Merge pull request #5083 from smowton/smowton/feature/sese-regions-pass

Add single-entry, single-exit regions pass

Author: smowton

jbmc/src/java_bytecode/java_local_variable_table.cpp

@@ -40,6 +40,7 @@ struct procedure_local_cfg_baset<
   typedef std::map<java_bytecode_convert_methodt::method_offsett,
                    java_bytecode_convert_methodt::method_offsett>
     entry_mapt;
+  typedef std::size_t entryt;
   entry_mapt entry_map;
 
   procedure_local_cfg_baset() {}

regression/goto-instrument/region-analysis-1/test.c

@@ -0,0 +1,22 @@
+int main(int argc, char **argv)
+{
+  // Function containing two SESE regions,
+  // one per if-block.
+  if(argc % 5)
+  {
+    if(argc % 2)
+    {
+      ++argc;
+    }
+    else
+    {
+      --argc;
+    }
+  }
+  else
+  {
+    ++argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-1/test.desc

@@ -0,0 +1,8 @@
+CORE
+test.c
+--show-sese-regions
+^Region starting at \(1, [0-9]+\) IF !\(argc % 2 != 0\) THEN GOTO 1 ends at \(5, [0-9]+\) 2: SKIP$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 3 ends at \(8, [0-9]+\) 4: SKIP$
+^Function contains 2 single-entry, single-exit regions:$
+^EXIT=0$
+^SIGNAL=0$

regression/goto-instrument/region-analysis-10/test.c

@@ -0,0 +1,20 @@
+int main(int argc, char **argv)
+{
+  // A function containing 2 SESE regions: one delimiting the outer
+  // if-block, and one delimiting the while-loop.
+
+  if(argc % 5)
+  {
+    int x = 0;
+    while(x < 10)
+    {
+      ++x;
+    }
+  }
+  else
+  {
+    --argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-10/test.desc

@@ -0,0 +1,8 @@
+CORE
+test.c
+--show-sese-regions
+^Function contains 2 single-entry, single-exit regions:$
+^Region starting at \(2, [0-9]+\) x = 0; ends at \(6, [0-9]+\) 2: SKIP$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 3 ends at \(10, [0-9]+\) 4: SKIP$
+^EXIT=0$
+^SIGNAL=0$

regression/goto-instrument/region-analysis-11/test.c

@@ -0,0 +1,22 @@
+int main(int argc, char **argv)
+{
+  // A function containing 2 SESE regions, one for the outer if-block
+  // and one for the while-loop, despite a break edge
+
+  if(argc % 5)
+  {
+    int x = 0;
+    while(x < 10)
+    {
+      if(x % 7)
+        break;
+      ++x;
+    }
+  }
+  else
+  {
+    --argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-11/test.desc

@@ -0,0 +1,8 @@
+CORE
+test.c
+--show-sese-regions
+^Function contains 2 single-entry, single-exit regions:$
+^Region starting at \(2, [0-9]+\) x = 0; ends at \(9, [0-9]+\) 2: SKIP$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 3 ends at \(13, [0-9]+\) 4: SKIP$
+^EXIT=0$
+^SIGNAL=0$

regression/goto-instrument/region-analysis-12/test.c

@@ -0,0 +1,26 @@
+int main(int argc, char **argv)
+{
+  // A function containing 3 SESE regions, one for the outer if-block
+  // and one for the while-loop, despite a continue edge, and one for
+  // the conditional inside the loop.
+
+  if(argc % 5)
+  {
+    int x = 0;
+    while(x < 10)
+    {
+      if(x % 7)
+      {
+        ++x;
+        continue;
+      }
+      ++x;
+    }
+  }
+  else
+  {
+    --argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-12/test.desc

