C99 modes, is a POSIX and ISO/IEC WDTR 24731-2 extension.  */
-+  scanf ("%ms", sp);
-+  scanf ("%mS", lsp);
-+  scanf ("%mls", lsp);
-+  scanf ("%m[bcd]", sp);
-+  scanf ("%ml[bcd]", lsp);
-+  scanf ("%mc", sp);
-+  scanf ("%mlc", lsp);
-+  scanf ("%mC", lsp);
-+  scanf ("%*ms");
-+  scanf ("%*mS");
-+  scanf ("%*mls");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*m[bcd]");
-+  scanf ("%*ml[bcd]");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*mc");
-+  scanf ("%*mlc");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*mC");
-+  scanf ("%10ms", sp);
-+  scanf ("%5mS", lsp);
-+  scanf ("%9mls", lsp);
-+  scanf ("%25m[bcd]", sp);
-+  scanf ("%41ml[bcd]", lsp);
-+  scanf ("%131mc", sp);
-+  scanf ("%27mlc", lsp);
-+  scanf ("%2mC", lsp);
-+  scanf ("%*10ms");
-+  scanf ("%*5mS");
-+  scanf ("%*9mls");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*25m[bcd]");
-+  scanf ("%*41ml[bcd]"); /* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*131mc");
-+  scanf ("%*27mlc");	/* { dg-warning "assignment suppression and length modifier" } */
-+  scanf ("%*2mC");
-+
-+  scanf ("%md", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%mi", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%mo", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%mu", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%mx", ip);	/* { dg-warning "flag used with" } */
-+  scanf ("%ma", fp);	/* { dg-warning "flag used with" } */
-+  scanf ("%mA", fp);	/* { dg-warning "flag used with" } */
-+  scanf ("%me", fp);	/* { dg-warning "flag used with" } */
-+  scanf ("%mf", fp);	/* { dg-warning "flag used with" } */
-+  scanf ("%mg", fp);	/* { dg-warning "flag used with" } */
-+  scanf ("%mp", pp);	/* { dg-warning "flag used with" } */
-+}
diff --git a/gcc41-sparc-niagara.patch b/gcc41-sparc-niagara.patch
deleted file mode 100644
index c9e61ba..0000000
--- a/gcc41-sparc-niagara.patch
+++ /dev/null
@@ -1,519 +0,0 @@
-2006-03-02  David S. Miller  <davem@sunset.davemloft.net>
-
-	Sun Niagara specific optimizations.
-	* config.gcc: Recognize niagara as target.
-	* config/sparc/sparc.h (SPARC_RELAXED_ORDERING): Mention Niagara.
-	(TARGET_CPU_niagara): Define.
-	(CPP_CPU64_DEFAULT_SPEC): Define __sparc_v9__ for Niagara.
-	(ASM_CPU64_DEFAULT_SPEC): Pass -Av9b for Niagara.
-	(CPP_CPU_SPEC): Handle -mcpu=niagara.
-	(ASM_CPU_SPEC): Likewise.
-	(PROCESSOR_NIAGARA): New enum entry.
-	(REGISTER_MOVE_COST): Handle Niagara.
-	(BRANCH_COST, PREFETCH_BLOCK, SIMULTANEOUS_PREFETCHES): Likewise.
-	* config/sparc/sparc.c (niagara_costs): New processor_costs entry.
-	(sparc_override_options): Recognize "niagara", set appropriate
-	default MASK_* values for it, and align functions to 32-bytes
-	by default just like ULTRASPARC/ULTRASPARC3.
-	(sparc_initialize_trampoline): Handle niagara like ultrasparc.
-	(sparc64_initialize_trampoline): Likewise.
-	(sparc_use_sched_lookahead): Use zero for niagara.
-	(sparc_issue_rate): Use one for niagara.
-	* config/sparc/niagara.md: New file.
-	* config/sparc/sparc.md: Include it.
-	* config/sparc/sol2-bi.h (CPP_CPU64_DEFAULT_SPEC,
-	ASM_CPU32_DEFAULT_SPEC, ASM_CPU64_DEFAULT_SPEC): Set appropriately
-	when default cpu is niagara.
-	(CPP_CPU_SPEC): Handle -mcpu=niagara.
-	(ASM_CPU_SPEC): Likewise.
-	* config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Set appropriately
-	when default cpu is niagara.
