gcc/

2008-08-06  H.J. Lu  <hongjiu.lu@intel.com>

	PR middle-end/37009
	* cfgexpand.c (expand_stack_alignment): Check parm_stack_boundary
	for incoming stack boundary.

	* function.c (assign_parm_find_entry_rtl): Update
	parm_stack_boundary.

	* function.h (rtl_data): Add parm_stack_boundary.

	* config/i386/i386.c (ix86_finalize_stack_realign_flags): Check
	parm_stack_boundary for incoming stack boundary.

gcc/testsuite/

2008-08-06  H.J. Lu  <hongjiu.lu@intel.com>

	PR middle-end/37009
	* gcc.dg/torture/stackalign/alloca-2.c: New.
	* gcc.dg/torture/stackalign/alloca-3.c: Likewise.
	* gcc.dg/torture/stackalign/alloca-4.c: Likewise.
	* gcc.dg/torture/stackalign/vararg-3.c: Likewise.
	* gcc.target/i386/incoming-1.c: Likewise.
	* gcc.target/i386/incoming-2.c: Likewise.
	* gcc.target/i386/incoming-3.c: Likewise.
	* gcc.target/i386/incoming-4.c: Likewise.
	* gcc.target/i386/incoming-5.c: Likewise.


git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@138806 138bc75d-0d04-0410-961f-82ee72b054a4

15 files changed, 380 insertions, 4 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 1e40f07e4f9..a22491f20e0 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,17 @@
+2008-08-06  H.J. Lu  <hongjiu.lu@intel.com>
+
+	PR middle-end/37009
+	* cfgexpand.c (expand_stack_alignment): Check parm_stack_boundary
+	for incoming stack boundary.
+
+	* function.c (assign_parm_find_entry_rtl): Update
+	parm_stack_boundary.
+
+	* function.h (rtl_data): Add parm_stack_boundary.
+
+	* config/i386/i386.c (ix86_finalize_stack_realign_flags): Check
+	parm_stack_boundary for incoming stack boundary.
+
 2008-08-06  Joseph Myers  <joseph@codesourcery.com>
 
 	* jump.c (rtx_renumbered_equal_p): Do not call subreg_regno_offset
diff --git a/gcc/cfgexpand.c b/gcc/cfgexpand.c
index a943eff6ec1..8d1e5d2e3e8 100644
--- a/gcc/cfgexpand.c
+++ b/gcc/cfgexpand.c
@@ -2184,7 +2184,7 @@ static void
 expand_stack_alignment (void)
 {
   rtx drap_rtx;
-  unsigned int preferred_stack_boundary;
+  unsigned int preferred_stack_boundary, incoming_stack_boundary;
 
   if (! SUPPORTS_STACK_ALIGNMENT)
     return;
@@ -2215,8 +2215,15 @@ expand_stack_alignment (void)
   if (preferred_stack_boundary > crtl->stack_alignment_needed)
     crtl->stack_alignment_needed = preferred_stack_boundary;
 
+  /* The incoming stack frame has to be aligned at least at
+     parm_stack_boundary.  */
+  if (crtl->parm_stack_boundary > INCOMING_STACK_BOUNDARY)
+    incoming_stack_boundary = crtl->parm_stack_boundary;
+  else
+    incoming_stack_boundary = INCOMING_STACK_BOUNDARY;
+
   crtl->stack_realign_needed
-    = INCOMING_STACK_BOUNDARY < crtl->stack_alignment_estimated;
+    = incoming_stack_boundary < crtl->stack_alignment_estimated;
   crtl->stack_realign_tried = crtl->stack_realign_needed;
 
   crtl->stack_realign_processed = true;
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index d31f176d9a3..7cd62110547 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -7614,7 +7614,10 @@ ix86_finalize_stack_realign_flags (void)
 {
   /* Check if stack realign is really needed after reload, and 
      stores result in cfun */
-  unsigned int stack_realign = (ix86_incoming_stack_boundary
+  unsigned int incoming_stack_boundary
+    = (crtl->parm_stack_boundary > ix86_incoming_stack_boundary
+       ? crtl->parm_stack_boundary : ix86_incoming_stack_boundary);
+  unsigned int stack_realign = (incoming_stack_boundary
 				< (current_function_is_leaf
 				   ? crtl->max_used_stack_slot_alignment
 				   : crtl->stack_alignment_needed));
diff --git a/gcc/function.c b/gcc/function.c
index 637775160eb..73600220c15 100644
--- a/gcc/function.c
+++ b/gcc/function.c
@@ -2261,6 +2261,11 @@ assign_parm_find_entry_rtl (struct assign_parm_data_all *all,
 		       entry_parm ? data->partial : 0, current_function_decl,
 		       &all->stack_args_size, &data->locate);
 
