Skip to content

[SDag] Notify listeners when deleting a node #66991

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

[SDag] Notify listeners when deleting a node #66991

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1087,6 +1087,9 @@ void SelectionDAG::RemoveDeadNode(SDNode *N){
}

void SelectionDAG::DeleteNode(SDNode *N) {
for (DAGUpdateListener *DUL = UpdateListeners; DUL; DUL = DUL->Next)
DUL->NodeDeleted(N, nullptr);

// First take this out of the appropriate CSE map.
RemoveNodeFromCSEMaps(N);

Expand Down
146 changes: 73 additions & 73 deletions llvm/test/CodeGen/AMDGPU/half.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2317,7 +2317,7 @@ define amdgpu_kernel void @global_extload_v16f16_to_v16f64(ptr addrspace(1) %out
; VI-NEXT: s_addc_u32 s3, s1, 0
; VI-NEXT: v_mov_b32_e32 v18, s3
; VI-NEXT: v_mov_b32_e32 v17, s2
; VI-NEXT: s_add_u32 s2, s0, 0x50
; VI-NEXT: s_add_u32 s2, s0, 0x70
; VI-NEXT: v_mov_b32_e32 v12, s1
; VI-NEXT: s_addc_u32 s3, s1, 0
; VI-NEXT: v_mov_b32_e32 v11, s0
Expand All @@ -2331,12 +2331,12 @@ define amdgpu_kernel void @global_extload_v16f16_to_v16f64(ptr addrspace(1) %out
; VI-NEXT: v_cvt_f32_f16_e32 v7, v6
; VI-NEXT: v_cvt_f32_f16_sdwa v8, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: s_waitcnt vmcnt(1)
; VI-NEXT: v_cvt_f32_f16_e32 v10, v2
; VI-NEXT: v_cvt_f32_f16_e32 v10, v0
; VI-NEXT: v_mov_b32_e32 v14, s3
; VI-NEXT: v_cvt_f64_f32_e32 v[6:7], v7
; VI-NEXT: v_cvt_f64_f32_e32 v[8:9], v8
; VI-NEXT: v_mov_b32_e32 v13, s2
; VI-NEXT: s_add_u32 s2, s0, 64
; VI-NEXT: s_add_u32 s2, s0, 0x60
; VI-NEXT: s_addc_u32 s3, s1, 0
; VI-NEXT: flat_store_dwordx4 v[15:16], v[6:9]
; VI-NEXT: v_mov_b32_e32 v16, s3
Expand All @@ -2347,37 +2347,37 @@ define amdgpu_kernel void @global_extload_v16f16_to_v16f64(ptr addrspace(1) %out
; VI-NEXT: v_cvt_f64_f32_e32 v[4:5], v6
; VI-NEXT: v_cvt_f64_f32_e32 v[6:7], v7
; VI-NEXT: v_mov_b32_e32 v15, s2
; VI-NEXT: s_add_u32 s2, s0, 0x70
; VI-NEXT: s_add_u32 s2, s0, 0x50
; VI-NEXT: s_addc_u32 s3, s1, 0
; VI-NEXT: flat_store_dwordx4 v[17:18], v[4:7]
; VI-NEXT: v_cvt_f32_f16_sdwa v17, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: v_cvt_f32_f16_sdwa v17, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: v_cvt_f64_f32_e32 v[4:5], v8
; VI-NEXT: v_cvt_f64_f32_e32 v[6:7], v9
; VI-NEXT: v_cvt_f32_f16_sdwa v9, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: v_cvt_f32_f16_sdwa v8, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: v_cvt_f32_f16_e32 v2, v1
; VI-NEXT: v_cvt_f32_f16_sdwa v8, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: v_cvt_f32_f16_e32 v0, v3
; VI-NEXT: s_add_u32 s0, s0, 64
; VI-NEXT: flat_store_dwordx4 v[11:12], v[4:7]
; VI-NEXT: v_cvt_f32_f16_sdwa v11, v0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: v_cvt_f32_f16_e32 v7, v3
; VI-NEXT: v_cvt_f64_f32_e32 v[3:4], v9
; VI-NEXT: v_cvt_f32_f16_e32 v9, v0
; VI-NEXT: v_cvt_f64_f32_e32 v[1:2], v2
; VI-NEXT: v_cvt_f64_f32_e32 v[5:6], v10
; VI-NEXT: v_cvt_f64_f32_e32 v[11:12], v11
; VI-NEXT: v_cvt_f64_f32_e32 v[9:10], v9
; VI-NEXT: s_add_u32 s0, s0, 0x60
; VI-NEXT: flat_store_dwordx4 v[13:14], v[1:4]
; VI-NEXT: v_cvt_f32_f16_sdwa v12, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: v_cvt_f64_f32_e32 v[5:6], v8
; VI-NEXT: v_cvt_f32_f16_e32 v8, v2
; VI-NEXT: v_cvt_f32_f16_sdwa v2, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
; VI-NEXT: v_cvt_f32_f16_e32 v7, v1
; VI-NEXT: v_cvt_f64_f32_e32 v[3:4], v0
; VI-NEXT: v_cvt_f64_f32_e32 v[0:1], v10
; VI-NEXT: v_cvt_f64_f32_e32 v[8:9], v8
; VI-NEXT: v_cvt_f64_f32_e32 v[10:11], v2
; VI-NEXT: flat_store_dwordx4 v[13:14], v[3:6]
; VI-NEXT: s_addc_u32 s1, s1, 0
; VI-NEXT: v_cvt_f64_f32_e32 v[0:1], v7
; VI-NEXT: v_cvt_f64_f32_e32 v[4:5], v7
; VI-NEXT: v_cvt_f64_f32_e32 v[6:7], v12
; VI-NEXT: v_cvt_f64_f32_e32 v[2:3], v17
; VI-NEXT: v_cvt_f64_f32_e32 v[7:8], v8
; VI-NEXT: v_mov_b32_e32 v20, s3
; VI-NEXT: v_mov_b32_e32 v14, s1
; VI-NEXT: v_mov_b32_e32 v13, s1
; VI-NEXT: v_mov_b32_e32 v19, s2
; VI-NEXT: v_mov_b32_e32 v13, s0
; VI-NEXT: flat_store_dwordx4 v[15:16], v[9:12]
; VI-NEXT: flat_store_dwordx4 v[19:20], v[0:3]
; VI-NEXT: flat_store_dwordx4 v[13:14], v[5:8]
; VI-NEXT: v_mov_b32_e32 v12, s0
; VI-NEXT: flat_store_dwordx4 v[15:16], v[8:11]
; VI-NEXT: flat_store_dwordx4 v[19:20], v[4:7]
; VI-NEXT: flat_store_dwordx4 v[12:13], v[0:3]
; VI-NEXT: s_endpgm
;
; GFX11-TRUE16-LABEL: global_extload_v16f16_to_v16f64:
Expand All @@ -2389,40 +2389,40 @@ define amdgpu_kernel void @global_extload_v16f16_to_v16f64(ptr addrspace(1) %out
; GFX11-TRUE16-NEXT: global_load_b128 v[0:3], v32, s[2:3]
; GFX11-TRUE16-NEXT: global_load_b128 v[4:7], v32, s[2:3] offset:16
; GFX11-TRUE16-NEXT: s_waitcnt vmcnt(1)
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v10, v1.l
; GFX11-TRUE16-NEXT: s_waitcnt vmcnt(0)