-	(ASM_CPU_SPEC): Handle -mcpu=niagara.
-	* config/sparc/linux64.h: Handle a default of TARGET_CPU_niagara
-	just like v9/ultrasparc/ultrasparc3.
-	* doc/invoke.texi: Add documentation for "niagara" and improve
-	existing documentation for ultrasparc variants.
-
---- gcc/doc/invoke.texi	(revision 111647)
-+++ gcc/doc/invoke.texi	(revision 111648)
-@@ -12268,8 +12268,8 @@ Set the instruction set, register set, a
- for machine type @var{cpu_type}.  Supported values for @var{cpu_type} are
- @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
- @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
--@samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
--@samp{ultrasparc3}.
-+@samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
-+@samp{ultrasparc3}, and @samp{niagara}.
- 
- Default instruction scheduling parameters are used for values that select
- an architecture and not an implementation.  These are @samp{v7}, @samp{v8},
-@@ -12283,7 +12283,7 @@ implementations.
-     v8:             supersparc, hypersparc
-     sparclite:      f930, f934, sparclite86x
-     sparclet:       tsc701
--    v9:             ultrasparc, ultrasparc3
-+    v9:             ultrasparc, ultrasparc3, niagara
- @end smallexample
- 
- By default (unless configured otherwise), GCC generates code for the V7
-@@ -12317,9 +12317,11 @@ With @option{-mcpu=v9}, GCC generates co
- architecture.  This adds 64-bit integer and floating-point move instructions,
- 3 additional floating-point condition code registers and conditional move
- instructions.  With @option{-mcpu=ultrasparc}, the compiler additionally
--optimizes it for the Sun UltraSPARC I/II chips.  With
-+optimizes it for the Sun UltraSPARC I/II/IIi chips.  With
- @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
--Sun UltraSPARC III chip.
-+Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips.  With
-+@option{-mcpu=niagara}, the compiler additionally optimizes it for
-+Sun UltraSPARC T1 chips.
- 
- @item -mtune=@var{cpu_type}
- @opindex mtune
-@@ -12331,8 +12333,8 @@ The same values for @option{-mcpu=@var{c
- @option{-mtune=@var{cpu_type}}, but the only useful values are those
- that select a particular cpu implementation.  Those are @samp{cypress},
- @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
--@samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
--@samp{ultrasparc3}.
-+@samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
-+@samp{ultrasparc3}, and @samp{niagara}.
- 
- @item -mv8plus
- @itemx -mno-v8plus
---- gcc/config.gcc	(revision 111647)
-+++ gcc/config.gcc	(revision 111648)
-@@ -2830,7 +2830,7 @@ case "${target}" in
- 			"" | sparc | sparcv9 | sparc64 | sparc86x \
- 			| v7 | cypress | v8 | supersparc | sparclite | f930 \
- 			| f934 | hypersparc | sparclite86x | sparclet | tsc701 \
--			| v9 | ultrasparc | ultrasparc3)
-+			| v9 | ultrasparc | ultrasparc3 | niagara)
- 				# OK
- 				;;
- 			*)
---- gcc/config/sparc/niagara.md	(revision 0)
-+++ gcc/config/sparc/niagara.md	(revision 111648)
-@@ -0,0 +1,119 @@
-+;; Scheduling description for Niagara.
-+;;   Copyright (C) 2006 Free Software Foundation, Inc.
-+;;
-+;; This file is part of GCC.
-+;;
-+;; GCC is free software; you can redistribute it and/or modify
-+;; it under the terms of the GNU General Public License as published by
-+;; the Free Software Foundation; either version 2, or (at your option)
-+;; any later version.
-+;;
-+;; GCC is distributed in the hope that it will be useful,
-+;; but WITHOUT ANY WARRANTY; without even the implied warranty of
-+;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-+;; GNU General Public License for more details.
-+;;
-+;; You should have received a copy of the GNU General Public License
-+;; along with GCC; see the file COPYING.  If not, write to
-+;; the Free Software Foundation, 51 Franklin Street, Fifth Floor,
-+;; Boston, MA 02110-1301, USA.
-+
-+;; Niagara is a single-issue processor.