@@ -0,0 +1,9 @@
+CORE
+test.c
+--show-sese-regions
+^Function contains 3 single-entry, single-exit regions:$
+^Region starting at \(4, [0-9]+\) IF !\(x % 7 != 0\) THEN GOTO 2 ends at \(9, [0-9]+\) 3: GOTO 1$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 5 ends at \(14, [0-9]+\) 6: SKIP$
+^Region starting at \(2, [0-9]+\) x = 0; ends at \(10, [0-9]+\) 4: SKIP$
+^EXIT=0$
+^SIGNAL=0$

regression/goto-instrument/region-analysis-13/test.c

@@ -0,0 +1,27 @@
+int main(int argc, char **argv)
+{
+  // A function containing 2 SESE regions, one for the outer if-block
+  // and one for the while-loop, despite a continue edge that branches
+  // directly to the top of the loop.
+
+  if(argc % 5)
+  {
+    int x = 0;
+  top_continue:
+    while(x < 10)
+    {
+      if(x % 7)
+      {
+        ++x;
+        goto top_continue;
+      }
+      ++x;
+    }
+  }
+  else
+  {
+    --argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-13/test.desc

@@ -0,0 +1,11 @@
+FUTURE
+test.c
+--show-sese-regions
+^Function contains 2 single-entry, single-exit regions:$
+^Region starting at \(2, [0-9]+\) x = 0; ends at \(10, [0-9]+\) 3: SKIP$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 4 ends at \(14, [0-9]+\) 5: SKIP$
+^EXIT=0$
+^SIGNAL=0$
+--
+--
+These regions cannot currently be identified because the function contains a loop with two backedges.

regression/goto-instrument/region-analysis-14/test.c

@@ -0,0 +1,20 @@
+int main(int argc, char **argv)
+{
+  // A function containing 2 SESE regions: one delimiting the outer
+  // if-block, and one delimiting the do-while loop.
+
+  if(argc % 5)
+  {
+    int x = 0;
+    do
+    {
+      ++x;
+    } while(x < 10);
+  }
+  else
+  {
+    --argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-14/test.desc

@@ -0,0 +1,8 @@
+CORE
+test.c
+--show-sese-regions
+^Function contains 2 single-entry, single-exit regions:$
+^Region starting at \(2, [0-9]+\) x = 0; ends at \(5, [0-9]+\) SKIP$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 2 ends at \(9, [0-9]+\) 3: SKIP$
+^EXIT=0$
+^SIGNAL=0$

regression/goto-instrument/region-analysis-15/test.c

@@ -0,0 +1,22 @@
+int main(int argc, char **argv)
+{
+  // A function containing 2 SESE regions, one for the outer if-block
+  // and one for the while-loop, despite a break edge
+
+  if(argc % 5)
+  {
+    int x = 0;
+    do
+    {
+      if(x % 7)
+        break;
+      ++x;
+    } while(x < 10);
+  }
+  else
+  {
+    --argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-15/test.desc

@@ -0,0 +1,8 @@
+CORE
+test.c
+--show-sese-regions
+^Function contains 2 single-entry, single-exit regions:$
+^Region starting at \(2, [0-9]+\) x = 0; ends at \(8, [0-9]+\) 2: SKIP$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 3 ends at \(12, [0-9]+\) 4: SKIP$
+^EXIT=0$
+^SIGNAL=0$

regression/goto-instrument/region-analysis-16/test.c

@@ -0,0 +1,26 @@
+int main(int argc, char **argv)
+{
+  // A function containing 3 SESE regions, one for the outer if-block
+  // and one for the while-loop, despite a continue edge, and one for
+  // the conditional inside the loop.
+
+  if(argc % 5)
+  {
+    int x = 0;
+    do
+    {
+      if(x % 7)
+      {
+        ++x;
+        continue;
+      }
+      ++x;
+    } while(x < 10);
+  }
+  else
+  {
+    --argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-16/test.desc

