obtained gcc-4.6.4.tar.bz2 from upstream website;upstream

verified gcc-4.6.4.tar.bz2.sig;
imported gcc-4.6.4 source tree from verified upstream tarball.

downloading a git-generated archive based on the 'upstream' tag
should provide you with a source tree that is binary identical
to the one extracted from the above tarball.

if you have obtained the source via the command 'git clone',
however, do note that line-endings of files in your working
directory might differ from line-endings of the respective
files in the upstream repository.

140 files changed, 8556 insertions, 0 deletions

diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.15.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.15.1.c
new file mode 100644
index 000000000..276ac6caa
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.15.1.c
@@ -0,0 +1,44 @@
+/* { dg-do run } */
+
+#include <stdio.h>
+
+void
+work (int n)
+{
+  printf ("[%d of %d], nested = %d, n = %d\n", omp_get_thread_num (), omp_get_num_threads(), omp_get_nested (), n);
+}
+
+void
+sub3 (int n)
+{
+  work (n);
+#pragma omp barrier
+  work (n);
+}
+
+void
+sub2 (int k)
+{
+#pragma omp parallel shared(k)
+  sub3 (k);
+}
+
+void
+sub1 (int n)
+{
+  int i;
+#pragma omp parallel private(i) shared(n)
+  {
+#pragma omp for
+    for (i = 0; i < n; i++)
+      sub2 (i);
+  }
+}
+int
+main ()
+{
+  sub1 (2);
+  sub2 (15);
+  sub3 (20);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.16.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.16.1.c
new file mode 100644
index 000000000..28a994a28
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.16.1.c
@@ -0,0 +1,47 @@
+/* { dg-do run } */
+
+#include <stdio.h>
+
+float
+work1 (int i)
+{
+  return 1.0 * i;
+}
+
+float
+work2 (int i)
+{
+  return 2.0 * i;
+}
+
+void
+a16 (float *x, float *y, int *index, int n)
+{
+  int i;
+#pragma omp parallel for shared(x, y, index, n)
+  for (i = 0; i < n; i++)
+    {
+#pragma omp atomic
+      x[index[i]] += work1 (i);
+      y[i] += work2 (i);
+    }
+}
+int
+main ()
+{
+  float x[1000];
+  float y[10000];
+  int index[10000];
+  int i;
+  for (i = 0; i < 10000; i++)
+    {
+      index[i] = i % 1000;
+      y[i] = 0.0;
+    }
+  for (i = 0; i < 1000; i++)
+    x[i] = 0.0;
+  a16 (x, y, index, 10000);
+  for (i = 0; i < 10; i++)
+    printf ("x[%d] = %f, y[%d] = %f\n", i, x[i], i, y[i]);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.18.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.18.1.c
new file mode 100644
index 000000000..388763e59
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.18.1.c
@@ -0,0 +1,67 @@
+/* { dg-do run } */
+
+#include <omp.h>
+#include <stdio.h>
+
+extern void abort (void);
+
+#define NUMBER_OF_THREADS 4
+
+int synch[NUMBER_OF_THREADS];
+int work[NUMBER_OF_THREADS];
+int result[NUMBER_OF_THREADS];
+int
+fn1 (int i)
+{
+  return i * 2;
+}
+
+int
+fn2 (int a, int b)
+{
+  return a + b;
+}
+
+int
+main ()
+{
+  int i, iam, neighbor;
+  omp_set_num_threads (NUMBER_OF_THREADS);
+#pragma omp parallel private(iam,neighbor) shared(work,synch)
+  {
+    iam = omp_get_thread_num ();
+    synch[iam] = 0;
+#pragma omp barrier
+    /*Do computation into my portion of work array */
+    work[iam] = fn1 (iam);
+    /* Announce that I am done with my work. The first flush
+     * ensures that my work is made visible before synch.
+     * The second flush ensures that synch is made visible.
+     */
+#pragma omp flush(work,synch)
+    synch[iam] = 1;
+#pragma omp flush(synch)
+    /* Wait for neighbor. The first flush ensures that synch is read
+     * from memory, rather than from the temporary view of memory.
+     * The second flush ensures that work is read from memory, and
+     * is done so after the while loop exits.
+     */
+    neighbor = (iam > 0 ? iam : omp_get_num_threads ()) - 1;
+    while (synch[neighbor] == 0)
+      {
+#pragma omp flush(synch)
+      }
+#pragma omp flush(work,synch)
+    /* Read neighbor's values of work array */
+    result[iam] = fn2 (work[neighbor], work[iam]);
+  }
+  /* output result here */
+  for (i = 0; i < NUMBER_OF_THREADS; i++)
+    {
+      neighbor = (i > 0 ? i : NUMBER_OF_THREADS) - 1;
+      if (result[i] != i * 2 + neighbor * 2)
+	abort ();
+    }
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.19.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.19.1.c
new file mode 100644
index 000000000..65ffe624c
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.19.1.c
@@ -0,0 +1,55 @@
+/* { dg-do run } */
+
+int x, *p = &x;
+extern void abort (void);
+void
+f1 (int *q)
+{
+  *q = 1;
+#pragma omp flush
+  /* x, p, and *q are flushed */
+  /* because they are shared and accessible */
+  /* q is not flushed because it is not shared. */
+}
+
+void
+f2 (int *q)
+{
+#pragma omp barrier
+  *q = 2;
+#pragma omp barrier
+  /*  a barrier implies a flush */
+  /*  x, p, and *q are flushed */
+  /*  because they are shared and accessible */
+  /*  q is not flushed because it is not shared. */
+}
+
+int
+g (int n)
+{
+  int i = 1, j, sum = 0;
+  *p = 1;
+#pragma omp parallel reduction(+: sum) num_threads(2)
+  {
+    f1 (&j);
+    /* i, n and sum were not flushed */
+    /* because they were not accessible in f1 */
+    /* j was flushed because it was accessible */
+    sum += j;
+    f2 (&j);
+    /* i, n, and sum were not flushed */
+    /* because they were not accessible in f2 */
+    /* j was flushed because it was accessible */
+    sum += i + j + *p + n;
+  }
+  return sum;
+}
+
+int
+main ()
+{
+  int result = g (10);
+  if (result != 30)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.2.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.2.1.c
new file mode 100644
index 000000000..f6ae4c7c5
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.2.1.c
@@ -0,0 +1,45 @@
+/* { dg-do run } */
+
+#include <stdio.h>
+#include <omp.h>
+extern void abort (void);
+int
+main ()
+{
+  int bad, x;
+  x = 2;
+  bad = 0;
+#pragma omp parallel num_threads(2) shared(x, bad)
+  {
+    if (omp_get_thread_num () == 0)
+      {
+	volatile int i;
+	for (i = 0; i < 100000000; i++)
+	  x = 5;
+      }
+    else
+      {
+	/* Print 1: the following read of x has a race */
+	if (x != 2 && x != 5)
+	  bad = 1;
+      }
+#pragma omp barrier
+    if (omp_get_thread_num () == 0)
+      {
+	/* x must be 5 now.  */
+	if (x != 5)
+	  bad = 1;
+      }
+    else
+      {
+	/* x must be 5 now.  */
+	if (x != 5)
+	  bad = 1;
+      }
+  }
+
+  if (bad)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.21.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.21.1.c
new file mode 100644
index 000000000..0c1c39a12
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.21.1.c
@@ -0,0 +1,25 @@
+/* { dg-do run } */
+
+#include <stdio.h>
+void
+work (int k)
+{
+#pragma omp ordered
+  printf (" %d\n", k);
+}
+
+void
+a21 (int lb, int ub, int stride)
+{
+  int i;
+#pragma omp parallel for ordered schedule(dynamic)
+  for (i = lb; i < ub; i += stride)
+    work (i);
+}
+
+int
+main ()
+{
+  a21 (0, 100, 5);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.26.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.26.1.c
new file mode 100644
index 000000000..e146fa204
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.26.1.c
@@ -0,0 +1,17 @@
+/* { dg-do run } */
+
+#include <stdio.h>
+int
+main ()
+{
+  int i, j;
+  i = 1;
+  j = 2;
+#pragma omp parallel private(i) firstprivate(j)
+  {
+    i = 3;
+    j = j + 2;
+  }
+  printf ("%d %d\n", i, j);	/* i and j are undefined */
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.29.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.29.1.c
new file mode 100644
index 000000000..6f0f65fa0
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.29.1.c
@@ -0,0 +1,30 @@
+/* { dg-do run } */
+
+#include <assert.h>
+int A[2][2] = { 1, 2, 3, 4 };
+void
+f (int n, int B[n][n], int C[])
+{
+  int D[2][2] = { 1, 2, 3, 4 };
+  int E[n][n];
+  assert (n >= 2);
+  E[1][1] = 4;
+#pragma omp parallel firstprivate(B, C, D, E)
+  {
+    assert (sizeof (B) == sizeof (int (*)[n]));
+    assert (sizeof (C) == sizeof (int *));
+    assert (sizeof (D) == 4 * sizeof (int));
+    assert (sizeof (E) == n * n * sizeof (int));
+    /* Private B and C have values of original B and C. */
+    assert (&B[1][1] == &A[1][1]);
+    assert (&C[3] == &A[1][1]);
+    assert (D[1][1] == 4);
+    assert (E[1][1] == 4);
+  }
+}
+int
+main ()
+{
+  f (2, A, A[0]);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.3.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.3.1.c
new file mode 100644
index 000000000..9e7c24f75
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.3.1.c
@@ -0,0 +1,11 @@
+/* { dg-do run } */
+
+#include <stdio.h>
+int
+main ()
+{
+# ifdef _OPENMP
+  printf ("Compiled by an OpenMP-compliant implementation.\n");
+# endif
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.33.3.c b/libgomp/testsuite/libgomp.c/appendix-a/a.33.3.c
new file mode 100644
index 000000000..0b7f0197c
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.33.3.c
@@ -0,0 +1,16 @@
+/* { dg-do compile } */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <omp.h>
+omp_lock_t *
+new_lock ()
+{
+  omp_lock_t *lock_ptr;
+#pragma omp single copyprivate(lock_ptr)
+  {
+    lock_ptr = (omp_lock_t *) malloc (sizeof (omp_lock_t));
+    omp_init_lock (lock_ptr);
+  }
+  return lock_ptr;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.36.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.36.1.c
new file mode 100644
index 000000000..b60534d11
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.36.1.c
@@ -0,0 +1,31 @@
+/* { dg-do run } */
+
+#include <omp.h>
+#include <stdlib.h>
+void
+do_by_16 (float *x, int iam, int ipoints)
+{
+}
+
+void
+a36 (float *x, int npoints)
+{
+  int iam, ipoints;
+  omp_set_dynamic (0);
+  omp_set_num_threads (16);
+#pragma omp parallel shared(x, npoints) private(iam, ipoints)
+  {
+    if (omp_get_num_threads () != 16)
+      abort ();
+    iam = omp_get_thread_num ();
+    ipoints = npoints / 16;
+    do_by_16 (x, iam, ipoints);
+  }
+}
+
+int main()
+{
+  float a[10];
+  a36 (a, 10);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.39.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.39.1.c
new file mode 100644
index 000000000..a129e87fc
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.39.1.c
@@ -0,0 +1,38 @@
+/* { dg-do run } */
+
+#include <stdio.h>
+#include <omp.h>
+void
+skip (int i)
+{
+}
+
+void
+work (int i)
+{
+}
+int
+main ()
+{
+  omp_lock_t lck;
+  int id;
+  omp_init_lock (&lck);
+#pragma omp parallel shared(lck) private(id)
+  {
+    id = omp_get_thread_num ();
+    omp_set_lock (&lck);
+    /* only one thread at a time can execute this printf */
+    printf ("My thread id is %d.\n", id);
+    omp_unset_lock (&lck);
+    while (!omp_test_lock (&lck))
+      {
+	skip (id);		/* we do not yet have the lock,
+				   so we must do something else */
+      }
+    work (id);			/* we now have the lock
+				   and can do the work */
+    omp_unset_lock (&lck);
+  }
+  omp_destroy_lock (&lck);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.4.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.4.1.c
new file mode 100644
index 000000000..c6be49090
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.4.1.c
@@ -0,0 +1,38 @@
+/* { dg-do run } */
+
+#include <omp.h>
+extern void abort (void);
+void
+subdomain (float *x, int istart, int ipoints)
+{
+  int i;
+  for (i = 0; i < ipoints; i++)
+    x[istart + i] = 123.456;
+}
+
+void
+sub (float *x, int npoints)
+{
+  int iam, nt, ipoints, istart;
+#pragma omp parallel default(shared) private(iam,nt,ipoints,istart)
+  {
+    iam = omp_get_thread_num ();
+    nt = omp_get_num_threads ();
+    ipoints = npoints / nt;	/* size of partition */
+    istart = iam * ipoints;	/* starting array index */
+    if (iam == nt - 1)		/* last thread may do more */
+      ipoints = npoints - istart;
+    subdomain (x, istart, ipoints);
+  }
+}
+int
+main ()
+{
+  int i;
+  float array[10000];
+  sub (array, 10000);
+  for (i = 0; i < 10000; i++)
+    if (array[i] < 123.45 || array[i] > 123.46)
+      abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.40.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.40.1.c
new file mode 100644
index 000000000..aa7b88d70
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.40.1.c
@@ -0,0 +1,48 @@
+/* { dg-do compile } */
+
+#include <omp.h>
+typedef struct
+{
+  int a, b;
+  omp_nest_lock_t lck;
+} pair;
+int work1 ();
+int work2 ();
+int work3 ();
+void
+incr_a (pair * p, int a)
+{
+  /* Called only from incr_pair, no need to lock. */
+  p->a += a;
+}
+
+void
+incr_b (pair * p, int b)
+{
+  /* Called both from incr_pair and elsewhere, */
+  /* so need a nestable lock. */
+  omp_set_nest_lock (&p->lck);
+  p->b += b;
+  omp_unset_nest_lock (&p->lck);
+}
+
+void
+incr_pair (pair * p, int a, int b)
+{
+  omp_set_nest_lock (&p->lck);
+  incr_a (p, a);
+  incr_b (p, b);
+  omp_unset_nest_lock (&p->lck);
+}
+
+void
+a40 (pair * p)
+{
+#pragma omp parallel sections
+  {
+#pragma omp section
+    incr_pair (p, work1 (), work2 ());
+#pragma omp section
+    incr_b (p, work3 ());
+  }
+}
diff --git a/libgomp/testsuite/libgomp.c/appendix-a/a.5.1.c b/libgomp/testsuite/libgomp.c/appendix-a/a.5.1.c
new file mode 100644
index 000000000..b909c4ddb
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/appendix-a/a.5.1.c
@@ -0,0 +1,13 @@
+/* { dg-do run } */
+
+#include <omp.h>
+int
+main ()
+{
+  omp_set_dynamic (1);
+#pragma omp parallel num_threads(10)
+  {
+    /* do work here */
+  }
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/atomic-1.c b/libgomp/testsuite/libgomp.c/atomic-1.c
new file mode 100644
index 000000000..b2be8f022
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/atomic-1.c
@@ -0,0 +1,62 @@
+/* { dg-do run } */
+/* { dg-options "-O2 -march=pentium" { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+
+#ifdef __i386__
+#include "cpuid.h"
+#endif
+
+extern void abort (void);
+double d;
+struct
+{
+  int i;
+  double e;
+  int j;
+} x;
+
+void
+f1 (void)
+{
+  #pragma omp atomic
+    d += 7.5;
+  #pragma omp atomic
+    d *= 2.5;
+  #pragma omp atomic
+    d /= 0.25;
+}
+
+void
+f2 (void)
+{
+  #pragma omp atomic
+    x.e += 7.5;
+  #pragma omp atomic
+    x.e *= 2.5;
+  #pragma omp atomic
+    x.e /= 0.25;
+}
+
+int
+main (void)
+{
+#ifdef __i386__
+  unsigned int eax, ebx, ecx, edx;
+
+  if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  if (!(edx & bit_CMPXCHG8B))
+    return 0;
+#endif
+
+  d = 1.0;
+  f1 ();
+  if (d != 85.0)
+    abort ();
+
+  x.e = 1.0;
+  f2 ();
+  if (x.i != 0 || x.e != 85.0 || x.j != 0)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/atomic-10.c b/libgomp/testsuite/libgomp.c/atomic-10.c
new file mode 100644
index 000000000..58edeed6c
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/atomic-10.c
@@ -0,0 +1,139 @@
+/* { dg-do run } */
+
+extern void abort (void);
+int x1, x2, x3, x4, x5;
+volatile int y6 = 9, y2, y3, y4, y5;
+volatile unsigned char z1, z2, z3, z4, z5;
+float a1, a2, a3, a4;
+
+void
+f1 (void)
+{
+  #pragma omp atomic
+    x1++;
+  #pragma omp atomic
+    x2--;
+  #pragma omp atomic
+    ++x3;
+  #pragma omp atomic
+    --x4;
+  #pragma omp atomic
+    x5 += 1;
+  #pragma omp atomic
+    x1 -= y6;
+  #pragma omp atomic
+    x2 |= 1;
+  #pragma omp atomic
+    x3 &= 1;
+  #pragma omp atomic
+    x4 ^= 1;
+  #pragma omp atomic
+    x5 *= 3;
+  #pragma omp atomic
+    x1 /= 3;
+  #pragma omp atomic
+    x2 /= 3;
+  #pragma omp atomic
+    x3 <<= 3;
+  #pragma omp atomic
+    x4 >>= 3;
+}
+
+void
+f2 (void)
+{
+  #pragma omp atomic
+    y6++;
+  #pragma omp atomic
+    y2--;
+  #pragma omp atomic
+    ++y3;
+  #pragma omp atomic
+    --y4;
+  #pragma omp atomic
+    y5 += 1;
+  #pragma omp atomic
+    y6 -= x1;
+  #pragma omp atomic
+    y2 |= 1;
+  #pragma omp atomic
+    y3 &= 1;
+  #pragma omp atomic
+    y4 ^= 1;
+  #pragma omp atomic
+    y5 *= 3;
+  #pragma omp atomic
+    y6 /= 3;
+  #pragma omp atomic
+    y2 /= 3;
+  #pragma omp atomic
+    y3 <<= 3;
+  #pragma omp atomic
+    y4 >>= 3;
+}
+
+void
+f3 (void)
+{
+  #pragma omp atomic
+    z1++;
+  #pragma omp atomic
+    z2--;
+  #pragma omp atomic
+    ++z3;
+  #pragma omp atomic
+    --z4;
+  #pragma omp atomic
+    z5 += 1;
+  #pragma omp atomic
+    z1 |= 1;
+  #pragma omp atomic
+    z2 &= 1;
+  #pragma omp atomic
+    z3 ^= 1;
+  #pragma omp atomic
+    z4 *= 3;
+  #pragma omp atomic
+    z5 /= 3;
+  #pragma omp atomic
+    z1 /= 3;
+  #pragma omp atomic
+    z2 <<= 3;
+  #pragma omp atomic
+    z3 >>= 3;
+}
+
+void
+f4 (void)
+{
+  #pragma omp atomic
+    a1 += 8.0;
+  #pragma omp atomic
+    a2 *= 3.5;
+  #pragma omp atomic
+    a3 -= a1 + a2;
+  #pragma omp atomic
+    a4 /= 2.0;
+}
+
+int
+main (void)
+{
+  f1 ();
+  if (x1 != -2 || x2 != 0 || x3 != 8 || x4 != -1 || x5 != 3)
+    abort ();
+  f2 ();
+  if (y6 != 4 || y2 != 0 || y3 != 8 || y4 != -1 || y5 != 3)
+    abort ();
+  f3 ();
+  if (z1 != 0 || z2 != 8 || z3 != 0 || z4 != 253 || z5 != 0)
+    abort ();
+  a1 = 7;
+  a2 = 10;
+  a3 = 11;
+  a4 = 13;
+  f4 ();
+  if (a1 != 15.0 || a2 != 35.0 || a3 != -39.0 || a4 != 6.5)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/atomic-2.c b/libgomp/testsuite/libgomp.c/atomic-2.c
new file mode 100644
index 000000000..c8779483b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/atomic-2.c
@@ -0,0 +1,37 @@
+/* { dg-do run } */
+/* { dg-options "-O2 -mcx16" { target { { i?86-*-* x86_64-*-* } && lp64 } } } */
+
+#ifdef __x86_64__
+#include "cpuid.h"
+#endif
+
+double d = 1.5;
+long double ld = 3;
+extern void abort (void);
+
+void
+test (void)
+{
+#pragma omp atomic
+  d *= 1.25;
+#pragma omp atomic
+  ld /= 0.75;
+  if (d != 1.875 || ld != 4.0L)
+    abort ();
+}
+
+int
+main (void)
+{
+#ifdef __x86_64__
+  unsigned int eax, ebx, ecx, edx;
+
+  if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  if (!(ecx & bit_CMPXCHG16B))
+    return 0;
+#endif
+  test ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/atomic-3.c b/libgomp/testsuite/libgomp.c/atomic-3.c
new file mode 100644
index 000000000..5b8fdc1a7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/atomic-3.c
@@ -0,0 +1,50 @@
+/* { dg-do run } */
+/* { dg-options "-fopenmp -O0" } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+short e[64];
+int g;
+_Complex double d, f;
+int num_threads;
+
+__attribute__((noinline)) void
+foo (int x, long long y)
+{
+#pragma omp parallel num_threads (4)
+  {
+    int i;
+    #pragma omp barrier
+    for (i = 0; i < 2400; i++)
+      {
+	if (i == 0)
+	  num_threads = omp_get_num_threads ();
+	#pragma omp atomic
+	  e[0] += x;
+	#pragma omp atomic
+	  e[16] += x;
+	#pragma omp atomic
+	  g += y;
+	#pragma omp atomic
+	  __real__ d += x;
+	#pragma omp atomic
+	  __imag__ f += x;
+      }
+  }
+}
+
+int
+main (void)
+{
+  int i;
+  foo (3, 3LL);
+  if (g != 3 * 2400 * num_threads
+      || __real__ d != g || __imag__ d != 0
+      || __real__ f != 0 || __imag__ f != g)
+    abort ();
+  for (i = 0; i < 64; i++)
+    if (e[i] != ((i && i != 16) ? 0 : g))
+      abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/atomic-4.c b/libgomp/testsuite/libgomp.c/atomic-4.c
new file mode 100644
index 000000000..10f8197b0
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/atomic-4.c
@@ -0,0 +1,18 @@
+/* PR middle-end/35611 */
+/* { dg-options "-O2" } */
+
+extern void abort (void);
+
+int
+main (void)
+{
+  long double d = .0L;
+  int i;
+  #pragma omp parallel for shared (d)
+    for (i = 0; i < 1000; i++)
+      #pragma omp atomic
+	d += 1.0L;
+  if (d != 1000.0L)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/atomic-5.c b/libgomp/testsuite/libgomp.c/atomic-5.c
new file mode 100644
index 000000000..168f68dd6
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/atomic-5.c
@@ -0,0 +1,41 @@
+/* PR middle-end/36106 */
+/* { dg-options "-O2" } */
+/* { dg-options "-O2 -mcx16" { target { { i?86-*-* x86_64-*-* } && lp64 } } } */
+
+#ifdef __x86_64__
+# include "cpuid.h"
+#endif
+
+extern void abort (void);
+
+int __attribute__((noinline))
+do_test (void)
+{
+  long double d = .0L;
+  int i;
+  #pragma omp parallel for shared (d)
+    for (i = 0; i < 10; i++)
+      #pragma omp atomic
+	d += 1.0L;
+  if (d != 10.0L)
+    abort ();
+  return 0;
+}
+
+int
+main (void)
+{
+#ifdef __x86_64__
+  unsigned int eax, ebx, ecx, edx;
+
+  if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  if (!(ecx & bit_CMPXCHG16B))
+    return 0;
+#endif
+
+  do_test ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/atomic-6.c b/libgomp/testsuite/libgomp.c/atomic-6.c
new file mode 100644
index 000000000..59baf7dd3
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/atomic-6.c
@@ -0,0 +1,38 @@
+/* PR middle-end/36106 */
+/* { dg-options "-O2" } */
+/* { dg-options "-O2 -mieee" { target alpha*-*-* } } */
+/* { dg-options "-O2 -march=i586" { target { { i?86-*-* x86_64-*-* } && ilp32 } } } */
+
+#ifdef __i386__
+# include "cpuid.h"
+#endif
+
+extern void abort (void);
+
+union { unsigned long long l; double d; } u = { .l = 0x7ff0000000072301ULL };
+
+int __attribute__((noinline))
+do_test (void)
+{
+#pragma omp atomic
+  u.d += 1.0L;
+  return 0;
+}
+
+int
+main (void)
+{
+#ifdef __i386__
+  unsigned int eax, ebx, ecx, edx;
+
+  if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
+    return 0;
+
+  if (!(edx & bit_CMPXCHG8B))
+    return 0;
+#endif
+
+  do_test ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/autopar-1.c b/libgomp/testsuite/libgomp.c/autopar-1.c
new file mode 100644
index 000000000..e56549b48
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/autopar-1.c
@@ -0,0 +1,44 @@
+/* { dg-do run } */
+/* { dg-options "-ftree-parallelize-loops=4 -O2 -ffast-math" } */
+
+extern void abort (void);
+
+double d[1024], e[1024];
+int f[1024], g[1024];
+
+double __attribute__((noinline))
+foo (void)
+{
+  double s = 0.0;
+  int i;
+  for (i = 0; i < 1024; i++)
+    s += d[i] - e[i];
+  return s;
+}
+
+int __attribute__((noinline))
+bar (void)
+{
+  int s = 0, i;
+  for (i = 0; i < 1024; i++)
+    s += f[i] - g[i];
+  return s;
+}
+
+int
+main (void)
+{
+  int i;
+  for (i = 0; i < 1024; i++)
+    {
+      d[i] = i * 2;
+      e[i] = i;
+      f[i] = i * 2;
+      g[i] = i;
+    }
+  if (foo () != 1023 * 1024 / 2)
+    abort ();
+  if (bar () != 1023 * 1024 / 2)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/barrier-1.c b/libgomp/testsuite/libgomp.c/barrier-1.c
new file mode 100644
index 000000000..1f8d1f0d3
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/barrier-1.c
@@ -0,0 +1,50 @@
+/* Trivial test of barrier.  */
+
+#include <omp.h>
+#include <sys/time.h>
+#include <unistd.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+struct timeval stamps[3][3];
+
+static void function(void *dummy)
+{
+  int iam = omp_get_thread_num ();
+
+  gettimeofday (&stamps[iam][0], NULL);
+  if (iam == 0)
+    usleep (10);
+
+  GOMP_barrier ();
+
+  if (iam == 0)
+    {
+      gettimeofday (&stamps[0][1], NULL);
+      usleep (10);
+    }
+
+  GOMP_barrier ();
+  
+  gettimeofday (&stamps[iam][2], NULL);
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  GOMP_parallel_start (function, NULL, 3);
+  function (NULL);
+  GOMP_parallel_end ();
+
+  assert (!timercmp (&stamps[0][0], &stamps[0][1], >));
+  assert (!timercmp (&stamps[1][0], &stamps[0][1], >));
+  assert (!timercmp (&stamps[2][0], &stamps[0][1], >));
+
+  assert (!timercmp (&stamps[0][1], &stamps[0][2], >));
+  assert (!timercmp (&stamps[0][1], &stamps[1][2], >));
+  assert (!timercmp (&stamps[0][1], &stamps[2][2], >));
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/c.exp b/libgomp/testsuite/libgomp.c/c.exp
new file mode 100644
index 000000000..980bb526f
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/c.exp
@@ -0,0 +1,30 @@
+if [info exists lang_library_path] then {
+    unset lang_library_path
+    unset lang_link_flags
+}
+if [info exists lang_test_file] then {
+    unset lang_test_file
+}
+
+load_lib libgomp-dg.exp
+
+# If a testcase doesn't have special options, use these.