-+
-+(define_automaton "niagara_0")
-+
-+(define_cpu_unit "niag_pipe" "niagara_0")
-+
-+(define_insn_reservation "niag_5cycle" 5
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "multi,flushw,iflush,trap"))
-+  "niag_pipe*5")
-+
-+(define_insn_reservation "niag_4cycle" 4
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "savew"))
-+  "niag_pipe*4")
-+
-+/* Most basic operations are single-cycle. */
-+(define_insn_reservation "niag_ialu" 1
-+ (and (eq_attr "cpu" "niagara")
-+   (eq_attr "type" "ialu,shift,compare,cmove"))
-+ "niag_pipe")
-+
-+(define_insn_reservation "niag_imul" 11
-+ (and (eq_attr "cpu" "niagara")
-+   (eq_attr "type" "imul"))
-+ "niag_pipe*11")
-+
-+(define_insn_reservation "niag_idiv" 72
-+ (and (eq_attr "cpu" "niagara")
-+   (eq_attr "type" "idiv"))
-+ "niag_pipe*72")
-+
-+(define_insn_reservation "niag_branch" 3
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "call,sibcall,call_no_delay_slot,uncond_branch,branch"))
-+  "niag_pipe*3")
-+
-+(define_insn_reservation "niag_3cycle_load" 3
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "load"))
-+  "niag_pipe*3")
-+
-+(define_insn_reservation "niag_9cycle_load" 9
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "fpload"))
-+  "niag_pipe*9")
-+
-+(define_insn_reservation "niag_1cycle_store" 1
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "store"))
-+  "niag_pipe")
-+
-+(define_insn_reservation "niag_8cycle_store" 8
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "fpstore"))
-+  "niag_pipe*8")
-+
-+/* Things incorrectly modelled here:
-+ *  FPADD{s,d}: 26 cycles
-+ *  FPSUB{s,d}: 26 cycles
-+ *  FABSD: 26 cycles
-+ *  F{s,d}TO{s,d}: 26 cycles
-+ *  F{s,d}TO{i,x}: 26 cycles
-+ *  FSMULD: 29 cycles
-+ */
-+(define_insn_reservation "niag_fmov" 8
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "fpmove,fpcmove,fpcrmove"))
-+  "niag_pipe*8")
-+
-+(define_insn_reservation "niag_fpcmp" 26
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "fpcmp"))
-+  "niag_pipe*26")
-+
-+(define_insn_reservation "niag_fmult" 29
-+ (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "fpmul"))
-+  "niag_pipe*29")
-+
-+(define_insn_reservation "niag_fdivs" 54
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "fpdivs"))
-+  "niag_pipe*54")
-+
-+(define_insn_reservation "niag_fdivd" 83
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "fpdivd"))
-+  "niag_pipe*83")
-+
-+/* Things incorrectly modelled here:
-+ *  FPADD{16,32}: 10 cycles
-+ *  FPSUB{16,32}: 10 cycles
-+ *  FALIGNDATA: 10 cycles
-+ */
-+(define_insn_reservation "niag_vis" 8
-+  (and (eq_attr "cpu" "niagara")
-+    (eq_attr "type" "fga,fgm_pack,fgm_mul,fgm_cmp,fgm_pdist"))
-+  "niag_pipe*8")
---- gcc/config/sparc/sparc.md	(revision 111647)
-+++ gcc/config/sparc/sparc.md	(revision 111648)
-@@ -94,7 +94,8 @@ (define_attr "cpu"
-    sparclet,tsc701,
-    v9,
-    ultrasparc,
--   ultrasparc3"
-+   ultrasparc3,
-+   niagara"
-   (const (symbol_ref "sparc_cpu_attr")))
- 
- ;; Attribute for the instruction set.
-@@ -315,6 +316,7 @@ (define_delay (eq_attr "type" "return")
- (include "sparclet.md")
- (include "ultra1_2.md")
- (include "ultra3.md")
-+(include "niagara.md")
- 
- 
- ;; Operand and operator predicates.
