Remove faulty assertion in llvm-pdbutil

[amccarth]

Summary

If a class's first data member is an instance of an empty class, then an assertion in the PrettyClassLayoutGraphicalDumper would fail. The storage is reserved, but it's not marked as in use.

As far as I understand, it's the assertion that's faulty, so I removed it and updated the nearby comment. If you think the MapUse is computed wrong in this case, let me know and I'll investigate further.

Found by running llvm-pdbutil against its own PDB, and this assertion would fail on HashAdjusters, which is a HashTable whose first data member is a TraitsT, which is a PdbHashTraits<T>, which is an empty struct. (The struct has a specialization for uint32_t, but that specialization doesn't apply here because the T is actually ulittle32_t.)

[amccarth]

I meant to mention that the instance of the empty class is reported as 8 bytes of padding (in a 64-bit build), in case that's relevant to anybody's understanding of the issue.

[zturner]

If the tests pass, this is probably fine. Honestly this code is really
complex and while I think it’s useful, it may not be useful enough to
justify the complexity so I’m thinking about getting rid of it

llvm/trunk/tools/llvm-pdbutil/PrettyClassLayoutGraphicalDumper.cpp

bool PrettyClassLayoutGraphicalDumper::start(const UDTLayoutBase &Layout) {
   const BitVector &UseMap = Layout.usedBytes();
   int NextPaddingByte = UseMap.find_first_unset();
 
   for (auto &Item : Layout.layout_items()) {
     // Calculate the absolute offset of the first byte of the next field.
     uint32_t RelativeOffset = Item->getOffsetInParent();
     CurrentAbsoluteOffset = ClassOffsetZero + RelativeOffset;
 
-    // Since there is storage there, it should be set!  However, this might
-    // be an empty base, in which case it could extend outside the bounds of
-    // the parent class.
+    // This might be an empty base, in which case it could extend outside the
+    // bounds of the parent class.
     if (RelativeOffset < UseMap.size() && (Item->getSize() > 0)) {
-      assert(UseMap.test(RelativeOffset));
-
       // If there is any remaining padding in this class, and the offset of the
       // new item is after the padding, then we must have just jumped over some
       // padding.  Print a padding row and then look for where the next block
       // of padding begins.
       if ((NextPaddingByte >= 0) &&
           (RelativeOffset > uint32_t(NextPaddingByte))) {
         printPaddingRow(RelativeOffset - NextPaddingByte);
         NextPaddingByte = UseMap.find_next_unset(RelativeOffset);