+  /* Update parm_stack_boundary if this parameter is passed in the
+     stack.  */
+  if (!in_regs && crtl->parm_stack_boundary < data->locate.boundary)
+    crtl->parm_stack_boundary = data->locate.boundary;
+
   /* Adjust offsets to include the pretend args.  */
   pretend_bytes = all->extra_pretend_bytes - pretend_bytes;
   data->locate.slot_offset.constant += pretend_bytes;
diff --git a/gcc/function.h b/gcc/function.h
index 1153fb0b4c4..eb85e3c37af 100644
--- a/gcc/function.h
+++ b/gcc/function.h
@@ -339,6 +339,9 @@ struct rtl_data GTY(())
      to call other functions.  */
   unsigned int preferred_stack_boundary;
 
+  /* The minimum alignment of parameter stack.  */
+  unsigned int parm_stack_boundary;
+
   /* The largest alignment of slot allocated on the stack.  */
   unsigned int max_used_stack_slot_alignment;
 
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index 1e035f3bf93..ead96385f80 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,4 +1,18 @@
+2008-08-06  H.J. Lu  <hongjiu.lu@intel.com>
+
+	PR middle-end/37009
+	* gcc.dg/torture/stackalign/alloca-2.c: New.
+	* gcc.dg/torture/stackalign/alloca-3.c: Likewise.
+	* gcc.dg/torture/stackalign/alloca-4.c: Likewise.
+	* gcc.dg/torture/stackalign/vararg-3.c: Likewise.
+	* gcc.target/i386/incoming-1.c: Likewise.
+	* gcc.target/i386/incoming-2.c: Likewise.
+	* gcc.target/i386/incoming-3.c: Likewise.
+	* gcc.target/i386/incoming-4.c: Likewise.
+	* gcc.target/i386/incoming-5.c: Likewise.
+
 2008-08-06  Aldy Hernandez  <aldyh@redhat.com>
+
 	PR middle-end/35432
 	* gcc.c-torture/compile/pr35432.c: New file.
 
@@ -9,7 +23,7 @@
 
 2008-08-06  Andreas Krebbel  <krebbel1@de.ibm.com>
 
- 	* gcc.c-torture/compile/20080806-1.c: New testcase.
+	* gcc.c-torture/compile/20080806-1.c: New testcase.
 