---- gcc/config/sparc/sparc.c	(revision 111647)
-+++ gcc/config/sparc/sparc.c	(revision 111648)
-@@ -197,6 +197,30 @@ struct processor_costs ultrasparc3_costs
-   0, /* shift penalty */
- };
- 
-+static const
-+struct processor_costs niagara_costs = {
-+  COSTS_N_INSNS (3), /* int load */
-+  COSTS_N_INSNS (3), /* int signed load */
-+  COSTS_N_INSNS (3), /* int zeroed load */
-+  COSTS_N_INSNS (9), /* float load */
-+  COSTS_N_INSNS (8), /* fmov, fneg, fabs */
-+  COSTS_N_INSNS (8), /* fadd, fsub */
-+  COSTS_N_INSNS (26), /* fcmp */
-+  COSTS_N_INSNS (8), /* fmov, fmovr */
-+  COSTS_N_INSNS (29), /* fmul */
-+  COSTS_N_INSNS (54), /* fdivs */
-+  COSTS_N_INSNS (83), /* fdivd */
-+  COSTS_N_INSNS (100), /* fsqrts - not implemented in hardware */
-+  COSTS_N_INSNS (100), /* fsqrtd - not implemented in hardware */
-+  COSTS_N_INSNS (11), /* imul */
-+  COSTS_N_INSNS (11), /* imulX */
-+  0, /* imul bit factor */
-+  COSTS_N_INSNS (72), /* idiv */
-+  COSTS_N_INSNS (72), /* idivX */
-+  COSTS_N_INSNS (1), /* movcc/movr */
-+  0, /* shift penalty */
-+};
-+
- const struct processor_costs *sparc_costs = &cypress_costs;
- 
- #ifdef HAVE_AS_RELAX_OPTION
-@@ -597,6 +621,7 @@ sparc_override_options (void)
-     { TARGET_CPU_v9, "v9" },
-     { TARGET_CPU_ultrasparc, "ultrasparc" },
-     { TARGET_CPU_ultrasparc3, "ultrasparc3" },
-+    { TARGET_CPU_niagara, "niagara" },
-     { 0, 0 }
-   };
-   const struct cpu_default *def;
-@@ -632,6 +657,8 @@ sparc_override_options (void)
-     /* TI ultrasparc III */
-     /* ??? Check if %y issue still holds true in ultra3.  */
-     { "ultrasparc3", PROCESSOR_ULTRASPARC3, MASK_ISA, MASK_V9|MASK_DEPRECATED_V8_INSNS},
-+    /* UltraSPARC T1 */
-+    { "niagara", PROCESSOR_NIAGARA, MASK_ISA, MASK_V9|MASK_DEPRECATED_V8_INSNS},
-     { 0, 0, 0, 0 }
-   };
-   const struct cpu_table *cpu;
-@@ -741,7 +768,8 @@ sparc_override_options (void)
-   /* Supply a default value for align_functions.  */
-   if (align_functions == 0
-       && (sparc_cpu == PROCESSOR_ULTRASPARC
--	  || sparc_cpu == PROCESSOR_ULTRASPARC3))
-+	  || sparc_cpu == PROCESSOR_ULTRASPARC3
-+	  || sparc_cpu == PROCESSOR_NIAGARA))
-     align_functions = 32;
- 
-   /* Validate PCC_STRUCT_RETURN.  */
-@@ -790,6 +818,9 @@ sparc_override_options (void)
-     case PROCESSOR_ULTRASPARC3:
-       sparc_costs = &ultrasparc3_costs;
-       break;
-+    case PROCESSOR_NIAGARA:
-+      sparc_costs = &niagara_costs;
-+      break;
-     };
- 
- #ifdef TARGET_DEFAULT_LONG_DOUBLE_128
-@@ -7099,7 +7130,8 @@ sparc_initialize_trampoline (rtx tramp, 
-      aligned on a 16 byte boundary so one flush clears it all.  */
-   emit_insn (gen_flush (validize_mem (gen_rtx_MEM (SImode, tramp))));
-   if (sparc_cpu != PROCESSOR_ULTRASPARC
--      && sparc_cpu != PROCESSOR_ULTRASPARC3)
-+      && sparc_cpu != PROCESSOR_ULTRASPARC3
-+      && sparc_cpu != PROCESSOR_NIAGARA)
-     emit_insn (gen_flush (validize_mem (gen_rtx_MEM (SImode,
- 						     plus_constant (tramp, 8)))));
- 
