[llvm-dev] Substitute instruction with a jump to a library code

[Rail Shafigulin via llvm-dev]

I'm trying to find a way to emulate a floating point instruction, say a
floating point add. My understanding is that in order to do that I need to
execute

setOperationAction(ISD::FADD, (MVT::f32, Expand);
setOperationAction(ISD::FADD, (MVT::f64, Expand);

in MyTargetISelLowering.cpp, MyTargetLowering::MyTargetLowering(...).

However for some reason I'm still seeing a floating point add in the final
assembly. I tried running my test code (provided below) on MSP430 and can
see it execute a jump instruction to a label that is clearly located in the
library code.

Here is my test code:

; ModuleID = 'simple-fadd.c'
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-pc-linux-gnu"

; Function Attrs: nounwind readnone uwtable
define float @fadd(float %a, float %b) #0 {
  %1 = fadd float %a, %b
  ret float %1
}

attributes #0 = { nounwind readnone uwtable "less-precise-fpmad"="false"
"no-frame-pointer-elim"="false" "no-infs-fp-math"="false"
"no-nans-fp-math"="false" "stack-protector-buffer-size"="8"
"unsafe-fp-math"="false" "use-soft-float"="false" }

!llvm.ident = !{!0}

!0 = metadata !{metadata !"Ubuntu clang version 3.4.0-2ubuntu1~trusty1
(tags/RELEASE_340/final) (based on LLVM 3.4.0)"}


And here is what I see in MSP430

.text
.syntax unified
.cpu msp430
.eabi_attribute 6, 1 @ Tag_CPU_arch
.eabi_attribute 8, 1 @ Tag_ARM_ISA_use
.eabi_attribute 20, 1 @ Tag_ABI_FP_denormal
.eabi_attribute 21, 1 @ Tag_ABI_FP_exceptions
.eabi_attribute 23, 3 @ Tag_ABI_FP_number_model
.eabi_attribute 24, 1 @ Tag_ABI_align_needed
.eabi_attribute 25, 1 @ Tag_ABI_align_preserved
.file "simple-fadd.ll"
.globl fadd
.align 2
.type fadd,%function
fadd:                                   @ @fadd
.fnstart
.Leh_func_begin0:
@ BB#0:
push {lr}
bl __addsf3
pop {lr}
mov pc, lr
.Ltmp0:
.size fadd, .Ltmp0-fadd
.Leh_func_end0:
.fnend


.ident "Ubuntu clang version 3.6.0-2ubuntu1~trusty1
(tags/RELEASE_360/final) (based on LLVM 3.6.0)"

One can clearly see a bl __addsf3 instruction. However for some reason my
target produces a lf.add.s (which is a floating point add) instead of
branching to a library code. Clearly I'm missing something, but I can't
figure out what exactly. Would anyone care to help?


-- 
Rail Shafigulin
Software Engineer
Esencia Technologies
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20151229/0e7d7d61/attachment.html>