+if ![info exists DEFAULT_CFLAGS] then {
+    set DEFAULT_CFLAGS "-O2"
+}
+
+# Initialize dg.
+dg-init
+
+# Gather a list of all tests.
+set tests [lsort [find $srcdir/$subdir *.c]]
+
+set ld_library_path $always_ld_library_path
+append ld_library_path [gcc-set-multilib-library-path $GCC_UNDER_TEST]
+set_ld_library_path_env_vars
+
+# Main loop.
+dg-runtest $tests "" $DEFAULT_CFLAGS
+
+# All done.
+dg-finish
diff --git a/libgomp/testsuite/libgomp.c/collapse-1.c b/libgomp/testsuite/libgomp.c/collapse-1.c
new file mode 100644
index 000000000..84277ff55
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/collapse-1.c
@@ -0,0 +1,30 @@
+/* { dg-do run } */
+
+#include <string.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i, j, k, l = 0;
+  int a[3][3][3];
+
+  memset (a, '\0', sizeof (a));
+  #pragma omp parallel for collapse(4 - 1) schedule(static, 4)
+    for (i = 0; i < 2; i++)
+      for (j = 0; j < 2; j++)
+	for (k = 0; k < 2; k++)
+	  a[i][j][k] = i + j * 4 + k * 16;
+  #pragma omp parallel
+    {
+      #pragma omp for collapse(2) reduction(|:l) private(k)
+	for (i = 0; i < 2; i++)
+	  for (j = 0; j < 2; j++)
+	    for (k = 0; k < 2; k++)
+	      if (a[i][j][k] != i + j * 4 + k * 16)
+		l = 1;
+    }
+  if (l)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/collapse-2.c b/libgomp/testsuite/libgomp.c/collapse-2.c
new file mode 100644
index 000000000..b5c77d461
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/collapse-2.c
@@ -0,0 +1,30 @@
+/* { dg-do run } */
+
+#include <stdlib.h>
+#include <omp.h>
+
+int
+main (void)
+{
+  int i, j, k, l = 0, f = 0;
+  int m1 = 4, m2 = -5, m3 = 17;
+
+  #pragma omp parallel for num_threads (8) collapse(3) \
+		       schedule(static, 9) reduction(+:l) \
+		       firstprivate(f)
+    for (i = -2; i < m1; i++)
+      for (j = m2; j < -2; j++)
+	{
+	  for (k = 13; k < m3; k++)
+	    {
+	      if (omp_get_num_threads () == 8
+		  && ((i + 2) * 12 + (j + 5) * 4 + (k - 13)
+		      != (omp_get_thread_num () * 9
+			  + f++)))
+		l++;
+	    }
+	}
+  if (l)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/collapse-3.c b/libgomp/testsuite/libgomp.c/collapse-3.c
new file mode 100644
index 000000000..4674f83f4
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/collapse-3.c
@@ -0,0 +1,31 @@
+/* { dg-do run } */
+/* { dg-options "-O2 -std=gnu99" } */
+
+#include <string.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i2, l = 0;
+  int a[3][3][3];
+
+  memset (a, '\0', sizeof (a));
+  #pragma omp parallel for collapse(4 - 1) schedule(static, 4)
+    for (int i = 0; i < 2; i++)
+      for (int j = 0; j < 2; j++)
+	for (int k = 0; k < 2; k++)
+	  a[i][j][k] = i + j * 4 + k * 16;
+  #pragma omp parallel
+    {
+      #pragma omp for collapse(2) reduction(|:l)
+	for (i2 = 0; i2 < 2; i2++)
+	  for (int j = 0; j < 2; j++)
+	    for (int k = 0; k < 2; k++)
+	      if (a[i2][j][k] != i2 + j * 4 + k * 16)
+		l = 1;
+    }
+  if (l)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/copyin-1.c b/libgomp/testsuite/libgomp.c/copyin-1.c
new file mode 100644
index 000000000..49c546004
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/copyin-1.c
@@ -0,0 +1,34 @@
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+/* { dg-require-effective-target tls_runtime } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int thr = 32;
+#pragma omp threadprivate (thr)
+
+int
+main (void)
+{
+  int l = 0;
+
+  omp_set_dynamic (0);
+  omp_set_num_threads (6);
+
+#pragma omp parallel copyin (thr) reduction (||:l)
+  {
+    l = thr != 32;
+    thr = omp_get_thread_num () + 11;
+  }
+
+  if (l || thr != 11)
+    abort ();
+
+#pragma omp parallel reduction (||:l)
+  l = thr != omp_get_thread_num () + 11;
+
+  if (l)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/copyin-2.c b/libgomp/testsuite/libgomp.c/copyin-2.c
new file mode 100644
index 000000000..ae2451ef1
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/copyin-2.c
@@ -0,0 +1,34 @@
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+/* { dg-require-effective-target tls_runtime } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+struct { int t; char buf[64]; } thr = { 32, "" };
+#pragma omp threadprivate (thr)
+
+int
+main (void)
+{
+  int l = 0;
+
+  omp_set_dynamic (0);
+  omp_set_num_threads (6);
+
+#pragma omp parallel copyin (thr) reduction (||:l)
+  {
+    l = thr.t != 32;
+    thr.t = omp_get_thread_num () + 11;
+  }
+
+  if (l || thr.t != 11)
+    abort ();
+
+#pragma omp parallel reduction (||:l)
+  l = thr.t != omp_get_thread_num () + 11;
+
+  if (l)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/copyin-3.c b/libgomp/testsuite/libgomp.c/copyin-3.c
new file mode 100644
index 000000000..86b0d691f
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/copyin-3.c
@@ -0,0 +1,42 @@
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+/* { dg-require-effective-target tls_runtime } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int thr;
+#pragma omp threadprivate (thr)
+
+int
+test (int l)
+{
+  return l || (thr != omp_get_thread_num () * 2);
+}
+
+int
+main (void)
+{
+  int l = 0;
+
+  omp_set_dynamic (0);
+  omp_set_num_threads (6);
+
+  thr = 8;
+  /* Broadcast the value to all threads.  */
+#pragma omp parallel copyin (thr)
+  ;
+
+#pragma omp parallel reduction (||:l)
+  {
+    /* Now test if the broadcast succeeded.  */
+    l = thr != 8;
+    thr = omp_get_thread_num () * 2;
+#pragma omp barrier
+    l = test (l);
+  }
+
+  if (l)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/critical-1.c b/libgomp/testsuite/libgomp.c/critical-1.c
new file mode 100644
index 000000000..2b8503be7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/critical-1.c
@@ -0,0 +1,39 @@
+/* Trivial test of critical sections.  */
+
+/* { dg-require-effective-target sync_int_long } */
+
+#include <omp.h>
+#include <sys/time.h>
+#include <unistd.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+static volatile int test = -1;
+
+static void function(void *dummy)
+{
+  int iam = omp_get_thread_num ();
+  int old;
+
+  GOMP_critical_start ();
+
+  old = __sync_lock_test_and_set (&test, iam);
+  assert (old == -1);
+
+  usleep (10);
+  test = -1;
+
+  GOMP_critical_end ();
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  GOMP_parallel_start (function, NULL, 3);
+  function (NULL);
+  GOMP_parallel_end ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/critical-2.c b/libgomp/testsuite/libgomp.c/critical-2.c
new file mode 100644
index 000000000..530a891f6
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/critical-2.c
@@ -0,0 +1,35 @@
+// { dg-do run }
+// Test several constructs within a parallel.  At one point in development,
+// the critical directive clobbered the shared clause of the parallel.
+
+#include <omp.h>
+#include <stdlib.h>
+
+#define N       2000
+
+int main()
+{
+  int A[N];
+  int nthreads;
+  int i;
+
+#pragma omp parallel shared (A, nthreads)
+  {
+    #pragma omp master
+      nthreads = omp_get_num_threads ();
+
+    #pragma omp for
+      for (i = 0; i < N; i++)
+        A[i] = 0;
+
+    #pragma omp critical
+      for (i = 0; i < N; i++)
+        A[i] += 1;
+  }
+
+  for (i = 0; i < N; i++)
+    if (A[i] != nthreads)
+      abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/debug-1.c b/libgomp/testsuite/libgomp.c/debug-1.c
new file mode 100644
index 000000000..09bcf7f3c
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/debug-1.c
@@ -0,0 +1,162 @@
+/* PR debug/36617 */
+/* { dg-do run } */
+/* { dg-options "-g -fopenmp -O0" } */
+
+int
+f1 (void)
+{
+  int v1i, v1j, v1k, v1l = 0;
+  v1i = 6;
+  v1j = 8;
+  #pragma omp parallel private (v1k) firstprivate (v1j) shared (v1i) reduction (+:v1l)
+  {
+    v1k = v1i + v1j;
+    {
+      int v1m = 1;
+      v1l = v1m;
+    }
+  }
+  return v1l;
+}
+
+int v2k = 9;
+
+int
+f2 (void)
+{
+  int v2i = 6, v2j = 7;
+  #pragma omp single private (v2i) firstprivate (v2k)
+  {
+    int v2l = v2j + v2k;
+    v2i = 8;
+    v2k = 10;
+    v2j = v2l + v2i;
+  }
+  return v2i + v2j;
+}
+
+int
+f3 (void)
+{
+  int v3i = 6, v3j = 7, v3k = 9;
+  #pragma omp parallel
+  {
+    #pragma omp master
+      v3i++;
+    #pragma omp single private (v3i) firstprivate (v3k)
+    {
+      int v3l = v3j + v3k;
+      v3i = 8;
+      v3k = 10;
+      v3j = v3l + v3i;
+    }
+    #pragma omp atomic
+      v3k++;
+  }
+  return v3i + v3j;
+}
+
+int v4k = 9, v4l = 0;
+
+int
+f4 (void)
+{
+  int v4i = 6, v4j = 7, v4n = 0;
+  #pragma omp sections private (v4i) firstprivate (v4k) reduction (+:v4l)
+  {
+    #pragma omp section
+    {
+      int v4m = v4j + v4k;
+      v4i = 8;
+      v4k = 10;
+      v4l++;
+      v4n = v4m + v4i;
+    }
+    #pragma omp section
+    {
+      int v4o = v4j + v4k;
+      v4i = 10;
+      v4k = 11;
+      v4l++;
+    }
+  }
+  return v4i + v4j + v4l + v4n;
+}
+
+int
+f5 (void)
+{
+  int v5i = 6, v5j = 7, v5k = 9, v5l = 0, v5n = 0, v5p = 0;
+  #pragma omp parallel
+  {
+    #pragma omp master
+      v5p++;
+    #pragma omp sections private (v5i) firstprivate (v5k) reduction (+:v5l)
+    {
+      #pragma omp section
+      {
+	int v5m = v5j + v5k;
+	v5i = 8;
+	v5k = 10;
+	v5l++;
+	v5n = v5m + v5i;
+      }
+      #pragma omp section
+      {
+	int v5o = v5j + v5k;
+	v5i = 10;
+	v5k = 11;
+	v5l++;
+      }
+    }
+  }
+  return v5i + v5j + v5l + v5n + v5p;
+}
+
+int v6k = 9, v6l = 0;
+
+int
+f6 (void)
+{
+  int v6i = 6, v6j = 7, v6n = 0;
+  #pragma omp for private (v6i) firstprivate (v6k) reduction (+:v6l)
+  for (v6n = 0; v6n < 3; v6n++)
+    {
+      int v6m = v6j + v6k;
+      v6i = 8;
+      v6l++;
+    }
+  return v6i + v6j + v6k + v6l + v6n;
+}
+
+int
+f7 (void)
+{
+  int v7i = 6, v7j = 7, v7k = 9, v7l = 0, v7n = 0, v7o = 1;
+  #pragma omp parallel
+  {
+    #pragma omp master
+      v7o++;
+    #pragma omp for private (v7i) firstprivate (v7k) reduction (+:v7l)
+    for (v7n = 0; v7n < 3; v7n++)
+      {
+	int v7m = v7j + v7k;
+	v7i = 8;
+	v7l++;
+      }
+  }
+  return v7i + v7j + v7k + v7l + v7n;
+}
+
+int
+main (void)
+{
+  f1 ();
+  f2 ();
+  f3 ();
+  f4 ();
+  f5 ();
+  f6 ();
+  f7 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/icv-1.c b/libgomp/testsuite/libgomp.c/icv-1.c
new file mode 100644
index 000000000..99708f823
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/icv-1.c
@@ -0,0 +1,33 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int err = 0;
+
+  omp_set_num_threads (4);
+  if (omp_get_max_threads () != 4)
+    abort ();
+  #pragma omp parallel reduction(|: err) num_threads(1)
+  {
+    if (omp_get_max_threads () != 4)
+      err |= 1;
+    omp_set_num_threads (6);
+    #pragma omp task if(0) shared(err)
+    {
+      if (omp_get_max_threads () != 6)
+	err |= 2;
+      omp_set_num_threads (5);
+      if (omp_get_max_threads () != 5)
+	err |= 4;
+    }
+    if (omp_get_max_threads () != 6)
+      err |= 8;
+  }
+  if (err)
+    abort ();
+  if (omp_get_max_threads () != 4)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/icv-2.c b/libgomp/testsuite/libgomp.c/icv-2.c
new file mode 100644
index 000000000..326f8eb40
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/icv-2.c
@@ -0,0 +1,46 @@
+/* { dg-do run { target *-*-linux* } } */
+
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE 1
+#endif
+#include <pthread.h>
+#include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+pthread_barrier_t bar;
+
+void *tf (void *p)
+{
+  int l;
+  if (p)
+    omp_set_num_threads (3);
+  pthread_barrier_wait (&bar);
+  if (!p)
+    omp_set_num_threads (6);
+  pthread_barrier_wait (&bar);
+  omp_set_dynamic (0);
+  if (omp_get_max_threads () != (p ? 3 : 6))
+    abort ();
+  l = 0;
+  #pragma omp parallel num_threads (6) reduction (|:l)
+    {
+      l |= omp_get_max_threads () != (p ? 3 : 6);
+      omp_set_num_threads ((p ? 3 : 6) + omp_get_thread_num ());
+      l |= omp_get_max_threads () != ((p ? 3 : 6) + omp_get_thread_num ());
+    }
+  if (l)
+    abort ();
+  return NULL;
+}
+
+int
+main (void)
+{
+  pthread_t th;
+  pthread_barrier_init (&bar, NULL, 2);
+  pthread_create (&th, NULL, tf, NULL);
+  tf ("");
+  pthread_join (th, NULL);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/lib-1.c b/libgomp/testsuite/libgomp.c/lib-1.c
new file mode 100644
index 000000000..4839cf936
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/lib-1.c
@@ -0,0 +1,99 @@
+#include <stdlib.h>
+#include <omp.h>
+
+int
+main (void)
+{
+  double d, e;
+  int l;
+  omp_lock_t lck;
+  omp_nest_lock_t nlck;
+
+  d = omp_get_wtime ();
+
+  omp_init_lock (&lck);
+  omp_set_lock (&lck);
+  if (omp_test_lock (&lck))
+    abort ();
+  omp_unset_lock (&lck);
+  if (! omp_test_lock (&lck))
+    abort ();
+  if (omp_test_lock (&lck))
+    abort ();
+  omp_unset_lock (&lck);
+  omp_destroy_lock (&lck);
+
+  omp_init_nest_lock (&nlck);
+  if (omp_test_nest_lock (&nlck) != 1)
+    abort ();
+  omp_set_nest_lock (&nlck);
+  if (omp_test_nest_lock (&nlck) != 3)
+    abort ();
+  omp_unset_nest_lock (&nlck);
+  omp_unset_nest_lock (&nlck);
+  if (omp_test_nest_lock (&nlck) != 2)
+    abort ();
+  omp_unset_nest_lock (&nlck);
+  omp_unset_nest_lock (&nlck);
+  omp_destroy_nest_lock (&nlck);
+
+  omp_set_dynamic (1);
+  if (! omp_get_dynamic ())
+    abort ();
+  omp_set_dynamic (0);
+  if (omp_get_dynamic ())
+    abort ();
+
+  omp_set_nested (1);
+  if (! omp_get_nested ())
+    abort ();
+  omp_set_nested (0);
+  if (omp_get_nested ())
+    abort ();
+
+  omp_set_num_threads (5);
+  if (omp_get_num_threads () != 1)
+    abort ();
+  if (omp_get_max_threads () != 5)
+    abort ();
+  if (omp_get_thread_num () != 0)
+    abort ();
+  omp_set_num_threads (3);
+  if (omp_get_num_threads () != 1)
+    abort ();
+  if (omp_get_max_threads () != 3)
+    abort ();
+  if (omp_get_thread_num () != 0)
+    abort ();
+  l = 0;
+#pragma omp parallel reduction (|:l)
+  {
+    l = omp_get_num_threads () != 3;
+    l |= omp_get_thread_num () < 0;
+    l |= omp_get_thread_num () >= 3;
+#pragma omp master
+    l |= omp_get_thread_num () != 0;
+  }
+  if (l)
+    abort ();
+
+  if (omp_get_num_procs () <= 0)
+    abort ();
+  if (omp_in_parallel ())
+    abort ();
+#pragma omp parallel reduction (|:l)
+  l = ! omp_in_parallel ();
+#pragma omp parallel reduction (|:l) if (1)
+  l = ! omp_in_parallel ();
+
+  e = omp_get_wtime ();
+  if (d > e)
+    abort ();
+  d = omp_get_wtick ();
+  /* Negative precision is definitely wrong,
+     bigger than 1s clock resolution is also strange.  */
+  if (d <= 0 || d > 1)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/lib-2.c b/libgomp/testsuite/libgomp.c/lib-2.c
new file mode 100644
index 000000000..3a3b3f655
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/lib-2.c
@@ -0,0 +1,25 @@
+#include <stdlib.h>
+#include <omp.h>
+
+int
+main (void)
+{
+  omp_sched_t kind;
+  int modifier;
+
+  omp_set_schedule (omp_sched_static, 32);
+  omp_get_schedule (&kind, &modifier);
+  if (kind != omp_sched_static || modifier != 32)
+    abort ();
+  omp_set_schedule (omp_sched_guided, 4);
+  omp_get_schedule (&kind, &modifier);
+  if (kind != omp_sched_guided || modifier != 4)
+    abort ();
+  if (omp_get_thread_limit () < 0)
+    abort ();
+  omp_set_max_active_levels (6);
+  if (omp_get_max_active_levels () != 6)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/lock-1.c b/libgomp/testsuite/libgomp.c/lock-1.c
new file mode 100644
index 000000000..e09645dbc
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/lock-1.c
@@ -0,0 +1,31 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int l = 0;
+  omp_nest_lock_t lock;
+  omp_init_nest_lock (&lock);
+  if (omp_test_nest_lock (&lock) != 1)
+    abort ();
+  if (omp_test_nest_lock (&lock) != 2)
+    abort ();
+#pragma omp parallel if (0) reduction (+:l)
+  {
+    /* In OpenMP 2.5 this was supposed to return 3,
+       but in OpenMP 3.0 the parallel region has a different
+       task and omp_*_lock_t are owned by tasks, not by threads.  */
+    if (omp_test_nest_lock (&lock) != 0)
+      l++;
+  }
+  if (l)
+    abort ();
+  if (omp_test_nest_lock (&lock) != 3)
+    abort ();
+  omp_unset_nest_lock (&lock);
+  omp_unset_nest_lock (&lock);
+  omp_unset_nest_lock (&lock);
+  omp_destroy_nest_lock (&lock);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/lock-2.c b/libgomp/testsuite/libgomp.c/lock-2.c
new file mode 100644
index 000000000..9009b12fe
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/lock-2.c
@@ -0,0 +1,32 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int l = 0;
+  omp_nest_lock_t lock;
+  omp_init_nest_lock (&lock);
+#pragma omp parallel reduction (+:l) num_threads (1)
+  {
+    if (omp_test_nest_lock (&lock) != 1)
+      l++;
+    if (omp_test_nest_lock (&lock) != 2)
+      l++;
+  #pragma omp task if (0) shared (lock, l)
+    {
+      if (omp_test_nest_lock (&lock) != 0)
+	l++;
+    }
+  #pragma omp taskwait
+    if (omp_test_nest_lock (&lock) != 3)
+      l++;
+    omp_unset_nest_lock (&lock);
+    omp_unset_nest_lock (&lock);
+    omp_unset_nest_lock (&lock);
+  }
+  if (l)
+    abort ();
+  omp_destroy_nest_lock (&lock);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/lock-3.c b/libgomp/testsuite/libgomp.c/lock-3.c
new file mode 100644
index 000000000..1fc83726d
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/lock-3.c
@@ -0,0 +1,60 @@
+/* { dg-do run { target *-*-linux* } } */
+
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE 1
+#endif
+#include <pthread.h>
+#include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+pthread_barrier_t bar;
+omp_nest_lock_t lock;
+
+void *tf (void *p)
+{
+  int l;
+  if (p)
+    {
+      if (omp_test_nest_lock (&lock) != 1)
+	abort ();
+      if (omp_test_nest_lock (&lock) != 2)
+	abort ();
+    }
+  pthread_barrier_wait (&bar);
+  if (!p && omp_test_nest_lock (&lock) != 0)
+    abort ();
+  pthread_barrier_wait (&bar);
+  if (p)
+    {
+      if (omp_test_nest_lock (&lock) != 3)
+	abort ();
+      omp_unset_nest_lock (&lock);
+      omp_unset_nest_lock (&lock);
+      omp_unset_nest_lock (&lock);
+    }
+  pthread_barrier_wait (&bar);
+  if (!p)
+    {
+      if (omp_test_nest_lock (&lock) != 1)
+	abort ();
+      if (omp_test_nest_lock (&lock) != 2)
+	abort ();
+      omp_unset_nest_lock (&lock);
+      omp_unset_nest_lock (&lock);
+    }
+  return NULL;
+}
+
+int
+main (void)
+{
+  pthread_t th;
+  omp_init_nest_lock (&lock);
+  pthread_barrier_init (&bar, NULL, 2);
+  pthread_create (&th, NULL, tf, NULL);
+  tf ("");
+  pthread_join (th, NULL);
+  omp_destroy_nest_lock (&lock);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-1.c b/libgomp/testsuite/libgomp.c/loop-1.c
new file mode 100644
index 000000000..11348dcfe
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-1.c
@@ -0,0 +1,140 @@
+/* Test that all loop iterations are touched.  This doesn't verify 
+   scheduling order, merely coverage.  */
+
+/* { dg-require-effective-target sync_int_long } */
+
+#include <omp.h>
+#include <string.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+#define N 10000
+static int S, E, INCR, CHUNK, NTHR;
+static int data[N];
+
+static void clean_data (void)
+{
+  memset (data, -1, sizeof (data));
+}
+
+static void test_data (void)
+{
+  int i, j;
+
+  for (i = 0; i < S; ++i)
+    assert (data[i] == -1);
+
+  for (j = 0; i < E; ++i, j = (j + 1) % INCR)
+    if (j == 0)
+      assert (data[i] != -1);
+    else
+      assert (data[i] == -1);
+
+  for (; i < N; ++i)
+    assert (data[i] == -1);
+}
+
+static void set_data (long i, int val)
+{
+  int old;
+  assert (i >= 0 && i < N);
+  old = __sync_lock_test_and_set (data+i, val);
+  assert (old == -1);
+}
+  
+
+#define TMPL_1(sched)						\
+static void f_##sched##_1 (void *dummy)				\
+{								\
+  int iam = omp_get_thread_num ();				\
+  long s0, e0, i;						\
+  if (GOMP_loop_##sched##_start (S, E, INCR, CHUNK, &s0, &e0))	\
+    do								\
+      {								\
+	for (i = s0; i < e0; i += INCR)				\
+	  set_data (i, iam);					\
+      }								\
+    while (GOMP_loop_##sched##_next (&s0, &e0));		\
+  GOMP_loop_end ();						\
+}								\
+static void t_##sched##_1 (void)				\
+{								\
+  clean_data ();						\
+  GOMP_parallel_start (f_##sched##_1, NULL, NTHR);		\
+  f_##sched##_1 (NULL);						\
+  GOMP_parallel_end ();						\
+  test_data ();							\
+}
+
+TMPL_1(static)
+TMPL_1(dynamic)
+TMPL_1(guided)
+
+#define TMPL_2(sched)					\
+static void f_##sched##_2 (void *dummy)			\
+{							\
+  int iam = omp_get_thread_num ();			\
+  long s0, e0, i;					\
+  while (GOMP_loop_##sched##_next (&s0, &e0))		\
+    {							\
+      for (i = s0; i < e0; i += INCR)			\
+	set_data (i, iam);				\
+    }							\
+  GOMP_loop_end_nowait ();				\
+}							\
+static void t_##sched##_2 (void)			\
+{							\
+  clean_data ();					\
+  GOMP_parallel_loop_##sched##_start			\
+    (f_##sched##_2, NULL, NTHR, S, E, INCR, CHUNK);	\
+  f_##sched##_2 (NULL);					\
+  GOMP_parallel_end ();					\
+  test_data ();						\
+}
+
+TMPL_2(static)
+TMPL_2(dynamic)
+TMPL_2(guided)
+
+static void test (void)
+{
+  t_static_1 ();
+  t_dynamic_1 ();
+  t_guided_1 ();
+  t_static_2 ();
+  t_dynamic_2 ();
+  t_guided_2 ();
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  NTHR = 4;
+
+  S = 0, E = N, INCR = 1, CHUNK = 4;
+  test ();
+
+  S = 0, E = N, INCR = 2, CHUNK = 4;
+  test ();
+
+  S = 1, E = N-1, INCR = 1, CHUNK = 5;
+  test ();
+
+  S = 1, E = N-1, INCR = 2, CHUNK = 5;
+  test ();
+
+  S = 2, E = 4, INCR = 1, CHUNK = 1;
+  test ();
+
+  S = 0, E = N, INCR = 1, CHUNK = 0;
+  t_static_1 ();
+  t_static_2 ();
+
+  S = 1, E = N-1, INCR = 1, CHUNK = 0;
+  t_static_1 ();
+  t_static_2 ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-10.c b/libgomp/testsuite/libgomp.c/loop-10.c
new file mode 100644
index 000000000..1b42c4bf1
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-10.c
@@ -0,0 +1,30 @@
+extern void abort (void);
+
+int i = 8;
+
+int main (void)
+{
+  int j = 7, k = 0;
+  #pragma omp for
+  for (i = 0; i < 10; i++)
+    ;
+  #pragma omp for
+  for (j = 0; j < 10; j++)
+    ;
+  /* OpenMP 3.0 newly guarantees that the original list items can't
+     be shared with the privatized omp for iterators, even when
+     the original list items are already private.  */
+  if (i != 8 || j != 7)
+    abort ();
+  #pragma omp parallel private (i) reduction (+:k)
+  {
+    i = 6;
+    #pragma omp for
+    for (i = 0; i < 10; i++)
+      ;
+    k = (i != 6);
+  }