 2008-08-06  Maxim Kuvyrkov  <maxim@codesourcery.com>
 
diff --git a/gcc/testsuite/gcc.dg/torture/stackalign/alloca-2.c b/gcc/testsuite/gcc.dg/torture/stackalign/alloca-2.c
new file mode 100644
index 00000000000..b52dcf06566
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/torture/stackalign/alloca-2.c
@@ -0,0 +1,56 @@
+/* PR middle-end/37009 */
+/* { dg-do run { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-msse2" } */
+
+#include <emmintrin.h>
+#include "cpuid.h"
+#include "check.h"
+
+#ifndef ALIGNMENT
+#define ALIGNMENT	16
+#endif
+
+typedef int aligned __attribute__((aligned(ALIGNMENT)));
+
+void
+bar (char *p, int size)
+{
+  __builtin_strncpy (p, "good", size);
+}
+
+void
+__attribute__ ((noinline))
+foo (__m128 x, __m128 y ,__m128 z , int size)
+{
+  char *p = __builtin_alloca (size + 1);
+  aligned i;
+
+  bar (p, size);
+  if (__builtin_strncmp (p, "good", size) != 0)
+    {
+#ifdef DEBUG
+      p[size] = '\0';
+      printf ("Failed: %s != good\n", p);
+#endif
+      abort ();
+    }
+
+  if (check_int (&i,  __alignof__(i)) != i)
+    abort ();
+}
+
+int
+main (void)
+{
+  __m128 x = { 1.0 };
+  unsigned int eax, ebx, ecx, edx;
+ 
+  if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  /* Run SSE2 test only if host has SSE2 support.  */
+  if (edx & bit_SSE2)
+    foo (x, x, x, 5);
+
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.dg/torture/stackalign/alloca-3.c b/gcc/testsuite/gcc.dg/torture/stackalign/alloca-3.c
new file mode 100644
index 00000000000..47f3607c2a5
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/torture/stackalign/alloca-3.c
@@ -0,0 +1,56 @@
+/* PR middle-end/37009 */
+/* { dg-do run { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-msse2" } */
+
+#include <emmintrin.h>
+#include "cpuid.h"
+#include "check.h"
+
+#ifndef ALIGNMENT
+#define ALIGNMENT	16
+#endif
+
+typedef int aligned __attribute__((aligned(ALIGNMENT)));
+
+void
+bar (char *p, int size)
+{
+  __builtin_strncpy (p, "good", size);
+}
+
+void
+__attribute__ ((noinline))
+foo (__m128 x, __m128 y ,__m128 z ,__m128 a, int size)
+{
+  char *p = __builtin_alloca (size + 1);
+  aligned i;
+
+  bar (p, size);
+  if (__builtin_strncmp (p, "good", size) != 0)
+    {
+#ifdef DEBUG
+      p[size] = '\0';
+      printf ("Failed: %s != good\n", p);
+#endif
+      abort ();
+    }
+
+  if (check_int (&i,  __alignof__(i)) != i)
+    abort ();
+}
+
+int
+main (void)
+{
+  __m128 x = { 1.0 };
+  unsigned int eax, ebx, ecx, edx;
+ 
+  if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  /* Run SSE2 test only if host has SSE2 support.  */
+  if (edx & bit_SSE2)
+    foo (x, x, x, x, 5);
+
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.dg/torture/stackalign/alloca-4.c b/gcc/testsuite/gcc.dg/torture/stackalign/alloca-4.c
new file mode 100644
index 00000000000..0ff0d02c43b
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/torture/stackalign/alloca-4.c
@@ -0,0 +1,41 @@
+/* PR middle-end/37009 */
+/* { dg-do run { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-m32 -mincoming-stack-boundary=2 -mpreferred-stack-boundary=2" } */
+
+#include "check.h"
+
+void
+bar (char *p, int size)
+{
+  __builtin_strncpy (p, "good", size);
+}
+
+void
+__attribute__ ((noinline))
+foo (double x, double y ,double z ,double a, int size)
+{
+  char *p = __builtin_alloca (size + 1);
+  double i;
+
+  bar (p, size);
+  if (__builtin_strncmp (p, "good", size) != 0)
+    {
+#ifdef DEBUG
+      p[size] = '\0';
+      printf ("Failed: %s != good\n", p);
+#endif
+     abort ();
+    }
+
+  check (&i, __alignof__(i));
+}
+
+int
+main (void)
+{
+  double x =  1.0 ;
+ 
+  foo (x, x, x, x, 5);
+
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.dg/torture/stackalign/vararg-3.c b/gcc/testsuite/gcc.dg/torture/stackalign/vararg-3.c
new file mode 100644
index 00000000000..cac206499ed
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/torture/stackalign/vararg-3.c
@@ -0,0 +1,84 @@
+/* PR middle-end/37009 */
+/* { dg-do run { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-msse2" } */
+
+#include <stdarg.h>
+#include <emmintrin.h>
+#include "cpuid.h"
+#include "check.h"
+
+#ifndef ALIGNMENT
+#define ALIGNMENT	16
+#endif
+
+typedef int aligned __attribute__((aligned(ALIGNMENT)));
+
+void
+bar (char *p, int size)
+{
