Index: gcc.fsf.stkalign2/gcc/doc/extend.texi
===================================================================
*** gcc.fsf.stkalign2/gcc/doc/extend.texi	(revision 110663)
--- gcc.fsf.stkalign2/gcc/doc/extend.texi	(working copy)
*************** SSE registers instead of on the stack.  
*** 2207,2212 ****
--- 2207,2226 ----
  variable number of arguments will continue to pass all of their
  floating point arguments on the stack.
  
+ @item force_align_arg_pointer
+ @cindex @code{force_align_arg_pointer} attribute
+ On the Intel x86, the @code{force_align_arg_pointer} attribute may be
+ applied to individual function definitions, generating an alternate
+ prologue and epilogue that realigns the runtime stack.  This supports
+ mixing legacy codes that run with a 4-byte aligned stack with modern
+ codes that keep a 16-byte stack for SSE compatibility.  The alternate
+ prologue/epilogue is slower and bigger than the regular one, and it
+ requires one dedicated register for the life of the function.  This
+ also lowers the number of registers available if used in conjunction
+ with the @code{regparm} attribute.  The @code{force_align_arg_pointer}
+ attribute is incompatible with nested functions; this is considered a
+ hard error.
+ 
  @item returns_twice
  @cindex @code{returns_twice} attribute
  The @code{returns_twice} attribute tells the compiler that a function may
Index: gcc.fsf.stkalign2/gcc/doc/invoke.texi
===================================================================
*** gcc.fsf.stkalign2/gcc/doc/invoke.texi	(revision 110663)
--- gcc.fsf.stkalign2/gcc/doc/invoke.texi	(working copy)
*************** Objective-C and Objective-C++ Dialects}.
*** 532,537 ****
--- 532,538 ----
  -mthreads  -mno-align-stringops  -minline-all-stringops @gol
  -mpush-args  -maccumulate-outgoing-args  -m128bit-long-double @gol
  -m96bit-long-double  -mregparm=@var{num}  -msseregparm @gol
+ -mstackrealign @gol
  -momit-leaf-frame-pointer  -mno-red-zone -mno-tls-direct-seg-refs @gol
  -mcmodel=@var{code-model} @gol
  -m32  -m64 -mlarge-data-threshold=@var{num}}
*************** function by using the function attribute
*** 9379,9384 ****
--- 9380,9399 ----
  modules with the same value, including any libraries.  This includes
  the system libraries and startup modules.
  
+ @item -mstackrealign
+ @opindex mstackrealign
+ Realign the stack at entry.  On the Intel x86, the
+ @option{-mstackrealign} option will generate an alternate
+ prologue/epilogue that realigns the runtime stack.  This supports
+ mixing legacy codes that keep a 4-byte aligned stack with modern codes
+ that keep a 16-byte stack for SSE compatibility.  The alternate
+ prologue and epilogue are slower and bigger than the regular ones, and
+ they require one dedicated register for the entire function.  This
+ also lowers the number of registers available if used in conjunction
+ with the @code{regparm} attribute.  The @code{force_align_arg_pointer}
+ attribute is incompatible with the nested function prologue; this is
+ considered a hard error.
+ 
  @item -mpreferred-stack-boundary=@var{num}
  @opindex mpreferred-stack-boundary
  Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
Index: gcc.fsf.stkalign2/gcc/config/i386/i386.opt
===================================================================
*** gcc.fsf.stkalign2/gcc/config/i386/i386.opt	(revision 110663)
--- gcc.fsf.stkalign2/gcc/config/i386/i386.opt	(working copy)
*************** msselibm
*** 205,210 ****
--- 205,214 ----
  Target Mask(SSELIBM)
  Use SSE2 ABI libgcc-math routines if using SSE math
  
