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 xcore-apple-darwin as ELF instead
of rejecting it outrightly.
51 lines
1.1 KiB
LLVM
51 lines
1.1 KiB
LLVM
; RUN: llc < %s -mtriple=xcore | FileCheck %s
|
|
|
|
define i32 @load32(ptr %p, i32 %offset) nounwind {
|
|
entry:
|
|
; CHECK-LABEL: load32:
|
|
; CHECK: ldw r0, r0[r1]
|
|
%0 = getelementptr i32, ptr %p, i32 %offset
|
|
%1 = load i32, ptr %0, align 4
|
|
ret i32 %1
|
|
}
|
|
|
|
define i32 @load32_imm(ptr %p) nounwind {
|
|
entry:
|
|
; CHECK-LABEL: load32_imm:
|
|
; CHECK: ldw r0, r0[11]
|
|
%0 = getelementptr i32, ptr %p, i32 11
|
|
%1 = load i32, ptr %0, align 4
|
|
ret i32 %1
|
|
}
|
|
|
|
define i32 @load16(ptr %p, i32 %offset) nounwind {
|
|
entry:
|
|
; CHECK-LABEL: load16:
|
|
; CHECK: ld16s r0, r0[r1]
|
|
; CHECK-NOT: sext
|
|
%0 = getelementptr i16, ptr %p, i32 %offset
|
|
%1 = load i16, ptr %0, align 2
|
|
%2 = sext i16 %1 to i32
|
|
ret i32 %2
|
|
}
|
|
|
|
define i32 @load8(ptr %p, i32 %offset) nounwind {
|
|
entry:
|
|
; CHECK-LABEL: load8:
|
|
; CHECK: ld8u r0, r0[r1]
|
|
; CHECK-NOT: zext
|
|
%0 = getelementptr i8, ptr %p, i32 %offset
|
|
%1 = load i8, ptr %0, align 1
|
|
%2 = zext i8 %1 to i32
|
|
ret i32 %2
|
|
}
|
|
|
|
@GConst = internal constant i32 42
|
|
define i32 @load_cp() nounwind {
|
|
entry:
|
|
; CHECK-LABEL: load_cp:
|
|
; CHECK: ldw r0, cp[GConst]
|
|
%0 = load i32, ptr @GConst
|
|
ret i32 %0
|
|
}
|