+  if (k)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-11.c b/libgomp/testsuite/libgomp.c/loop-11.c
new file mode 100644
index 000000000..c5ac3c434
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-11.c
@@ -0,0 +1,276 @@
+#include <omp.h>
+#include <stdlib.h>
+#include <string.h>
+
+int
+test1 (void)
+{
+  short int buf[64], *p;
+  int i;
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[10]; &buf[54] > p; p++)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[3]; &buf[63] >= p; p += 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[16]; &buf[51] > p; p = 4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[16]; &buf[40] >= p; p = p + 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[53]; &buf[9] < p; --p)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[63]; &buf[3] <= p; p -= 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[48]; &buf[15] < p; p = -4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[40]; &buf[16] <= p; p = p - 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  return 0;
+}
+
+int
+test2 (void)
+{
+  int buf[64], *p;
+  int i;
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[10]; &buf[54] > p; p++)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[3]; &buf[63] >= p; p += 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[16]; &buf[51] > p; p = 4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[16]; &buf[40] >= p; p = p + 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[53]; &buf[9] < p; --p)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[63]; &buf[3] <= p; p -= 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[48]; &buf[15] < p; p = -4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[40]; &buf[16] <= p; p = p - 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  return 0;
+}
+
+int
+test3 (void)
+{
+  int buf[64], *p;
+  int i;
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[10]; &buf[54] > p; p++)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[3]; &buf[63] >= p; p += 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[16]; &buf[51] > p; p = 4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[16]; &buf[40] >= p; p = p + 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[53]; &buf[9] < p; --p)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[63]; &buf[3] <= p; p -= 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[48]; &buf[15] < p; p = -4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[40]; &buf[16] <= p; p = p - 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  return 0;
+}
+
+int
+test4 (void)
+{
+  int buf[64], *p;
+  int i;
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[10]; &buf[54] > p; p++)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[3]; &buf[63] >= p; p += 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[16]; &buf[51] > p; p = 4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[16]; &buf[40] >= p; p = p + 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[53]; &buf[9] < p; --p)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[63]; &buf[3] <= p; p -= 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[48]; &buf[15] < p; p = -4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[40]; &buf[16] <= p; p = p - 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  return 0;
+}
+
+int
+main (void)
+{
+  test1 ();
+  test2 ();
+  test3 ();
+  omp_set_schedule (omp_sched_static, 0);
+  test4 ();
+  omp_set_schedule (omp_sched_static, 3);
+  test4 ();
+  omp_set_schedule (omp_sched_dynamic, 5);
+  test4 ();
+  omp_set_schedule (omp_sched_guided, 2);
+  test4 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-12.c b/libgomp/testsuite/libgomp.c/loop-12.c
new file mode 100644
index 000000000..395da363e
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-12.c
@@ -0,0 +1,387 @@
+/* { dg-do run } */
+
+#include <omp.h>
+
+extern void abort (void);
+
+#define LLONG_MAX __LONG_LONG_MAX__
+#define ULLONG_MAX (LLONG_MAX * 2ULL + 1)
+#define INT_MAX __INT_MAX__
+
+int arr[6 * 5];
+
+void
+set (int loopidx, int idx)
+{
+#pragma omp atomic
+  arr[loopidx * 5 + idx]++;
+}
+
+#define check(var, val, loopidx, idx) \
+  if (var == (val)) set (loopidx, idx); else
+#define test(loopidx, count) \
+  for (idx = 0; idx < 5; idx++) \
+    if (arr[loopidx * 5 + idx] != idx < count) \
+      abort (); \
+    else \
+      arr[loopidx * 5 + idx] = 0
+
+int
+test1 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(dynamic,1) nowait
+    for (i = LLONG_MAX - 30001; LLONG_MAX - 10001 >= i; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (i = -LLONG_MAX + 30000; -LLONG_MAX + 10000 <= i; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (j = 20; LLONG_MAX - 70 >= j; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (j = ULLONG_MAX - 3; LLONG_MAX + 70ULL <= j; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (j = LLONG_MAX - 20000ULL; LLONG_MAX + 10000ULL >= j; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (i = -3LL * INT_MAX - 20000LL; INT_MAX + 10000LL >= i; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+test2 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(guided,1) nowait
+    for (i = LLONG_MAX - 30001; LLONG_MAX - 10001 >= i; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (i = -LLONG_MAX + 30000; -LLONG_MAX + 10000 <= i; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (j = 20; LLONG_MAX - 70 >= j; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (j = ULLONG_MAX - 3; LLONG_MAX + 70ULL <= j; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (j = LLONG_MAX - 20000ULL; LLONG_MAX + 10000ULL >= j; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (i = -3LL * INT_MAX - 20000LL; INT_MAX + 10000LL >= i; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+test3 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(static) nowait
+    for (i = LLONG_MAX - 30001; LLONG_MAX - 10001 >= i; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (i = -LLONG_MAX + 30000; -LLONG_MAX + 10000 <= i; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (j = 20; LLONG_MAX - 70 >= j; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (j = ULLONG_MAX - 3; LLONG_MAX + 70ULL <= j; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (j = LLONG_MAX - 20000ULL; LLONG_MAX + 10000ULL >= j; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (i = -3LL * INT_MAX - 20000LL; INT_MAX + 10000LL >= i; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+test4 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(static,1) nowait
+    for (i = LLONG_MAX - 30001; LLONG_MAX - 10001 >= i; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (i = -LLONG_MAX + 30000; -LLONG_MAX + 10000 <= i; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (j = 20; LLONG_MAX - 70 >= j; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (j = ULLONG_MAX - 3; LLONG_MAX + 70ULL <= j; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (j = LLONG_MAX - 20000ULL; LLONG_MAX + 10000ULL >= j; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (i = -3LL * INT_MAX - 20000LL; INT_MAX + 10000LL >= i; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+test5 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(runtime) nowait
+    for (i = LLONG_MAX - 30001; LLONG_MAX - 10001 >= i; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (i = -LLONG_MAX + 30000; -LLONG_MAX + 10000 <= i; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (j = 20; LLONG_MAX - 70 >= j; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (j = ULLONG_MAX - 3; LLONG_MAX + 70ULL <= j; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (j = LLONG_MAX - 20000ULL; LLONG_MAX + 10000ULL >= j; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (i = -3LL * INT_MAX - 20000LL; INT_MAX + 10000LL >= i; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+main (void)
+{
+  if (2 * sizeof (int) != sizeof (long long))
+    return 0;
+  test1 ();
+  test2 ();
+  test3 ();
+  test4 ();
+  omp_set_schedule (omp_sched_static, 0);
+  test5 ();
+  omp_set_schedule (omp_sched_static, 3);
+  test5 ();
+  omp_set_schedule (omp_sched_dynamic, 5);
+  test5 ();
+  omp_set_schedule (omp_sched_guided, 2);
+  test5 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-2.c b/libgomp/testsuite/libgomp.c/loop-2.c
new file mode 100644
index 000000000..4bae023c7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-2.c
@@ -0,0 +1,114 @@
+/* Validate static scheduling iteration dispatch.  We only test with
+   even thread distributions here; there are multiple valid solutions
+   for uneven thread distributions.  */
+
+/* { dg-require-effective-target sync_int_long } */
+
+#include <omp.h>
+#include <string.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+#define N 360
+static int data[N][2];
+static int INCR, NTHR, CHUNK;
+
+static void clean_data (void)
+{
+  memset (data, -1, sizeof (data));
+}
+
+static void test_data (void)
+{
+  int n, i, c, thr, iter, chunk;
+
+  chunk = CHUNK;
+  if (chunk == 0)
+    chunk = N / INCR / NTHR;
+
+  thr = iter = c = i = 0;
+
+  for (n = 0; n < N; ++n)
+    {
+      if (i == 0)
+	{
+	  assert (data[n][0] == thr);
+	  assert (data[n][1] == iter);
+	}
+      else
+	{
+	  assert (data[n][0] == -1);
+	  assert (data[n][1] == -1);
+	}
+
+      if (++i == INCR)
+	{
+	  i = 0;
+	  if (++c == chunk)
+	    {
+	      c = 0;
+	      if (++thr == NTHR)
+		{
+		  thr = 0;
+		  ++iter;
+		}
+	    }
+	}
+    }
+}
+
+static void set_data (long i, int thr, int iter)
+{
+  int old;
+  assert (i >= 0 && i < N);
+  old = __sync_lock_test_and_set (&data[i][0], thr);
+  assert (old == -1);
+  old = __sync_lock_test_and_set (&data[i][1], iter);
+  assert (old == -1);
+}
+  
+static void f_static_1 (void *dummy)
+{
+  int iam = omp_get_thread_num ();
+  long s0, e0, i, count = 0;
+  if (GOMP_loop_static_start (0, N, INCR, CHUNK, &s0, &e0))
+    do
+      {
+	for (i = s0; i < e0; i += INCR)
+	  set_data (i, iam, count);
+	++count;	
+      }
+    while (GOMP_loop_static_next (&s0, &e0));
+  GOMP_loop_end ();
+}
+
+static void test (void)
+{
+  clean_data ();
+  GOMP_parallel_start (f_static_1, NULL, NTHR);
+  f_static_1 (NULL);
+  GOMP_parallel_end ();
+  test_data ();
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  NTHR = 5;
+
+  INCR = 1, CHUNK = 0;	/* chunk = 360 / 5 = 72 */
+  test ();
+
+  INCR = 4, CHUNK = 0;	/* chunk = 360 / 4 / 5 = 18 */
+  test ();
+
+  INCR = 1, CHUNK = 4;	/* 1 * 4 * 5 = 20 -> 360 / 20 = 18 iterations.  */
+  test ();
+
+  INCR = 3, CHUNK = 4;	/* 3 * 4 * 5 = 60 -> 360 / 60 = 6 iterations.  */
+  test ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-3.c b/libgomp/testsuite/libgomp.c/loop-3.c
new file mode 100644
index 000000000..f0f9b4705
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-3.c
@@ -0,0 +1,25 @@
+/* { dg-do run } */
+
+extern void abort (void);
+
+volatile int count;
+static int test(void)
+{
+  return ++count > 0;
+}
+
+int i;
+
+int main()
+{
+  #pragma omp for lastprivate (i)
+  for (i = 0; i < 10; ++i)
+    {
+      if (test())
+	continue;
+      abort ();
+    }
+  if (i != count)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-4.c b/libgomp/testsuite/libgomp.c/loop-4.c
new file mode 100644
index 000000000..bc57c043a
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-4.c
@@ -0,0 +1,28 @@
+/* { dg-do run } */
+
+extern void abort (void);
+
+int
+main (void)
+{
+  int e = 0;
+#pragma omp parallel num_threads (4) reduction(+:e)
+  {
+    long i;
+    #pragma omp for schedule(dynamic,1)
+    for (i = __LONG_MAX__ - 30001; i <= __LONG_MAX__ - 10001; i += 10000)
+      if (i != __LONG_MAX__ - 30001
+	  && i != __LONG_MAX__ - 20001
+	  && i != __LONG_MAX__ - 10001)
+	e = 1;
+    #pragma omp for schedule(dynamic,1)
+    for (i = -__LONG_MAX__ + 30000; i >= -__LONG_MAX__ + 10000; i -= 10000)
+      if (i != -__LONG_MAX__ + 30000
+	  && i != -__LONG_MAX__ + 20000
+	  && i != -__LONG_MAX__ + 10000)
+	e = 1;
+  }
+  if (e)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-5.c b/libgomp/testsuite/libgomp.c/loop-5.c
new file mode 100644
index 000000000..3a5c7cf45
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-5.c
@@ -0,0 +1,276 @@
+#include <omp.h>
+#include <stdlib.h>
+#include <string.h>
+
+int
+test1 (void)
+{
+  short int buf[64], *p;
+  int i;
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[10]; p < &buf[54]; p++)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[3]; p <= &buf[63]; p += 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[16]; p < &buf[51]; p = 4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[16]; p <= &buf[40]; p = p + 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[53]; p > &buf[9]; --p)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[63]; p >= &buf[3]; p -= 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[48]; p > &buf[15]; p = -4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for
+  for (p = &buf[40]; p >= &buf[16]; p = p - 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  return 0;
+}
+
+int
+test2 (void)
+{
+  int buf[64], *p;
+  int i;
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[10]; p < &buf[54]; p++)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[3]; p <= &buf[63]; p += 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[16]; p < &buf[51]; p = 4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[16]; p <= &buf[40]; p = p + 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[53]; p > &buf[9]; --p)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[63]; p >= &buf[3]; p -= 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[48]; p > &buf[15]; p = -4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (static, 3)
+  for (p = &buf[40]; p >= &buf[16]; p = p - 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  return 0;
+}
+
+int
+test3 (void)
+{
+  int buf[64], *p;
+  int i;
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[10]; p < &buf[54]; p++)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[3]; p <= &buf[63]; p += 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[16]; p < &buf[51]; p = 4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[16]; p <= &buf[40]; p = p + 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[53]; p > &buf[9]; --p)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[63]; p >= &buf[3]; p -= 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[48]; p > &buf[15]; p = -4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (dynamic, 3)
+  for (p = &buf[40]; p >= &buf[16]; p = p - 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  return 0;
+}
+
+int
+test4 (void)
+{
+  int buf[64], *p;
+  int i;
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[10]; p < &buf[54]; p++)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[3]; p <= &buf[63]; p += 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[16]; p < &buf[51]; p = 4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[16]; p <= &buf[40]; p = p + 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[53]; p > &buf[9]; --p)
+    *p = 5;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 5 * (i >= 10 && i < 54))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[63]; p >= &buf[3]; p -= 2)
+    p[-2] = 6;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 6 * ((i & 1) && i <= 61))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[48]; p > &buf[15]; p = -4 + p)
+    p[2] = 7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != 7 * ((i & 3) == 2 && i >= 18 && i < 53))
+      abort ();
+  memset (buf, '\0', sizeof (buf));
+#pragma omp parallel for schedule (runtime)
+  for (p = &buf[40]; p >= &buf[16]; p = p - 4ULL)
+    p[2] = -7;
+  for (i = 0; i < 64; i++)
+    if (buf[i] != -7 * ((i & 3) == 2 && i >= 18 && i <= 42))
+      abort ();
+  return 0;
+}
+
+int
+main (void)
+{
+  test1 ();
+  test2 ();
+  test3 ();
+  omp_set_schedule (omp_sched_static, 0);
+  test4 ();
+  omp_set_schedule (omp_sched_static, 3);
+  test4 ();
+  omp_set_schedule (omp_sched_dynamic, 5);
+  test4 ();
+  omp_set_schedule (omp_sched_guided, 2);
+  test4 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-6.c b/libgomp/testsuite/libgomp.c/loop-6.c
new file mode 100644
index 000000000..9029e181b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-6.c
@@ -0,0 +1,387 @@
+/* { dg-do run } */
+
+#include <omp.h>
+
+extern void abort (void);
+
+#define LLONG_MAX __LONG_LONG_MAX__
+#define ULLONG_MAX (LLONG_MAX * 2ULL + 1)
+#define INT_MAX __INT_MAX__
+
+int arr[6 * 5];
+
+void
+set (int loopidx, int idx)
+{
+#pragma omp atomic
+  arr[loopidx * 5 + idx]++;
+}
+
+#define check(var, val, loopidx, idx) \
+  if (var == (val)) set (loopidx, idx); else
+#define test(loopidx, count) \
+  for (idx = 0; idx < 5; idx++) \
+    if (arr[loopidx * 5 + idx] != idx < count) \
+      abort (); \
+    else \
+      arr[loopidx * 5 + idx] = 0
+
+int
+test1 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(dynamic,1) nowait
+    for (i = LLONG_MAX - 30001; i <= LLONG_MAX - 10001; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (i = -LLONG_MAX + 30000; i >= -LLONG_MAX + 10000; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (j = 20; j <= LLONG_MAX - 70; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (j = ULLONG_MAX - 3; j >= LLONG_MAX + 70ULL; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (j = LLONG_MAX - 20000ULL; j <= LLONG_MAX + 10000ULL; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(dynamic,1) nowait
+    for (i = -3LL * INT_MAX - 20000LL; i <= INT_MAX + 10000LL; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+test2 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(guided,1) nowait
+    for (i = LLONG_MAX - 30001; i <= LLONG_MAX - 10001; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (i = -LLONG_MAX + 30000; i >= -LLONG_MAX + 10000; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (j = 20; j <= LLONG_MAX - 70; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (j = ULLONG_MAX - 3; j >= LLONG_MAX + 70ULL; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (j = LLONG_MAX - 20000ULL; j <= LLONG_MAX + 10000ULL; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(guided,1) nowait
+    for (i = -3LL * INT_MAX - 20000LL; i <= INT_MAX + 10000LL; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+test3 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(static) nowait
+    for (i = LLONG_MAX - 30001; i <= LLONG_MAX - 10001; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (i = -LLONG_MAX + 30000; i >= -LLONG_MAX + 10000; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (j = 20; j <= LLONG_MAX - 70; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (j = ULLONG_MAX - 3; j >= LLONG_MAX + 70ULL; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (j = LLONG_MAX - 20000ULL; j <= LLONG_MAX + 10000ULL; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(static) nowait
+    for (i = -3LL * INT_MAX - 20000LL; i <= INT_MAX + 10000LL; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+test4 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(static,1) nowait
+    for (i = LLONG_MAX - 30001; i <= LLONG_MAX - 10001; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (i = -LLONG_MAX + 30000; i >= -LLONG_MAX + 10000; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (j = 20; j <= LLONG_MAX - 70; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (j = ULLONG_MAX - 3; j >= LLONG_MAX + 70ULL; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (j = LLONG_MAX - 20000ULL; j <= LLONG_MAX + 10000ULL; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(static,1) nowait
+    for (i = -3LL * INT_MAX - 20000LL; i <= INT_MAX + 10000LL; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+test5 (void)
+{
+  int e = 0, idx;
+
+#pragma omp parallel reduction(+:e)
+  {
+    long long i;
+    unsigned long long j;
+    #pragma omp for schedule(runtime) nowait
+    for (i = LLONG_MAX - 30001; i <= LLONG_MAX - 10001; i += 10000)
+      {
+	check (i, LLONG_MAX - 30001, 0, 0)
+	check (i, LLONG_MAX - 20001, 0, 1)
+	check (i, LLONG_MAX - 10001, 0, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (i = -LLONG_MAX + 30000; i >= -LLONG_MAX + 10000; i -= 10000)
+      {
+	check (i, -LLONG_MAX + 30000, 1, 0)
+	check (i, -LLONG_MAX + 20000, 1, 1)
+	check (i, -LLONG_MAX + 10000, 1, 2)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (j = 20; j <= LLONG_MAX - 70; j += LLONG_MAX + 50ULL)
+      {
+	check (j, 20, 2, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (j = ULLONG_MAX - 3; j >= LLONG_MAX + 70ULL; j -= LLONG_MAX + 50ULL)
+      {
+	check (j, ULLONG_MAX - 3, 3, 0)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (j = LLONG_MAX - 20000ULL; j <= LLONG_MAX + 10000ULL; j += 10000ULL)
+      {
+	check (j, LLONG_MAX - 20000ULL, 4, 0)
+	check (j, LLONG_MAX - 10000ULL, 4, 1)
+	check (j, LLONG_MAX, 4, 2)
+	check (j, LLONG_MAX + 10000ULL, 4, 3)
+	e = 1;
+      }
+    #pragma omp for schedule(runtime) nowait
+    for (i = -3LL * INT_MAX - 20000LL; i <= INT_MAX + 10000LL; i += INT_MAX + 200LL)
+      {
+	check (i, -3LL * INT_MAX - 20000LL, 5, 0)
+	check (i, -2LL * INT_MAX - 20000LL + 200LL, 5, 1)
+	check (i, -INT_MAX - 20000LL + 400LL, 5, 2)
+	check (i, -20000LL + 600LL, 5, 3)
+	check (i, INT_MAX - 20000LL + 800LL, 5, 4)
+	e = 1;
+      }
+  }
+  if (e)
+    abort ();
+  test (0, 3);
+  test (1, 3);
+  test (2, 1);
+  test (3, 1);
+  test (4, 4);
+  test (5, 5);
+  return 0;
+}
+
+int
+main (void)
+{
+  if (2 * sizeof (int) != sizeof (long long))
+    return 0;
+  test1 ();
+  test2 ();
+  test3 ();
+  test4 ();
+  omp_set_schedule (omp_sched_static, 0);
+  test5 ();
+  omp_set_schedule (omp_sched_static, 3);
+  test5 ();
+  omp_set_schedule (omp_sched_dynamic, 5);
+  test5 ();
+  omp_set_schedule (omp_sched_guided, 2);
+  test5 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-7.c b/libgomp/testsuite/libgomp.c/loop-7.c
new file mode 100644
index 000000000..fc97f4a29
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-7.c
@@ -0,0 +1,105 @@
+/* { dg-do run } */
+
+#include <omp.h>
+
+extern void abort (void);
+
+#define LLONG_MAX __LONG_LONG_MAX__
+#define ULLONG_MAX (LLONG_MAX * 2ULL + 1)
+#define INT_MAX __INT_MAX__
+
+int v;
+
+int
+test1 (void)
+{
+  int e = 0, cnt = 0;
+  long long i;
+  unsigned long long j;
+  char buf[6], *p;
+
+  #pragma omp for schedule(dynamic,1) collapse(2) nowait
+  for (i = LLONG_MAX - 30001; i <= LLONG_MAX - 10001; i += 10000)
+    for (j = 20; j <= LLONG_MAX - 70; j += LLONG_MAX + 50ULL)
