Files
clang-p2996/llvm/test/CodeGen/SPARC/zerostructcall.ll
Fangrui Song 728490257e [Sparc,test] Change llc -march= to -mtriple=
Similar to 806761a762

-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.
2024-12-15 10:29:34 -08:00

52 lines
1.1 KiB
LLVM

; RUN: llc < %s -mtriple=sparc | FileCheck %s
; CHECK-LABEL: struct_ptr_test
; CHECK: call struct_ptr_fn
; CHECK-NEXT: st %i0, [%fp+-4]
; CHECK-NEXT: ret
%struct.S = type {}
define void @struct_ptr_test(i32 %i) {
entry:
%i.addr = alloca i32, align 4
store i32 %i, ptr %i.addr, align 4
%0 = bitcast ptr %i.addr to ptr
call void @struct_ptr_fn(ptr byval(%struct.S) align 1 %0)
ret void
}
declare void @struct_ptr_fn(ptr byval(%struct.S) align 1)
; CHECK-LABEL: struct_test
; CHECK: call struct_fn
; CHECK-NEXT: nop
; CHECK-NEXT: ret
%struct.U = type {}
@a = internal global [1 x %struct.U] zeroinitializer, align 1
define void @struct_test() {
entry:
tail call void @struct_fn(ptr byval(%struct.U) align 1 @a)
ret void
}
; CHECK-LABEL: struct_arg_test
; CHECK: call struct_arg_fn
; CHECK-NEXT: nop
; CHECK-NEXT: ret
declare void @struct_fn(ptr byval(%struct.U) align 1)
@b = internal global [1 x %struct.U] zeroinitializer, align 1
define void @struct_arg_test() {
entry:
tail call void @struct_arg_fn(ptr byval(%struct.U) align 1 @b)
ret void
}
declare void @struct_arg_fn(ptr byval(%struct.U) align 1)