-@@ -7141,7 +7173,8 @@ sparc64_initialize_trampoline (rtx tramp
-   emit_insn (gen_flushdi (validize_mem (gen_rtx_MEM (DImode, tramp))));
- 
-   if (sparc_cpu != PROCESSOR_ULTRASPARC
--      && sparc_cpu != PROCESSOR_ULTRASPARC3)
-+      && sparc_cpu != PROCESSOR_ULTRASPARC3
-+      && sparc_cpu != PROCESSOR_NIAGARA)
-     emit_insn (gen_flushdi (validize_mem (gen_rtx_MEM (DImode, plus_constant (tramp, 8)))));
- 
-   /* Call __enable_execute_stack after writing onto the stack to make sure
-@@ -7321,6 +7354,8 @@ sparc_sched_init (FILE *dump ATTRIBUTE_U
- static int
- sparc_use_sched_lookahead (void)
- {
-+  if (sparc_cpu == PROCESSOR_NIAGARA)
-+    return 0;
-   if (sparc_cpu == PROCESSOR_ULTRASPARC
-       || sparc_cpu == PROCESSOR_ULTRASPARC3)
-     return 4;
-@@ -7336,6 +7371,7 @@ sparc_issue_rate (void)
- {
-   switch (sparc_cpu)
-     {
-+    case PROCESSOR_NIAGARA:
-     default:
-       return 1;
-     case PROCESSOR_V9:
---- gcc/config/sparc/sol2-bi.h	(revision 111647)
-+++ gcc/config/sparc/sol2-bi.h	(revision 111648)
-@@ -39,6 +39,15 @@
- #define ASM_CPU64_DEFAULT_SPEC AS_SPARC64_FLAG "b"
- #endif
- 
-+#if TARGET_CPU_DEFAULT == TARGET_CPU_niagara
-+#undef CPP_CPU64_DEFAULT_SPEC
-+#define CPP_CPU64_DEFAULT_SPEC ""
-+#undef ASM_CPU32_DEFAULT_SPEC
-+#define ASM_CPU32_DEFAULT_SPEC "-xarch=v8plusb"
-+#undef ASM_CPU64_DEFAULT_SPEC
-+#define ASM_CPU64_DEFAULT_SPEC AS_SPARC64_FLAG "b"
-+#endif
-+
- #if DEFAULT_ARCH32_P
- #define DEF_ARCH32_SPEC(__str) "%{!m64:" __str "}"
- #define DEF_ARCH64_SPEC(__str) "%{m64:" __str "}"
-@@ -57,7 +66,7 @@
- %{mcpu=sparclite|mcpu-f930|mcpu=f934:-D__sparclite__} \
- %{mcpu=v8:" DEF_ARCH32_SPEC("-D__sparcv8") "} \
- %{mcpu=supersparc:-D__supersparc__ " DEF_ARCH32_SPEC("-D__sparcv8") "} \
--%{mcpu=v9|mcpu=ultrasparc|mcpu=ultrasparc3:" DEF_ARCH32_SPEC("-D__sparcv8") "} \
-+%{mcpu=v9|mcpu=ultrasparc|mcpu=ultrasparc3|mcpu=niagara:" DEF_ARCH32_SPEC("-D__sparcv8") "} \
- %{!mcpu*:%{!mcypress:%{!msparclite:%{!mf930:%{!mf934:%{!mv8:%{!msupersparc:%(cpp_cpu_default)}}}}}}} \
- "
- 
-@@ -66,7 +75,8 @@
- %{mcpu=v9:" DEF_ARCH32_SPEC("-xarch=v8plus") DEF_ARCH64_SPEC(AS_SPARC64_FLAG) "} \
- %{mcpu=ultrasparc:" DEF_ARCH32_SPEC("-xarch=v8plusa") DEF_ARCH64_SPEC(AS_SPARC64_FLAG "a") "} \
- %{mcpu=ultrasparc3:" DEF_ARCH32_SPEC("-xarch=v8plusb") DEF_ARCH64_SPEC(AS_SPARC64_FLAG "b") "} \
--%{!mcpu=ultrasparc3:%{!mcpu=ultrasparc:%{!mcpu=v9:%{mcpu*:" DEF_ARCH32_SPEC("-xarch=v8") DEF_ARCH64_SPEC(AS_SPARC64_FLAG) "}}}} \
-+%{mcpu=niagara:" DEF_ARCH32_SPEC("-xarch=v8plusb") DEF_ARCH64_SPEC(AS_SPARC64_FLAG "b") "} \
-+%{!mcpu=niagara:%{!mcpu=ultrasparc3:%{!mcpu=ultrasparc:%{!mcpu=v9:%{mcpu*:" DEF_ARCH32_SPEC("-xarch=v8") DEF_ARCH64_SPEC(AS_SPARC64_FLAG) "}}}}} \
- %{!mcpu*:%(asm_cpu_default)} \
- "
- 
---- gcc/config/sparc/sparc.h	(revision 111647)
-+++ gcc/config/sparc/sparc.h	(revision 111648)
-@@ -206,7 +206,7 @@ extern enum cmodel sparc_cmodel;
-    which requires the following macro to be true if enabled.  Prior to V9,
-    there are no instructions to even talk about memory synchronization.
