[PATCH] D48779: [llvm-exegesis] Delegate the decision of cycle counter name to the target

[Clement Courbet via Phabricator via llvm-commits]

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

Just two minor comments.



================
Comment at: tools/llvm-exegesis/lib/Latency.cpp:98
+const char *LatencyBenchmarkRunner::getCounterName() const {
+  // FIXME: Do not require SchedModel for latency.
+  if (!State.getSubtargetInfo().getSchedModel().hasExtraProcessorInfo())
----------------
I think you can remove this fixme now.


================
Comment at: tools/llvm-exegesis/lib/Latency.cpp:117
   int64_t MinLatency = std::numeric_limits<int64_t>::max();
-  const char *CounterName = State.getSubtargetInfo()
-                                .getSchedModel()
-                                .getExtraProcessorInfo()
-                                .PfmCounters.CycleCounter;
-  if (!CounterName)
-    llvm::report_fatal_error("sched model does not define a cycle counter");
+  const char *CounterName = getCounterName();
   const pfm::PerfEvent CyclesPerfEvent(CounterName);
----------------
You need to keep the test for nullity, just in case the implementation returns nullptr.


Repository:
  rL LLVM

https://reviews.llvm.org/D48779