[PATCH] D140460: [RISCV][MC] Add support for experimental zfa extension

Fri Jan 13 00:58:08 PST 2023

craig.topper added a comment.

Can we split this into two patches? One for fli and one for everything else. I think everything but fli is ready to go in. Split the CodeGen patch the same way. That way we can get some stuff committed.

================
Comment at: llvm/lib/Target/RISCV/AsmParser/RISCVAsmParser.cpp:1633
+      APFloat SpecialVal = APFloat::getInf(APFloat::IEEEsingle());
+      Operands.push_back(RISCVOperand::createFPImm(
+          SpecialVal.bitcastToAPInt().getZExtValue(), S, isRV64()));
----------------
craig.topper wrote:
> Rather than converting to FPImm here. I think maybe should store the 5-bit immediate in the Operands array. This would require validating the floating point values and converting to integers in this function instead of in addFPImmOperands.
I guess we shouldn’t do that because we need may need multiple isFPImm to handle the different min values for the different sizes.

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D140460/new/

https://reviews.llvm.org/D140460