[Lldb-commits] [PATCH] D42083: [lldb][PPC64] Fixed long double variables dump

[Pavel Labath via Phabricator via lldb-commits]

labath added a comment.

This adds a special case to "long double" logic, which is already a special case compared to "float" and "double" cases. This was written this way (see http://reviews.llvm.org/D8417) because the x86 long double type is special, but if I understand this correctly, for the ppc type, the "default" logic of just using the item_byte_size would be correct.

What if we reverse the conditions here to be something like?

  offset_t byte_size = item_byte_size;
  if (&semantics == &x87DoubleExtended)
    byte_size = (APFloat::getSizeInBits(semantics)+7)/8;


Repository:
  rL LLVM

https://reviews.llvm.org/D42083