Fix populateFuncOpConversionPattern in nvidia after llvm-update

Author: karupayun

third_party/nvidia/lib/TritonNVIDIAGPUToLLVM/TritonGPUToLLVM.cpp

@@ -97,7 +97,8 @@ struct ConvertTritonGPUToLLVM
     TritonLLVMFunctionConversionTarget funcTarget(*context);
     RewritePatternSet funcPatterns(context);
     mlir::triton::populateFuncOpConversionPattern(
-        typeConverter, funcPatterns, targetInfo, patternBenefitDefault);
+        typeConverter, funcPatterns, targetInfo, patternBenefitDefault,
+        /*symTable=*/nullptr);
     if (failed(
             applyPartialConversion(mod, funcTarget, std::move(funcPatterns))))
       return signalPassFailure();