+      if ((i != LLONG_MAX - 30001
+	   && i != LLONG_MAX - 20001
+	   && i != LLONG_MAX - 10001)
+	  || j != 20)
+	e = 1;
+      else
+	cnt++;
+  if (e || cnt != 3)
+    abort ();
+  else
+    cnt = 0;
+
+  #pragma omp for schedule(guided,1) collapse(2) nowait
+  for (i = -LLONG_MAX + 30000; i >= -LLONG_MAX + 10000; i -= 10000)
+    for (j = ULLONG_MAX - 3; j >= LLONG_MAX + 70ULL; j -= LLONG_MAX + 50ULL)
+      if ((i != -LLONG_MAX + 30000
+	   && i != -LLONG_MAX + 20000
+	   && i != -LLONG_MAX + 10000)
+	  || j != ULLONG_MAX - 3)
+	e = 1;
+      else
+	cnt++;
+  if (e || cnt != 3)
+    abort ();
+  else
+    cnt = 0;
+
+  #pragma omp for schedule(static,1) collapse(2) nowait
+  for (i = LLONG_MAX - 30001; i <= LLONG_MAX - 10001; i += 10000)
+    for (j = 20; j <= LLONG_MAX - 70 + v; j += LLONG_MAX + 50ULL)
+      if ((i != LLONG_MAX - 30001
+	   && i != LLONG_MAX - 20001
+	   && i != LLONG_MAX - 10001)
+	  || j != 20)
+	e = 1;
+      else
+	cnt++;
+  if (e || cnt != 3)
+    abort ();
+  else
+    cnt = 0;
+
+  #pragma omp for schedule(static) collapse(2) nowait
+  for (i = -LLONG_MAX + 30000 + v; i >= -LLONG_MAX + 10000; i -= 10000)
+    for (j = ULLONG_MAX - 3; j >= LLONG_MAX + 70ULL; j -= LLONG_MAX + 50ULL)
+      if ((i != -LLONG_MAX + 30000
+	   && i != -LLONG_MAX + 20000
+	   && i != -LLONG_MAX + 10000)
+	  || j != ULLONG_MAX - 3)
+	e = 1;
+      else
+	cnt++;
+  if (e || cnt != 3)
+    abort ();
+  else
+    cnt = 0;
+
+  #pragma omp for schedule(runtime) collapse(2) nowait
+  for (i = 10; i < 30; i++)
+    for (p = buf; p <= buf + 4; p += 2)
+      if (i < 10 || i >= 30 || (p != buf && p != buf + 2 && p != buf + 4))
+	e = 1;
+      else
+	cnt++;
+  if (e || cnt != 60)
+    abort ();
+  else
+    cnt = 0;
+
+  return 0;
+}
+
+int
+main (void)
+{
+  if (2 * sizeof (int) != sizeof (long long))
+    return 0;
+  asm volatile ("" : "+r" (v));
+  omp_set_schedule (omp_sched_dynamic, 1);
+  test1 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-8.c b/libgomp/testsuite/libgomp.c/loop-8.c
new file mode 100644
index 000000000..25db25c3b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-8.c
@@ -0,0 +1,27 @@
+extern void abort (void);
+
+int buf[256];
+
+void __attribute__((noinline))
+foo (void)
+{
+  int i;
+  #pragma omp for schedule (auto)
+    for (i = 0; i < 256; i++)
+      buf[i] += i;
+}
+
+int
+main (void)
+{
+  int i;
+  #pragma omp parallel for schedule (auto)
+    for (i = 0; i < 256; i++)
+      buf[i] = i;
+  #pragma omp parallel num_threads (4)
+    foo ();
+  for (i = 0; i < 256; i++)
+    if (buf[i] != 2 * i)
+      abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/loop-9.c b/libgomp/testsuite/libgomp.c/loop-9.c
new file mode 100644
index 000000000..1f789e12e
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/loop-9.c
@@ -0,0 +1,18 @@
+extern void abort (void);
+
+char buf[8] = "01234567";
+char buf2[8] = "23456789";
+
+int
+main (void)
+{
+  char *p, *q;
+  int sum = 0;
+  #pragma omp parallel for collapse (2) reduction (+:sum) lastprivate (p, q)
+  for (p = buf; p < &buf[8]; p++)
+    for (q = &buf2[0]; q <= buf2 + 7; q++)
+      sum += (*p - '0') + (*q - '0');
+  if (p != &buf[8] || q != buf2 + 8 || sum != 576)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nested-1.c b/libgomp/testsuite/libgomp.c/nested-1.c
new file mode 100644
index 000000000..d3cfb0100
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nested-1.c
@@ -0,0 +1,30 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i = -1, j = -1;
+
+  omp_set_nested (1);
+  omp_set_dynamic (0);
+#pragma omp parallel num_threads (4)
+  {
+#pragma omp single
+    {
+      i = omp_get_thread_num () + omp_get_num_threads () * 256;
+#pragma omp parallel num_threads (2)
+      {
+#pragma omp single
+        {
+          j = omp_get_thread_num () + omp_get_num_threads () * 256;
+        }
+      }
+    }
+  }
+  if (i < 4 * 256 || i >= 4 * 256 + 4)
+    abort ();
+  if (j < 2 * 256 || j >= 2 * 256 + 2)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nested-2.c b/libgomp/testsuite/libgomp.c/nested-2.c
new file mode 100644
index 000000000..f52b074ff
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nested-2.c
@@ -0,0 +1,30 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i = -1, j = -1;
+
+  omp_set_nested (0);
+  omp_set_dynamic (0);
+#pragma omp parallel num_threads (4)
+  {
+#pragma omp single
+    {
+      i = omp_get_thread_num () + omp_get_num_threads () * 256;
+#pragma omp parallel num_threads (2)
+      {
+#pragma omp single
+        {
+          j = omp_get_thread_num () + omp_get_num_threads () * 256;
+        }
+      }
+    }
+  }
+  if (i < 4 * 256 || i >= 4 * 256 + 4)
+    abort ();
+  if (j != 256 + 0)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nested-3.c b/libgomp/testsuite/libgomp.c/nested-3.c
new file mode 100644
index 000000000..618600633
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nested-3.c
@@ -0,0 +1,89 @@
+#include <omp.h>
+#include <stdlib.h>
+#include <string.h>
+
+int
+main (void)
+{
+  int e[3];
+
+  memset (e, '\0', sizeof (e));
+  omp_set_nested (1);
+  omp_set_dynamic (0);
+  if (omp_in_parallel ()
+      || omp_get_level () != 0
+      || omp_get_ancestor_thread_num (0) != 0
+      || omp_get_ancestor_thread_num (-1) != -1
+      || omp_get_ancestor_thread_num (1) != -1
+      || omp_get_team_size (0) != 1
+      || omp_get_team_size (-1) != -1
+      || omp_get_team_size (1) != -1
+      || omp_get_active_level () != 0)
+    abort ();
+#pragma omp parallel num_threads (4)
+  {
+    int tn1 = omp_get_thread_num ();
+    if (omp_in_parallel () != 1
+	|| omp_get_num_threads () != 4
+	|| tn1 >= 4 || tn1 < 0
+	|| omp_get_level () != 1
+	|| omp_get_ancestor_thread_num (0) != 0
+	|| omp_get_ancestor_thread_num (1) != tn1
+	|| omp_get_ancestor_thread_num (-1) != -1
+	|| omp_get_ancestor_thread_num (2) != -1
+	|| omp_get_team_size (0) != 1
+	|| omp_get_team_size (1) != omp_get_num_threads ()
+	|| omp_get_team_size (-1) != -1
+	|| omp_get_team_size (2) != -1
+	|| omp_get_active_level () != 1)
+      #pragma omp atomic
+	e[0] += 1;
+    #pragma omp parallel if (0) num_threads(5) firstprivate(tn1)
+    {
+      int tn2 = omp_get_thread_num ();
+      if (omp_in_parallel () != 1
+	  || omp_get_num_threads () != 1
+	  || tn2 != 0
+	  || omp_get_level () != 2
+	  || omp_get_ancestor_thread_num (0) != 0
+	  || omp_get_ancestor_thread_num (1) != tn1
+	  || omp_get_ancestor_thread_num (2) != tn2
+	  || omp_get_ancestor_thread_num (-1) != -1
+	  || omp_get_ancestor_thread_num (3) != -1
+	  || omp_get_team_size (0) != 1
+	  || omp_get_team_size (1) != 4
+	  || omp_get_team_size (2) != 1
+	  || omp_get_team_size (-1) != -1
+	  || omp_get_team_size (3) != -1
+	  || omp_get_active_level () != 1)
+	#pragma omp atomic
+	  e[1] += 1;
+      #pragma omp parallel num_threads(2) firstprivate(tn1, tn2)
+      {
+	int tn3 = omp_get_thread_num ();
+	if (omp_in_parallel () != 1
+	    || omp_get_num_threads () != 2
+	    || tn3 > 1 || tn3 < 0
+	    || omp_get_level () != 3
+	    || omp_get_ancestor_thread_num (0) != 0
+	    || omp_get_ancestor_thread_num (1) != tn1
+	    || omp_get_ancestor_thread_num (2) != tn2
+	    || omp_get_ancestor_thread_num (3) != tn3
+	    || omp_get_ancestor_thread_num (-1) != -1
+	    || omp_get_ancestor_thread_num (4) != -1
+	    || omp_get_team_size (0) != 1
+	    || omp_get_team_size (1) != 4
+	    || omp_get_team_size (2) != 1
+	    || omp_get_team_size (3) != 2
+	    || omp_get_team_size (-1) != -1
+	    || omp_get_team_size (4) != -1
+	    || omp_get_active_level () != 2)
+	  #pragma omp atomic
+	    e[2] += 1;
+      }
+    }
+  }
+  if (e[0] || e[1] || e[2])
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nestedfn-1.c b/libgomp/testsuite/libgomp.c/nestedfn-1.c
new file mode 100644
index 000000000..26c0d237c
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nestedfn-1.c
@@ -0,0 +1,49 @@
+/* { dg-do run } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int a = 1, b = 2, c = 3;
+  void
+  foo (void)
+  {
+    int l = 0;
+#pragma omp parallel shared (a) private (b) firstprivate (c) \
+		     num_threads (2) reduction (||:l)
+    {
+      if (a != 1 || c != 3) l = 1;
+#pragma omp barrier
+      if (omp_get_thread_num () == 0)
+	{
+	  a = 4;
+	  b = 5;
+	  c = 6;
+	}
+#pragma omp barrier
+      if (omp_get_thread_num () == 1)
+	{
+	  if (a != 4 || c != 3) l = 1;
+	  a = 7;
+	  b = 8;
+	  c = 9;
+	}
+      else if (omp_get_num_threads () == 1)
+	a = 7;
+#pragma omp barrier
+      if (omp_get_thread_num () == 0)
+	if (a != 7 || b != 5 || c != 6) l = 1;
+#pragma omp barrier
+      if (omp_get_thread_num () == 1)
+	if (a != 7 || b != 8 || c != 9) l = 1;
+    }
+    if (l)
+      abort ();
+  }
+  foo ();
+  if (a != 7)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nestedfn-2.c b/libgomp/testsuite/libgomp.c/nestedfn-2.c
new file mode 100644
index 000000000..fdbbe0f73
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nestedfn-2.c
@@ -0,0 +1,20 @@
+/* { dg-do run } */
+
+extern void abort (void);
+
+int
+main (void)
+{
+  int i;
+  void
+  foo (void)
+  {
+#pragma omp master
+    i += 8;
+  }
+  i = 4;
+  foo ();
+  if (i != 12)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nestedfn-3.c b/libgomp/testsuite/libgomp.c/nestedfn-3.c
new file mode 100644
index 000000000..8f8847f97
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nestedfn-3.c
@@ -0,0 +1,52 @@
+/* { dg-do run } */
+
+#include <omp.h>
+
+extern void abort (void);
+
+int
+main (void)
+{
+  int i = 5, l = 0;
+  int foo (void) { return i == 6; }
+  int bar (void) { return i - 3; }
+
+  omp_set_dynamic (0);
+
+#pragma omp parallel if (foo ()) num_threads (bar ()) reduction (|:l)
+  if (omp_get_num_threads () != 1)
+    l = 1;
+
+  i++;
+
+#pragma omp parallel if (foo ()) num_threads (bar ()) reduction (|:l)
+  if (omp_get_num_threads () != 3)
+    l = 1;
+
+  i++;
+
+#pragma omp master
+  if (bar () != 4)
+    abort ();
+
+#pragma omp single
+  {
+    if (foo ())
+      abort ();
+    i--;
+    if (! foo ())
+      abort ();
+  }
+
+  if (l)
+    abort ();
+
+  i = 8;
+#pragma omp atomic
+  l += bar ();
+
+  if (l != 5)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nestedfn-4.c b/libgomp/testsuite/libgomp.c/nestedfn-4.c
new file mode 100644
index 000000000..dbe1062bd
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nestedfn-4.c
@@ -0,0 +1,65 @@
+/* PR middle-end/25261 */
+/* { dg-do run } */
+
+#include <omp.h>
+
+extern void abort (void);
+
+int
+main (void)
+{
+  int i = 5, j, l = 0;
+  int foo (void)
+  {
+    return i == 6;
+  }
+  int bar (void)
+  {
+    return i - 3;
+  }
+
+  omp_set_dynamic (0);
+
+#pragma omp parallel if (foo ()) num_threads (2)
+  if (omp_get_num_threads () != 1)
+#pragma omp atomic
+    l++;
+
+#pragma omp parallel for schedule (static, bar ()) num_threads (2) \
+		     reduction (|:l)
+  for (j = 0; j < 4; j++)
+    if (omp_get_thread_num () != (j >= 2))
+#pragma omp atomic
+      l++;
+
+  i++;
+
+#pragma omp parallel if (foo ()) num_threads (2)
+  if (omp_get_num_threads () != 2)
+#pragma omp atomic
+    l++;
+
+#pragma omp parallel for schedule (static, bar ()) num_threads (2) \
+		     reduction (|:l)
+  for (j = 0; j < 6; j++)
+    if (omp_get_thread_num () != (j >= 3))
+#pragma omp atomic
+      l++;
+
+#pragma omp parallel num_threads (4) reduction (|:l)
+  if (!foo () || bar () != 3)
+#pragma omp atomic
+      l++;
+
+  i++;
+
+#pragma omp parallel num_threads (4) reduction (|:l)
+  if (foo () || bar () != 4)
+#pragma omp atomic
+      l++;
+
+  if (l)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nestedfn-5.c b/libgomp/testsuite/libgomp.c/nestedfn-5.c
new file mode 100644
index 000000000..6072b1fe3
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nestedfn-5.c
@@ -0,0 +1,38 @@
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+
+extern void abort (void);
+
+void
+foo (int *j)
+{
+  int i = 5;
+  int bar (void) { return i + 1; }
+#pragma omp sections
+  {
+    #pragma omp section
+      {
+	if (bar () != 6)
+	#pragma omp atomic
+	  ++*j;
+      }
+    #pragma omp section
+      {
+	if (bar () != 6)
+	#pragma omp atomic
+	  ++*j;
+      }
+  }
+}
+
+int
+main (void)
+{
+  int j = 0;
+#pragma omp parallel num_threads (2)
+  foo (&j);
+  if (j)
+    abort ();
+  return 0;
+}
+
diff --git a/libgomp/testsuite/libgomp.c/nestedfn-6.c b/libgomp/testsuite/libgomp.c/nestedfn-6.c
new file mode 100644
index 000000000..c0ace6b3f
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nestedfn-6.c
@@ -0,0 +1,21 @@
+extern void abort (void);
+
+int j;
+
+int
+main (void)
+{
+  int i;
+  void nested (void) { i = 0; }
+#pragma omp parallel for lastprivate (i)
+  for (i = 0; i < 50; i += 3)
+    ;
+  if (i != 51)
+    abort ();
+#pragma omp parallel for lastprivate (j)
+  for (j = -50; j < 70; j += 7)
+    ;
+  if (j != 76)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/nqueens-1.c b/libgomp/testsuite/libgomp.c/nqueens-1.c
new file mode 100644
index 000000000..1fdc67b29
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/nqueens-1.c
@@ -0,0 +1,66 @@
+/* { dg-do run } */
+/* { dg-options "-O2 -fopenmp" } */
+/* { dg-require-effective-target tls_runtime } */
+
+#include <omp.h>
+#include <stdio.h>
+#include <string.h>
+
+int cnt;
+#pragma omp threadprivate (cnt)
+
+void
+nqueens (char *a, int n, int pos)
+{
+  /* b[i] = j means the queen in i-th row is in column j.  */
+  char b[pos + 1];
+  int i, j;
+  memcpy (b, a, pos);
+  for (i = 0; i < n; i++)
+    {
+      for (j = 0; j < pos; j++)
+	if (b[j] == i || b[j] == i + pos - j || i == b[j] + pos - j)
+	  break;
+      if (j < pos)
+	continue;
+      if (pos == n - 1)
+	/* Found a solution.  Could output it here.  */
+	++cnt;
+      else
+	{
+	  b[pos] = i;
+	  #pragma omp task
+	    nqueens (b, n, pos + 1);
+	}
+    }
+}
+
+int
+main (int argc, char **argv)
+{
+  int n = 8;
+  if (argc >= 2)
+    n = strtoul (argv[1], NULL, 0);
+  if (n < 1 || n > 127)
+    {
+      fprintf (stderr, "invalid count %d\n", n);
+      return 1;
+    }
+  cnt = 0;
+  double stime = omp_get_wtime ();
+  nqueens ("", n, 0);
+  printf ("serial   N %d solutions # %d time %f\n", n, cnt, omp_get_wtime () - stime);
+  #pragma omp parallel
+    cnt = 0;
+  stime = omp_get_wtime ();
+  int tempcnt = 0;
+  #pragma omp parallel reduction (+:tempcnt)
+    {
+      #pragma omp single
+	nqueens ("", n, 0);
+      tempcnt = cnt;
+    }
+  cnt = tempcnt;
+  printf ("parallel N %d solutions # %d time %f\n", n, cnt, omp_get_wtime () - stime);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-loop01.c b/libgomp/testsuite/libgomp.c/omp-loop01.c
new file mode 100644
index 000000000..0e83c9583
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-loop01.c
@@ -0,0 +1,96 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <omp.h>
+
+#define MAX	1000
+
+void main1()
+{
+  int i, N1, N2, step;
+  int a[MAX], b[MAX];
+
+  N1 = rand () % 13;
+  N2 = rand () % (MAX - 51) + 50;
+  step = rand () % 7 + 1;
+
+  printf ("N1 = %d\nN2 = %d\nstep = %d\n", N1, N2, step);
+
+  for (i = N1; i <= N2; i += step)
+    a[i] = 42+ i;
+
+  /* COUNTING UP (<).  Fill in array 'b' in parallel.  */
+  memset (b, 0, sizeof b);
+#pragma omp parallel shared(a,b,N1,N2,step) private(i)
+  {
+#pragma omp for
+    for (i = N1; i < N2; i += step)
+      b[i] = a[i];
+  }
+
+  /* COUNTING UP (<).  Check that all the cells were filled in properly.  */
+  for (i = N1; i < N2; i += step)
+    if (a[i] != b[i])
+      abort ();
+
+  printf ("for (i = %d; i < %d; i += %d) [OK]\n", N1, N2, step);
+
+  /* COUNTING UP (<=).  Fill in array 'b' in parallel.  */
+  memset (b, 0, sizeof b);
+#pragma omp parallel shared(a,b,N1,N2,step) private(i)
+  {
+#pragma omp for
+    for (i = N1; i <= N2; i += step)
+      b[i] = a[i];
+  }
+
+  /* COUNTING UP (<=).  Check that all the cells were filled in properly.  */
+  for (i = N1; i <= N2; i += step)
+    if (a[i] != b[i])
+      abort ();
+
+  printf ("for (i = %d; i <= %d; i += %d) [OK]\n", N1, N2, step);
+
+  /* COUNTING DOWN (>).  Fill in array 'b' in parallel.  */
+  memset (b, 0, sizeof b);
+#pragma omp parallel shared(a,b,N1,N2,step) private(i)
+  {
+#pragma omp for
+    for (i = N2; i > N1; i -= step)
+      b[i] = a[i];
+  }
+
+  /* COUNTING DOWN (>).  Check that all the cells were filled in properly.  */
+  for (i = N2; i > N1; i -= step)
+    if (a[i] != b[i])
+      abort ();
+
+  printf ("for (i = %d; i > %d; i -= %d) [OK]\n", N2, N1, step);
+
+  /* COUNTING DOWN (>=).  Fill in array 'b' in parallel.  */
+  memset (b, 0, sizeof b);
+#pragma omp parallel shared(a,b,N1,N2,step) private(i)
+  {
+#pragma omp for
+    for (i = N2; i >= N1; i -= step)
+      b[i] = a[i];
+  }
+
+  /* COUNTING DOWN (>=).  Check that all the cells were filled in properly.  */
+  for (i = N2; i >= N1; i -= step)
+    if (a[i] != b[i])
+      abort ();
+
+  printf ("for (i = %d; i >= %d; i -= %d) [OK]\n", N2, N1, step);
+}
+
+int
+main ()
+{
+  int i;
+
+  srand (0);
+  for (i = 0; i < 10; ++i)
+    main1();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-loop02.c b/libgomp/testsuite/libgomp.c/omp-loop02.c
new file mode 100644
index 000000000..04aaea2e3
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-loop02.c
@@ -0,0 +1,32 @@
+#include <omp.h>
+
+/* Orphaned work sharing.  */
+
+extern void abort (void);
+
+#define N 10
+
+void parloop (int *a)
+{
+  int i;
+
+#pragma omp for
+  for (i = 0; i < N; i++)
+    a[i] = i + 3;
+}
+
+main()
+{
+  int i, a[N];
+
+#pragma omp parallel shared(a)
+  {
+    parloop (a);
+  }
+
+  for (i = 0; i < N; i++)
+    if (a[i] != i + 3)
+      abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-loop03.c b/libgomp/testsuite/libgomp.c/omp-loop03.c
new file mode 100644
index 000000000..7bb9a1943
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-loop03.c
@@ -0,0 +1,26 @@
+extern void abort (void);
+int a;
+
+void
+foo ()
+{
+  int i;
+  a = 30;
+#pragma omp barrier
+#pragma omp for lastprivate (a)
+  for (i = 0; i < 1024; i++)
+    {
+      a = i;
+    }
+  if (a != 1023)
+    abort ();
+}
+
+int
+main (void)
+{
+#pragma omp parallel num_threads (64)
+  foo ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-nested-1.c b/libgomp/testsuite/libgomp.c/omp-nested-1.c
new file mode 100644
index 000000000..655ef26fa
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-nested-1.c
@@ -0,0 +1,28 @@
+// { dg-do run }
+
+extern void abort(void);
+#define N 1000
+
+int foo()
+{
+  int i = 0, j;
+
+  #pragma omp parallel for num_threads(2) shared (i)
+  for (j = 0; j < N; ++j)
+    {
+      #pragma omp parallel num_threads(1) shared (i)
+      {
+	#pragma omp atomic
+	i++;
+      }
+    }
+
+  return i;
+}
+
+int main()
+{
+  if (foo() != N)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-parallel-for.c b/libgomp/testsuite/libgomp.c/omp-parallel-for.c
new file mode 100644
index 000000000..c6631a0a7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-parallel-for.c
@@ -0,0 +1,20 @@
+extern void abort (void);
+
+main()
+{
+  int i, a;
+
+  a = 30;
+
+#pragma omp parallel for firstprivate (a) lastprivate (a) \
+	num_threads (2) schedule(static)
+  for (i = 0; i < 10; i++)
+    a = a + i;
+
+  /* The thread that owns the last iteration will have computed
+     30 + 5 + 6 + 7 + 8 + 9 = 65.  */
+  if (a != 65)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-parallel-if.c b/libgomp/testsuite/libgomp.c/omp-parallel-if.c
new file mode 100644
index 000000000..5e378359a
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-parallel-if.c
@@ -0,0 +1,40 @@
+#include <omp.h>
+
+extern void abort (void);
+
+int
+foo (void)
+{
+  return 10;
+}
+
+main ()
+{
+  int A = 0;
+
+  #pragma omp parallel if (foo () > 10) shared (A)
+    {
+      A = omp_get_num_threads ();
+    }
+
+  if (A != 1)
+    abort ();
+
+  #pragma omp parallel if (foo () == 10) num_threads (3) shared (A)
+    {
+      A = omp_get_num_threads ();
+    }
+
+  if (A != 3)
+    abort ();
+
+  #pragma omp parallel if (foo () == 10) num_threads (foo ()) shared (A)
+    {
+      A = omp_get_num_threads ();
+    }
+
+  if (A != 10)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-single-1.c b/libgomp/testsuite/libgomp.c/omp-single-1.c
new file mode 100644
index 000000000..a44ca54ac
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-single-1.c
@@ -0,0 +1,19 @@
+extern void abort (void);
+
+main()
+{
+  int i = 0;
+
+  #pragma omp parallel shared (i)
+    {
+      #pragma omp single
+	{
+	  i++;
+	}
+    }
+
+  if (i != 1)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-single-2.c b/libgomp/testsuite/libgomp.c/omp-single-2.c
new file mode 100644
index 000000000..687855973
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-single-2.c
@@ -0,0 +1,38 @@
+#include <omp.h>
+
+extern void abort (void);
+
+struct X
+{
+  int a;
+  char b;
+  int c;
+};
+
+main()
+{
+  int i = 0;
+  struct X x;
+  int bad = 0;
+
+  #pragma omp parallel private (i, x) shared (bad)
+    {
+      i = 5;
+
+      #pragma omp single copyprivate (i, x)
+	{
+	  i++;
+	  x.a = 23;
+	  x.b = 42;
+	  x.c = 26;
+	}
+
+      if (i != 6 || x.a != 23 || x.b != 42 || x.c != 26)
+	bad = 1;
+    }
+
+  if (bad)
+    abort ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp-single-3.c b/libgomp/testsuite/libgomp.c/omp-single-3.c
new file mode 100644
index 000000000..5a0653244
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp-single-3.c
@@ -0,0 +1,21 @@
+extern void abort (void);
+
+void
+single (int a, int b)
+{
+  #pragma omp single copyprivate(a) copyprivate(b)
+    {
+      a = b = 5;
+    }
+
+  if (a != b)
+    abort ();
+}
+
+int main()
+{
+  #pragma omp parallel
+    single (1, 2);
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp_hello.c b/libgomp/testsuite/libgomp.c/omp_hello.c
new file mode 100644
index 000000000..8d58cd43b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp_hello.c
@@ -0,0 +1,39 @@
+/******************************************************************************
+* FILE: omp_hello.c
+* DESCRIPTION:
+*   OpenMP Example - Hello World - C/C++ Version
+*   In this simple example, the master thread forks a parallel region.
+*   All threads in the team obtain their unique thread number and print it.
+*   The master thread only prints the total number of threads.  Two OpenMP
+*   library routines are used to obtain the number of threads and each
+*   thread's number.
