Skip to content

[MachineScheduler] Fix physreg dependencies of ExitSU #123541

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

Merged
merged 7 commits into from
Feb 1, 2025
Merged

[MachineScheduler] Fix physreg dependencies of ExitSU #123541

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
05cca1b
[MachineScheduler] Fix physreg dependencies of ExitSU
s-barannikov Jan 20, 2025
8b87e2f
Comment the check
s-barannikov Jan 20, 2025
d709f4b
Update remaining tests
s-barannikov Jan 20, 2025
806fa90
Update LSR test
s-barannikov Jan 20, 2025
b7cfe8b
Try to improve FIXME
s-barannikov Jan 20, 2025
e51732a
Remove FIXME as it turned out that there is nothing to fix
s-barannikov Jan 20, 2025
38cf1e4
Merge remote-tracking branch 'upstream/main' into misched/exit-su-dep
s-barannikov Feb 1, 2025
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
16 changes: 14 additions & 2 deletions llvm/lib/CodeGen/ScheduleDAGInstrs.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -209,13 +209,25 @@ void ScheduleDAGInstrs::addSchedBarrierDeps() {
ExitSU.setInstr(ExitMI);
// Add dependencies on the defs and uses of the instruction.
if (ExitMI) {
const MCInstrDesc &MIDesc = ExitMI->getDesc();
for (const MachineOperand &MO : ExitMI->all_uses()) {
unsigned OpIdx = MO.getOperandNo();
Register Reg = MO.getReg();
if (Reg.isPhysical()) {
// addPhysRegDataDeps uses the provided operand index to retrieve
// the operand use cycle from the scheduling model. If the operand
// is "fake" (e.g., an operand of a call instruction used to pass
// an argument to the called function.), the scheduling model may not
// have an entry for it. If this is the case, pass -1 as operand index,
// which will cause addPhysRegDataDeps to add an artificial dependency.
// FIXME: Using hasImplicitUseOfPhysReg here is inaccurate as it misses
// aliases. When fixing, make sure to update addPhysRegDataDeps, too.
bool IsRealUse = OpIdx < MIDesc.getNumOperands() ||
MIDesc.hasImplicitUseOfPhysReg(Reg);
Comment on lines +225 to +226
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Comment this usage. I'm assuming getting this accurate isn't required for correctness?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

MIDesc.hasImplicitUseOfPhysReg also looks buggy to me. It doesn't have the MCRI argument like hasImplicitDefOfPhysReg, so it will miss any aliases

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm assuming getting this accurate isn't required for correctness?

As far as I'm aware, getting this accurate isn't required for correctness. My motivation for fixing this was to move comparisons away from branches if the latency between them is high (which is true on my target).

MIDesc.hasImplicitUseOfPhysReg also looks buggy to me. It doesn't have the MCRI argument like hasImplicitDefOfPhysReg, so it will miss any aliases

Good point, I'll add a FIXME.

for (MCRegUnit Unit : TRI->regunits(Reg))
Uses.insert(PhysRegSUOper(&ExitSU, -1, Unit));
Uses.insert(PhysRegSUOper(&ExitSU, IsRealUse ? OpIdx : -1, Unit));
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is where an operand index gets a negative value. There is one more place at the end of this function.

