Fix DoNotOptimize() GCC compile error with some types (#1340) (#1424)

Non-const DoNotOptimize() can't compile when used with some types.
Example of code which can't compile:

char buffer3[3] = "";
benchmark::DoNotOptimize(buffer3);

Error message:

error: impossible constraint in 'asm'
  asm volatile("" : "+r"(value) : : "memory");

Introduced in 8545dfb (Fix DoNotOptimize() GCC copy overhead (#1340) (#1410))

The cause is compiler can't work with the +r constraint for types that can't
be placed perfectly in registers. For example, char array[3] can't be perfectly
fit in register on x86_64 so it requires placed in memory but constraint
doesn't allow that.

Solution
- Use +m,r constraint for the small objects so the compiler can decide to use
  register or/and memory
- For the big objects +m constraint is used which allows avoiding extra copy
  bug(see #1340)
- The same approach is used for the const version of DoNotOptimize()
  although the const version works fine with the "r" constraint only.
  Using mixed r,m constraint looks more general solution.

See
- Issue #1340 ([BUG] DoNotOptimize() adds overhead with extra copy of argument(gcc))
- Pull request #1410 (Fix DoNotOptimize() GCC copy overhead (#1340) #1410)
- Commit 8545dfb (Fix DoNotOptimize() GCC copy overhead (#1340) (#1410))

include/benchmark/benchmark.h

@@ -468,7 +468,7 @@ inline BENCHMARK_ALWAYS_INLINE
     typename std::enable_if<std::is_trivially_copyable<Tp>::value &&
                             (sizeof(Tp) <= sizeof(Tp*))>::type
     DoNotOptimize(Tp const& value) {
-  asm volatile("" : : "r"(value) : "memory");
+  asm volatile("" : : "r,m"(value) : "memory");
 }
 
 template <class Tp>
@@ -484,7 +484,7 @@ inline BENCHMARK_ALWAYS_INLINE
     typename std::enable_if<std::is_trivially_copyable<Tp>::value &&
                             (sizeof(Tp) <= sizeof(Tp*))>::type
     DoNotOptimize(Tp& value) {
-  asm volatile("" : "+r"(value) : : "memory");
+  asm volatile("" : "+m,r"(value) : : "memory");
 }
 
 template <class Tp>

test/donotoptimize_test.cc

@@ -29,6 +29,15 @@ struct BitRef {
 int main(int, char*[]) {
   // this test verifies compilation of DoNotOptimize() for some types
 
+  char buffer1[1] = "";
+  benchmark::DoNotOptimize(buffer1);
+
+  char buffer2[2] = "";
+  benchmark::DoNotOptimize(buffer2);
+
+  char buffer3[3] = "";
+  benchmark::DoNotOptimize(buffer3);
+
   char buffer8[8] = "";
   benchmark::DoNotOptimize(buffer8);
 
@@ -39,6 +48,25 @@ int main(int, char*[]) {
   benchmark::DoNotOptimize(buffer1024);
   benchmark::DoNotOptimize(&buffer1024[0]);
 
+  const char const_buffer1[1] = "";
+  benchmark::DoNotOptimize(const_buffer1);
+
+  const char const_buffer2[2] = "";
+  benchmark::DoNotOptimize(const_buffer2);
+
+  const char const_buffer3[3] = "";
+  benchmark::DoNotOptimize(const_buffer3);
+
+  const char const_buffer8[8] = "";
+  benchmark::DoNotOptimize(const_buffer8);
+
+  const char const_buffer20[20] = "";
+  benchmark::DoNotOptimize(const_buffer20);
+
+  const char const_buffer1024[1024] = "";
+  benchmark::DoNotOptimize(const_buffer1024);
+  benchmark::DoNotOptimize(&const_buffer1024[0]);
+
   int x = 123;
   benchmark::DoNotOptimize(x);
   benchmark::DoNotOptimize(&x);