re PR target/66697 (Feature request: -mstackrealign and force_align_arg_pointer for x86_64)

	PR target/66697
	* config/i386/i386.c (ix86_option_override_internal): Always use
	8-byte minimum stack boundary in 64-bit mode.
	(ix86_compute_frame_layout): Remove assert on INCOMING_STACK_BOUNDARY.
	(ix86_emit_save_reg_using_mov): Support unaligned SSE store.
	Add a REG_CFA_EXPRESSION note if needed.
	(ix86_emit_restore_sse_regs_using_mov): Support unaligned SSE load.
	(ix86_handle_force_align_arg_pointer_attribute): New.
	(ix86_minimum_incoming_stack_boundary): Remove TARGET_64BIT check.
	(ix86_attribute_table): Set ix86_force_align_arg_pointer_string
	with ix86_handle_force_align_arg_pointer_attribute.
	* config/i386/i386.h (MIN_STACK_BOUNDARY): Set to BITS_PER_WORD.

testsuite/ChangeLog:

	PR target/66697
	* gcc.target/i386/20060512-1.c: Remove ia32 requirement.
	(PUSH, POP): New defines.
	(sse2_test): Use PUSH and POP to misalign runtime stack.
	* gcc.target/i386/20060512-2.c: Remove ia32 requirement.

From-SVN: r228577

6 files changed, 97 insertions, 21 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 2d51ce8..7504c56 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,18 @@
+2015-10-07  Uros Bizjak  <ubizjak@gmail.com>
+
+	PR target/66697
+	* config/i386/i386.c (ix86_option_override_internal): Always use
+	8-byte minimum stack boundary in 64-bit mode.
+	(ix86_compute_frame_layout): Remove assert on INCOMING_STACK_BOUNDARY.
+	(ix86_emit_save_reg_using_mov): Support unaligned SSE store.
+	Add a REG_CFA_EXPRESSION note if needed.
+	(ix86_emit_restore_sse_regs_using_mov): Support unaligned SSE load.
+	(ix86_handle_force_align_arg_pointer_attribute): New.
+	(ix86_minimum_incoming_stack_boundary): Remove TARGET_64BIT check.
+	(ix86_attribute_table): Set ix86_force_align_arg_pointer_string
+	with ix86_handle_force_align_arg_pointer_attribute.
+	* config/i386/i386.h (MIN_STACK_BOUNDARY): Set to BITS_PER_WORD.
+
 2015-10-07  Aditya Kumar  <aditya.k7@samsung.com>
 	    Sebastian Pop  <s.pop@samsung.com>
 
@@ -287,7 +302,8 @@
 	* graphite-sese-to-poly.c (build_loop_iteration_domains): Only loops
 	which are in this region are passed so gcc_assert and remove redundant
 	computation.
-	* sese.c (sese_build_liveouts): Pass only those bbs which are not in region.
+	* sese.c (sese_build_liveouts): Pass only those bbs which are not
+	in region.
 	(sese_bad_liveouts_use): Only BBs which are not in region are passed so
 	gcc_assert on that and remove unnecessary computation.
 	(sese_build_liveouts_use): Same.
@@ -500,7 +516,7 @@
 2015-10-05  Aditya Kumar  <aditya.k7@samsung.com>
 	    Sebastian Pop  <s.pop@samsung.com>
 
-        * params.def (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Increase to 7.
+	* params.def (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Increase to 7.
 
 2015-10-05  Aditya Kumar  <aditya.k7@samsung.com>
 	    Sebastian Pop  <s.pop@samsung.com>
@@ -1111,7 +1127,8 @@
 	* graphite-poly.c (new_poly_bb): Same.
 	* graphite-poly.h (gbb_from_bb): Same.
 	* sese.h: Same.
-	* graphite-sese-to-poly.c (new_gimple_bb): gimple_bb_p -> gimple_poly_bb_p
+	* graphite-sese-to-poly.c (new_gimple_bb):
+	gimple_bb_p -> gimple_poly_bb_p
 	(build_scop_scattering): Same.
 	(find_params_in_bb): Same.
 	(add_conditions_to_domain): Same.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index c5ebff5..a24bd26 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -5209,8 +5209,7 @@ ix86_option_override_internal (bool main_args_p,
   ix86_incoming_stack_boundary = ix86_default_incoming_stack_boundary;
   if (opts_set->x_ix86_incoming_stack_boundary_arg)
     {
-      int min = (TARGET_64BIT_P (opts->x_ix86_isa_flags)
-		 ? (TARGET_SSE_P (opts->x_ix86_isa_flags) ? 4 : 3) : 2);
+      int min = TARGET_64BIT_P (opts->x_ix86_isa_flags) ? 3 : 2;
 