+ mstackrealign
+ Target Report Var(ix86_force_align_arg_pointer)
+ Realign stack in prologue
+ 
  msvr3-shlib
  Target Report Mask(SVR3_SHLIB)
  Uninitialized locals in .bss
Index: gcc.fsf.stkalign2/gcc/config/i386/i386.c
===================================================================
*** gcc.fsf.stkalign2/gcc/config/i386/i386.c	(revision 110663)
--- gcc.fsf.stkalign2/gcc/config/i386/i386.c	(working copy)
*************** int x86_prefetch_sse;
*** 1036,1041 ****
--- 1036,1045 ----
  /* ix86_regparm_string as a number */
  static int ix86_regparm;
  
+ /* -mstackrealign option */
+ extern int ix86_force_align_arg_pointer;
+ static const char ix86_force_align_arg_pointer_string[] = "force_align_arg_pointer";
+ 
  /* Preferred alignment for stack boundary in bits.  */
  unsigned int ix86_preferred_stack_boundary;
  
*************** const struct attribute_spec ix86_attribu
*** 2202,2207 ****
--- 2206,2214 ----
    /* Sseregparm attribute says we are using x86_64 calling conventions
       for FP arguments.  */
    { "sseregparm", 0, 0, false, true, true, ix86_handle_cconv_attribute },
+   /* force_align_arg_pointer says this function realigns the stack at entry.  */
+   { (const char *)&ix86_force_align_arg_pointer_string, 0, 0,
+     false, true,  true, ix86_handle_cconv_attribute },
  #if TARGET_DLLIMPORT_DECL_ATTRIBUTES
    { "dllimport", 0, 0, false, false, false, handle_dll_attribute },
    { "dllexport", 0, 0, false, false, false, handle_dll_attribute },
*************** ix86_function_regparm (tree type, tree d
*** 2483,2488 ****
--- 2490,2499 ----
  		  && decl_function_context (decl)
  		  && !DECL_NO_STATIC_CHAIN (decl))
  		local_regparm = 2;
+ 	      /* If the function realigns its stackpointer, it
+ 		 uses %ecx to reach unaligned arguments.  */
+ 	      if (cfun->machine->force_align_arg_pointer)
+ 		local_regparm = 2;
  	      /* Each global register variable increases register preassure,
  		 so the more global reg vars there are, the smaller regparm
  		 optimization use, unless requested by the user explicitly.  */
*************** pro_epilogue_adjust_stack (rtx dest, rtx
*** 5132,5142 ****
  static rtx
  ix86_internal_arg_pointer (void)
  {
!   if (FORCE_PREFERRED_STACK_BOUNDARY_IN_MAIN
!       && DECL_NAME (current_function_decl)
!       && MAIN_NAME_P (DECL_NAME (current_function_decl))
!       && DECL_FILE_SCOPE_P (current_function_decl))
!     {
        cfun->machine->force_align_arg_pointer = gen_rtx_REG (Pmode, 2);
        return copy_to_reg (cfun->machine->force_align_arg_pointer);
      }
--- 5143,5167 ----
  static rtx
  ix86_internal_arg_pointer (void)
  {
!   bool has_force_align_arg_pointer =
!     (0 != lookup_attribute (ix86_force_align_arg_pointer_string,
! 			    TYPE_ATTRIBUTES (TREE_TYPE (current_function_decl))));
!   if ((FORCE_PREFERRED_STACK_BOUNDARY_IN_MAIN
!        && DECL_NAME (current_function_decl)
!        && MAIN_NAME_P (DECL_NAME (current_function_decl))
!        && DECL_FILE_SCOPE_P (current_function_decl))
!       || ix86_force_align_arg_pointer
!       || has_force_align_arg_pointer)
!     {
!       if (!DECL_FILE_SCOPE_P (current_function_decl))
! 	{
! 	  if (ix86_force_align_arg_pointer)
! 	    warning (0, "-mstackrealign ignored for nested functions");
! 	  if (has_force_align_arg_pointer)
! 	    error ("%s not supported for nested functions",
! 		   ix86_force_align_arg_pointer_string);
! 	  return virtual_incoming_args_rtx;
! 	}
        cfun->machine->force_align_arg_pointer = gen_rtx_REG (Pmode, 2);
        return copy_to_reg (cfun->machine->force_align_arg_pointer);
      }