} else if (Reg.isVirtual() && MO.readsReg()) {
addVRegUseDeps(&ExitSU, MO.getOperandNo());
addVRegUseDeps(&ExitSU, OpIdx);
}
}
}
Expand Down
10 changes: 5 additions & 5 deletions llvm/test/CodeGen/AMDGPU/GlobalISel/assert-align.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,12 +15,12 @@ define ptr addrspace(1) @call_assert_align() {
; CHECK-NEXT: v_writelane_b32 v40, s16, 2
; CHECK-NEXT: s_addk_i32 s32, 0x400
; CHECK-NEXT: v_writelane_b32 v40, s30, 0
; CHECK-NEXT: v_mov_b32_e32 v0, 0
; CHECK-NEXT: v_mov_b32_e32 v1, 0
; CHECK-NEXT: v_writelane_b32 v40, s31, 1
; CHECK-NEXT: s_getpc_b64 s[16:17]
; CHECK-NEXT: s_add_u32 s16, s16, ext@rel32@lo+4
; CHECK-NEXT: s_addc_u32 s17, s17, ext@rel32@hi+12
; CHECK-NEXT: v_mov_b32_e32 v0, 0
; CHECK-NEXT: v_mov_b32_e32 v1, 0
; CHECK-NEXT: v_writelane_b32 v40, s31, 1
; CHECK-NEXT: s_swappc_b64 s[30:31], s[16:17]
; CHECK-NEXT: v_mov_b32_e32 v2, 0
; CHECK-NEXT: global_store_dword v[0:1], v2, off
Expand All @@ -45,11 +45,11 @@ define ptr addrspace(1) @tail_call_assert_align() {
; CHECK-LABEL: tail_call_assert_align:
; CHECK: ; %bb.0: ; %entry
; CHECK-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT: v_mov_b32_e32 v0, 0
; CHECK-NEXT: v_mov_b32_e32 v1, 0
; CHECK-NEXT: s_getpc_b64 s[16:17]
; CHECK-NEXT: s_add_u32 s16, s16, ext@rel32@lo+4
; CHECK-NEXT: s_addc_u32 s17, s17, ext@rel32@hi+12
; CHECK-NEXT: v_mov_b32_e32 v0, 0
; CHECK-NEXT: v_mov_b32_e32 v1, 0
; CHECK-NEXT: s_setpc_b64 s[16:17]
entry:
%call = tail call align 4 ptr addrspace(1) @ext(ptr addrspace(1) null)
Expand Down
14 changes: 7 additions & 7 deletions llvm/test/CodeGen/AMDGPU/GlobalISel/call-outgoing-stack-args.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,8 +44,8 @@ define amdgpu_kernel void @kernel_caller_stack() {
; FLATSCR-NEXT: scratch_store_dword off, v0, s0
; FLATSCR-NEXT: s_add_u32 s0, s32, 12
; FLATSCR-NEXT: v_mov_b32_e32 v0, 11
; FLATSCR-NEXT: scratch_store_dword off, v0, s0
; FLATSCR-NEXT: s_add_u32 s2, s32, 16
; FLATSCR-NEXT: scratch_store_dword off, v0, s0
; FLATSCR-NEXT: v_mov_b32_e32 v0, 12
; FLATSCR-NEXT: s_getpc_b64 s[0:1]
; FLATSCR-NEXT: s_add_u32 s0, s0, external_void_func_v16i32_v16i32_v4i32@rel32@lo+4
Expand Down Expand Up @@ -239,11 +239,11 @@ define void @func_caller_stack() {
; MUBUF-NEXT: v_writelane_b32 v40, s30, 0
; MUBUF-NEXT: buffer_store_dword v0, off, s[0:3], s32 offset:12
; MUBUF-NEXT: v_mov_b32_e32 v0, 12
; MUBUF-NEXT: v_writelane_b32 v40, s31, 1
; MUBUF-NEXT: buffer_store_dword v0, off, s[0:3], s32 offset:16
; MUBUF-NEXT: s_getpc_b64 s[4:5]
; MUBUF-NEXT: s_add_u32 s4, s4, external_void_func_v16i32_v16i32_v4i32@rel32@lo+4
; MUBUF-NEXT: s_addc_u32 s5, s5, external_void_func_v16i32_v16i32_v4i32@rel32@hi+12
; MUBUF-NEXT: v_writelane_b32 v40, s31, 1
; MUBUF-NEXT: buffer_store_dword v0, off, s[0:3], s32 offset:16
; MUBUF-NEXT: s_swappc_b64 s[30:31], s[4:5]
; MUBUF-NEXT: v_readlane_b32 s31, v40, 1