       if (opts->x_ix86_incoming_stack_boundary_arg < min
 	  || opts->x_ix86_incoming_stack_boundary_arg > 12)
@@ -11386,7 +11385,6 @@ ix86_compute_frame_layout (struct ix86_frame *frame)
       /* The only ABI that has saved SSE registers (Win64) also has a
          16-byte aligned default stack, and thus we don't need to be
 	 within the re-aligned local stack frame to save them.  */
-      gcc_assert (INCOMING_STACK_BOUNDARY >= 128);
       offset = ROUND_UP (offset, 16);
       offset += frame->nsseregs * 16;
     }
@@ -11611,14 +11609,26 @@ ix86_emit_save_reg_using_mov (machine_mode mode, unsigned int regno,
   struct machine_function *m = cfun->machine;
   rtx reg = gen_rtx_REG (mode, regno);
   rtx mem, addr, base, insn;
+  unsigned int align;
 
   addr = choose_baseaddr (cfa_offset);
   mem = gen_frame_mem (mode, addr);
 
-  /* For SSE saves, we need to indicate the 128-bit alignment.  */
-  set_mem_align (mem, GET_MODE_ALIGNMENT (mode));
+  /* The location is aligned up to INCOMING_STACK_BOUNDARY.  */
+  align = MIN (GET_MODE_ALIGNMENT (mode), INCOMING_STACK_BOUNDARY);
+  set_mem_align (mem, align);
+
+  /* SSE saves are not within re-aligned local stack frame.
+     In case INCOMING_STACK_BOUNDARY is misaligned, we have
+     to emit unaligned store.  */
+  if (mode == V4SFmode && align < 128)
+    {
+      rtx unspec = gen_rtx_UNSPEC (mode, gen_rtvec (1, reg), UNSPEC_STOREU);
+      insn = emit_insn (gen_rtx_SET (mem, unspec));
+    }
+  else
+    insn = emit_insn (gen_rtx_SET (mem, reg));
 
-  insn = emit_move_insn (mem, reg);
   RTX_FRAME_RELATED_P (insn) = 1;
 
   base = addr;
@@ -11665,6 +11675,8 @@ ix86_emit_save_reg_using_mov (machine_mode mode, unsigned int regno,
       mem = gen_rtx_MEM (mode, addr);
       add_reg_note (insn, REG_CFA_OFFSET, gen_rtx_SET (mem, reg));
     }
+  else
+    add_reg_note (insn, REG_CFA_EXPRESSION, gen_rtx_SET (mem, reg));
 }
 
 /* Emit code to save registers using MOV insns.
@@ -11881,6 +11893,25 @@ find_drap_reg (void)
     }
 }
 
+/* Handle a "force_align_arg_pointer" attribute.  */
+
+static tree
+ix86_handle_force_align_arg_pointer_attribute (tree *node, tree name,
+					       tree, int, bool *no_add_attrs)
+{
+  if (TREE_CODE (*node) != FUNCTION_TYPE
+      && TREE_CODE (*node) != METHOD_TYPE
+      && TREE_CODE (*node) != FIELD_DECL
+      && TREE_CODE (*node) != TYPE_DECL)
+    {
+      warning (OPT_Wattributes, "%qE attribute only applies to functions",
+	       name);
+      *no_add_attrs = true;
+    }
+
+  return NULL_TREE;
+}
+
 /* Return minimum incoming stack alignment.  */
 
 static unsigned int
@@ -11895,7 +11926,6 @@ ix86_minimum_incoming_stack_boundary (bool sibcall)
      if -mstackrealign is used, it isn't used for sibcall check and
      estimated stack alignment is 128bit.  */
   else if (!sibcall
-	   && !TARGET_64BIT
 	   && ix86_force_align_arg_pointer
 	   && crtl->stack_alignment_estimated == 128)
     incoming_stack_boundary = MIN_STACK_BOUNDARY;
@@ -13179,11 +13209,26 @@ ix86_emit_restore_sse_regs_using_mov (HOST_WIDE_INT cfa_offset,
       {
 	rtx reg = gen_rtx_REG (V4SFmode, regno);
 	rtx mem;
+	unsigned int align;
 
 	mem = choose_baseaddr (cfa_offset);
 	mem = gen_rtx_MEM (V4SFmode, mem);
-	set_mem_align (mem, 128);
-	emit_move_insn (reg, mem);
+
+	/* The location is aligned up to INCOMING_STACK_BOUNDARY.  */
+	align = MIN (GET_MODE_ALIGNMENT (V4SFmode), INCOMING_STACK_BOUNDARY);
+	set_mem_align (mem, align);
+
+	/* SSE saves are not within re-aligned local stack frame.
+	   In case INCOMING_STACK_BOUNDARY is misaligned, we have
+	   to emit unaligned load.  */
+	if (align < 128)
+	  {
+	    rtx unspec = gen_rtx_UNSPEC (V4SFmode, gen_rtvec (1, mem),
+					 UNSPEC_LOADU);
+	    emit_insn (gen_rtx_SET (reg, unspec));
+	  }
+	else
+	  emit_insn (gen_rtx_SET (reg, mem));
 
 	ix86_add_cfa_restore_note (NULL, reg, cfa_offset);
 
