[PATCH] D81852: [DebugInfo] Update MachineInstr interface to better support variadic DBG_VALUE instructions

[Stephen Tozer via Phabricator via llvm-commits]

StephenTozer marked 2 inline comments as done.
StephenTozer added inline comments.


================
Comment at: llvm/include/llvm/CodeGen/MachineInstr.h:466
   const MachineOperand& getOperand(unsigned i) const {
     assert(i < getNumOperands() && "getOperand() out of range!");
     return Operands[i];
----------------
vsk wrote:
> Is it possible to take this further and assert that getOperand() is not called on a debug instruction?
`getOperand()` is still used (internally) by the debug-specific getters, although it could be replaced by `Operands[i]`. I'm not sure if reducing the interface so drastically would be ideal though; code that is agnostic about the type of instruction it operates on - such as replacing register usage - would break for debug values, and have to be rewritten to accomodate. That wouldn't necessarily be a bad thing though, since we might actually want different code in all those places.


================
Comment at: llvm/test/CodeGen/MIR/Generic/dbg-value-missing-loc.mir:41
   bb.0.entry:
-    DBG_VALUE 1, 2, 3, 4
+    DBG_VALUE 0, $noreg, !12, !DIExpression()
+    $eax = XOR32rr undef $eax, undef $eax, implicit-def dead $eflags, debug-location !14
----------------
vsk wrote:
> What necessitates this change?
Some of the previous code dealing specifically with debug values had some statements along the lines of `if (MI.isDebugValue() && MI.getOperand(2).isMetadata()) {...}`. As of this patch we're now using specific functions, i.e. `getDebugVariable()`, that assert that the operand is the right type and then cast to it. AFAIUI this shouldn't be a breaking change as we never produce DBG_VALUEs without the "right" operands, like this test had.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D81852/new/

https://reviews.llvm.org/D81852