; MUBUF-NEXT: v_readlane_b32 s30, v40, 0
Expand Down Expand Up @@ -274,15 +274,15 @@ define void @func_caller_stack() {
; FLATSCR-NEXT: scratch_store_dword off, v0, s0
; FLATSCR-NEXT: s_add_u32 s0, s32, 12
; FLATSCR-NEXT: v_mov_b32_e32 v0, 11
; FLATSCR-NEXT: v_writelane_b32 v40, s30, 0
; FLATSCR-NEXT: scratch_store_dword off, v0, s0
; FLATSCR-NEXT: s_add_u32 s0, s32, 16
; FLATSCR-NEXT: v_mov_b32_e32 v0, 12
; FLATSCR-NEXT: v_writelane_b32 v40, s31, 1
; FLATSCR-NEXT: v_writelane_b32 v40, s30, 0
; FLATSCR-NEXT: scratch_store_dword off, v0, s0
; FLATSCR-NEXT: s_getpc_b64 s[0:1]
; FLATSCR-NEXT: s_add_u32 s0, s0, external_void_func_v16i32_v16i32_v4i32@rel32@lo+4
; FLATSCR-NEXT: s_addc_u32 s1, s1, external_void_func_v16i32_v16i32_v4i32@rel32@hi+12
; FLATSCR-NEXT: v_writelane_b32 v40, s31, 1
; FLATSCR-NEXT: s_swappc_b64 s[30:31], s[0:1]
; FLATSCR-NEXT: v_readlane_b32 s31, v40, 1
; FLATSCR-NEXT: v_readlane_b32 s30, v40, 0
Expand Down Expand Up @@ -312,10 +312,10 @@ define void @func_caller_byval(ptr addrspace(5) %argptr) {
; MUBUF-NEXT: s_addk_i32 s32, 0x400
; MUBUF-NEXT: v_writelane_b32 v40, s4, 2
; MUBUF-NEXT: v_writelane_b32 v40, s30, 0
; MUBUF-NEXT: v_writelane_b32 v40, s31, 1
; MUBUF-NEXT: s_getpc_b64 s[4:5]
; MUBUF-NEXT: s_add_u32 s4, s4, external_void_func_byval@rel32@lo+4
; MUBUF-NEXT: s_addc_u32 s5, s5, external_void_func_byval@rel32@hi+12
; MUBUF-NEXT: v_writelane_b32 v40, s31, 1
; MUBUF-NEXT: s_waitcnt vmcnt(1)
; MUBUF-NEXT: buffer_store_dword v1, off, s[0:3], s32
; MUBUF-NEXT: s_waitcnt vmcnt(1)
Expand Down Expand Up @@ -394,8 +394,8 @@ define void @func_caller_byval(ptr addrspace(5) %argptr) {
; FLATSCR-NEXT: v_add_u32_e32 v3, 8, v0
; FLATSCR-NEXT: v_writelane_b32 v40, s0, 2
; FLATSCR-NEXT: s_add_u32 s0, s32, 8
; FLATSCR-NEXT: v_writelane_b32 v40, s30, 0
; FLATSCR-NEXT: s_add_u32 s2, s32, 56
; FLATSCR-NEXT: v_writelane_b32 v40, s30, 0
; FLATSCR-NEXT: v_writelane_b32 v40, s31, 1
; FLATSCR-NEXT: s_waitcnt vmcnt(0)
; FLATSCR-NEXT: scratch_store_dwordx2 off, v[1:2], s32
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/CodeGen/AMDGPU/GlobalISel/divergence-divergent-i1-used-outside-loop.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -191,9 +191,9 @@ define void @divergent_i1_xor_used_outside_loop_larger_loop_body(i32 %num.elts,
; GFX10-LABEL: divergent_i1_xor_used_outside_loop_larger_loop_body:
; GFX10: ; %bb.0: ; %entry
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX10-NEXT: s_mov_b32 s5, 0
; GFX10-NEXT: s_mov_b32 s6, -1
; GFX10-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX10-NEXT: s_and_saveexec_b32 s4, vcc_lo
; GFX10-NEXT: s_cbranch_execz .LBB3_6
; GFX10-NEXT: ; %bb.1: ; %loop.start.preheader
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-mui.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -387,8 +387,8 @@ define amdgpu_ps void @and_i1_scc(i32 inreg %a, i32 inreg %b, ptr addrspace(1) %
