libcompiler_rt: Update Makefile for arm64 (part 1)

[emaste]

lib/libcompiler_rt/Makefile

202–203 ↗

(On Diff #4469)

these first two already appear in SRCF above

[emaste]

lib/libcompiler_rt/Makefile

198–210 ↗

(On Diff #4469)

Discussed with @dim on IRC; this whole section is probably incorrect.
These implement PPC's "double double" long double support.

[emaste]

lib/libcompiler_rt/Makefile

40 ↗

(On Diff #4469)

// // This file implements the following soft-float comparison routines:      
//
//   __eqtf2   __getf2   __unordtf2
//   __letf2   __gttf2
//   __lttf2
//   __netf2

54 ↗

(On Diff #4469)

//===-- lib/extenddftf2.c - double -> quad conversion -------------*- C -*-===//

55 ↗

(On Diff #4469)

//===-- lib/extendsftf2.c - single -> quad conversion -------------*- C -*-===//

133 ↗

(On Diff #4469)

//===-- lib/truncdfsf2.c - quad -> double conversion --------------*- C -*-===//

134 ↗

(On Diff #4469)

//===-- lib/trunctfsf2.c - quad -> single conversion --------------*- C -*-===//

156 ↗

(On Diff #4469)

//===-- lib/floatsitf.c - integer -> quad-precision conversion ----*- C -*-===//

[emaste]

lib/libcompiler_rt/Makefile

156 ↗

(On Diff #4469)

This should move the common case above I think: it is not provided by libc.a on arm/mips AFAICT

[emaste]

• move floatsitf to common list: it is not provided by libc.a on arm/mips
• remove use of PPC's long double "double double" functions - this will need to be fixed by new functionality implemented in compiler-rt

[emaste]

Missing functionality in these reviews:
D2173 multc3
D2174 floatditf, floatunditf

[emaste]

Closed by commit rS280853 (authored by @emaste).