[PATCH] D118601: [DebugInfo][InstrRef] Add a max-stack-slots-to-track limitation / cut-out

[Orlando Cazalet-Hyams via Phabricator via llvm-commits]

Orlando accepted this revision.
Orlando added a comment.
This revision is now accepted and ready to land.

LGTM



================
Comment at: llvm/test/DebugInfo/MIR/InstrRef/spill-slot-limits.mir:26-27
+## Test that if there's no limit, we _do_ get some locations.
+# NOLIMIT:      DBG_INSTR_REF 1, 0
+# NOLIMIT-NEXT: DBG_VALUE $esi
+#
----------------
jmorse wrote:
> Orlando wrote:
> > I'm confused - why don't we see this initial non-spill location for the `-livedebugvalues-max-stack-slots=0` case (in regular `CHECK` directives above)?
> The test spills to the stack and restores to $esi -- if we don't track it through the stack, there'll be no location. I suppose a clearer test would test each stage.
Oops what I really meant was "why don't we see a $eax location" and, now I look again, I can answer the myself: $eax is clobbered before we get to the debug instruction (and, as per the point of this patch, we don't follow the value into the stack). Sorry for the noise!


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D118601/new/

https://reviews.llvm.org/D118601