[PATCH] D30941: Better testing of schedule model instruction latencies/throughputs

[Hal Finkel via Phabricator via llvm-commits]

hfinkel added inline comments.


================
Comment at: lib/CodeGen/AsmPrinter/AsmPrinter.cpp:1413
+                             // TODO: uncomment when it gets matured
+                             : false /*STI.supportPrintSchedInfo()*/;
+  if (EnablePrintSchedInfo)
----------------
Why is this commented out? You have a target hook and it defaults to false, so that should be fine.


================
Comment at: lib/Target/X86/X86Subtarget.cpp:394
+
+std::string X86Subtarget::getSchedInfoStr(const MachineInstr &MI) const {
+  if (MI.isPseudo() || MI.isTerminator())
----------------
Why are these functions X86-specific?


================
Comment at: lib/Target/X86/X86Subtarget.h:628
 
+  bool supportPrintSchedInfo() const override { return true; }
+
----------------
Put the false and the TODO here.


https://reviews.llvm.org/D30941