[RISCV] Store fli min/nan/inf in index form in RISCVOperand.
authorCraig Topper <craig.topper@sifive.com>
Wed, 8 Mar 2023 17:10:18 +0000 (09:10 -0800)
committerCraig Topper <craig.topper@sifive.com>
Wed, 8 Mar 2023 17:24:58 +0000 (09:24 -0800)
commit61ae2aaf52b4c40bf297daa4bbf36d46bfecd1d3
tree8affa7e73fc2aa595e0c316514f15bf86d06c469
parentd23b9fa61d190a7a95a83303987b186450d5c58a
[RISCV] Store fli min/nan/inf in index form in RISCVOperand.

Instead of converting to FP value, store them as Index using an
immediate operand.

Do the same for the explicit index form.

This avoids using the FP32 version of these special values as the
representation for fli.h and fli.d. inf/nan aren't so bad, but
"min" is problematic as the current implementation allows fli.d and
fli.h to accept the decimal version of the fp32 minimum value. I
will submit another patch to fix that.

Differential Revision: https://reviews.llvm.org/D145535
llvm/lib/Target/RISCV/AsmParser/RISCVAsmParser.cpp
llvm/lib/Target/RISCV/MCTargetDesc/RISCVBaseInfo.h