[llvm-stress] Ensure that the C++11 random device respects its min/max values (PR32585)

[RKSimon]

Summary

As noted on PR32585, the change in D29780/rL295325 resulted in calls to Rand32() (values 0 -> 0xFFFFFFFF) but the min()/max() operators indicated it would be (0 -> 0x7FFFF).

This patch changes the random operator to call Rand() instead which does respect the 0 -> 0x7FFFF range.

Not sure the best way to test this - would an assert in the operator be enough?

[vsk]

Asserting 'r <= max()' in the operator would be a good check.

[RKSimon]

Added assertion that the random value doesn't exceed the maximum.

Stripped UTF8 BOM from my previous patch

[mclow.lists]

After thinking about this, I agree with @bogner 's comment that we should just remove the whole Random class, and use the C++11 random number facilities instead.

[RKSimon]

In D34089#779023, @mclow.lists wrote:

After thinking about this, I agree with @bogner 's comment that we should just remove the whole Random class, and use the C++11 random number facilities instead.

Is there a C++11 random class that covers the features we need (same results on different targets, drivable by input seed)?

[mclow.lists]

Is there a C++11 random class that covers the features we need (same results on different targets, drivable by input seed)?

Given that the libc++ mersenne_twister_engine tests test specifically that, yes.

see <libc++ root>/test/std/numerics/rand/rand.eng/rand.eng.mers/ctor_sseq.pass.cpp

[RKSimon]

In D34089#779145, @mclow.lists wrote:

Is there a C++11 random class that covers the features we need (same results on different targets, drivable by input seed)?

Given that the libc++ mersenne_twister_engine tests test specifically that, yes.

see <libc++ root>/test/std/numerics/rand/rand.eng/rand.eng.mers/ctor_sseq.pass.cpp

I've created D34157 for the refactor which is going to take a few days, in the meantime are there any objections to committing this patch to fix PR32585's current assertion in fuzz testing?

[RKSimon]

ping - D34157 pretty much needs to be started again from scratch and it'd be nice to at least stop the assert from happening from long run fuzz tests.....

[hfinkel]

Yes, go ahead. Let's fix the current code in the mean time.

llvm/trunk/tools/llvm-stress/llvm-stress.cpp

public:
 
   /// Rand operator for STL algorithms.
   ptrdiff_t operator()(ptrdiff_t y) {
     return  Rand64() % y;
   }
 
   /// Make this like a C++11 random device
   typedef uint32_t result_type;
-  uint32_t operator()() { return Rand32(); }
   static constexpr result_type min() { return 0; }
   static constexpr result_type max() { return 0x7ffff; }
-  
+  uint32_t operator()() {
+    uint32_t Val = Rand();
+    assert(Val <= max() && "Random value out of range");
+    return Val;
+  }
 private:
   unsigned Seed;
 };
 
 /// Generate an empty function with a default argument list.
 Function *GenEmptyFunction(Module *M) {
   // Define a few arguments
   LLVMContext &Context = M->getContext();