+* AUTHOR: Blaise Barney  5/99
+* LAST REVISED: 04/06/05
+******************************************************************************/
+#include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+int main (int argc, char *argv[]) {
+
+int nthreads, tid;
+
+/* Fork a team of threads giving them their own copies of variables */
+#pragma omp parallel private(nthreads, tid)
+  {
+
+  /* Obtain thread number */
+  tid = omp_get_thread_num();
+  printf("Hello World from thread = %d\n", tid);
+
+  /* Only master thread does this */
+  if (tid == 0) 
+    {
+    nthreads = omp_get_num_threads();
+    printf("Number of threads = %d\n", nthreads);
+    }
+
+  }  /* All threads join master thread and disband */
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp_matvec.c b/libgomp/testsuite/libgomp.c/omp_matvec.c
new file mode 100644
index 000000000..12b8c6896
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp_matvec.c
@@ -0,0 +1,72 @@
+/******************************************************************************
+* OpenMP Example - Matrix-vector multiplication - C/C++ Version
+* FILE: omp_matvec.c
+* DESCRIPTION:
+*   This example multiplies all row i elements of matrix A with vector
+*   element b(i) and stores the summed products in vector c(i).  A total is
+*   maintained for the entire matrix.  Performed by using the OpenMP loop
+*   work-sharing construct.  The update of the shared global total is
+*   serialized by using the OpenMP critical directive.
+* SOURCE: Blaise Barney  5/99
+* LAST REVISED:
+******************************************************************************/
+
+#include <omp.h>
+#include <stdio.h>
+#define SIZE 10
+
+
+main ()
+{
+
+float A[SIZE][SIZE], b[SIZE], c[SIZE], total;
+int i, j, tid;
+
+/* Initializations */
+total = 0.0;
+for (i=0; i < SIZE; i++)
+  {
+  for (j=0; j < SIZE; j++)
+    A[i][j] = (j+1) * 1.0;
+  b[i] = 1.0 * (i+1);
+  c[i] = 0.0;
+  }
+printf("\nStarting values of matrix A and vector b:\n");
+for (i=0; i < SIZE; i++)
+  {
+  printf("  A[%d]= ",i);
+  for (j=0; j < SIZE; j++)
+    printf("%.1f ",A[i][j]);
+  printf("  b[%d]= %.1f\n",i,b[i]);
+  }
+printf("\nResults by thread/row:\n");
+
+/* Create a team of threads and scope variables */
+#pragma omp parallel shared(A,b,c,total) private(tid,i)
+  {
+  tid = omp_get_thread_num();
+
+/* Loop work-sharing construct - distribute rows of matrix */
+#pragma omp for private(j)
+  for (i=0; i < SIZE; i++)
+    {
+    for (j=0; j < SIZE; j++)
+      c[i] += (A[i][j] * b[i]);
+
+    /* Update and display of running total must be serialized */
+    #pragma omp critical
+      {
+      total = total + c[i];
+      printf("  thread %d did row %d\t c[%d]=%.2f\t",tid,i,i,c[i]);
+      printf("Running total= %.2f\n",total);
+      }
+
+    }   /* end of parallel i loop */
+
+  } /* end of parallel construct */
+
+printf("\nMatrix-vector total - sum of all c[] = %.2f\n\n",total);
+
+  return 0;
+}
+
diff --git a/libgomp/testsuite/libgomp.c/omp_orphan.c b/libgomp/testsuite/libgomp.c/omp_orphan.c
new file mode 100644
index 000000000..cbf7abf37
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp_orphan.c
@@ -0,0 +1,47 @@
+/******************************************************************************
+* FILE: omp_orphan.c
+* DESCRIPTION:
+*   OpenMP Example - Parallel region with an orphaned directive - C/C++ Version
+*   This example demonstrates a dot product  being performed by an orphaned
+*   loop reduction construct.  Scoping of the reduction variable is critical.
+* AUTHOR: Blaise Barney  5/99
+* LAST REVISED: 04/06/05
+******************************************************************************/
+#include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+#define VECLEN 100
+
+float a[VECLEN], b[VECLEN], sum;
+
+float dotprod ()
+{
+int i,tid;
+
+tid = omp_get_thread_num();
+#pragma omp for reduction(+:sum)
+  for (i=0; i < VECLEN; i++)
+    {
+    sum = sum + (a[i]*b[i]);
+    printf("  tid= %d i=%d\n",tid,i);
+    }
+
+return(sum);
+}
+
+
+int main (int argc, char *argv[]) 
+{
+int i;
+
+for (i=0; i < VECLEN; i++)
+  a[i] = b[i] = 1.0 * i;
+sum = 0.0;
+
+#pragma omp parallel
+  sum = dotprod();
+
+printf("Sum = %f\n",sum);
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp_reduction.c b/libgomp/testsuite/libgomp.c/omp_reduction.c
new file mode 100644
index 000000000..5c9c41ec7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp_reduction.c
@@ -0,0 +1,35 @@
+/******************************************************************************
+* FILE: omp_reduction.c
+* DESCRIPTION:
+*   OpenMP Example - Combined Parallel Loop Reduction - C/C++ Version
+*   This example demonstrates a sum reduction within a combined parallel loop
+*   construct.  Notice that default data element scoping is assumed - there
+*   are no clauses specifying shared or private variables.  OpenMP will 
+*   automatically make loop index variables private within team threads, and
+*   global variables shared.
+* AUTHOR: Blaise Barney  5/99
+* LAST REVISED: 04/06/05
+******************************************************************************/
+#include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+int main (int argc, char *argv[]) {
+
+int   i, n;
+float a[100], b[100], sum; 
+
+/* Some initializations */
+n = 100;
+for (i=0; i < n; i++)
+  a[i] = b[i] = i * 1.0;
+sum = 0.0;
+
+#pragma omp parallel for reduction(+:sum)
+  for (i=0; i < n; i++)
+    sum = sum + (a[i] * b[i]);
+
+printf("   Sum = %f\n",sum);
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp_workshare1.c b/libgomp/testsuite/libgomp.c/omp_workshare1.c
new file mode 100644
index 000000000..e33bef316
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp_workshare1.c
@@ -0,0 +1,47 @@
+/******************************************************************************
+* FILE: omp_workshare1.c
+* DESCRIPTION:
+*   OpenMP Example - Loop Work-sharing - C/C++ Version
+*   In this example, the iterations of a loop are scheduled dynamically
+*   across the team of threads.  A thread will perform CHUNK iterations
+*   at a time before being scheduled for the next CHUNK of work.
+* AUTHOR: Blaise Barney  5/99
+* LAST REVISED: 04/06/05
+******************************************************************************/
+#include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+#define CHUNKSIZE   10
+#define N       100
+
+int main (int argc, char *argv[]) {
+
+int nthreads, tid, i, chunk;
+float a[N], b[N], c[N];
+
+/* Some initializations */
+for (i=0; i < N; i++)
+  a[i] = b[i] = i * 1.0;
+chunk = CHUNKSIZE;
+
+#pragma omp parallel shared(a,b,c,nthreads,chunk) private(i,tid)
+  {
+  tid = omp_get_thread_num();
+  if (tid == 0)
+    {
+    nthreads = omp_get_num_threads();
+    printf("Number of threads = %d\n", nthreads);
+    }
+  printf("Thread %d starting...\n",tid);
+
+  #pragma omp for schedule(dynamic,chunk)
+  for (i=0; i<N; i++)
+    {
+    c[i] = a[i] + b[i];
+    printf("Thread %d: c[%d]= %f\n",tid,i,c[i]);
+    }
+
+  }  /* end of parallel section */
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp_workshare2.c b/libgomp/testsuite/libgomp.c/omp_workshare2.c
new file mode 100644
index 000000000..32c93dbde
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp_workshare2.c
@@ -0,0 +1,64 @@
+/******************************************************************************
+* FILE: omp_workshare2.c
+* DESCRIPTION:
+*   OpenMP Example - Sections Work-sharing - C/C++ Version
+*   In this example, the OpenMP SECTION directive is used to assign
+*   different array operations to threads that execute a SECTION. Each 
+*   thread receives its own copy of the result array to work with.
+* AUTHOR: Blaise Barney  5/99
+* LAST REVISED: 04/06/05
+******************************************************************************/
+#include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+#define N     50
+
+int main (int argc, char *argv[]) {
+
+int i, nthreads, tid;
+float a[N], b[N], c[N];
+
+/* Some initializations */
+for (i=0; i<N; i++)
+  a[i] = b[i] = i * 1.0;
+
+#pragma omp parallel shared(a,b,nthreads) private(c,i,tid)
+  {
+  tid = omp_get_thread_num();
+  if (tid == 0)
+    {
+    nthreads = omp_get_num_threads();
+    printf("Number of threads = %d\n", nthreads);
+    }
+  printf("Thread %d starting...\n",tid);
+
+  #pragma omp sections nowait
+    {
+    #pragma omp section
+      {
+      printf("Thread %d doing section 1\n",tid);
+      for (i=0; i<N; i++)
+        {
+        c[i] = a[i] + b[i];
+        printf("Thread %d: c[%d]= %f\n",tid,i,c[i]);
+        }
+      }
+
+    #pragma omp section
+      {
+      printf("Thread %d doing section 2\n",tid);
+      for (i=0; i<N; i++)
+        {
+        c[i] = a[i] * b[i];
+        printf("Thread %d: c[%d]= %f\n",tid,i,c[i]);
+        }
+      }
+
+    }  /* end of sections */
+
+    printf("Thread %d done.\n",tid); 
+
+  }  /* end of parallel section */
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp_workshare3.c b/libgomp/testsuite/libgomp.c/omp_workshare3.c
new file mode 100644
index 000000000..913f1f731
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp_workshare3.c
@@ -0,0 +1,43 @@
+/* { dg-do compile } */
+
+/******************************************************************************
+* OpenMP Example - Combined Parallel Loop Work-sharing - C/C++ Version
+* FILE: omp_workshare3.c
+* DESCRIPTION:
+*   This example attempts to show use of the parallel for construct.  However
+*   it will generate errors at compile time.  Try to determine what is causing
+*   the error.  See omp_workshare4.c for a corrected version.
+* SOURCE: Blaise Barney  5/99
+* LAST REVISED: 03/03/2002
+******************************************************************************/
+
+#include <omp.h>
+#include <stdio.h>
+#define N       50
+#define CHUNKSIZE   5
+
+main ()  {
+
+int i, chunk, tid;
+float a[N], b[N], c[N];
+
+/* Some initializations */
+for (i=0; i < N; i++)
+  a[i] = b[i] = i * 1.0;
+chunk = CHUNKSIZE;
+
+#pragma omp parallel for     \
+  shared(a,b,c,chunk)            \
+  private(i,tid)             \
+  schedule(static,chunk)
+  {				/* { dg-error "expected" } */
+  tid = omp_get_thread_num();
+  for (i=0; i < N; i++)
+    {
+    c[i] = a[i] + b[i];
+    printf("tid= %d i= %d c[i]= %f\n", tid, i, c[i]);
+    }
+  }  /* end of parallel for construct */
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/omp_workshare4.c b/libgomp/testsuite/libgomp.c/omp_workshare4.c
new file mode 100644
index 000000000..67605e38b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/omp_workshare4.c
@@ -0,0 +1,48 @@
+/******************************************************************************
+* OpenMP Example - Combined Parallel Loop Work-sharing - C/C++ Version
+* FILE: omp_workshare4.c
+* DESCRIPTION:
+*   This is a corrected version of the omp_workshare3.c example. Corrections
+*   include removing all statements between the parallel for construct and
+*   the actual for loop, and introducing logic to preserve the ability to 
+*   query a thread's id and print it from inside the for loop.
+* SOURCE: Blaise Barney  5/99
+* LAST REVISED: 03/03/2002
+******************************************************************************/
+
+#include <omp.h>
+#include <stdio.h>
+#define N       50
+#define CHUNKSIZE   5
+
+main ()  {
+
+int i, chunk, tid;
+float a[N], b[N], c[N];
+char first_time;
+
+/* Some initializations */
+for (i=0; i < N; i++)
+  a[i] = b[i] = i * 1.0;
+chunk = CHUNKSIZE;
+first_time = 'y';
+
+#pragma omp parallel for     \
+  shared(a,b,c,chunk)            \
+  private(i,tid)             \
+  schedule(static,chunk)     \
+  firstprivate(first_time)
+
+  for (i=0; i < N; i++)
+    {
+    if (first_time == 'y')
+      {
+      tid = omp_get_thread_num();
+      first_time = 'n';
+      }
+    c[i] = a[i] + b[i];
+    printf("tid= %d i= %d c[i]= %f\n", tid, i, c[i]);
+    }
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/ordered-1.c b/libgomp/testsuite/libgomp.c/ordered-1.c
new file mode 100644
index 000000000..c6143fac2
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/ordered-1.c
@@ -0,0 +1,115 @@
+/* Test that all loop iterations are touched.  This doesn't verify 
+   scheduling order, merely coverage.  */
+/* Note that we never call GOMP_ordered_start in here.  AFAICS, this is
+   valid; the only requirement is "not more than once per iteration".  */
+
+/* { dg-require-effective-target sync_int_long } */
+
+#include <omp.h>
+#include <string.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+#define N 1000
+static int S, E, INCR, CHUNK, NTHR;
+static int data[N];
+
+static void clean_data (void)
+{
+  memset (data, -1, sizeof (data));
+}
+
+static void test_data (void)
+{
+  int i, j;
+
+  for (i = 0; i < S; ++i)
+    assert (data[i] == -1);
+
+  for (j = 0; i < E; ++i, j = (j + 1) % INCR)
+    if (j == 0)
+      assert (data[i] != -1);
+    else
+      assert (data[i] == -1);
+
+  for (; i < N; ++i)
+    assert (data[i] == -1);
+}
+
+static void set_data (long i, int val)
+{
+  int old;
+  assert (i >= 0 && i < N);
+  old = __sync_lock_test_and_set (data+i, val);
+  assert (old == -1);
+}
+  
+
+#define TMPL_1(sched)							\
+static void f_##sched##_1 (void *dummy)					\
+{									\
+  int iam = omp_get_thread_num ();					\
+  long s0, e0, i;							\
+  if (GOMP_loop_ordered_##sched##_start (S, E, INCR, CHUNK, &s0, &e0))	\
+    do									\
+      {									\
+	for (i = s0; i < e0; i += INCR)					\
+	  set_data (i, iam);						\
+      }									\
+    while (GOMP_loop_ordered_##sched##_next (&s0, &e0));		\
+  GOMP_loop_end ();							\
+}									\
+static void t_##sched##_1 (void)					\
+{									\
+  clean_data ();							\
+  GOMP_parallel_start (f_##sched##_1, NULL, NTHR);			\
+  f_##sched##_1 (NULL);							\
+  GOMP_parallel_end ();							\
+  test_data ();								\
+}
+
+TMPL_1(static)
+TMPL_1(dynamic)
+TMPL_1(guided)
+
+static void test (void)
+{
+  t_static_1 ();
+  t_dynamic_1 ();
+  t_guided_1 ();
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  NTHR = 4;
+
+  S = 0, E = N, INCR = 1, CHUNK = 4;
+  test ();
+
+  S = 0, E = N, INCR = 2, CHUNK = 4;
+  test ();
+
+  S = 1, E = N-1, INCR = 1, CHUNK = 5;
+  test ();
+
+  S = 1, E = N-1, INCR = 2, CHUNK = 5;
+  test ();
+
+  S = 2, E = 4, INCR = 1, CHUNK = 1;
+  test ();
+
+  S = 0, E = N, INCR = 1, CHUNK = 0;
+  t_static_1 ();
+
+  S = 1, E = N-1, INCR = 1, CHUNK = 0;
+  t_static_1 ();
+
+  NTHR = 10;
+  S = 1, E = 9, INCR = 1, CHUNK = 0;
+  t_static_1 ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/ordered-2.c b/libgomp/testsuite/libgomp.c/ordered-2.c
new file mode 100644
index 000000000..91564371a
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/ordered-2.c
@@ -0,0 +1,82 @@
+/* Trivial test of ordered.  */
+
+/* { dg-require-effective-target sync_int_long } */
+
+#include <omp.h>
+#include <string.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+#define N 100
+static int next;
+static int CHUNK, NTHR;
+
+static void clean_data (void)
+{
+  next = 0;
+}
+
+static void set_data (long i)
+{
+  int n = __sync_fetch_and_add (&next, 1);
+  assert (n == i);
+}
+
+
+#define TMPL_1(sched)							\
+static void f_##sched##_1 (void *dummy)					\
+{									\
+  long s0, e0, i;							\
+  if (GOMP_loop_ordered_##sched##_start (0, N, 1, CHUNK, &s0, &e0))	\
+    do									\
+      {									\
+	for (i = s0; i < e0; ++i)					\
+	  {								\
+	    GOMP_ordered_start ();					\
+	    set_data (i);						\
+	    GOMP_ordered_end ();					\
+	  }								\
+      }									\
+    while (GOMP_loop_ordered_##sched##_next (&s0, &e0));		\
+  GOMP_loop_end ();							\
+}									\
+static void t_##sched##_1 (void)					\
+{									\
+  clean_data ();							\
+  GOMP_parallel_start (f_##sched##_1, NULL, NTHR);			\
+  f_##sched##_1 (NULL);							\
+  GOMP_parallel_end ();							\
+}
+
+TMPL_1(static)
+TMPL_1(dynamic)
+TMPL_1(guided)
+
+static void test (void)
+{
+  t_static_1 ();
+  t_dynamic_1 ();
+  t_guided_1 ();
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  NTHR = 4;
+
+  CHUNK = 1;
+  test ();
+
+  CHUNK = 5;
+  test ();
+
+  CHUNK = 7;
+  test ();
+
+  CHUNK = 0;
+  t_static_1 ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/ordered-3.c b/libgomp/testsuite/libgomp.c/ordered-3.c
new file mode 100644
index 000000000..2a2f21970
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/ordered-3.c
@@ -0,0 +1,82 @@
+#include <stdlib.h>
+
+int cnt;
+
+void
+check (int x)
+{
+  if (cnt++ != x)
+    abort ();
+}
+
+int
+main (void)
+{
+  int j;
+
+  cnt = 0;
+#pragma omp parallel for ordered schedule (static, 1) num_threads (4) if (0)
+  for (j = 0; j < 1000; j++)
+    {
+#pragma omp ordered
+      check (j);
+    }
+
+  cnt = 0;
+#pragma omp parallel for ordered schedule (static, 1) num_threads (4) if (1)
+  for (j = 0; j < 1000; j++)
+    {
+#pragma omp ordered
+      check (j);
+    }
+
+  cnt = 0;
+#pragma omp parallel for ordered schedule (runtime) num_threads (4) if (0)
+  for (j = 0; j < 1000; j++)
+    {
+#pragma omp ordered
+      check (j);
+    }
+
+  cnt = 0;
+#pragma omp parallel for ordered schedule (runtime) num_threads (4) if (1)
+  for (j = 0; j < 1000; j++)
+    {
+#pragma omp ordered
+      check (j);
+    }
+
+  cnt = 0;
+#pragma omp parallel for ordered schedule (dynamic) num_threads (4) if (0)
+  for (j = 0; j < 1000; j++)
+    {
+#pragma omp ordered
+      check (j);
+    }
+
+  cnt = 0;
+#pragma omp parallel for ordered schedule (dynamic) num_threads (4) if (1)
+  for (j = 0; j < 1000; j++)
+    {
+#pragma omp ordered
+      check (j);
+    }
+
+  cnt = 0;
+#pragma omp parallel for ordered schedule (guided) num_threads (4) if (0)
+  for (j = 0; j < 1000; j++)
+    {
+#pragma omp ordered
+      check (j);
+    }
+
+  cnt = 0;
+#pragma omp parallel for ordered schedule (guided) num_threads (4) if (1)
+  for (j = 0; j < 1000; j++)
+    {
+#pragma omp ordered
+      check (j);
+    }
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/parallel-1.c b/libgomp/testsuite/libgomp.c/parallel-1.c
new file mode 100644
index 000000000..031f5bf88
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/parallel-1.c
@@ -0,0 +1,48 @@
+/* Trivial test of thread startup.  */
+
+#include <omp.h>
+#include <string.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+static int nthr;
+static int saw[4];
+
+static void function(void *dummy)
+{
+  int iam = omp_get_thread_num ();
+
+  if (iam == 0)
+    nthr = omp_get_num_threads ();
+
+  saw[iam] = 1;
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  GOMP_parallel_start (function, NULL, 2);
+  function (NULL);
+  GOMP_parallel_end ();
+
+  assert (nthr == 2);
+  assert (saw[0] != 0);
+  assert (saw[1] != 0);
+  assert (saw[2] == 0);
+
+  memset (saw, 0, sizeof (saw));
+  
+  GOMP_parallel_start (function, NULL, 3);
+  function (NULL);
+  GOMP_parallel_end ();
+
+  assert (nthr == 3);
+  assert (saw[0] != 0);
+  assert (saw[1] != 0);
+  assert (saw[2] != 0);
+  assert (saw[3] == 0);
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr24455-1.c b/libgomp/testsuite/libgomp.c/pr24455-1.c
new file mode 100644
index 000000000..c39068f80
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr24455-1.c
@@ -0,0 +1,6 @@
+/* { dg-do compile } */
+/* { dg-require-effective-target tls } */
+extern int i;
+#pragma omp threadprivate (i)
+
+int i;
diff --git a/libgomp/testsuite/libgomp.c/pr24455.c b/libgomp/testsuite/libgomp.c/pr24455.c
new file mode 100644
index 000000000..8af449e7b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr24455.c
@@ -0,0 +1,23 @@
+/* { dg-do run } */
+/* { dg-additional-sources pr24455-1.c } */
+/* { dg-require-effective-target tls_runtime } */
+
+extern void abort (void);
+
+extern int i;
+#pragma omp threadprivate(i)
+
+int main()
+{
+  i = 0;
+
+#pragma omp parallel default(none) num_threads(10)
+    {
+      i++;
+#pragma omp barrier
+      if (i != 1)
+	abort ();
+    }
+
+    return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr26171.c b/libgomp/testsuite/libgomp.c/pr26171.c
new file mode 100644
index 000000000..eacc9a71d
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr26171.c
@@ -0,0 +1,14 @@
+/* PR c/26171 */
+/* { dg-do run } */
+/* { dg-options "-fopenmp" } */
+/* { dg-require-effective-target tls_runtime } */
+
+int thrv = 0;
+#pragma omp threadprivate (thrv)
+
+int
+main ()
+{
+  thrv = 1;
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr26943-1.c b/libgomp/testsuite/libgomp.c/pr26943-1.c
new file mode 100644
index 000000000..86c43f04b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr26943-1.c
@@ -0,0 +1,24 @@
+/* PR c++/26943 */
+/* { dg-do run } */
+
+extern void abort (void);
+extern void omp_set_dynamic (int);
+int n = 6;
+
+int
+main (void)
+{
+  int i, x = 0;
+  omp_set_dynamic (0);
+#pragma omp parallel for num_threads (16) firstprivate (n) lastprivate (n) \
+			 schedule (static, 1) reduction (+: x)
+  for (i = 0; i < 16; i++)
+    {
+      if (n != 6)
+	++x;
+      n = i;
+    }
+  if (x || n != 15)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr26943-2.c b/libgomp/testsuite/libgomp.c/pr26943-2.c
new file mode 100644
index 000000000..c052e8112
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr26943-2.c
@@ -0,0 +1,47 @@
+/* PR c++/26943 */
+/* { dg-do run } */
+
+extern int omp_set_dynamic (int);
+extern void abort (void);
+
+int a = 8, b = 12, c = 16, d = 20, j = 0;
+char e[10] = "a", f[10] = "b", g[10] = "c", h[10] = "d";
+
+int
+main (void)
+{
+  int i;
+  omp_set_dynamic (0);
+#pragma omp parallel for shared (a, e) firstprivate (b, f) \
+			 lastprivate (c, g) private (d, h) \
+			 schedule (static, 1) num_threads (4) \
+			 reduction (+:j)
+  for (i = 0; i < 4; i++)
+    {
+      if (a != 8 || b != 12 || e[0] != 'a' || f[0] != 'b')
+	j++;
+#pragma omp barrier	/* { dg-warning "may not be closely nested" } */
+#pragma omp atomic
+      a += i;
+      b += i;
+      c = i;
+      d = i;
+#pragma omp atomic
+      e[0] += i;
+      f[0] += i;
+      g[0] = 'g' + i;
+      h[0] = 'h' + i;
+#pragma omp barrier	/* { dg-warning "may not be closely nested" } */
+      if (a != 8 + 6 || b != 12 + i || c != i || d != i)
+	j += 8;
+      if (e[0] != 'a' + 6 || f[0] != 'b' + i || g[0] != 'g' + i)
+	j += 64;
+      if (h[0] != 'h' + i)
+	j += 512;
+    }
+  if (j || a != 8 + 6 || b != 12 || c != 3 || d != 20)