define amdgpu_ps void @divergent_phi_with_uniform_inputs(i32 %a, ptr addrspace(1) %out) {
; OLD_RBS-LABEL: divergent_phi_with_uniform_inputs:
; OLD_RBS: ; %bb.0: ; %A
; OLD_RBS-NEXT: s_mov_b32 s0, 0
; OLD_RBS-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; OLD_RBS-NEXT: s_mov_b32 s0, 0
; OLD_RBS-NEXT: s_and_saveexec_b32 s1, vcc_lo
; OLD_RBS-NEXT: ; %bb.1: ; %B
; OLD_RBS-NEXT: s_mov_b32 s0, 1
Expand Down
26 changes: 13 additions & 13 deletions llvm/test/CodeGen/AMDGPU/abi-attribute-hints-undefined-behavior.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,10 +25,10 @@ define void @parent_func_missing_inputs() #0 {
; FIXEDABI-NEXT: v_writelane_b32 v40, s16, 2
; FIXEDABI-NEXT: s_addk_i32 s32, 0x400
; FIXEDABI-NEXT: v_writelane_b32 v40, s30, 0
; FIXEDABI-NEXT: v_writelane_b32 v40, s31, 1
; FIXEDABI-NEXT: s_getpc_b64 s[16:17]
; FIXEDABI-NEXT: s_add_u32 s16, s16, requires_all_inputs@rel32@lo+4
; FIXEDABI-NEXT: s_addc_u32 s17, s17, requires_all_inputs@rel32@hi+12
; FIXEDABI-NEXT: v_writelane_b32 v40, s31, 1
; FIXEDABI-NEXT: s_swappc_b64 s[30:31], s[16:17]
; FIXEDABI-NEXT: v_readlane_b32 s31, v40, 1
; FIXEDABI-NEXT: v_readlane_b32 s30, v40, 0
Expand All @@ -49,43 +49,43 @@ define amdgpu_kernel void @parent_kernel_missing_inputs() #0 {
; FIXEDABI-SDAG: ; %bb.0:
; FIXEDABI-SDAG-NEXT: s_add_i32 s4, s4, s9
; FIXEDABI-SDAG-NEXT: s_lshr_b32 flat_scratch_hi, s4, 8
; FIXEDABI-SDAG-NEXT: v_lshlrev_b32_e32 v1, 10, v1
; FIXEDABI-SDAG-NEXT: s_add_u32 s0, s0, s9
; FIXEDABI-SDAG-NEXT: v_lshlrev_b32_e32 v1, 10, v1
; FIXEDABI-SDAG-NEXT: s_addc_u32 s1, s1, 0
; FIXEDABI-SDAG-NEXT: v_lshlrev_b32_e32 v2, 20, v2
; FIXEDABI-SDAG-NEXT: v_or_b32_e32 v0, v0, v1
; FIXEDABI-SDAG-NEXT: s_addc_u32 s1, s1, 0
; FIXEDABI-SDAG-NEXT: s_mov_b32 flat_scratch_lo, s5
; FIXEDABI-SDAG-NEXT: s_mov_b32 s14, s8
; FIXEDABI-SDAG-NEXT: s_getpc_b64 s[4:5]
; FIXEDABI-SDAG-NEXT: s_add_u32 s4, s4, requires_all_inputs@rel32@lo+4
; FIXEDABI-SDAG-NEXT: s_addc_u32 s5, s5, requires_all_inputs@rel32@hi+12
; FIXEDABI-SDAG-NEXT: v_or_b32_e32 v31, v0, v2
; FIXEDABI-SDAG-NEXT: s_mov_b64 s[8:9], 0
; FIXEDABI-SDAG-NEXT: s_mov_b32 s12, s6
; FIXEDABI-SDAG-NEXT: s_mov_b32 s13, s7
; FIXEDABI-SDAG-NEXT: s_mov_b32 s32, 0
; FIXEDABI-SDAG-NEXT: s_mov_b32 flat_scratch_lo, s5
; FIXEDABI-SDAG-NEXT: s_getpc_b64 s[4:5]
; FIXEDABI-SDAG-NEXT: s_add_u32 s4, s4, requires_all_inputs@rel32@lo+4
; FIXEDABI-SDAG-NEXT: s_addc_u32 s5, s5, requires_all_inputs@rel32@hi+12
; FIXEDABI-SDAG-NEXT: s_swappc_b64 s[30:31], s[4:5]
; FIXEDABI-SDAG-NEXT: s_endpgm
;
; FIXEDABI-GISEL-LABEL: parent_kernel_missing_inputs:
; FIXEDABI-GISEL: ; %bb.0:
; FIXEDABI-GISEL-NEXT: s_add_i32 s4, s4, s9
; FIXEDABI-GISEL-NEXT: s_lshr_b32 flat_scratch_hi, s4, 8
; FIXEDABI-GISEL-NEXT: v_lshlrev_b32_e32 v1, 10, v1
; FIXEDABI-GISEL-NEXT: s_add_u32 s0, s0, s9
; FIXEDABI-GISEL-NEXT: v_lshlrev_b32_e32 v1, 10, v1
; FIXEDABI-GISEL-NEXT: s_addc_u32 s1, s1, 0
; FIXEDABI-GISEL-NEXT: v_or_b32_e32 v0, v0, v1