@@ -48154,7 +48199,7 @@ static const struct attribute_spec ix86_attribute_table[] =
     true },
   /* 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, false },
+    false, true,  true, ix86_handle_force_align_arg_pointer_attribute, false },
 #if TARGET_DLLIMPORT_DECL_ATTRIBUTES
   { "dllimport", 0, 0, false, false, false, handle_dll_attribute, false },
   { "dllexport", 0, 0, false, false, false, handle_dll_attribute, false },
diff --git a/gcc/config/i386/i386.h b/gcc/config/i386/i386.h
index ad17460..4a84fb9 100644
--- a/gcc/config/i386/i386.h
+++ b/gcc/config/i386/i386.h
@@ -755,7 +755,7 @@ extern const char *host_detect_local_cpu (int argc, const char **argv);
 #define MAIN_STACK_BOUNDARY (TARGET_64BIT ? 128 : 32)
 
 /* Minimum stack boundary.  */
-#define MIN_STACK_BOUNDARY (TARGET_64BIT ? (TARGET_SSE ? 128 : 64) : 32)
+#define MIN_STACK_BOUNDARY BITS_PER_WORD
 
 /* Boundary (in *bits*) on which the stack pointer prefers to be
    aligned; the compiler cannot rely on having this alignment.  */
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index d06a25f..cc3cd6b 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,6 +1,14 @@
+2015-10-07  Uros Bizjak  <ubizjak@gmail.com>
+
+	PR target/66697
+	* gcc.target/i386/20060512-1.c: Remove ia32 requirement.
+	(PUSH, POP): New defines.
+	(sse2_test): Use PUSH and POP to misalign runtime stack.
+	* gcc.target/i386/20060512-2.c: Remove ia32 requirement.
+
 2015-10-07  James Norris  <jnorris@codesourcery.com>
 
-	* testsuite/lib/atomic-dg.exp (atomic_link_flags): Move flag setting
+	* lib/atomic-dg.exp (atomic_link_flags): Move flag setting
 	to atomic_init. (atomic_init): Restrict flags usage.
 
 2015-10-07  Andre Vehreschild  <vehre@gcc.gnu.org>
@@ -62,12 +70,12 @@
 2015-10-05  Aditya Kumar  <aditya.k7@samsung.com>
 	    Sebastian Pop  <s.pop@samsung.com>
 
-        * gcc.dg/graphite/scop-sor.c: Un-xfail.
+	* gcc.dg/graphite/scop-sor.c: Un-xfail.
 
 2015-10-05  Aditya Kumar  <aditya.k7@samsung.com>
 	    Sebastian Pop  <s.pop@samsung.com>
 
-        * gcc.dg/graphite/scop-sor.c: Xfail.
+	* gcc.dg/graphite/scop-sor.c: Xfail.
 
 2015-10-05  Martin Jambor  <mjambor@suse.cz>
 	    Jan Hubicka  <hubicka@ucw.cz>
diff --git a/gcc/testsuite/gcc.target/i386/20060512-1.c b/gcc/testsuite/gcc.target/i386/20060512-1.c
index 374d18a..ec163a9 100644
--- a/gcc/testsuite/gcc.target/i386/20060512-1.c
+++ b/gcc/testsuite/gcc.target/i386/20060512-1.c
@@ -1,5 +1,4 @@
 /* { dg-do run } */
-/* { dg-require-effective-target ia32 } */
 /* { dg-options "-std=gnu99 -msse2 -mpreferred-stack-boundary=4" } */
 /* { dg-require-effective-target sse2 } */
 
@@ -7,6 +6,14 @@
 
 #include <emmintrin.h>
 
+#ifdef __x86_64__
+# define PUSH "pushq %rsi"
+# define POP "popq %rsi"
+#else
+# define PUSH "pushl %esi"
+# define POP "popl %esi"
+#endif
+
 __m128i __attribute__ ((__noinline__))
 vector_using_function ()
 {
@@ -27,9 +34,9 @@ static void
 sse2_test (void)
 {
   int result;
-  asm ("pushl %esi");		/* Disalign runtime stack.  */
+  asm (PUSH);                  /* Misalign runtime stack.  */
   result = self_aligning_function (g_1, g_2);
   if (result != 42)
     abort ();
-  asm ("popl %esi");
+  asm (POP);
 }
diff --git a/gcc/testsuite/gcc.target/i386/20060512-2.c b/gcc/testsuite/gcc.target/i386/20060512-2.c
index d3a779c..8ce4bd7 100644
--- a/gcc/testsuite/gcc.target/i386/20060512-2.c
+++ b/gcc/testsuite/gcc.target/i386/20060512-2.c
@@ -1,5 +1,4 @@
 /* { dg-do compile } */
-/* { dg-require-effective-target ia32 } */
 /* { dg-options "-std=gnu99 -mpreferred-stack-boundary=4" } */
 int
 outer_function (int x, int y)