This final patch simplifies the logic in SelectionDAGISel.cpp with no functional change. This construct was in use:

    if (I.getType()->isFloatingPoint())
      visitFPBinary(I, ISD::FADD, ISD::VADD); 
    else
      visitIntBinary(I, ISD::ADD, ISD::VADD); 

Which confusingly sent even floating-point vector operations through the visitIntBinary routine. However, this was not a bug because visitFPBinary and visitIntBinary were identical (as is the vector opcode parameter).

This patch streamlines the logic and eliminates redundant methods.

— Gordon


Attachment: pr970-5.patch
Description: Binary data


_______________________________________________
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

Reply via email to