; FIXEDABI-GISEL-NEXT: v_lshlrev_b32_e32 v1, 20, v2
; FIXEDABI-GISEL-NEXT: s_addc_u32 s1, s1, 0
; FIXEDABI-GISEL-NEXT: s_mov_b32 flat_scratch_lo, s5
; FIXEDABI-GISEL-NEXT: s_mov_b32 s14, s8
; FIXEDABI-GISEL-NEXT: v_or_b32_e32 v31, v0, v1
; FIXEDABI-GISEL-NEXT: s_getpc_b64 s[4:5]
; FIXEDABI-GISEL-NEXT: s_add_u32 s4, s4, requires_all_inputs@rel32@lo+4
; FIXEDABI-GISEL-NEXT: s_addc_u32 s5, s5, requires_all_inputs@rel32@hi+12
; FIXEDABI-GISEL-NEXT: s_mov_b64 s[8:9], 0
; FIXEDABI-GISEL-NEXT: s_mov_b32 s12, s6
; FIXEDABI-GISEL-NEXT: s_mov_b32 s13, s7
; FIXEDABI-GISEL-NEXT: s_mov_b32 s32, 0
; FIXEDABI-GISEL-NEXT: s_mov_b32 flat_scratch_lo, s5
; FIXEDABI-GISEL-NEXT: s_getpc_b64 s[4:5]
; FIXEDABI-GISEL-NEXT: s_add_u32 s4, s4, requires_all_inputs@rel32@lo+4
; FIXEDABI-GISEL-NEXT: s_addc_u32 s5, s5, requires_all_inputs@rel32@hi+12
; FIXEDABI-GISEL-NEXT: s_swappc_b64 s[30:31], s[4:5]
; FIXEDABI-GISEL-NEXT: s_endpgm
call void @requires_all_inputs()
Expand Down
20 changes: 12 additions & 8 deletions llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1286,9 +1286,9 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1032_DPP-NEXT: s_or_saveexec_b32 s4, -1
; GFX1032_DPP-NEXT: v_writelane_b32 v3, s5, 16
; GFX1032_DPP-NEXT: s_mov_b32 exec_lo, s4
; GFX1032_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1032_DPP-NEXT: s_mov_b32 s4, s6
; GFX1032_DPP-NEXT: s_mov_b32 s6, -1
; GFX1032_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1032_DPP-NEXT: ; implicit-def: $vgpr0
; GFX1032_DPP-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1032_DPP-NEXT: s_cbranch_execz .LBB2_2
Expand Down Expand Up @@ -1412,9 +1412,10 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1132_DPP-NEXT: s_or_saveexec_b32 s4, -1
; GFX1132_DPP-NEXT: v_writelane_b32 v3, s5, 16
; GFX1132_DPP-NEXT: s_mov_b32 exec_lo, s4
; GFX1132_DPP-NEXT: s_delay_alu instid0(VALU_DEP_2)
; GFX1132_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1132_DPP-NEXT: s_mov_b32 s4, s6
; GFX1132_DPP-NEXT: s_mov_b32 s6, -1
; GFX1132_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1132_DPP-NEXT: ; implicit-def: $vgpr0
; GFX1132_DPP-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1132_DPP-NEXT: s_cbranch_execz .LBB2_2
Expand Down Expand Up @@ -1540,9 +1541,10 @@ define amdgpu_kernel void @add_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1232_DPP-NEXT: s_or_saveexec_b32 s4, -1
; GFX1232_DPP-NEXT: v_writelane_b32 v3, s5, 16
; GFX1232_DPP-NEXT: s_mov_b32 exec_lo, s4
; GFX1232_DPP-NEXT: s_delay_alu instid0(VALU_DEP_2)
; GFX1232_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1232_DPP-NEXT: s_mov_b32 s4, s6
; GFX1232_DPP-NEXT: s_mov_b32 s6, -1
; GFX1232_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1232_DPP-NEXT: ; implicit-def: $vgpr0
; GFX1232_DPP-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1232_DPP-NEXT: s_cbranch_execz .LBB2_2