+    abort ();
+  if (e[0] != 'a' + 6 || f[0] != 'b' || g[0] != 'g' + 3 || h[0] != 'd')
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr26943-3.c b/libgomp/testsuite/libgomp.c/pr26943-3.c
new file mode 100644
index 000000000..dc3d5010d
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr26943-3.c
@@ -0,0 +1,56 @@
+/* PR c++/26943 */
+/* { dg-do run } */
+
+extern int omp_set_dynamic (int);
+extern int omp_get_thread_num (void);
+extern void abort (void);
+
+int a = 8, b = 12, c = 16, d = 20, j = 0, l = 0;
+char e[10] = "a", f[10] = "b", g[10] = "c", h[10] = "d";
+volatile int k;
+
+int
+main (void)
+{
+  int i;
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+#pragma omp parallel num_threads (2) reduction (+:l)
+  if (k == omp_get_thread_num ())
+    {
+#pragma omp parallel for shared (a, e) firstprivate (b, f) \
+			 lastprivate (c, g) private (d, h) \
+			 schedule (static, 1) num_threads (4) \
+			 reduction (+:j)
+      for (i = 0; i < 4; i++)
+	{
+	  if (a != 8 || b != 12 || e[0] != 'a' || f[0] != 'b')
+	    j++;
+#pragma omp barrier	/* { dg-warning "may not be closely nested" } */
+#pragma omp atomic
+	  a += i;
+	  b += i;
+	  c = i;
+	  d = i;
+#pragma omp atomic
+	  e[0] += i;
+	  f[0] += i;
+	  g[0] = 'g' + i;
+	  h[0] = 'h' + i;
+#pragma omp barrier	/* { dg-warning "may not be closely nested" } */
+	  if (a != 8 + 6 || b != 12 + i || c != i || d != i)
+	    j += 8;
+	  if (e[0] != 'a' + 6 || f[0] != 'b' + i || g[0] != 'g' + i)
+	    j += 64;
+	  if (h[0] != 'h' + i)
+	    j += 512;
+	}
+      if (j || a != 8 + 6 || b != 12 || c != 3 || d != 20)
+	++l;
+      if (e[0] != 'a' + 6 || f[0] != 'b' || g[0] != 'g' + 3 || h[0] != 'd')
+	l += 8;
+    }
+  if (l)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr26943-4.c b/libgomp/testsuite/libgomp.c/pr26943-4.c
new file mode 100644
index 000000000..0f1d4197a
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr26943-4.c
@@ -0,0 +1,61 @@
+/* PR c++/26943 */
+/* { dg-do run } */
+
+extern int omp_set_dynamic (int);
+extern int omp_get_thread_num (void);
+extern void abort (void);
+
+int a = 8, b = 12, c = 16, d = 20, j = 0, l = 0;
+char e[10] = "a", f[10] = "b", g[10] = "c", h[10] = "d";
+volatile int k;
+
+int
+main (void)
+{
+  int i;
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+#pragma omp parallel num_threads (2) reduction (+:l) \
+		     firstprivate (a, b, c, d, e, f, g, h, j)
+  if (k == omp_get_thread_num ())
+    {
+#pragma omp parallel for shared (a, e) firstprivate (b, f) \
+			 lastprivate (c, g) private (d, h) \
+			 schedule (static, 1) num_threads (4) \
+			 reduction (+:j)
+      for (i = 0; i < 4; i++)
+	{
+	  if (a != 8 || b != 12 || e[0] != 'a' || f[0] != 'b')
+	    j++;
+#pragma omp barrier	/* { dg-warning "may not be closely nested" } */
+#pragma omp atomic
+	  a += i;
+	  b += i;
+	  c = i;
+	  d = i;
+#pragma omp atomic
+	  e[0] += i;
+	  f[0] += i;
+	  g[0] = 'g' + i;
+	  h[0] = 'h' + i;
+#pragma omp barrier	/* { dg-warning "may not be closely nested" } */
+	  if (a != 8 + 6 || b != 12 + i || c != i || d != i)
+	    j += 8;
+	  if (e[0] != 'a' + 6 || f[0] != 'b' + i || g[0] != 'g' + i)
+	    j += 64;
+	  if (h[0] != 'h' + i)
+	    j += 512;
+	}
+      if (j || a != 8 + 6 || b != 12 || c != 3 || d != 20)
+	++l;
+      if (e[0] != 'a' + 6 || f[0] != 'b' || g[0] != 'g' + 3 || h[0] != 'd')
+	l += 8;
+    }
+  if (l)
+    abort ();
+  if (a != 8 || b != 12 || c != 16 || d != 20)
+    abort ();
+  if (e[0] != 'a' || f[0] != 'b' || g[0] != 'c' || h[0] != 'd')
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr29947-1.c b/libgomp/testsuite/libgomp.c/pr29947-1.c
new file mode 100644
index 000000000..509c63229
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr29947-1.c
@@ -0,0 +1,328 @@
+/* PR libgomp/29947 */
+
+/* { dg-do run } */
+
+extern void abort (void);
+
+int cnt;
+
+void
+test1 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (dynamic)
+    for (i = j1; i <= k1; ++i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test2 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (dynamic)
+    for (i = k1; i >= j1; --i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test3 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (guided)
+    for (i = j1; i <= k1; ++i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test4 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (guided)
+    for (i = k1; i >= j1; --i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test5 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (dynamic) ordered
+    for (i = j1; i <= k1; ++i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+#pragma omp ordered
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test6 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (dynamic) ordered
+    for (i = k1; i >= j1; --i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+#pragma omp ordered
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test7 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (guided) ordered
+    for (i = j1; i <= k1; ++i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+#pragma omp ordered
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test8 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (guided) ordered
+    for (i = k1; i >= j1; --i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+#pragma omp ordered
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test9 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (dynamic)
+  for (i = j1; i <= k1; ++i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test10 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (dynamic)
+  for (i = k1; i >= j1; --i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test11 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (guided)
+  for (i = j1; i <= k1; ++i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test12 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (guided)
+  for (i = k1; i >= j1; --i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test13 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (dynamic) ordered
+  for (i = j1; i <= k1; ++i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+#pragma omp ordered
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test14 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (dynamic) ordered
+  for (i = k1; i >= j1; --i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+#pragma omp ordered
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test15 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (guided) ordered
+  for (i = j1; i <= k1; ++i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+#pragma omp ordered
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test16 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (guided) ordered
+  for (i = k1; i >= j1; --i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+#pragma omp ordered
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+int
+__attribute__((noinline))
+test (long j1, long k1, long j2, long k2)
+{
+  test1 (j1, k1, j2, k2);
+  test2 (j1, k1, j2, k2);
+  test3 (j1, k1, j2, k2);
+  test4 (j1, k1, j2, k2);
+  test5 (j1, k1, j2, k2);
+  test6 (j1, k1, j2, k2);
+  test7 (j1, k1, j2, k2);
+  test8 (j1, k1, j2, k2);
+  test9 (j1, k1, j2, k2);
+  test10 (j1, k1, j2, k2);
+  test11 (j1, k1, j2, k2);
+  test12 (j1, k1, j2, k2);
+  test13 (j1, k1, j2, k2);
+  test14 (j1, k1, j2, k2);
+  test15 (j1, k1, j2, k2);
+  test16 (j1, k1, j2, k2);
+  return cnt;
+}
+
+int
+main (void)
+{
+  test (1, 5, 1, 5);
+  test (5, 5, 5, 5);
+  test (5, 4, 5, 4);
+  test (5, 1, 5, 1);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr29947-2.c b/libgomp/testsuite/libgomp.c/pr29947-2.c
new file mode 100644
index 000000000..097498311
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr29947-2.c
@@ -0,0 +1,328 @@
+/* PR libgomp/29947 */
+
+/* { dg-do run } */
+
+extern void abort (void);
+
+int cnt;
+
+void
+test1 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (static)
+    for (i = j1; i <= k1; ++i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test2 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (static)
+    for (i = k1; i >= j1; --i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test3 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (static, 1)
+    for (i = j1; i <= k1; ++i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test4 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (static, 1)
+    for (i = k1; i >= j1; --i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test5 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (static) ordered
+    for (i = j1; i <= k1; ++i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+#pragma omp ordered
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test6 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (static) ordered
+    for (i = k1; i >= j1; --i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+#pragma omp ordered
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test7 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (static, 1) ordered
+    for (i = j1; i <= k1; ++i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+#pragma omp ordered
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test8 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel reduction (+:e,c)
+  {
+#pragma omp for schedule (static, 1) ordered
+    for (i = k1; i >= j1; --i)
+      {
+	if (i < j2 || i > k2)
+	  ++e;
+#pragma omp ordered
+	++c;
+      }
+#pragma omp atomic
+    ++cnt;
+  }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test9 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (static)
+  for (i = j1; i <= k1; ++i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test10 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (static)
+  for (i = k1; i >= j1; --i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test11 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (static, 1)
+  for (i = j1; i <= k1; ++i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test12 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (static, 1)
+  for (i = k1; i >= j1; --i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test13 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (static) ordered
+  for (i = j1; i <= k1; ++i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+#pragma omp ordered
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test14 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (static) ordered
+  for (i = k1; i >= j1; --i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+#pragma omp ordered
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test15 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (static, 1) ordered
+  for (i = j1; i <= k1; ++i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+#pragma omp ordered
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+void
+test16 (long j1, long k1, long j2, long k2)
+{
+  long i, e = 0, c = 0;
+#pragma omp parallel for reduction (+:e,c) schedule (static, 1) ordered
+  for (i = k1; i >= j1; --i)
+    {
+      if (i < j2 || i > k2)
+	++e;
+#pragma omp ordered
+      ++c;
+    }
+  if (e || (c != j2 > k2 ? 0 : k2 - j2 + 1))
+    abort ();
+}
+
+int
+__attribute__((noinline))
+test (long j1, long k1, long j2, long k2)
+{
+  test1 (j1, k1, j2, k2);
+  test2 (j1, k1, j2, k2);
+  test3 (j1, k1, j2, k2);
+  test4 (j1, k1, j2, k2);
+  test5 (j1, k1, j2, k2);
+  test6 (j1, k1, j2, k2);
+  test7 (j1, k1, j2, k2);
+  test8 (j1, k1, j2, k2);
+  test9 (j1, k1, j2, k2);
+  test10 (j1, k1, j2, k2);
+  test11 (j1, k1, j2, k2);
+  test12 (j1, k1, j2, k2);
+  test13 (j1, k1, j2, k2);
+  test14 (j1, k1, j2, k2);
+  test15 (j1, k1, j2, k2);
+  test16 (j1, k1, j2, k2);
+  return cnt;
+}
+
+int
+main (void)
+{
+  test (1, 5, 1, 5);
+  test (5, 5, 5, 5);
+  test (5, 4, 5, 4);
+  test (5, 1, 5, 1);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr30494.c b/libgomp/testsuite/libgomp.c/pr30494.c
new file mode 100644
index 000000000..ec6828e44
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr30494.c
@@ -0,0 +1,64 @@
+/* PR middle-end/30494 */
+/* { dg-do run } */
+
+#include <omp.h>
+
+int errors;
+
+int
+check (int m, int i, int *v, int *w)
+{
+  int j;
+  int n = omp_get_thread_num ();
+  for (j = 0; j < m; j++)
+    if (v[j] != j + n)
+      #pragma omp atomic
+	errors += 1;
+  for (j = 0; j < m * 3 + i; j++)
+    if (w[j] != j + 10 + n)
+      #pragma omp atomic
+	errors += 1;
+}
+
+int
+foo (int n, int m)
+{
+  int i;
+#pragma omp for
+  for (i = 0; i < 6; i++)
+    {
+      int v[n], w[n * 3 + i], j;
+      for (j = 0; j < n; j++)
+	v[j] = j + omp_get_thread_num ();
+      for (j = 0; j < n * 3 + i; j++)
+	w[j] = j + 10 + omp_get_thread_num ();
+      check (m, i, v, w);
+    }
+  return 0;
+}
+
+int
+bar (int n, int m)
+{
+  int i;
+#pragma omp parallel for num_threads (4)
+  for (i = 0; i < 6; i++)
+    {
+      int v[n], w[n * 3 + i], j;
+      for (j = 0; j < n; j++)
+	v[j] = j + omp_get_thread_num ();
+      for (j = 0; j < n * 3 + i; j++)
+	w[j] = j + 10 + omp_get_thread_num ();
+      check (m, i, v, w);
+    }
+  return 0;
+}
+
+int
+main (void)
+{
+#pragma omp parallel num_threads (3)
+  foo (128, 128);
+  bar (256, 256);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr32362-1.c b/libgomp/testsuite/libgomp.c/pr32362-1.c
new file mode 100644
index 000000000..3c62d4bdb
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr32362-1.c
@@ -0,0 +1,32 @@
+/* PR middle-end/32362 */
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main ()
+{
+  int n[4] = { -1, -1, -1, -1 };
+  static int a = 2, b = 4;
+  omp_set_num_threads (4);
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+#pragma omp parallel private(b)
+  {
+    b = omp_get_thread_num ();
+#pragma omp parallel firstprivate(a)
+    {
+      a = (omp_get_thread_num () + a) + 1;
+      if (b == omp_get_thread_num ())
+	n[omp_get_thread_num ()] = a + (b << 4);
+    }
+  }
+  if (n[0] != 3)
+    abort ();
+  if (n[3] != -1
+      && (n[1] != 0x14 || n[2] != 0x25 || n[3] != 0x36))
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr32362-2.c b/libgomp/testsuite/libgomp.c/pr32362-2.c
new file mode 100644
index 000000000..43f36e0e9
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr32362-2.c
@@ -0,0 +1,33 @@
+/* PR middle-end/32362 */
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int a = 2, b = 4;
+
+int
+main ()
+{
+  int n[4] = { -1, -1, -1, -1 };
+  omp_set_num_threads (4);
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+#pragma omp parallel private(b)
+  {
+    b = omp_get_thread_num ();
+#pragma omp parallel firstprivate(a)
+    {
+      a = (omp_get_thread_num () + a) + 1;
+      if (b == omp_get_thread_num ())
+	n[omp_get_thread_num ()] = a + (b << 4);
+    }
+  }
+  if (n[0] != 3)
+    abort ();
+  if (n[3] != -1
+      && (n[1] != 0x14 || n[2] != 0x25 || n[3] != 0x36))
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr32362-3.c b/libgomp/testsuite/libgomp.c/pr32362-3.c
new file mode 100644
index 000000000..09a88f52a
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr32362-3.c
@@ -0,0 +1,34 @@
+/* PR middle-end/32362 */
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int a = 2;
+
+int
+main ()
+{
+  int n[4] = { -1, -1, -1, -1 };
+  int b = 4;
+  omp_set_num_threads (4);
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+#pragma omp parallel private(b)
+  {
+    b = omp_get_thread_num ();
+#pragma omp parallel firstprivate(a)
+    {
+      a = (omp_get_thread_num () + a) + 1;
+      if (b == omp_get_thread_num ())
+	n[omp_get_thread_num ()] = a + (b << 4);
+    }
+  }
+  if (n[0] != 3)
+    abort ();
+  if (n[3] != -1
+      && (n[1] != 0x14 || n[2] != 0x25 || n[3] != 0x36))
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr32468.c b/libgomp/testsuite/libgomp.c/pr32468.c
new file mode 100644
index 000000000..f20f660bf
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr32468.c
@@ -0,0 +1,26 @@
+/* PR libgomp/32468 */
+/* { dg-do run } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int res[2] = { -1, -1 };
+  omp_set_dynamic (0);
+  omp_set_num_threads (4);
+#pragma omp parallel
+  {
+    #pragma omp sections
+      {
+	#pragma omp section
+	res[0] = omp_get_num_threads () != 4;
+	#pragma omp section
+	res[1] = omp_get_num_threads () != 4;
+      }
+  }
+  if (res[0] != 0 || res[1] != 0)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr33880.c b/libgomp/testsuite/libgomp.c/pr33880.c
new file mode 100644
index 000000000..5d719cd63
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr33880.c
@@ -0,0 +1,123 @@
+/* PR middle-end/33880 */
+/* { dg-do run } */
+
+extern void abort (void);
+
+void
+test1 (void)
+{
+  int i = 0, j = 0;
+  void bar (void)
+  {
+    i++;
+    j++;
+  }
+  bar ();
+  #pragma omp parallel for num_threads(4)
+    for (i = 0; i < 100; i++)
+      #pragma omp atomic
+	j += 1;
+  if (j != 101)
+    abort ();
+  #pragma omp parallel for lastprivate(i) num_threads(2)
+    for (i = 0; i < 100; i++)
+      #pragma omp atomic
+	j += 1;
+  if (i != 100)
+    abort ();
+  i = 3;
+  bar ();
+  if (j != 202)
+    abort ();
+  if (i != 4)
+    abort ();
+}
+
+void
+test2 (void)
+{
+  int i = -1, j = 99, k, l = 9, m = 0;
+  void bar (void)
+  {
+    i++;
+    j++;
+    l++;
+    m++;
+  }
+  bar ();
+  #pragma omp parallel for num_threads(4)
+    for (k = i; k < j; k += l)
+      #pragma omp atomic
+	m += 1;
+  bar ();
+  if (i != 1 || j != 101 || l != 11 || m != 12)
+    abort ();
+}
+
+void
+test3 (void)
+{
+  int i, j, k, l, m;
+  void bar (void)
+  {
+  #pragma omp parallel for num_threads(4)
+    for (i = j; i < k; i += l)
+      #pragma omp atomic
+        m += 1;
+  }
+  void baz (void)
+  {
+  #pragma omp parallel for num_threads(2) lastprivate(i)
+    for (i = j; i < k * 2; i += l / 2)
+      #pragma omp atomic
+        m += 1;
+  }
+  i = 7;
+  j = 0;
+  k = 100;
+  l = 2;
+  m = 0;
+  bar ();
+  if (j != 0 || k != 100 || l != 2 || m != 50)
+    abort ();
+  baz ();
+  if (i != 200 || j != 0 || k != 100 || l != 2 || m != 250)
+    abort ();
+}
+
+void
+test4 (void)
+{
+  int i, j, k, l, m = 0;
+  int foo (void)
+  {
+    return j;
+  }
+  int bar (void)
+  {
+    return k;
+  }
+  int baz (void)
+  {
+    return l;
+  }
+  j = 0;
+  k = 1000;
+  l = 2;
+  #pragma omp parallel for num_threads(8) lastprivate(i)
+  for (i = foo (); i < bar (); i += baz ())
+    #pragma omp atomic
+      m += 1;
+  if (i != 1000 || m != 500 || j != 0 || k != 1000 || l != 2)
+    abort ();
+}
+
+int
+main (void)
+{
+  test1 ();
+  test2 ();
+  test3 ();
+  test4 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr34513.c b/libgomp/testsuite/libgomp.c/pr34513.c
new file mode 100644
index 000000000..76c7ac1f3
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr34513.c
@@ -0,0 +1,33 @@
+/* PR c++/34513 */
+/* { dg-do run } */
+
+#include <omp.h>
+
+extern void abort ();
+
+static int errors = 0;
+static int thrs = 4;
+
+int
+main ()
+{
+  omp_set_dynamic (0);
+
+  #pragma omp parallel num_threads (thrs)
+  {
+    static int shrd = 0;
+
+    #pragma omp atomic
+      shrd += 1;
+
+    #pragma omp barrier
+
+    if (shrd != thrs)
+      #pragma omp atomic
+	errors += 1;
+  }
+
+  if (errors)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr35130.c b/libgomp/testsuite/libgomp.c/pr35130.c
new file mode 100644
index 000000000..12962d807
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr35130.c
@@ -0,0 +1,131 @@
+/* PR middle-end/35130 */
+
+extern void abort (void);
+
+void
+f1 (void)
+{
+  int a[4], k;
+  void nested (int x)
+  {
+    a[x] = 42;
+  }
+
+  for (k = 0; k < 4; k++)
+    a[k] = 0;
+#pragma omp parallel for
+  for (k = 0; k < 4; k++)
+    nested (k);
+
+  if (a[0] != 42 || a[1] != 42 || a[2] != 42 || a[3] != 42)
+    abort ();
+}
+
+void
+f2 (void)
+{
+  int a[4], k;
+  void nested (void)
+  {
+    int l;
+    void nested2 (int x)
+    {
+      a[x] = 42;
+    }
+#pragma omp parallel for
+    for (l = 0; l < 4; l++)
+      nested2 (l);
+  }
+
+  for (k = 0; k < 4; k++)
+    a[k] = 0;
+
+  nested ();
+
+  if (a[0] != 42 || a[1] != 42 || a[2] != 42 || a[3] != 42)
+    abort ();
+}
+
+void
+f3 (void)
+{
+  int a[4], b[4], c[4], k;
+  void nested (int x)
+  {
+    a[x] = b[x] = c[x] = 42;
+  }
+
+  for (k = 0; k < 4; k++)
+    a[k] = b[k] = c[k] = 0;
+  nested (0);
+
+#pragma omp parallel
+  {
+  #pragma omp single
+    {
+      a[1] = 43;
+      b[1] = 43;
+    }
+  #pragma omp parallel
+    {
+    #pragma omp single
+      {
+	b[2] = 44;
+	c[2] = 44;
+      }
+    }
+  }
+
+  if (a[0] != 42 || a[1] != 43 || a[2] != 0 || a[3] != 0)
+    abort ();
+  if (b[0] != 42 || b[1] != 43 || b[2] != 44 || b[3] != 0)
+    abort ();
+  if (c[0] != 42 || c[1] != 0 || c[2] != 44 || c[3] != 0)
+    abort ();
+}
+
+void
+f4 (void)
+{
+  int a[4], b[4], c[4], k;
+  void nested ()
+  {
+  #pragma omp parallel
+    {
+    #pragma omp single
+      {
+	a[1] = 43;
+	b[1] = 43;
+      }
+    #pragma omp parallel
+      {
+      #pragma omp single
+	{
+	  b[2] = 44;
+	  c[2] = 44;
+	}
+      }
+    }
+  }
+
+  for (k = 0; k < 4; k++)
+    a[k] = b[k] = c[k] = k == 0 ? 42 : 0;
+  nested ();
+
+  if (a[0] != 42 || a[1] != 43 || a[2] != 0 || a[3] != 0)
+    abort ();
+  if (b[0] != 42 || b[1] != 43 || b[2] != 44 || b[3] != 0)
+    abort ();
+  if (c[0] != 42 || c[1] != 0 || c[2] != 44 || c[3] != 0)
+    abort ();
+}
+
+int
+main (void)
+{
+  f1 ();
+  f2 ();
+  f3 ();
+  f4 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr35196.c b/libgomp/testsuite/libgomp.c/pr35196.c
new file mode 100644
index 000000000..e92d97629
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr35196.c
@@ -0,0 +1,43 @@
+/* PR middle-end/35196 */
+/* { dg-do run } */
+
+extern void abort (void);
+extern void omp_set_dynamic (int);
+
+int
+main (void)
+{
+  int i, j;
+  omp_set_dynamic (0);
+#pragma omp parallel for lastprivate (i, j) num_threads (8) schedule (static)
+  for (i = 0; i < 5; i++)
+    j = i;
+  if (i != 5 || j != 4)
+    abort ();
+#pragma omp parallel for lastprivate (i, j) num_threads (8) schedule (static, 2)
+  for (i = 0; i < 5; i++)
+    j = i;
+  if (i != 5 || j != 4)
+    abort ();
+#pragma omp parallel for lastprivate (i, j) num_threads (8) schedule (dynamic)
+  for (i = 0; i < 5; i++)
+    j = i;
+  if (i != 5 || j != 4)
+    abort ();
+#pragma omp parallel for lastprivate (i, j) num_threads (8) schedule (static)
