[lld] r299748 - [Core] Fix parallel_for for Linux

[James Henderson via llvm-commits]

Author: jhenderson
Date: Fri Apr  7 03:11:28 2017
New Revision: 299748

URL: http://llvm.org/viewvc/llvm-project?rev=299748&view=rev
Log:
[Core] Fix parallel_for for Linux

r299635 exposed a latent bug in the Linux implementation of parallel_for, which
resulted in it calling the function outside of the range requested, resulting
later in a segmentation fault. This change fixes this issue and adds a unit test.

Modified:
    lld/trunk/include/lld/Core/Parallel.h
    lld/trunk/unittests/CoreTests/ParallelTest.cpp

Modified: lld/trunk/include/lld/Core/Parallel.h
URL: http://llvm.org/viewvc/llvm-project/lld/trunk/include/lld/Core/Parallel.h?rev=299748&r1=299747&r2=299748&view=diff
==============================================================================
--- lld/trunk/include/lld/Core/Parallel.h (original)
+++ lld/trunk/include/lld/Core/Parallel.h Fri Apr  7 03:11:28 2017
@@ -318,12 +318,11 @@ void parallel_for(IndexTy Begin, IndexTy
 
   TaskGroup Tg;
   IndexTy I = Begin;
-  for (; I < End; I += TaskSize) {
+  for (; I + TaskSize < End; I += TaskSize) {
     Tg.spawn([=, &Fn] {
       for (IndexTy J = I, E = I + TaskSize; J != E; ++J)
         Fn(J);
     });
-    Begin += TaskSize;
   }
   Tg.spawn([=, &Fn] {
     for (IndexTy J = I; J < End; ++J)

Modified: lld/trunk/unittests/CoreTests/ParallelTest.cpp
URL: http://llvm.org/viewvc/llvm-project/lld/trunk/unittests/CoreTests/ParallelTest.cpp?rev=299748&r1=299747&r2=299748&view=diff
==============================================================================
--- lld/trunk/unittests/CoreTests/ParallelTest.cpp (original)
+++ lld/trunk/unittests/CoreTests/ParallelTest.cpp Fri Apr  7 03:11:28 2017
@@ -29,3 +29,18 @@ TEST(Parallel, sort) {
   lld::parallel_sort(std::begin(array), std::end(array));
   ASSERT_TRUE(std::is_sorted(std::begin(array), std::end(array)));
 }
+
+TEST(Parallel, parallel_for) {
+  // We need to test the case with a TaskSize > 1. We are white-box testing
+  // here. The TaskSize is calculated as (End - Begin) / 1024 at the time of
+  // writing.
+  uint32_t range[2050];
+  std::fill(range, range + 2050, 1);
+  lld::parallel_for(0, 2049, [&range](size_t I) { ++range[I]; });
+
+  uint32_t expected[2049];
+  std::fill(expected, expected + 2049, 2);
+  ASSERT_TRUE(std::equal(range, range + 2049, expected));
+  // Check that we don't write past the end of the requested range.
+  ASSERT_EQ(range[2049], 1);
+}