95
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
1 ; RUN: llc -verify-machineinstrs -march=amdgcn -mcpu=SI -mattr=-promote-alloca < %s | FileCheck -check-prefix=SI-ALLOCA -check-prefix=SI %s
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
2 ; RUN: llc -verify-machineinstrs -march=amdgcn -mcpu=SI -mattr=+promote-alloca < %s | FileCheck -check-prefix=SI-PROMOTE -check-prefix=SI %s
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
3
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
4 declare i32 @llvm.SI.tid() nounwind readnone
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
5 declare void @llvm.AMDGPU.barrier.local() nounwind noduplicate
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
6
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
7 ; The required pointer calculations for the alloca'd actually requires
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
8 ; an add and won't be folded into the addressing, which fails with a
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
9 ; 64-bit pointer add. This should work since private pointers should
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
10 ; be 32-bits.
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
11
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
12 ; SI-LABEL: {{^}}test_private_array_ptr_calc:
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
13
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
14 ; FIXME: We end up with zero argument for ADD, because
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
15 ; SIRegisterInfo::eliminateFrameIndex() blindly replaces the frame index
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
16 ; with the appropriate offset. We should fold this into the store.
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
17 ; SI-ALLOCA: v_add_i32_e32 [[PTRREG:v[0-9]+]], vcc, 0, v{{[0-9]+}}
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
18 ; SI-ALLOCA: buffer_store_dword {{v[0-9]+}}, [[PTRREG]], s[{{[0-9]+:[0-9]+}}]
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
19 ;
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
20 ; FIXME: The AMDGPUPromoteAlloca pass should be able to convert this
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
21 ; alloca to a vector. It currently fails because it does not know how
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
22 ; to interpret:
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
23 ; getelementptr [4 x i32], [4 x i32]* %alloca, i32 1, i32 %b
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
24
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
25 ; SI-PROMOTE: v_add_i32_e32 [[PTRREG:v[0-9]+]], vcc, 16
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
26 ; SI-PROMOTE: ds_write_b32 [[PTRREG]]
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
27 define void @test_private_array_ptr_calc(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %inA, i32 addrspace(1)* noalias %inB) {
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
28 %alloca = alloca [4 x i32], i32 4, align 16
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
29 %tid = call i32 @llvm.SI.tid() readnone
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
30 %a_ptr = getelementptr i32, i32 addrspace(1)* %inA, i32 %tid
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
31 %b_ptr = getelementptr i32, i32 addrspace(1)* %inB, i32 %tid
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
32 %a = load i32, i32 addrspace(1)* %a_ptr
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
33 %b = load i32, i32 addrspace(1)* %b_ptr
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
34 %result = add i32 %a, %b
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
35 %alloca_ptr = getelementptr [4 x i32], [4 x i32]* %alloca, i32 1, i32 %b
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
36 store i32 %result, i32* %alloca_ptr, align 4
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
37 ; Dummy call
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
38 call void @llvm.AMDGPU.barrier.local() nounwind noduplicate
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
39 %reload = load i32, i32* %alloca_ptr, align 4
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
40 %out_ptr = getelementptr i32, i32 addrspace(1)* %out, i32 %tid
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
41 store i32 %reload, i32 addrspace(1)* %out_ptr, align 4
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
42 ret void
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
43 }
|
Kaito Tokumori <e105711@ie.u-ryukyu.ac.jp>
parents:
diff
changeset
|
44
|