+  __builtin_strncpy (p, "good", size);
+}
+
+__m128 a = { 1.0 };
+
+void
+test (va_list arg)
+{
+  char *p;
+  aligned i;
+  int size;
+  double x;
+  __m128 e;
+
+  size = va_arg (arg, int);
+  if (size != 5)
+    abort ();
+
+  p = __builtin_alloca (size + 1);
+
+  x = va_arg (arg, double);
+  if (x != 5.0)
+    abort ();
+
+  bar (p, size);
+  if (__builtin_strncmp (p, "good", size) != 0)
+    {
+#ifdef DEBUG
+      p[size] = '\0';
+      printf ("Failed: %s != good\n", p);
+#endif
+      abort ();
+    }
+
+  if (check_int (&i,  __alignof__(i)) != i)
+    abort ();
+
+  e = va_arg (arg, __m128);
+  if (__builtin_memcmp (&e, &a, sizeof (e)))
+    abort ();
+}
+
+void
+foo (const char *fmt, ...)
+{
+  va_list arg;
+  va_start (arg, fmt);
+  test (arg);
+  va_end (arg);
+}
+
+int
+main (void)
+{
+  __m128 x = { 1.0 };
+  unsigned int eax, ebx, ecx, edx;
+ 
+  if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  /* Run SSE2 test only if host has SSE2 support.  */
+  if (edx & bit_SSE2)
+    foo ("foo", 5, 5.0, x);
+
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.target/i386/incoming-1.c b/gcc/testsuite/gcc.target/i386/incoming-1.c
new file mode 100644
index 00000000000..fcc80f1ec2a
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/incoming-1.c
@@ -0,0 +1,19 @@
+/* PR middle-end/37009 */
+/* { dg-do compile { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-w -msse2 -mpreferred-stack-boundary=2" } */
+
+#include <emmintrin.h>
+
+extern void bar (int *);
+
+int
+foo(__m128 x, __m128 y, __m128 z, int size)
+{
+  int __attribute((aligned(16))) xxx;
+
+  xxx = 2;
+  bar (&xxx);
+  return size;
+}
+
+/* { dg-final { scan-assembler "andl\[\\t \]*\\$-16,\[\\t \]*%esp" } } */
diff --git a/gcc/testsuite/gcc.target/i386/incoming-2.c b/gcc/testsuite/gcc.target/i386/incoming-2.c
new file mode 100644
index 00000000000..cc6c3934253
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/incoming-2.c
@@ -0,0 +1,19 @@
+/* PR middle-end/37009 */
+/* { dg-do compile { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-w -msse2 -mpreferred-stack-boundary=2" } */
+
+#include <emmintrin.h>
+
+extern void bar (int *);
+
+int
+foo(__m128 x, __m128 y, __m128 z, __m128 a, int size)
+{
+  int __attribute((aligned(16))) xxx;
+
+  xxx = 2;
+  bar (&xxx);
+  return size;
+}
+
+/* { dg-final { scan-assembler-not "and\[l\]\[ \t\]" } } */
diff --git a/gcc/testsuite/gcc.target/i386/incoming-3.c b/gcc/testsuite/gcc.target/i386/incoming-3.c
new file mode 100644
index 00000000000..aad38b53ab4
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/incoming-3.c
@@ -0,0 +1,19 @@
+/* PR middle-end/37009 */
+/* { dg-do compile { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-w -msse2 -mpreferred-stack-boundary=2" } */
+
+#include <emmintrin.h>
+
+extern void bar (int *);
+
+int
+foo(__m128 y, int size, ...)
+{
+  int __attribute((aligned(16))) xxx;
+
+  xxx = 2;
+  bar (&xxx);
+  return size;
+}
+
+/* { dg-final { scan-assembler-not "and\[l\]\[ \t\]" } } */
diff --git a/gcc/testsuite/gcc.target/i386/incoming-4.c b/gcc/testsuite/gcc.target/i386/incoming-4.c
new file mode 100644
index 00000000000..270024bc15c
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/incoming-4.c
@@ -0,0 +1,20 @@
+/* PR middle-end/37009 */
+/* { dg-do compile { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-w -msse2 -mpreferred-stack-boundary=2" } */
+
+#include <stdarg.h>
+#include <emmintrin.h>
+
+extern void bar (int *);
+
+__m128
+foo(va_list arg) 
+{
+  int __attribute((aligned(16))) xxx;
+
+  xxx = 2;
+  bar (&xxx);
+  return va_arg (arg, __m128);
+}
+
+/* { dg-final { scan-assembler "andl\[\\t \]*\\$-16,\[\\t \]*%esp" } } */
diff --git a/gcc/testsuite/gcc.target/i386/incoming-5.c b/gcc/testsuite/gcc.target/i386/incoming-5.c
new file mode 100644
index 00000000000..38620bfc938
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/incoming-5.c
@@ -0,0 +1,16 @@
+/* PR middle-end/37009 */
+/* { dg-do compile { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+/* { dg-options "-m32 -mincoming-stack-boundary=2 -mpreferred-stack-boundary=2" } */
+
+extern void bar (double *);
+
+double
+foo(double x)
+{
+  double xxx = x + 13.0;
+
+  bar (&xxx);
+  return xxx;
+}
+
+/* { dg-final { scan-assembler "andl\[\\t \]*\\$-8,\[\\t \]*%esp" } } */