[AArch64][FPenv] Update chain of int to fp conversion

[dnsampaio]

Summary

When using strict fp, it is required to update the
chain when performing integer type promotion of a
operand to a integer to floating point conversion.

[john.brawn]

llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp

1244–1245

Instead of setting this per-opcode in the switch, it would be simpler to do

bool IsStrict = N->isStrictFPOpcode()

and then use IsStrict to decide if there's another result. This is also more similar to how we handle strict opcodes elsewhere in legalization.

llvm/test/CodeGen/AArch64/strict-fp-int-promote.mir

67 ↗

(On Diff #244596)

It would be good to have a test of uitfp as well. Also, is there any specific reason this is a .mir test instead of a .ll test?

[dnsampaio]

Changes requested: IR test, chain changed depending on isStrictFPOpcode, signed and unsigned tests

[craig.topper]

LGTM