+  for (i = -12; i < 21; i += 3)
+    j = i;
+  if (i != 21 || j != 18)
+    abort ();
+#pragma omp parallel for lastprivate (i, j) num_threads (8) schedule (static, 2)
+  for (i = -12; i < 21; i += 3)
+    j = i;
+  if (i != 21 || j != 18)
+    abort ();
+#pragma omp parallel for lastprivate (i, j) num_threads (8) schedule (dynamic, 3)
+  for (i = -12; i < 21; i += 3)
+    j = i;
+  if (i != 21 || j != 18)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr35549.c b/libgomp/testsuite/libgomp.c/pr35549.c
new file mode 100644
index 000000000..269a0c262
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr35549.c
@@ -0,0 +1,30 @@
+/* PR middle-end/35549 */
+/* { dg-do run } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i = 6, n = 0;
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+  #pragma omp parallel shared (i) num_threads (3)
+  {
+    if (omp_get_num_threads () != 3)
+      #pragma omp atomic
+	n += 1;
+    #pragma omp parallel shared (i) num_threads (4)
+    {
+      if (omp_get_num_threads () != 4)
+	#pragma omp atomic
+	  n += 1;
+      #pragma omp critical
+	i += 1;
+    }
+  }
+  if (n == 0 && i != 6 + 3 * 4)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr35625.c b/libgomp/testsuite/libgomp.c/pr35625.c
new file mode 100644
index 000000000..f2978f911
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr35625.c
@@ -0,0 +1,18 @@
+/* PR libgomp/35625 */
+/* { dg-do run } */
+/* { dg-options "-std=c99" } */
+
+int
+main (void)
+{
+#pragma omp parallel
+  {
+    #pragma omp for schedule (guided, 10)
+      for (int i = 0; i < 1826; i += 10)
+	;
+    #pragma omp for schedule (guided, 10)
+      for (int i = 0; i > -1826; i -= 10)
+	;
+  }
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr36802-1.c b/libgomp/testsuite/libgomp.c/pr36802-1.c
new file mode 100644
index 000000000..4ed5e1276
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr36802-1.c
@@ -0,0 +1,34 @@
+/* PR middle-end/36802 */
+
+extern void abort (void);
+
+int
+foo (int k)
+{
+  int i = 0;
+#pragma omp parallel
+  #pragma omp single
+    {
+      if (!k)
+	{
+	  int j;
+	  for (j = 0; j < 10; j++)
+	  #pragma omp task
+	    if (j == 4)  
+	      i++;
+	}
+      else
+	i++;
+    }
+  return i;
+}
+
+int
+main (void)
+{
+  if (foo (0) != 1)
+    abort ();
+  if (foo (1) != 1)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr36802-2.c b/libgomp/testsuite/libgomp.c/pr36802-2.c
new file mode 100644
index 000000000..06e792f0f
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr36802-2.c
@@ -0,0 +1,46 @@
+/* PR middle-end/36802 */
+
+extern void abort (void);
+
+int q;
+
+int
+foo (int k)
+{
+  int i = 6, n = 0;
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+#pragma omp parallel shared (i) num_threads (3)
+  {
+    int l;
+
+    if (omp_get_num_threads () != 3)
+    #pragma omp atomic
+      n += 1;
+    else
+    #pragma omp for
+      for (l = 0; l < 3; l++)
+	if (k)
+	#pragma omp atomic
+	  q += i;
+	else
+	#pragma omp parallel shared (i) num_threads (4)
+	  {
+	    if (omp_get_num_threads () != 4)
+	    #pragma omp atomic
+	      n += 1;
+	    #pragma omp critical
+	      i += 1;
+	  }
+  }
+  if (n == 0 && i != 6 + 3 * 4)
+    abort ();
+  return 0;
+}
+
+int
+main (void)
+{
+  foo (0);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr36802-3.c b/libgomp/testsuite/libgomp.c/pr36802-3.c
new file mode 100644
index 000000000..f11baa09f
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr36802-3.c
@@ -0,0 +1,46 @@
+/* PR middle-end/36802 */
+
+extern void abort (void);
+
+int q;
+
+int
+foo (int k)
+{
+  int i = 6, n = 0;
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+#pragma omp parallel shared (i) num_threads (3)
+  {
+    int l;
+
+    if (omp_get_num_threads () != 3)
+    #pragma omp atomic
+      n += 1;
+    else
+    #pragma omp for
+      for (l = 0; l < 3; l++)
+	if (!k)
+	#pragma omp parallel shared (i) num_threads (4)
+	  {
+	    if (omp_get_num_threads () != 4)
+	    #pragma omp atomic
+	      n += 1;
+	    #pragma omp critical
+	      i += 1;
+	  }
+	else
+	#pragma omp atomic
+	  q += i;
+  }
+  if (n == 0 && i != 6 + 3 * 4)
+    abort ();
+  return 0;
+}
+
+int
+main (void)
+{
+  foo (0);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr38650.c b/libgomp/testsuite/libgomp.c/pr38650.c
new file mode 100644
index 000000000..7066239b0
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr38650.c
@@ -0,0 +1,49 @@
+/* PR c++/38650 */
+/* { dg-do run } */
+
+#include <stdlib.h>
+
+int e;
+
+int
+main ()
+{
+  volatile int i, j = 10;
+  e = 0;
+#pragma omp parallel for reduction(+:e)
+  for (i = 0; i < j; i += 1)
+    e++;
+  if (e != 10)
+    abort ();
+  e = 0;
+#pragma omp parallel for reduction(+:e)
+  for (i = 0; i < j; ++i)
+    e++;
+  if (e != 10)
+    abort ();
+  e = 0;
+#pragma omp parallel for reduction(+:e)
+  for (i = 0; i < j; i++)
+    e++;
+  if (e != 10)
+    abort ();
+  e = 0;
+#pragma omp parallel for reduction(+:e)
+  for (i = 0; i < 10; i += 1)
+    e++;
+  if (e != 10)
+    abort ();
+  e = 0;
+#pragma omp parallel for reduction(+:e)
+  for (i = 0; i < 10; ++i)
+    e++;
+  if (e != 10)
+    abort ();
+  e = 0;
+#pragma omp parallel for reduction(+:e)
+  for (i = 0; i < 10; i++)
+    e++;
+  if (e != 10)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr39154.c b/libgomp/testsuite/libgomp.c/pr39154.c
new file mode 100644
index 000000000..5a4c89e13
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr39154.c
@@ -0,0 +1,105 @@
+/* PR middle-end/39154 */
+/* { dg-do compile } */
+/* { dg-options "-O2 -std=gnu99" } */
+
+extern void abort (void);
+
+int n = 20;
+
+int
+main (void)
+{
+  int a[n], b[n][n];
+
+#pragma omp parallel for
+    for (int i = 0; i < n; i++)
+      {
+	a[i] = i + 1;
+#pragma omp parallel for
+	for (int j = 0; j < n; j++)
+	  b[i][j] = a[i];
+      }
+
+  for (int i = 0; i < n; i++)
+    {
+      for (int j = 0; j < n; j++)
+	if (b[i][j] != i + 1)
+	  abort ();
+      if (a[i] != i + 1)
+	abort ();
+    }
+
+#pragma omp parallel for shared (n, a, b)
+    for (int i = 0; i < n; i++)
+      {
+	a[i] = i + 3;
+#pragma omp parallel for
+	for (int j = 0; j < n; j++)
+	  b[i][j] = a[i];
+      }
+
+  for (int i = 0; i < n; i++)
+    {
+      for (int j = 0; j < n; j++)
+	if (b[i][j] != i + 3)
+	  abort ();
+      if (a[i] != i + 3)
+	abort ();
+    }
+
+#pragma omp parallel for
+    for (int i = 0; i < n; i++)
+      {
+	a[i] = i + 5;
+#pragma omp parallel for shared (n, a, b)
+	for (int j = 0; j < n; j++)
+	  b[i][j] = a[i];
+      }
+
+  for (int i = 0; i < n; i++)
+    {
+      for (int j = 0; j < n; j++)
+	if (b[i][j] != i + 5)
+	  abort ();
+      if (a[i] != i + 5)
+	abort ();
+    }
+
+#pragma omp parallel for shared (n, a, b)
+    for (int i = 0; i < n; i++)
+      {
+	a[i] = i + 7;
+#pragma omp parallel for shared (n, a, b)
+	for (int j = 0; j < n; j++)
+	  b[i][j] = a[i];
+      }
+
+  for (int i = 0; i < n; i++)
+    {
+      for (int j = 0; j < n; j++)
+	if (b[i][j] != i + 7)
+	  abort ();
+      if (a[i] != i + 7)
+	abort ();
+    }
+
+#pragma omp parallel for private (a, b)
+    for (int i = 0; i < n; i++)
+      {
+	a[i] = i + 1;
+#pragma omp parallel for
+	for (int j = 0; j < n; j++)
+	  b[i][j] = a[i];
+      }
+
+#pragma omp parallel for private (a, b)
+    for (int i = 0; i < n; i++)
+      {
+	a[i] = i + 1;
+#pragma omp parallel for private (b)
+	for (int j = 0; j < n; j++)
+	  b[i][j] = a[i];
+      }
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr39591-1.c b/libgomp/testsuite/libgomp.c/pr39591-1.c
new file mode 100644
index 000000000..dfd8d9e8a
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr39591-1.c
@@ -0,0 +1,33 @@
+/* PR other/39591 */
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+
+extern void abort (void);
+
+int err;
+
+int
+main (void)
+{
+#pragma omp parallel
+  {
+    int array[40];
+    int i;
+    for (i = 0; i < sizeof array / sizeof array[0]; i++)
+      array[i] = 0x55555555;
+
+#pragma omp for schedule(dynamic)
+    for (i = 0; i < 50; i++)
+#pragma omp task shared(array)
+      {
+	int j;
+	for (j = 0; j < sizeof array / sizeof array[0]; j++)
+	  if (array[j] != 0x55555555)
+#pragma omp atomic
+	    err++;
+      }
+  }
+  if (err)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr39591-2.c b/libgomp/testsuite/libgomp.c/pr39591-2.c
new file mode 100644
index 000000000..b5f8f9cc7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr39591-2.c
@@ -0,0 +1,39 @@
+/* PR other/39591 */
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+
+extern void abort (void);
+
+int err;
+
+void __attribute__((noinline))
+foo (int *array)
+{
+#pragma omp task
+  {
+    int j;
+    for (j = 0; j < sizeof array / sizeof array[0]; j++)
+      if (array[j] != 0x55555555)
+#pragma omp atomic
+	err++;
+  }
+}
+
+int
+main (void)
+{
+#pragma omp parallel
+  {
+    int array[40];
+    int i;
+    for (i = 0; i < sizeof array / sizeof array[0]; i++)
+      array[i] = 0x55555555;
+
+#pragma omp for schedule (dynamic)
+    for (i = 0; i < 50; i++)
+      foo (array);
+  }
+  if (err)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr39591-3.c b/libgomp/testsuite/libgomp.c/pr39591-3.c
new file mode 100644
index 000000000..a9aeea7c8
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr39591-3.c
@@ -0,0 +1,40 @@
+/* PR other/39591 */
+/* { dg-do run } */
+/* { dg-options "-O2" } */
+
+extern void abort (void);
+
+int err, a[40];
+
+void __attribute__((noinline))
+foo (int *array)
+{
+#pragma omp task
+  {
+    int j;
+    for (j = 0; j < sizeof array / sizeof array[0]; j++)
+      if (array[j] != 0x55555555)
+#pragma omp atomic
+	err++;
+  }
+}
+
+int
+main (void)
+{
+  int k;
+  for (k = 0; k < sizeof a / sizeof a[0]; k++)
+    a[k] = 0x55555555;
+
+#pragma omp parallel
+  {
+    int i;
+
+#pragma omp for schedule (dynamic)
+    for (i = 0; i < 50; i++)
+      foo (a);
+  }
+  if (err)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr42029.c b/libgomp/testsuite/libgomp.c/pr42029.c
new file mode 100644
index 000000000..ea7ac2c09
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr42029.c
@@ -0,0 +1,19 @@
+/* PR middle-end/42029 */
+/* { dg-do run } */
+
+extern void abort (void);
+
+int
+main ()
+{
+  int i;
+  _Complex int c = 0;
+
+#pragma omp parallel for private(i) reduction(+:c)
+  for (i = 0; i < 8; ++i)
+    c += 1;
+
+  if (c != 8)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr42942.c b/libgomp/testsuite/libgomp.c/pr42942.c
new file mode 100644
index 000000000..5d5785245
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr42942.c
@@ -0,0 +1,61 @@
+/* PR libgomp/42942 */
+/* { dg-do run } */
+
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int e = 0;
+  omp_set_dynamic (0);
+  omp_set_nested (1);
+  omp_set_max_active_levels (1);
+  if (omp_get_max_active_levels () != 1)
+    abort ();
+#pragma omp parallel num_threads(2) reduction(|:e)
+  if (!omp_in_parallel ()
+      || omp_get_num_threads () != 2)
+    e = 1;
+  else
+#pragma omp parallel num_threads(2) reduction(|:e)
+    if (!omp_in_parallel ()
+	|| omp_get_num_threads () != 1)
+      e = 1;
+  if (e)
+    abort ();
+  omp_set_max_active_levels (0);
+  if (omp_get_max_active_levels () != 0)
+    abort ();
+#pragma omp parallel num_threads(2) reduction(|:e)
+  if (omp_in_parallel ()
+      || omp_get_num_threads () != 1)
+    e = 1;
+  else
+#pragma omp parallel num_threads(2) reduction(|:e)
+    if (omp_in_parallel ()
+	|| omp_get_num_threads () != 1)
+      e = 1;
+  if (e)
+    abort ();
+  omp_set_max_active_levels (2);
+  if (omp_get_max_active_levels () != 2)
+    abort ();
+#pragma omp parallel num_threads(2) reduction(|:e)
+  if (!omp_in_parallel ()
+      || omp_get_num_threads () != 2)
+    e = 1;
+  else
+#pragma omp parallel num_threads(2) reduction(|:e)
+    if (!omp_in_parallel ()
+	|| omp_get_num_threads () != 2)
+      e = 1;
+    else
+#pragma omp parallel num_threads(2) reduction(|:e)
+      if (!omp_in_parallel ()
+	  || omp_get_num_threads () != 1)
+	e = 1;
+  if (e)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr43893.c b/libgomp/testsuite/libgomp.c/pr43893.c
new file mode 100644
index 000000000..b85e9242b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr43893.c
@@ -0,0 +1,61 @@
+/* PR c/43893 */
+/* { dg-do run } */
+
+extern void abort (void);
+
+int
+main ()
+{
+  int c;
+  unsigned int i;
+  int j;
+  c = 0;
+#pragma omp parallel for reduction(+:c)
+  for (i = 0; i < 1; i++)
+    c++;
+  if (c != 1)
+    abort ();
+  c = 0;
+#pragma omp parallel for reduction(+:c)
+  for (i = 0; i <= 0; i++)
+    c++;
+  if (c != 1)
+    abort ();
+  c = 0;
+#pragma omp parallel for reduction(+:c)
+  for (j = - __INT_MAX__ - 1; j < - __INT_MAX__; j++)
+    c++;
+  if (c != 1)
+    abort ();
+  c = 0;
+#pragma omp parallel for reduction(+:c)
+  for (j = - __INT_MAX__ - 1; j <= - __INT_MAX__ - 1; j++)
+    c++;
+  if (c != 1)
+    abort ();
+  c = 0;
+#pragma omp parallel for reduction(+:c)
+  for (i = 2U * __INT_MAX__ + 1; i > 2U * __INT_MAX__; i--)
+    c++;
+  if (c != 1)
+    abort ();
+  c = 0;
+#pragma omp parallel for reduction(+:c)
+  for (i = 2U * __INT_MAX__ + 1; i >= 2U * __INT_MAX__ + 1; i--)
+    c++;
+  if (c != 1)
+    abort ();
+  c = 0;
+#pragma omp parallel for reduction(+:c)
+  for (j = __INT_MAX__; j > __INT_MAX__ - 1; j--)
+    c++;
+  if (c != 1)
+    abort ();
+  c = 0;
+#pragma omp parallel for reduction(+:c)
+  for (j = __INT_MAX__; j >= __INT_MAX__; j--)
+    c++;
+  if (c != 1)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr46886.c b/libgomp/testsuite/libgomp.c/pr46886.c
new file mode 100644
index 000000000..fbdc4e130
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr46886.c
@@ -0,0 +1,28 @@
+/* { dg-do run } */
+/* { dg-options "-O -ftree-parallelize-loops=4 -fno-tree-ch -fno-tree-dominator-opts" } */
+
+void abort(void);
+
+int d[1024], e[1024];
+
+int foo (void)
+{
+  int s = 0;
+  int i;
+  for (i = 0; i < 1024; i++)
+    s += d[i] - e[i];
+  return s;
+}
+
+int main ()
+{
+  int i;
+  for (i = 0; i < 1024; i++)
+    {
+      d[i] = i * 2;
+      e[i] = i;
+    }
+  if (foo () != 1023 * 1024 / 2)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr48591.c b/libgomp/testsuite/libgomp.c/pr48591.c
new file mode 100644
index 000000000..18dfd7f47
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr48591.c
@@ -0,0 +1,22 @@
+/* PR middle-end/48591 */
+/* { dg-do run { target i?86-*-linux* x86_64-*-linux* ia64-*-linux* } } */
+/* { dg-options "-fopenmp" } */
+
+extern void abort (void);
+
+int
+main ()
+{
+  __float128 f = 0.0;
+  int i;
+  #pragma omp parallel for reduction(+:f)
+    for (i = 0; i < 128; i++)
+      f += 0.5Q;
+  if (f != 64.0Q)
+    abort ();
+  #pragma omp atomic
+    f += 8.5Q;
+  if (f != 72.5Q)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr49897-1.c b/libgomp/testsuite/libgomp.c/pr49897-1.c
new file mode 100644
index 000000000..d21a26252
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr49897-1.c
@@ -0,0 +1,31 @@
+/* PR middle-end/49897 */
+/* { dg-do run } */
+
+extern void abort (void);
+
+int
+main ()
+{
+  int i, j, x = 0, y, sum = 0;
+#pragma omp parallel reduction(+:sum)
+  {
+  #pragma omp for firstprivate(x) lastprivate(x, y)
+    for (i = 0; i < 10; i++)
+      {
+	x = i;
+	y = 0;
+      #pragma omp parallel reduction(+:sum)
+	{
+	#pragma omp for firstprivate(y) lastprivate(y)
+	  for (j = 0; j < 10; j++)
+	    {
+	      y = j;
+	      sum += y;
+	    }
+	}
+      }
+  }
+  if (x != 9 || y != 9 || sum != 450)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr49897-2.c b/libgomp/testsuite/libgomp.c/pr49897-2.c
new file mode 100644
index 000000000..c9ea5eced
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr49897-2.c
@@ -0,0 +1,25 @@
+/* PR middle-end/49897 */
+/* { dg-do run } */
+
+extern void abort (void);
+
+int
+main ()
+{
+  int i, j, x = 0, y, sum = 0;
+#pragma omp parallel for reduction(+:sum) firstprivate(x) lastprivate(x, y)
+  for (i = 0; i < 10; i++)
+    {
+      x = i;
+      y = 0;
+    #pragma omp parallel for reduction(+:sum) firstprivate(y) lastprivate(y)
+      for (j = 0; j < 10; j++)
+	{
+	  y = j;
+	  sum += y;
+	}
+    }
+  if (x != 9 || y != 9 || sum != 450)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr49898-1.c b/libgomp/testsuite/libgomp.c/pr49898-1.c
new file mode 100644
index 000000000..175426d40
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr49898-1.c
@@ -0,0 +1,26 @@
+/* PR middle-end/49898 */
+/* { dg-do run } */
+
+extern void abort (void);
+
+int
+main ()
+{
+  int i, j, sum = 0;
+#pragma omp parallel
+  {
+  #pragma omp for reduction(+:sum)
+    for (i = 0; i < 10; i++)
+      {
+      #pragma omp parallel
+	{
+	#pragma omp for reduction(+:sum)
+	  for (j = 0; j < 10; j++)
+	    sum += j;
+	}
+      }
+  }
+  if (sum != 450)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr49898-2.c b/libgomp/testsuite/libgomp.c/pr49898-2.c
new file mode 100644
index 000000000..03ba0f8ff
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr49898-2.c
@@ -0,0 +1,18 @@
+/* PR middle-end/49898 */
+/* { dg-do run } */
+
+extern void abort (void);
+
+int
+main ()
+{
+  int i, j, sum = 0;
+#pragma omp parallel for reduction(+:sum)
+  for (i = 0; i < 10; i++)
+    #pragma omp parallel for reduction(+:sum)
+    for (j = 0; j < 10; j++)
+      sum += j;
+  if (sum != 450)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/pr52547.c b/libgomp/testsuite/libgomp.c/pr52547.c
new file mode 100644
index 000000000..f746e2ec4
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/pr52547.c
@@ -0,0 +1,36 @@
+/* PR middle-end/52547 */
+/* { dg-do run } */
+
+extern void abort (void);
+
+__attribute__((noinline, noclone)) int
+baz (int *x, int (*fn) (int *))
+{
+  return fn (x);
+}
+
+__attribute__((noinline, noclone)) int
+foo (int x, int *y)
+{
+  int i, e = 0;
+#pragma omp parallel for reduction(|:e)
+  for (i = 0; i < x; ++i)
+    {
+      __label__ lab;
+      int bar (int *z) { return z - y; }
+      if (baz (&y[i], bar) != i)
+	e |= 1;
+    }
+  return e;
+}
+
+int
+main ()
+{
+  int a[100], i;
+  for (i = 0; i < 100; i++)
+    a[i] = i;
+  if (foo (100, a))
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/private-1.c b/libgomp/testsuite/libgomp.c/private-1.c
new file mode 100644
index 000000000..1d3659b25
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/private-1.c
@@ -0,0 +1,54 @@
+extern void abort (void);
+
+int a = 18;
+
+void
+f1 (int i, int j, int k)
+{
+  int l = 6, m = 7, n = 8;
+#pragma omp parallel private(j, m) shared(k, n) firstprivate(i, l) \
+	    num_threads(1)
+  {
+    j = 6;
+    m = 5;
+    if (++a != 19 || ++i != 9 || j != 6 || ++l != 7 || m != 5 || ++n != 9)
+      #pragma omp atomic
+	k++;
+  }
+  if (a != 19 || i != 8 || j != 26 || k != 0 || l != 6 || m != 7 || n != 9)
+    abort ();
+}
+
+int v1 = 1, v2 = 2, v5 = 5;
+int err;
+
+void
+f2 (void)
+{
+  int v3 = 3;
+#pragma omp sections private (v1) firstprivate (v2)
+  {
+  #pragma omp section
+    {
+      int v4 = 4;
+      v1 = 7;
+      #pragma omp parallel num_threads(1) firstprivate(v1, v2, v3, v4)
+	{
+	  if (++v1 != 8 || ++v2 != 3 || ++v3 != 4 || ++v4 != 5 || ++v5 != 6)
+	    err = 1;
+	}
+      if (v1 != 7 || v2 != 2 || v3 != 3 || v4 != 4 || v5 != 6)
+	abort ();
+      if (err)
+	abort ();
+    }
+  }
+}
+
+int
+main (void)
+{
+  f1 (8, 26, 0);
+  f2 ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/reduction-1.c b/libgomp/testsuite/libgomp.c/reduction-1.c
new file mode 100644
index 000000000..665163af0
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/reduction-1.c
@@ -0,0 +1,36 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i = 0, j = 0, k = ~0;
+  double d = 1.0;
+#pragma omp parallel num_threads(4) reduction(+:i) reduction(*:d) reduction(&:k)
+  {
+    if (i != 0 || d != 1.0 || k != ~0)
+#pragma omp atomic
+      j |= 1;
+  
+    if (omp_get_num_threads () != 4)
+#pragma omp atomic
+      j |= 2;
+
+    i = omp_get_thread_num ();
+    d = i + 1;
+    k = ~(1 << (2 * i));
+  }
+
+  if (j & 1)
+    abort ();
+  if ((j & 2) == 0)
+    {
+      if (i != (0 + 1 + 2 + 3))
+	abort ();
+      if (d != (1.0 * 2.0 * 3.0 * 4.0))
+	abort ();
+      if (k != (~0 ^ 0x55))
+	abort ();
+    }
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/reduction-2.c b/libgomp/testsuite/libgomp.c/reduction-2.c
new file mode 100644
index 000000000..52b3faff7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/reduction-2.c
@@ -0,0 +1,50 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i = 0, j = 0, k = ~0, l;
+  double d = 1.0;
+#pragma omp parallel num_threads(4)
+  {
+#pragma omp single
+    {
+      i = 16;
+      k ^= (1 << 16);
+      d += 32.0;
+    }
+
+#pragma omp for reduction(+:i) reduction(*:d) reduction(&:k)
+    for (l = 0; l < 4; l++)
+      {
+	if (omp_get_num_threads () == 4 && (i != 0 || d != 1.0 || k != ~0))
+#pragma omp atomic
+	  j |= 1;
+  
+	if (l == omp_get_thread_num ())
+	  {
+	    i = omp_get_thread_num ();
+	    d = i + 1;
+	    k = ~(1 << (2 * i));
+	  }
+      }
+
+    if (omp_get_num_threads () == 4)
+      {
+	if (i != (16 + 0 + 1 + 2 + 3))
+#pragma omp atomic
+	  j |= 2;
+	if (d != (33.0 * 1.0 * 2.0 * 3.0 * 4.0))
+#pragma omp atomic
+	  j |= 4;
+	if (k != (~0 ^ 0x55 ^ (1 << 16)))
+#pragma omp atomic
+	  j |= 8;
+      }
+  }
+
+  if (j)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/reduction-3.c b/libgomp/testsuite/libgomp.c/reduction-3.c
new file mode 100644
index 000000000..4f8f2fc12
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/reduction-3.c
@@ -0,0 +1,51 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i = 0, j = 0, k = ~0, l;
+  double d = 1.0;
+#pragma omp parallel num_threads(4)
+  {
+#pragma omp single
+    {
+      i = 16;
+      k ^= (1 << 16);
+      d += 32.0;
+    }
+
+#pragma omp for reduction(+:i) reduction(*:d) reduction(&:k) nowait
+    for (l = 0; l < 4; l++)
+      {
+	if (omp_get_num_threads () == 4 && (i != 0 || d != 1.0 || k != ~0))
+#pragma omp atomic
+	  j |= 1;
+  
+	if (l == omp_get_thread_num ())
+	  {
+	    i = omp_get_thread_num ();
+	    d = i + 1;
+	    k = ~(1 << (2 * i));
+	  }
+      }
+
+    if (omp_get_num_threads () == 4)
+      {
+#pragma omp barrier
+	if (i != (16 + 0 + 1 + 2 + 3))
+#pragma omp atomic
+	  j |= 2;
+	if (d != (33.0 * 1.0 * 2.0 * 3.0 * 4.0))
+#pragma omp atomic
+	  j |= 4;
+	if (k != (~0 ^ 0x55 ^ (1 << 16)))