Expand Down Expand Up @@ -3129,8 +3131,8 @@ define amdgpu_kernel void @add_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1032_DPP-NEXT: v_writelane_b32 v8, s8, 16
; GFX1032_DPP-NEXT: v_writelane_b32 v7, s7, 16
; GFX1032_DPP-NEXT: s_mov_b32 exec_lo, s6
; GFX1032_DPP-NEXT: s_mov_b32 s6, -1
; GFX1032_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1032_DPP-NEXT: s_mov_b32 s6, -1
; GFX1032_DPP-NEXT: ; implicit-def: $vgpr9_vgpr10
; GFX1032_DPP-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1032_DPP-NEXT: s_cbranch_execz .LBB5_2
Expand Down Expand Up @@ -4839,9 +4841,9 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1032_DPP-NEXT: s_or_saveexec_b32 s4, -1
; GFX1032_DPP-NEXT: v_writelane_b32 v3, s5, 16
; GFX1032_DPP-NEXT: s_mov_b32 exec_lo, s4
; GFX1032_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1032_DPP-NEXT: s_mov_b32 s4, s6
; GFX1032_DPP-NEXT: s_mov_b32 s6, -1
; GFX1032_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1032_DPP-NEXT: ; implicit-def: $vgpr0
; GFX1032_DPP-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1032_DPP-NEXT: s_cbranch_execz .LBB8_2
Expand Down Expand Up @@ -4965,9 +4967,10 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1132_DPP-NEXT: s_or_saveexec_b32 s4, -1
; GFX1132_DPP-NEXT: v_writelane_b32 v3, s5, 16
; GFX1132_DPP-NEXT: s_mov_b32 exec_lo, s4
; GFX1132_DPP-NEXT: s_delay_alu instid0(VALU_DEP_2)
; GFX1132_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1132_DPP-NEXT: s_mov_b32 s4, s6
; GFX1132_DPP-NEXT: s_mov_b32 s6, -1
; GFX1132_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1132_DPP-NEXT: ; implicit-def: $vgpr0
; GFX1132_DPP-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1132_DPP-NEXT: s_cbranch_execz .LBB8_2
Expand Down Expand Up @@ -5093,9 +5096,10 @@ define amdgpu_kernel void @sub_i32_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1232_DPP-NEXT: s_or_saveexec_b32 s4, -1
; GFX1232_DPP-NEXT: v_writelane_b32 v3, s5, 16
; GFX1232_DPP-NEXT: s_mov_b32 exec_lo, s4
; GFX1232_DPP-NEXT: s_delay_alu instid0(VALU_DEP_2)
; GFX1232_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1232_DPP-NEXT: s_mov_b32 s4, s6
; GFX1232_DPP-NEXT: s_mov_b32 s6, -1
; GFX1232_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1232_DPP-NEXT: ; implicit-def: $vgpr0
; GFX1232_DPP-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1232_DPP-NEXT: s_cbranch_execz .LBB8_2
Expand Down Expand Up @@ -6715,8 +6719,8 @@ define amdgpu_kernel void @sub_i64_varying(ptr addrspace(1) %out, ptr addrspace(
; GFX1032_DPP-NEXT: v_writelane_b32 v8, s8, 16
; GFX1032_DPP-NEXT: v_writelane_b32 v7, s7, 16
; GFX1032_DPP-NEXT: s_mov_b32 exec_lo, s6
; GFX1032_DPP-NEXT: s_mov_b32 s6, -1
; GFX1032_DPP-NEXT: v_cmp_eq_u32_e32 vcc_lo, 0, v0
; GFX1032_DPP-NEXT: s_mov_b32 s6, -1
; GFX1032_DPP-NEXT: ; implicit-def: $vgpr9_vgpr10
; GFX1032_DPP-NEXT: s_and_saveexec_b32 s8, vcc_lo
; GFX1032_DPP-NEXT: s_cbranch_execz .LBB11_2
Expand Down
Loading