-    Note that the UltraSPARC III processors don't implement RMO, unlike the
--   UltraSPARC II processors.
-+   UltraSPARC II processors.  Niagara does not implement RMO either.
- 
-    Default to false; for example, Solaris never enables RMO, only ever uses
-    total memory ordering (TMO).  */
-@@ -238,10 +238,12 @@ extern enum cmodel sparc_cmodel;
- #define TARGET_CPU_sparc64	7	/* alias */
- #define TARGET_CPU_ultrasparc	8
- #define TARGET_CPU_ultrasparc3	9
-+#define TARGET_CPU_niagara	10
- 
- #if TARGET_CPU_DEFAULT == TARGET_CPU_v9 \
-  || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc \
-- || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc3
-+ || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc3 \
-+ || TARGET_CPU_DEFAULT == TARGET_CPU_niagara
- 
- #define CPP_CPU32_DEFAULT_SPEC ""
- #define ASM_CPU32_DEFAULT_SPEC ""
-@@ -262,6 +264,10 @@ extern enum cmodel sparc_cmodel;
- #define CPP_CPU64_DEFAULT_SPEC "-D__sparc_v9__"
- #define ASM_CPU64_DEFAULT_SPEC "-Av9b"
- #endif
-+#if TARGET_CPU_DEFAULT == TARGET_CPU_niagara
-+#define CPP_CPU64_DEFAULT_SPEC "-D__sparc_v9__"
-+#define ASM_CPU64_DEFAULT_SPEC "-Av9b"
-+#endif
- 
- #else
- 
-@@ -352,6 +358,7 @@ extern enum cmodel sparc_cmodel;
- %{mcpu=v9:-D__sparc_v9__} \
- %{mcpu=ultrasparc:-D__sparc_v9__} \
- %{mcpu=ultrasparc3:-D__sparc_v9__} \
-+%{mcpu=niagara:-D__sparc_v9__} \
- %{!mcpu*:%{!mcypress:%{!msparclite:%{!mf930:%{!mf934:%{!mv8:%{!msupersparc:%(cpp_cpu_default)}}}}}}} \
- "
- #define CPP_ARCH32_SPEC ""
-@@ -401,6 +408,7 @@ extern enum cmodel sparc_cmodel;
- %{mcpu=v9:-Av9} \
- %{mcpu=ultrasparc:%{!mv8plus:-Av9a}} \
- %{mcpu=ultrasparc3:%{!mv8plus:-Av9b}} \
-+%{mcpu=niagara:%{!mv8plus:-Av9b}} \
- %{!mcpu*:%{!mcypress:%{!msparclite:%{!mf930:%{!mf934:%{!mv8:%{!msupersparc:%(asm_cpu_default)}}}}}}} \
- "
- 
-@@ -524,7 +532,8 @@ enum processor_type {
-   PROCESSOR_TSC701,
-   PROCESSOR_V9,
-   PROCESSOR_ULTRASPARC,
--  PROCESSOR_ULTRASPARC3
-+  PROCESSOR_ULTRASPARC3,
-+  PROCESSOR_NIAGARA
- };
- 
- /* This is set from -m{cpu,tune}=xxx.  */
-@@ -2137,7 +2146,8 @@ do {                                    
-     || (GENERAL_OR_I64 (CLASS1) && FP_REG_CLASS_P (CLASS2)) \
-     || (CLASS1) == FPCC_REGS || (CLASS2) == FPCC_REGS)		\
-    ? ((sparc_cpu == PROCESSOR_ULTRASPARC \
--       || sparc_cpu == PROCESSOR_ULTRASPARC3) ? 12 : 6) : 2)
-+       || sparc_cpu == PROCESSOR_ULTRASPARC3 \
-+       || sparc_cpu == PROCESSOR_NIAGARA) ? 12 : 6) : 2)
- 
- /* Provide the cost of a branch.  For pre-v9 processors we use
-    a value of 3 to take into account the potential annulling of
-@@ -2147,22 +2157,30 @@ do {                                    
- 
-    On v9 and later, which have branch prediction facilities, we set
-    it to the depth of the pipeline as that is the cost of a
--   mispredicted branch.  */
-+   mispredicted branch.