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v23, 16, v5
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v11, 16, v1
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v19, 16, v4
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v15, v7.l
; GFX11-TRUE16-NEXT: s_waitcnt vmcnt(0)
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v23, v7.l
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v7, 16, v7
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v14, v6.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v22, v6.l
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v6, 16, v6
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v19, 16, v5
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v10, v1.l
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v15, 16, v4
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v13, v3.l
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v3, 16, v3
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v12, v2.l
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v2, 16, v2
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v18, v4.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v22, v5.l
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[4:5], v10
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v10, v23.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v34, v11.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v11, v19.l
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v9, 16, v0
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v7, v7.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v6, v6.l
; GFX11-TRUE16-NEXT: v_lshrrev_b32_e32 v9, 16, v0
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v18, v5.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v34, v11.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v11, v19.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v14, v4.l
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[4:5], v10
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v10, v15.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v8, v0.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v3, v3.l
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v2, v2.l
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[28:29], v22
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[30:31], v10
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[24:25], v18
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[26:27], v11
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[28:29], v23
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[30:31], v7
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[24:25], v22
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[26:27], v6
; GFX11-TRUE16-NEXT: v_cvt_f32_f16_e32 v33, v9.l
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[20:21], v15
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[22:23], v7
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[20:21], v18
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[22:23], v11
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[16:17], v14
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[18:19], v6
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[18:19], v10
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[0:1], v8
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[8:9], v12
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[12:13], v13
Expand All @@ -2431,10 +2431,10 @@ define amdgpu_kernel void @global_extload_v16f16_to_v16f64(ptr addrspace(1) %out
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[6:7], v34
; GFX11-TRUE16-NEXT: v_cvt_f64_f32_e32 v[2:3], v33
; GFX11-TRUE16-NEXT: s_clause 0x7
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[28:31], s[0:1] offset:80
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[24:27], s[0:1] offset:64
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[20:23], s[0:1] offset:112
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[16:19], s[0:1] offset:96
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[28:31], s[0:1] offset:112
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[24:27], s[0:1] offset:96
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[20:23], s[0:1] offset:80
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[16:19], s[0:1] offset:64
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[12:15], s[0:1] offset:48
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[8:11], s[0:1] offset:32
; GFX11-TRUE16-NEXT: global_store_b128 v32, v[4:7], s[0:1] offset:16
Expand All @@ -2450,40 +2450,40 @@ define amdgpu_kernel void @global_extload_v16f16_to_v16f64(ptr addrspace(1) %out
; GFX11-FAKE16-NEXT: global_load_b128 v[0:3], v32, s[2:3]