@@ -0,0 +1,9 @@
+CORE
+test.c
+--show-sese-regions
+^Function contains 3 single-entry, single-exit regions:$
+^Region starting at \(3, [0-9]+\) 1: IF !\(x % 7 != 0\) THEN GOTO 2 ends at \(8, [0-9]+\) 3: IF x < 10 THEN GOTO 1$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 4 ends at \(13, [0-9]+\) 5: SKIP$
+^Region starting at \(2, [0-9]+\) x = 0; ends at \(9, [0-9]+\) SKIP$
+^EXIT=0$
+^SIGNAL=0$

regression/goto-instrument/region-analysis-17/test.c

@@ -0,0 +1,27 @@
+int main(int argc, char **argv)
+{
+  // A function containing 2 SESE regions, one for the outer if-block
+  // and one for the while-loop, despite a continue edge that branches
+  // directly to the top of the loop.
+
+  if(argc % 5)
+  {
+    int x = 0;
+  top_continue:
+    do
+    {
+      if(x % 7)
+      {
+        ++x;
+        goto top_continue;
+      }
+      ++x;
+    } while(x < 10);
+  }
+  else
+  {
+    --argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-17/test.desc

@@ -0,0 +1,14 @@
+FUTURE
+test.c
+--show-sese-regions
+^Function contains 2 single-entry, single-exit regions:$
+^Region starting at \(2, [0-9]+\) x = 0; ends at \(10, [0-9]+\) 3: SKIP$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 4 ends at \(14, [0-9]+\) 5: SKIP$
+^EXIT=0$
+^SIGNAL=0$
+--
+--
+The program test.c produces what is effectively two nested loops, but natural-loop
+analysis conflates the two. This would risk identifying a region that straddles the
+boundary between those two loops, and so SESE analysis currently conservatively refrains
+from identifying the regions.

regression/goto-instrument/region-analysis-2/test.c

@@ -0,0 +1,27 @@
+int main(int argc, char **argv)
+{
+  // Function containing two SESE regions: the inner if-block and the whole
+  // function. The outer if-block is specifically *not* a region due to
+  // a spoiling incoming edge.
+  if(argc % 7)
+    goto target;
+
+  if(argc % 5)
+  {
+    if(argc % 2)
+    {
+      ++argc;
+    }
+    else
+    {
+      --argc;
+    }
+  }
+  else
+  {
+  target:
+    ++argc;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-2/test.desc

@@ -0,0 +1,8 @@
+CORE
+test.c
+--show-sese-regions
+^Function contains 2 single-entry, single-exit regions:$
+^Region starting at \(4, [0-9]+\) IF !\(argc % 2 != 0\) THEN GOTO 1 ends at \(8, [0-9]+\) 2: SKIP$
+^Region starting at \(0, [0-9]+\) IF argc % 7 != 0 THEN GOTO 3 ends at \(11, [0-9]+\) 4: SKIP$
+^EXIT=0$
+^SIGNAL=0$

regression/goto-instrument/region-analysis-3/test.c

@@ -0,0 +1,23 @@
+int main(int argc, char **argv)
+{
+  // Function containing two SESE regions: the inner if-block and the whole
+  // function. The outer if-block is specifically *not* a region due to
+  // a spoiling outgoing edge,
+  if(argc % 5)
+  {
+    if(argc % 2)
+    {
+      ++argc;
+    }
+    else
+    {
+      --argc;
+    }
+  }
+  else
+  {
+    return argc - 1;
+  }
+
+  return argc;
+}

regression/goto-instrument/region-analysis-3/test.desc

@@ -0,0 +1,8 @@
+CORE
+test.c
+--show-sese-regions
+^Function contains 2 single-entry, single-exit regions:$
+^Region starting at \(1, [0-9]+\) IF !\(argc % 2 != 0\) THEN GOTO 1 ends at \(5, [0-9]+\) 2: SKIP$
+^Region starting at \(0, [0-9]+\) IF !\(argc % 5 != 0\) THEN GOTO 3 ends at \(12, [0-9]+\) 5: END_FUNCTION$
+^EXIT=0$
+^SIGNAL=0$