-+
-+   On Niagara, normal branches insert 3 bubbles into the pipe
-+   and annulled branches insert 4 bubbles.  */
- 
- #define BRANCH_COST \
- 	((sparc_cpu == PROCESSOR_V9 \
- 	  || sparc_cpu == PROCESSOR_ULTRASPARC) \
- 	 ? 7 \
-          : (sparc_cpu == PROCESSOR_ULTRASPARC3 \
--            ? 9 : 3))
-+            ? 9 \
-+	 : (sparc_cpu == PROCESSOR_NIAGARA \
-+	    ? 4 \
-+	 : 3)))
- 
- #define PREFETCH_BLOCK \
- 	((sparc_cpu == PROCESSOR_ULTRASPARC \
--          || sparc_cpu == PROCESSOR_ULTRASPARC3) \
-+          || sparc_cpu == PROCESSOR_ULTRASPARC3 \
-+	  || sparc_cpu == PROCESSOR_NIAGARA) \
-          ? 64 : 32)
- 
- #define SIMULTANEOUS_PREFETCHES \
--	((sparc_cpu == PROCESSOR_ULTRASPARC) \
-+	((sparc_cpu == PROCESSOR_ULTRASPARC \
-+	  || sparc_cpu == PROCESSOR_NIAGARA) \
-          ? 2 \
-          : (sparc_cpu == PROCESSOR_ULTRASPARC3 \
-             ? 8 : 3))
---- gcc/config/sparc/linux64.h	(revision 111647)
-+++ gcc/config/sparc/linux64.h	(revision 111648)
-@@ -43,7 +43,8 @@ Boston, MA 02110-1301, USA.  */
- 
- #if TARGET_CPU_DEFAULT == TARGET_CPU_v9 \
-     || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc \
--    || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc3
-+    || TARGET_CPU_DEFAULT == TARGET_CPU_ultrasparc3 \
-+    || TARGET_CPU_DEFAULT == TARGET_CPU_niagara
- /* A 64 bit v9 compiler with stack-bias,
-    in a Medium/Low code model environment.  */
- 
---- gcc/config/sparc/sol2.h	(revision 111647)
-+++ gcc/config/sparc/sol2.h	(revision 111648)
-@@ -41,11 +41,17 @@ Boston, MA 02110-1301, USA.  */
- #define ASM_CPU_DEFAULT_SPEC "-xarch=v8plusb"
- #endif
- 
-+#if TARGET_CPU_DEFAULT == TARGET_CPU_niagara
-+#undef ASM_CPU_DEFAULT_SPEC
-+#define ASM_CPU_DEFAULT_SPEC "-xarch=v8plusb"
-+#endif
-+
- #undef ASM_CPU_SPEC
- #define ASM_CPU_SPEC "\
- %{mcpu=v9:-xarch=v8plus} \
- %{mcpu=ultrasparc:-xarch=v8plusa} \
- %{mcpu=ultrasparc3:-xarch=v8plusb} \
-+%{mcpu=niagara:-xarch=v8plusb} \
- %{!mcpu*:%(asm_cpu_default)} \
- "
- 
diff --git a/gcc41.spec b/gcc41.spec
index 42c3142..0e08180 100644
--- a/gcc41.spec
+++ b/gcc41.spec
@@ -1,6 +1,6 @@
-%define DATE 20070821
+%define DATE 20070925
 %define gcc_version 4.1.2
-%define gcc_release 26
+%define gcc_release 28
 %define _unpackaged_files_terminate_build 0
 %define multilib_64_archs sparc64 ppc64 s390x x86_64
 %define include_gappletviewer 1
@@ -138,24 +138,13 @@ Patch21: gcc41-rh235008.patch
 Patch22: gcc41-build-id.patch
 Patch23: gcc41-pr28690.patch
 Patch24: gcc41-rh247256.patch
-Patch25: gcc41-pr22244.patch
-Patch26: gcc41-pr32678.patch
-Patch27: gcc41-pr32912.patch
-Patch28: gcc41-sparc-niagara.patch
-Patch29: gcc41-ppc-tramp.patch
-Patch30: gcc41-rh253102.patch
-Patch31: gcc41-c++-gnu_inline.patch
-Patch32: gcc41-ppc-sync-qihi.patch
-Patch33: gcc41-ppc64-ia64-GNU-stack.patch
-Patch34: gcc41-builtin-chk-anticipated.patch
-Patch35: gcc41-builtin-throw.patch
-Patch36: gcc41-builtin-va-arg-pack.patch
-Patch37: gcc41-builtin-va-arg-pack-len.patch
-Patch38: gcc41-pr27954.patch
-Patch39: gcc41-pr33423.patch
-Patch40: gcc41-scanf-fmt-check.patch
-Patch41: gcc41-pr33506.patch
-Patch42: gcc41-makeinfo-ver.patch
+Patch25: gcc41-ppc64-ia64-GNU-stack.patch
+Patch26: gcc41-pr33506.patch
+Patch27: gcc41-artificial-attrib.patch
+Patch28: gcc41-error-attrib.patch
+Patch29: gcc41-jdwp.patch
+Patch30: gcc41-pr20880.patch
+Patch31: gcc41-pr32694.patch
 
