[PATCH] D48408: [Debugify] Diagnose mis-sized dbg.values

[Vedant Kumar via Phabricator via llvm-commits]

vsk added a comment.

In https://reviews.llvm.org/D48408#1141148, @probinson wrote:

> I remember there being a difference of opinion about the "correct" size of an 80-bit float in a 128-bit container.  How would that case be handled here?


That's a good question. Currently, we emit IR that would not trigger the diagnostic this patch introduces (https://godbolt.org/g/7DXyJy):

  call void @llvm.dbg.value(metadata x86_fp80 %10, metadata !14, metadata !DIExpression())
  !10 = !DIBasicType(name: "long double", size: 128, encoding: DW_ATE_float)
  !14 = !DILocalVariable(name: "x", scope: !7, file: !1, line: 2, type: !10)

getSizeInBits() reports back 128 for an x86_fp80.


https://reviews.llvm.org/D48408