+#pragma omp atomic
+	  j |= 8;
+      }
+  }
+
+  if (j)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/reduction-4.c b/libgomp/testsuite/libgomp.c/reduction-4.c
new file mode 100644
index 000000000..23e9d6d5b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/reduction-4.c
@@ -0,0 +1,36 @@
+#include <omp.h>
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i = 0, j = 0, k = 0, l = 0;
+#pragma omp parallel num_threads(4) reduction(-:i) reduction(|:k) \
+		     reduction(^:l)
+  {
+    if (i != 0 || k != 0 || l != 0)
+#pragma omp atomic
+      j |= 1;
+  
+    if (omp_get_num_threads () != 4)
+#pragma omp atomic
+      j |= 2;
+
+    i = omp_get_thread_num ();
+    k = 1 << (2 * i);
+    l = 0xea << (3 * i);
+  }
+
+  if (j & 1)
+    abort ();
+  if ((j & 2) == 0)
+    {
+      if (i != (0 + 1 + 2 + 3))
+	abort ();
+      if (k != 0x55)
+	abort ();
+      if (l != 0x1e93a)
+	abort ();
+    }
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/reduction-5.c b/libgomp/testsuite/libgomp.c/reduction-5.c
new file mode 100644
index 000000000..de87d9f6d
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/reduction-5.c
@@ -0,0 +1,78 @@
+/* PR middle-end/36506 */
+
+extern void abort (void);
+
+int
+main (void)
+{
+  int sum = 0, prod = 1;
+#pragma omp parallel
+  #pragma omp sections reduction (+:sum)
+    {
+    #pragma omp section
+      sum += 2;
+    #pragma omp section
+      sum += 2;
+    #pragma omp section
+      sum += 2;
+    }
+  if (sum != 6)
+    abort ();
+  sum = 0;
+#pragma omp parallel sections reduction (+:sum)
+  {
+  #pragma omp section
+    sum += 2;
+  #pragma omp section
+    sum += 2;
+  #pragma omp section
+    sum += 2;
+  }
+  if (sum != 6)
+    abort ();
+  sum = 0;
+#pragma omp parallel
+  #pragma omp sections reduction (+:sum) reduction (*:prod)
+    {
+    #pragma omp section
+      {
+	sum += 2;
+	prod *= 2;
+      }
+    #pragma omp section
+      {
+	sum += 2;
+	prod *= 2;
+      }
+    #pragma omp section
+      {
+	sum += 2;
+	prod *= 2;
+      }
+    }
+  if (sum != 6 || prod != 8)
+    abort ();
+  sum = 0;
+  prod = 1;
+#pragma omp parallel sections reduction (+:sum) reduction (*:prod)
+  {
+  #pragma omp section
+    {
+      sum += 2;
+      prod *= 2;
+    }
+  #pragma omp section
+    {
+      sum += 2;
+      prod *= 2;
+    }
+  #pragma omp section
+    {
+      sum += 2;
+      prod *= 2;
+    }
+  }
+  if (sum != 6 || prod != 8)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/sections-1.c b/libgomp/testsuite/libgomp.c/sections-1.c
new file mode 100644
index 000000000..3a6584cb7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/sections-1.c
@@ -0,0 +1,85 @@
+/* Test that all sections are touched.  */
+
+/* { dg-require-effective-target sync_int_long } */
+
+#include <omp.h>
+#include <string.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+#define N 100
+static int data[N];
+static int NTHR;
+
+static void clean_data (void)
+{
+  memset (data, -1, sizeof (data));
+}
+
+static void test_data (void)
+{
+  int i;
+
+  for (i = 0; i < N; ++i)
+    assert (data[i] != -1);
+}
+
+static void set_data (unsigned i, int val)
+{
+  int old;
+  assert (i >= 1 && i <= N);
+  old = __sync_lock_test_and_set (data+i-1, val);
+  assert (old == -1);
+}
+  
+
+static void f_1 (void *dummy)
+{
+  int iam = omp_get_thread_num ();
+  unsigned long s;
+
+  for (s = GOMP_sections_start (N); s ; s = GOMP_sections_next ())
+    set_data (s, iam);
+  GOMP_sections_end ();
+}
+
+static void test_1 (void)
+{
+  clean_data ();
+  GOMP_parallel_start (f_1, NULL, NTHR);
+  f_1 (NULL);
+  GOMP_parallel_end ();
+  test_data ();
+}
+
+static void f_2 (void *dummy)
+{
+  int iam = omp_get_thread_num ();
+  unsigned s;
+
+  while ((s = GOMP_sections_next ()))
+    set_data (s, iam);
+  GOMP_sections_end_nowait ();
+}
+
+static void test_2 (void)
+{
+  clean_data ();
+  GOMP_parallel_sections_start (f_2, NULL, NTHR, N);
+  f_2 (NULL);
+  GOMP_parallel_end ();
+  test_data ();
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  NTHR = 4;
+
+  test_1 ();
+  test_2 ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/shared-1.c b/libgomp/testsuite/libgomp.c/shared-1.c
new file mode 100644
index 000000000..f7d4fb2d6
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/shared-1.c
@@ -0,0 +1,58 @@
+extern void abort (void);
+
+struct Y
+{
+  int l[5][10];
+};
+
+struct X
+{
+  struct Y y;
+  float b[10];
+};
+
+void
+parallel (int a, int b)
+{
+  int i, j;
+  struct X A[10][5];
+  a = b = 3;
+
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 5; j++)
+      A[i][j].y.l[3][3] = -10;
+
+  #pragma omp parallel shared (a, b, A) num_threads (5)
+    {
+      int i, j;
+
+      #pragma omp atomic
+      a += omp_get_num_threads ();
+
+      #pragma omp atomic
+      b += omp_get_num_threads ();
+
+      #pragma omp for private (j)
+      for (i = 0; i < 10; i++)
+	for (j = 0; j < 5; j++)
+	  A[i][j].y.l[3][3] += 20;
+
+    }
+
+  for (i = 0; i < 10; i++)
+    for (j = 0; j < 5; j++)
+      if (A[i][j].y.l[3][3] != 10)
+	abort ();
+
+  if (a != 28)
+    abort ();
+
+  if (b != 28)
+    abort ();
+}
+
+main()
+{
+  parallel (1, 2);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/shared-2.c b/libgomp/testsuite/libgomp.c/shared-2.c
new file mode 100644
index 000000000..56c88ecc7
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/shared-2.c
@@ -0,0 +1,50 @@
+#include <stdio.h>
+#include <omp.h>
+
+extern void abort (void);
+
+void
+parallel (int a, int b)
+{
+  int bad, LASTPRIV, LASTPRIV_SEC;
+  int i;
+
+  a = b = 3;
+
+  bad = 0;
+
+  #pragma omp parallel firstprivate (a,b) shared (bad) num_threads (5)
+    {
+      if (a != 3 || b != 3)
+	bad = 1;
+
+      #pragma omp for lastprivate (LASTPRIV)
+      for (i = 0; i < 10; i++)
+	LASTPRIV = i;
+
+      #pragma omp sections lastprivate (LASTPRIV_SEC)
+	{
+	  #pragma omp section
+	    { LASTPRIV_SEC = 3; }
+
+	  #pragma omp section
+	    { LASTPRIV_SEC = 42; }
+	}
+
+    }
+
+  if (LASTPRIV != 9)
+    abort ();
+
+  if (LASTPRIV_SEC != 42)
+    abort ();
+
+  if (bad)
+    abort ();
+}
+
+int main()
+{
+  parallel (1, 2);
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/shared-3.c b/libgomp/testsuite/libgomp.c/shared-3.c
new file mode 100644
index 000000000..494a970ad
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/shared-3.c
@@ -0,0 +1,19 @@
+/* { dg-do run } */
+
+void abort (void);
+
+int main()
+{
+  int x;
+  int *p;
+
+  p = &x;
+
+  #pragma omp parallel
+    {
+      if (p != &x)
+        abort ();
+    }
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/single-1.c b/libgomp/testsuite/libgomp.c/single-1.c
new file mode 100644
index 000000000..1ce89118d
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/single-1.c
@@ -0,0 +1,53 @@
+/* Trivial test of single.  */
+
+/* { dg-require-effective-target sync_int_long } */
+
+#include <omp.h>
+#include <sys/time.h>
+#include <unistd.h>
+#include <assert.h>
+#include "libgomp_g.h"
+
+
+static int test;
+
+static void f_nocopy (void *dummy)
+{
+  if (GOMP_single_start ())
+    {
+      int iam = omp_get_thread_num ();
+      int old = __sync_lock_test_and_set (&test, iam);
+      assert (old == -1);
+    }
+}
+
+static void f_copy (void *dummy)
+{
+  int *x = GOMP_single_copy_start ();
+  if (x == NULL)
+    {
+      int iam = omp_get_thread_num ();
+      int old = __sync_lock_test_and_set (&test, iam);
+      assert (old == -1);
+      GOMP_single_copy_end (&test);
+    }
+  else
+    assert (x == &test);
+}
+
+int main()
+{
+  omp_set_dynamic (0);
+
+  test = -1;
+  GOMP_parallel_start (f_nocopy, NULL, 3);
+  f_nocopy (NULL);
+  GOMP_parallel_end ();
+
+  test = -1;
+  GOMP_parallel_start (f_copy, NULL, 3);
+  f_copy (NULL);
+  GOMP_parallel_end ();
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/single-2.c b/libgomp/testsuite/libgomp.c/single-2.c
new file mode 100644
index 000000000..b510ce735
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/single-2.c
@@ -0,0 +1,15 @@
+#include <stdlib.h>
+
+int
+main (void)
+{
+  int i;
+  i = 4;
+#pragma omp single copyprivate (i)
+  {
+    i = 6;
+  }
+  if (i != 6)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/sort-1.c b/libgomp/testsuite/libgomp.c/sort-1.c
new file mode 100644
index 000000000..269d69da1
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/sort-1.c
@@ -0,0 +1,379 @@
+/* Test and benchmark of a couple of parallel sorting algorithms.
+   Copyright (C) 2008 Free Software Foundation, Inc.
+
+   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 3, 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 COPYING3.  If not see
+   <http://www.gnu.org/licenses/>.  */
+
+#include <limits.h>
+#include <omp.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+int failures;
+
+#define THRESHOLD 100
+
+static void
+verify (const char *name, double stime, int *array, int count)
+{
+  int i;
+  double etime = omp_get_wtime ();
+
+  printf ("%s: %g\n", name, etime - stime);
+  for (i = 1; i < count; i++)
+    if (array[i] < array[i - 1])
+      {
+	printf ("%s: incorrectly sorted\n", name);
+	failures = 1;
+      }
+}
+
+static void
+insertsort (int *array, int s, int e)
+{
+  int i, j, val;
+  for (i = s + 1; i <= e; i++)
+    {
+      val = array[i];
+      j = i;
+      while (j-- > s && val < array[j])
+	array[j + 1] = array[j];
+      array[j + 1] = val;
+    }
+}
+
+struct int_pair
+{
+  int lo;
+  int hi;
+};
+
+struct int_pair_stack
+{
+  struct int_pair *top;
+#define STACK_SIZE 4 * CHAR_BIT * sizeof (int)
+  struct int_pair arr[STACK_SIZE];
+};
+
+static inline void
+init_int_pair_stack (struct int_pair_stack *stack)
+{
+  stack->top = &stack->arr[0];
+}
+
+static inline void
+push_int_pair_stack (struct int_pair_stack *stack, int lo, int hi)
+{
+  stack->top->lo = lo;
+  stack->top->hi = hi;
+  stack->top++;
+}
+
+static inline void
+pop_int_pair_stack (struct int_pair_stack *stack, int *lo, int *hi)
+{
+  stack->top--;
+  *lo = stack->top->lo;
+  *hi = stack->top->hi;
+}
+
+static inline int
+size_int_pair_stack (struct int_pair_stack *stack)
+{
+  return stack->top - &stack->arr[0];
+}
+
+static inline void
+busy_wait (void)
+{
+#if defined __i386__ || defined __x86_64__
+  __asm volatile ("rep; nop" : : : "memory");
+#elif defined __ia64__
+  __asm volatile ("hint @pause" : : : "memory");
+#elif defined __sparc__ && (defined __arch64__ || defined __sparc_v9__)
+  __asm volatile ("membar #LoadLoad" : : : "memory");
+#else
+  __asm volatile ("" : : : "memory");
+#endif
+}
+
+static inline void
+swap (int *array, int a, int b)
+{
+  int val = array[a];
+  array[a] = array[b];
+  array[b] = val;
+}
+
+static inline int
+choose_pivot (int *array, int lo, int hi)
+{
+  int mid = (lo + hi) / 2;
+
+  if (array[mid] < array[lo])
+    swap (array, lo, mid);
+  if (array[hi] < array[mid])
+    {
+      swap (array, mid, hi);
+      if (array[mid] < array[lo])
+	swap (array, lo, mid);
+    }
+  return array[mid];
+}
+
+static inline int
+partition (int *array, int lo, int hi)
+{
+  int pivot = choose_pivot (array, lo, hi);
+  int left = lo;
+  int right = hi;
+
+  for (;;)
+    {
+      while (array[++left] < pivot);
+      while (array[--right] > pivot);
+      if (left >= right)
+	break;
+      swap (array, left, right);
+    }
+  return left;
+}
+
+static void
+sort1 (int *array, int count)
+{
+  omp_lock_t lock;
+  struct int_pair_stack global_stack;
+  int busy = 1;
+  int num_threads;
+
+  omp_init_lock (&lock);
+  init_int_pair_stack (&global_stack);
+  #pragma omp parallel firstprivate (array, count)
+  {
+    int lo = 0, hi = 0, mid, next_lo, next_hi;
+    bool idle = true;
+    struct int_pair_stack local_stack;
+
+    init_int_pair_stack (&local_stack);
+    if (omp_get_thread_num () == 0)
+      {
+	num_threads = omp_get_num_threads ();
+	hi = count - 1;
+	idle = false;
+      }
+
+    for (;;)
+      {
+	if (hi - lo < THRESHOLD)
+	  {
+	    insertsort (array, lo, hi);
+	    lo = hi;
+	  }
+	if (lo >= hi)
+	  {
+	    if (size_int_pair_stack (&local_stack) == 0)
+	      {
+	      again:
+		omp_set_lock (&lock);
+		if (size_int_pair_stack (&global_stack) == 0)
+		  {
+		    if (!idle)
+		      busy--;
+		    if (busy == 0)
+		      {
+			omp_unset_lock (&lock);
+			break;
+		      }
+		    omp_unset_lock (&lock);
+		    idle = true;
+		    while (size_int_pair_stack (&global_stack) == 0
+			   && busy)
+		      busy_wait ();
+		    goto again;
+		  }
+		if (idle)
+		  busy++;
+		pop_int_pair_stack (&global_stack, &lo, &hi);
+		omp_unset_lock (&lock);
+		idle = false;
+	      }
+	    else
+	      pop_int_pair_stack (&local_stack, &lo, &hi);
+	  }
+
+	mid = partition (array, lo, hi);
+	if (mid - lo < hi - mid)
+	  {
+	    next_lo = mid;
+	    next_hi = hi;
+	    hi = mid - 1;
+	  }
+	else
+	  {
+	    next_lo = lo;
+	    next_hi = mid - 1;
+	    lo = mid;
+	  }
+
+	if (next_hi - next_lo < THRESHOLD)
+	  insertsort (array, next_lo, next_hi);
+	else
+	  {
+	    if (size_int_pair_stack (&global_stack) < num_threads - 1)
+	      {
+		int size;
+
+		omp_set_lock (&lock);
+		size = size_int_pair_stack (&global_stack);
+		if (size < num_threads - 1 && size < STACK_SIZE)
+		  push_int_pair_stack (&global_stack, next_lo, next_hi);
+		else
+		  push_int_pair_stack (&local_stack, next_lo, next_hi);
+		omp_unset_lock (&lock);
+	      }
+	    else
+	      push_int_pair_stack (&local_stack, next_lo, next_hi);
+	  }
+      }
+    }
+  omp_destroy_lock (&lock);
+}
+
+static void
+sort2_1 (int *array, int lo, int hi, int num_threads, int *busy)
+{
+  int mid;
+
+  if (hi - lo < THRESHOLD)
+    {
+      insertsort (array, lo, hi);
+      return;
+    }
+
+  mid = partition (array, lo, hi);
+
+  if (*busy >= num_threads)
+    {
+      sort2_1 (array, lo, mid - 1, num_threads, busy);
+      sort2_1 (array, mid, hi, num_threads, busy);
+      return;
+    }
+
+  #pragma omp atomic
+    *busy += 1;
+
+  #pragma omp parallel num_threads (2) \
+		       firstprivate (array, lo, hi, mid, num_threads, busy)
+  {
+    if (omp_get_thread_num () == 0)
+      sort2_1 (array, lo, mid - 1, num_threads, busy);
+    else
+      {
+	sort2_1 (array, mid, hi, num_threads, busy);
+	#pragma omp atomic
+	  *busy -= 1;
+      }
+  }
+}
+
+static void
+sort2 (int *array, int count)
+{
+  int num_threads;
+  int busy = 1;
+
+  #pragma omp parallel
+    #pragma omp single nowait
+      num_threads = omp_get_num_threads ();
+
+  sort2_1 (array, 0, count - 1, num_threads, &busy);
+}
+
+#if _OPENMP >= 200805
+static void
+sort3_1 (int *array, int lo, int hi)
+{
+  int mid;
+
+  if (hi - lo < THRESHOLD)
+    {
+      insertsort (array, lo, hi);
+      return;
+    }
+
+  mid = partition (array, lo, hi);
+  #pragma omp task
+    sort3_1 (array, lo, mid - 1);
+  sort3_1 (array, mid, hi);
+}
+
+static void
+sort3 (int *array, int count)
+{
+  #pragma omp parallel
+    #pragma omp single
+      sort3_1 (array, 0, count - 1);
+}
+#endif
+
+int
+main (int argc, char **argv)
+{
+  int i, count = 1000000;
+  double stime;
+  int *unsorted, *sorted, num_threads;
+  if (argc >= 2)
+    count = strtoul (argv[1], NULL, 0);
+
+  unsorted = malloc (count * sizeof (int));
+  sorted = malloc (count * sizeof (int));
+  if (unsorted == NULL || sorted == NULL)
+    {
+      puts ("allocation failure");
+      exit (1);
+    }
+
+  srand (0xdeadbeef);
+  for (i = 0; i < count; i++)
+    unsorted[i] = rand ();
+
+  omp_set_nested (1);
+  omp_set_dynamic (0);
+  #pragma omp parallel
+    #pragma omp single nowait
+      num_threads = omp_get_num_threads ();
+  printf ("Threads: %d\n", num_threads);
+
+  memcpy (sorted, unsorted, count * sizeof (int));
+  stime = omp_get_wtime ();
+  sort1 (sorted, count);
+  verify ("sort1", stime, sorted, count);
+
+  memcpy (sorted, unsorted, count * sizeof (int));
+  stime = omp_get_wtime ();
+  sort2 (sorted, count);
+  verify ("sort2", stime, sorted, count);
+
+#if _OPENMP >= 200805
+  memcpy (sorted, unsorted, count * sizeof (int));
+  stime = omp_get_wtime ();
+  sort3 (sorted, count);
+  verify ("sort3", stime, sorted, count);
+#endif
+
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/task-1.c b/libgomp/testsuite/libgomp.c/task-1.c
new file mode 100644
index 000000000..66f58a29b
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/task-1.c
@@ -0,0 +1,84 @@
+extern void abort (void);
+
+int a = 18;
+
+void
+f1 (int i, int j, int k)
+{
+  int l = 6, m = 7, n = 8;
+#pragma omp task private(j, m) shared(k, n)
+  {
+    j = 6;
+    m = 5;
+    if (++a != 19 || ++i != 9 || j != 6 || ++l != 7 || m != 5 || ++n != 9)
+      #pragma omp atomic
+	k++;
+  }
+#pragma omp taskwait
+  if (a != 19 || i != 8 || j != 26 || k != 0 || l != 6 || m != 7 || n != 9)
+    abort ();
+}
+
+int v1 = 1, v2 = 2, v5 = 5;
+int err;
+
+void
+f2 (void)
+{
+  int v3 = 3;
+#pragma omp sections private (v1) firstprivate (v2)
+  {
+  #pragma omp section
+    {
+      int v4 = 4;
+      v1 = 7;
+      #pragma omp task
+	{
+	  if (++v1 != 8 || ++v2 != 3 || ++v3 != 4 || ++v4 != 5 || ++v5 != 6)
+	    err = 1;
+	}
+      #pragma omp taskwait
+      if (v1 != 7 || v2 != 2 || v3 != 3 || v4 != 4 || v5 != 6)
+	abort ();
+      if (err)
+	abort ();
+    }
+  }
+}
+
+void
+f3 (int i, int j, int k)
+{
+  int l = 6, m = 7, n = 8;
+#pragma omp task private(j, m) shared(k, n) untied
+  {
+    j = 6;
+    m = 5;
+    if (++a != 19 || ++i != 9 || j != 6 || ++l != 7 || m != 5 || ++n != 9)
+      #pragma omp atomic
+	k++;
+  }
+#pragma omp taskwait
+  if (a != 19 || i != 8 || j != 26 || k != 0 || l != 6 || m != 7 || n != 9)
+    abort ();
+}
+
+int
+main (void)
+{
+  f1 (8, 26, 0);
+  f2 ();
+  a = 18;
+  f3 (8, 26, 0);
+  a = 18;
+#pragma omp parallel num_threads(4)
+  {
+    #pragma omp master
+      {
+	f1 (8, 26, 0);
+	a = 18;
+	f3 (8, 26, 0);
+      }
+  }
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/task-2.c b/libgomp/testsuite/libgomp.c/task-2.c
new file mode 100644
index 000000000..ed6a09c35
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/task-2.c
@@ -0,0 +1,53 @@
+extern void abort (void);
+
+int
+f1 (void)
+{
+  int a = 6, e = 0;
+  int nested (int x)
+  {
+    return x + a;
+  }
+  #pragma omp task
+  {
+    int n = nested (5);
+    if (n != 11)
+      #pragma omp atomic
+	e += 1;
+  }
+  #pragma omp taskwait
+  return e;
+}
+
+int
+f2 (void)
+{
+  int a = 6, e = 0;
+  int nested (int x)
+  {
+    return x + a;
+  }
+  a = nested (4);
+  #pragma omp task
+  {
+    if (a != 10)
+      #pragma omp atomic
+	e += 1;
+  }
+  #pragma omp taskwait
+  return e;
+}
+
+int
+main (void)
+{
+  int e = 0;
+  #pragma omp parallel num_threads(4) reduction(+:e)
+  {
+    e += f1 ();
+    e += f2 ();
+  }
+  if (e)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/task-3.c b/libgomp/testsuite/libgomp.c/task-3.c
new file mode 100644
index 000000000..5657346bd
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/task-3.c
@@ -0,0 +1,70 @@
+/* { dg-do run } */
+
+#include <omp.h>
+extern void abort ();
+
+int l = 5;
+
+int
+foo (int i)
+{
+  int j = 7;
+  const int k = 8;
+  #pragma omp task firstprivate (i) shared (j, l)
+  {
+    #pragma omp critical
+      {
+	j += i;
+	l += k;
+      }
+  }
+  i++;
+  #pragma omp task firstprivate (i) shared (j, l)
+  {
+    #pragma omp critical
+      {
+	j += i;
+	l += k;
+      }
+  }
+  i++;
+  #pragma omp task firstprivate (i) shared (j, l)
+  {
+    #pragma omp critical
+      {
+	j += i;
+	l += k;
+      }
+  }
+  i++;
+  #pragma omp task firstprivate (i) shared (j, l)
+  {
+    #pragma omp critical
+      {
+	j += i;
+	l += k;
+      }
+  }
+  i++;
+  #pragma omp taskwait
+  return (i != 8 * omp_get_thread_num () + 4
+	  || j != 4 * i - 3
+	  || k != 8);
+}
+
+int
+main (void)
+{
+  int r = 0;
+  #pragma omp parallel num_threads (4) reduction(+:r)
+    if (omp_get_num_threads () != 4)
+      {
+	#pragma omp master
+	  l = 133;
+      }
+    else if (foo (8 * omp_get_thread_num ()))
+      r++;
+  if (r || l != 133)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/task-4.c b/libgomp/testsuite/libgomp.c/task-4.c
new file mode 100644
index 000000000..184359300
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/task-4.c
@@ -0,0 +1,40 @@
+/* { dg-do run } */
+
+#include <omp.h>
+#include <stdlib.h>
+#include <string.h>
+
+int e;
+
+void __attribute__((noinline))
+baz (int i, int *p, int j, int *q)
+{
+  if (p[0] != 1 || p[i] != 3 || q[0] != 2 || q[j] != 4)
+    #pragma omp atomic
+      e++;
+}
+
+void __attribute__((noinline))
+foo (int i, int j)
+{
+  int p[i + 1];
+  int q[j + 1];
+  memset (p, 0, sizeof (p));
+  memset (q, 0, sizeof (q));
+  p[0] = 1;
+  p[i] = 3;
+  q[0] = 2;
+  q[j] = 4;
+  #pragma omp task firstprivate (p, q)
+    baz (i, p, j, q);
+}
+
+int
+main (void)
+{
+  #pragma omp parallel num_threads (4)
+    foo (5 + omp_get_thread_num (), 7 + omp_get_thread_num ());
+  if (e)
+    abort ();
+  return 0;
+}
diff --git a/libgomp/testsuite/libgomp.c/vla-1.c b/libgomp/testsuite/libgomp.c/vla-1.c
new file mode 100644
index 000000000..bdacdbbe8
--- /dev/null
+++ b/libgomp/testsuite/libgomp.c/vla-1.c
@@ -0,0 +1,60 @@
+/* { dg-do run } */
+
+#include <omp.h>
+#include <stdlib.h>
+#include <string.h>
+
+int
+main (int argc, char **argv[])
+{
+  int n = argc < 5 ? 12 : 31, i, m, l;
+  char a[n + 3];
+  unsigned short b[n / 2 - 1];
+  int c[n * 2 + 1];
+
+  for (i = 0; i < n + 3; i++)
+    a[i] = i;
+  for (i = 0; i < n / 2 - 1; i++)
+    b[i] = (i << 8) | i;
+  for (i = 0; i < n * 2 + 1; i++)
+    c[i] = (i << 24) | i;
+  l = 0;
+  m = n;
+#pragma omp parallel default (shared) num_threads (4) \
+  firstprivate (a, m) private (b, i) reduction (+:l)
+  {
+    for (i = 0; i < m + 3; i++)
+      if (a[i] != i)
+	l++;
+    for (i = 0; i < m * 2 + 1; i++)
+      if (c[i] != ((i << 24) | i))
+	l++;
+#pragma omp barrier
+    memset (a, omp_get_thread_num (), m + 3);
+    for (i = 0; i < m / 2 - 1; i++)
+      b[i] = a[0] + 7;
+#pragma omp master
+    {
+      for (i = 0; i < m * 2 + 1; i++)
+	c[i] = a[0] + 16;
+    }
+#pragma omp barrier
+    if (a[0] != omp_get_thread_num ())
+      l++;
+    for (i = 1; i < m + 3; i++)
+      if (a[i] != a[0])
+	l++;
+    for (i = 0; i < m / 2 - 1; i++)
+      if (b[i] != a[0] + 7)
+	l++;
+    for (i = 0; i < m * 2 + 1; i++)
+      if (c[i] != 16)
+	l++;
+  }
+  if (l)
+    abort ();
+  for (i = 0; i < n * 2 + 1; i++)
+    if (c[i] != 16)
+      l++;
+  return 0;
+}