 # On ARM EABI systems, we do want -gnueabi to be part of the
 # target triple.
@@ -463,24 +452,13 @@ which are required to run programs compiled with the GNAT.
 %patch22 -p0 -b .build-id~
 %patch23 -p0 -b .pr28690~
 %patch24 -p0 -b .rh247256~
-%patch25 -p0 -b .pr22244~
-%patch26 -p0 -b .pr32678~
-%patch27 -p0 -b .pr32912~
-%patch28 -p0 -b .sparc-niagara~
-%patch29 -p0 -b .ppc-tramp~
-%patch30 -p0 -b .rh253102~
-%patch31 -p0 -b .c++-gnu_inline~
-%patch32 -p0 -b .ppc-sync-qihi~
-%patch33 -p0 -b .ppc64-ia64-GNU-stack~
-%patch34 -p0 -b .builtin-chk-anticipated~
-%patch35 -p0 -b .builtin-throw~
-%patch36 -p0 -b .builtin-va-arg-pack~
-%patch37 -p0 -b .builtin-va-arg-pack-len~
-%patch38 -p0 -b .pr27954~
-%patch39 -p0 -b .pr33423~
-%patch40 -p0 -b .scanf-fmt-check~
-%patch41 -p0 -b .pr33506~
-%patch42 -p0 -b .makeinfo-ver~
+%patch25 -p0 -b .ppc64-ia64-GNU-stack~
+%patch26 -p0 -b .pr33506~
+%patch27 -p0 -b .artificial-attrib~
+%patch28 -p0 -b .error-attrib~
+%patch29 -p0 -b .jdwp~
+%patch30 -p0 -b .pr20880~
+%patch31 -p0 -b .pr32694~
 
 sed -i -e 's/4\.1\.3/4.1.2/' gcc/BASE-VER gcc/version.c
 sed -i -e 's/" (Red Hat[^)]*)"/" (Red Hat %{version}-%{gcc_release})"/' gcc/version.c
@@ -1635,6 +1613,17 @@ fi
 %doc rpm.doc/changelogs/libmudflap/ChangeLog*
 
 %changelog
+* Tue Sep 25 2007 Jakub Jelinek <jakub@redhat.com> 4.1.2-28
+- update from gcc-4_1-branch (-r127672:128736)
+  - PRs bootstrap/33418, c++/31941, c++/32113, java/31842, target/33256,
+	tree-optimization/33142
+- add support for artificial, error and warning attributes
+- restore Java CNI ABI compatibility broken by JDWP changes (Keith Seitz)
+- fix ICE with set_rhs allowing non-gimple (Roger Sayle, #247407,
+  PR tree-optimization/32694)
+- fix ICE on Fortran interface-body declaring current subroutine name
+  (Paul Thomas, #300851, PR fortran/20880)
+
 * Sat Sep 22 2007 Jakub Jelinek <jakub@redhat.com> 4.1.2-26
 - don't ignore throw specification of function types in type hashing
   (PR c++/33506)
diff --git a/sources b/sources
index 941bcb0..64eed40 100644
--- a/sources
+++ b/sources
@@ -1 +1 @@
-65778706d6b9c029a06fca968a45ab7f  gcc-4.1.2-20070821.tar.bz2
+562ab2446c60a9145da385ac56cf7715  gcc-4.1.2-20070925.tar.bz2