The only supported calling convention for X86-64 uses
commit6985f085869c6b99be0278e2d24c54be3e5a1612
authorDale Johannesen <dalej@apple.com>
Fri, 23 Jul 2010 00:30:35 +0000 (23 00:30 +0000)
committerDale Johannesen <dalej@apple.com>
Fri, 23 Jul 2010 00:30:35 +0000 (23 00:30 +0000)
tree412ac434690af7e7a59e0de528dd3b2282ea8fe2
parent4c45d5c0c0d505628f9332828e79019c2c20c684
The only supported calling convention for X86-64 uses
SSE, so we can't return floating point values if this
is disabled.  Detect this error for clang.

With SSE1 only, f64 is a problem; it can be done, but
neither llvm-gcc nor clang has ever generated correct
code for it.  Since nobody noticed this I think it's
OK to treat it as an error for now.

This also handles SSE-sized vectors of floating point.
82076868204109.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109201 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/X86ISelLowering.cpp