[AMDGPU] Fix running ResourceUsageAnalysis

[sebastian-ne]

Summary

Clear the map when running the analysis multiple times.
The assertion below that should ensure that every function is only
analyzed once triggered sometimes (once every ~70 compiles of some
graphics pipelines).

If someone knows how to add a testcase, that would be nice, because
I don't know how to test for this.

[arsenm]

I don't think there's a way to test this short of a unit test running a custom pass pipeline

[madhur13490]

If you want to you can emit some debugging message after clearing the map and have a llc lit test with the debug flag to CHECK the debugging message.

[sebastian-ne]

The previous fix here was actually wrong (and failed lit tests), resetting the map should be done in doInitialization.

llvm/lib/Target/AMDGPU/AMDGPUResourceUsageAnalysis.cpp

if (HasIndirectCall)
     propagateIndirectCallRegisterUsage();
 
   return false;
 }
 
 AMDGPUResourceUsageAnalysis::SIFunctionResourceInfo
 AMDGPUResourceUsageAnalysis::analyzeResourceUsage(
     const MachineFunction &MF, const TargetMachine &TM) const {
+  CallGraphResourceInfo.clear();
+
   SIFunctionResourceInfo Info;
 
   const SIMachineFunctionInfo *MFI = MF.getInfo<SIMachineFunctionInfo>();
   const GCNSubtarget &ST = MF.getSubtarget<GCNSubtarget>();
   const MachineFrameInfo &FrameInfo = MF.getFrameInfo();
   const MachineRegisterInfo &MRI = MF.getRegInfo();
   const SIInstrInfo *TII = ST.getInstrInfo();
   const SIRegisterInfo &TRI = TII->getRegisterInfo();