728490257e
Similar to 806761a7629df268c8aed49657aeccffa6bca449 -mtriple= specifies the full target triple while -march= merely sets the architecture part of the default target triple (e.g. Windows, macOS), leaving a target triple which may not make sense. Therefore, -march= is error-prone and not recommended for tests without a target triple. The issue has been benign as we recognize sparc*-apple-darwin as ELF instead of rejecting it outrightly.
17 lines
750 B
LLVM
17 lines
750 B
LLVM
; RUN: llc < %s -mtriple=sparc
|
|
; PR 1557
|
|
|
|
target datalayout = "E-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-f128:128:128"
|
|
@llvm.global_ctors = appending global [1 x { i32, ptr, ptr }] [ { i32, ptr, ptr } { i32 65535, ptr @set_fast_math, ptr null } ] ; <[1 x { i32, void ()*, i8* }]*> [#uses=0]
|
|
|
|
define internal void @set_fast_math() nounwind {
|
|
entry:
|
|
%fsr = alloca i32 ; <i32*> [#uses=4]
|
|
call void asm "st %fsr, $0", "=*m"(ptr elementtype(i32) %fsr) nounwind
|
|
%0 = load i32, ptr %fsr, align 4 ; <i32> [#uses=1]
|
|
%1 = or i32 %0, 4194304 ; <i32> [#uses=1]
|
|
store i32 %1, ptr %fsr, align 4
|
|
call void asm sideeffect "ld $0, %fsr", "*m"(ptr elementtype(i32) %fsr) nounwind
|
|
ret void
|
|
}
|