; GFX11-FAKE16-NEXT: global_load_b128 v[4:7], v32, s[2:3] offset:16
; GFX11-FAKE16-NEXT: s_waitcnt vmcnt(1)
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v10, v1
; GFX11-FAKE16-NEXT: s_waitcnt vmcnt(0)
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v23, 16, v5
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v11, 16, v1
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v19, 16, v4
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v15, v7
; GFX11-FAKE16-NEXT: s_waitcnt vmcnt(0)
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v23, v7
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v7, 16, v7
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v14, v6
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v22, v6
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v6, 16, v6
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v19, 16, v5
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v10, v1
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v15, 16, v4
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v13, v3
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v3, 16, v3
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v12, v2
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v2, 16, v2
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v18, v4
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v22, v5
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[4:5], v10
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v10, v23
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v34, v11
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v11, v19
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v9, 16, v0
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v7, v7
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v6, v6
; GFX11-FAKE16-NEXT: v_lshrrev_b32_e32 v9, 16, v0
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v18, v5
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v34, v11
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v11, v19
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v14, v4
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[4:5], v10
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v10, v15
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v8, v0
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v3, v3
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v2, v2
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[28:29], v22
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[30:31], v10
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[24:25], v18
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[26:27], v11
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[28:29], v23
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[30:31], v7
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[24:25], v22
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[26:27], v6
; GFX11-FAKE16-NEXT: v_cvt_f32_f16_e32 v33, v9
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[20:21], v15
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[22:23], v7
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[20:21], v18
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[22:23], v11
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[16:17], v14
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[18:19], v6
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[18:19], v10
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[0:1], v8
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[8:9], v12
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[12:13], v13
Expand All @@ -2492,10 +2492,10 @@ define amdgpu_kernel void @global_extload_v16f16_to_v16f64(ptr addrspace(1) %out
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[6:7], v34
; GFX11-FAKE16-NEXT: v_cvt_f64_f32_e32 v[2:3], v33
; GFX11-FAKE16-NEXT: s_clause 0x7
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[28:31], s[0:1] offset:80
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[24:27], s[0:1] offset:64
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[20:23], s[0:1] offset:112
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[16:19], s[0:1] offset:96
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[28:31], s[0:1] offset:112
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[24:27], s[0:1] offset:96
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[20:23], s[0:1] offset:80
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[16:19], s[0:1] offset:64
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[12:15], s[0:1] offset:48
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[8:11], s[0:1] offset:32
; GFX11-FAKE16-NEXT: global_store_b128 v32, v[4:7], s[0:1] offset:16
Expand Down
Loading
Loading