7a98f084c4
Currently, the custom SGPR spill lowering pass spills SGPRs into physical VGPR lanes and the remaining VGPRs are used by regalloc for vector regclass allocation. This imposes many restrictions that we ended up with unsuccessful SGPR spilling when there won't be enough VGPRs and we are forced to spill the leftover into memory during PEI. The custom spill handling during PEI has many edge cases and often breaks the compiler time to time. This patch implements spilling SGPRs into virtual VGPR lanes. Since we now split the register allocation for SGPRs and VGPRs, the virtual registers introduced for the spill lanes would get allocated automatically in the subsequent regalloc invocation for VGPRs. Spill to virtual registers will always be successful, even in the high-pressure situations, and hence it avoids most of the edge cases during PEI. We are now left with only the custom SGPR spills during PEI for special registers like the frame pointer which is an unproblematic case. Differential Revision: https://reviews.llvm.org/D124196
159 lines
7.1 KiB
YAML
159 lines
7.1 KiB
YAML
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
|
|
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -verify-machineinstrs -stress-regalloc=2 -start-before=greedy -stop-after=virtregmap -o - %s | FileCheck %s
|
|
|
|
# Test that a spill of a copy of exec is not folded to be a spill of exec directly.
|
|
|
|
---
|
|
|
|
name: merge_sgpr_spill_into_copy_from_exec_lo
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; CHECK-LABEL: name: merge_sgpr_spill_into_copy_from_exec_lo
|
|
; CHECK: renamable $vgpr0 = IMPLICIT_DEF
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $exec_lo
|
|
; CHECK-NEXT: $sgpr0 = S_MOV_B32 $exec_lo
|
|
; CHECK-NEXT: renamable $vgpr0 = V_WRITELANE_B32 killed $sgpr0, 0, killed $vgpr0
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 $vgpr0, 0
|
|
; CHECK-NEXT: S_NOP 0, implicit-def dead renamable $sgpr1, implicit-def dead renamable $sgpr0, implicit killed renamable $sgpr0
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 killed $vgpr0, 0
|
|
; CHECK-NEXT: $exec_lo = S_MOV_B32 killed $sgpr0
|
|
; CHECK-NEXT: S_SENDMSG 0, implicit $m0, implicit $exec
|
|
S_NOP 0, implicit-def $exec_lo
|
|
%0:sreg_32 = COPY $exec_lo
|
|
S_NOP 0, implicit-def %1:sreg_32, implicit-def %2:sreg_32, implicit %0
|
|
$exec_lo = COPY %0
|
|
S_SENDMSG 0, implicit $m0, implicit $exec
|
|
|
|
...
|
|
---
|
|
|
|
name: merge_sgpr_spill_into_copy_from_exec_hi
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; CHECK-LABEL: name: merge_sgpr_spill_into_copy_from_exec_hi
|
|
; CHECK: renamable $vgpr0 = IMPLICIT_DEF
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $exec_hi
|
|
; CHECK-NEXT: $sgpr0 = S_MOV_B32 $exec_hi
|
|
; CHECK-NEXT: renamable $vgpr0 = V_WRITELANE_B32 killed $sgpr0, 0, killed $vgpr0
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 $vgpr0, 0
|
|
; CHECK-NEXT: S_NOP 0, implicit-def dead renamable $sgpr1, implicit-def dead renamable $sgpr0, implicit killed renamable $sgpr0
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 killed $vgpr0, 0
|
|
; CHECK-NEXT: $exec_hi = S_MOV_B32 killed $sgpr0
|
|
; CHECK-NEXT: S_SENDMSG 0, implicit $m0, implicit $exec
|
|
S_NOP 0, implicit-def $exec_hi
|
|
%0:sreg_32 = COPY $exec_hi
|
|
S_NOP 0, implicit-def %1:sreg_32, implicit-def %2:sreg_32, implicit %0
|
|
$exec_hi = COPY %0
|
|
S_SENDMSG 0, implicit $m0, implicit $exec
|
|
|
|
...
|
|
---
|
|
|
|
name: merge_sgpr_spill_into_copy_from_exec
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; CHECK-LABEL: name: merge_sgpr_spill_into_copy_from_exec
|
|
; CHECK: renamable $vgpr0 = IMPLICIT_DEF
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $exec
|
|
; CHECK-NEXT: $sgpr0_sgpr1 = S_MOV_B64 $exec
|
|
; CHECK-NEXT: renamable $vgpr0 = V_WRITELANE_B32 killed $sgpr0, 0, killed $vgpr0, implicit-def $sgpr0_sgpr1, implicit $sgpr0_sgpr1
|
|
; CHECK-NEXT: renamable $vgpr0 = V_WRITELANE_B32 killed $sgpr1, 1, killed $vgpr0, implicit $sgpr0_sgpr1
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 $vgpr0, 0, implicit-def $sgpr0_sgpr1
|
|
; CHECK-NEXT: $sgpr1 = V_READLANE_B32 $vgpr0, 1
|
|
; CHECK-NEXT: S_NOP 0, implicit-def dead renamable $sgpr2_sgpr3, implicit-def dead renamable $sgpr0_sgpr1, implicit killed renamable $sgpr0_sgpr1
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 $vgpr0, 0, implicit-def $sgpr0_sgpr1
|
|
; CHECK-NEXT: $sgpr1 = V_READLANE_B32 killed $vgpr0, 1
|
|
; CHECK-NEXT: $exec = S_MOV_B64 killed $sgpr0_sgpr1
|
|
; CHECK-NEXT: S_SENDMSG 0, implicit $m0, implicit $exec
|
|
S_NOP 0, implicit-def $exec
|
|
%0:sreg_64 = COPY $exec
|
|
S_NOP 0, implicit-def %1:sreg_64, implicit-def %2:sreg_64, implicit %0
|
|
$exec = COPY %0
|
|
S_SENDMSG 0, implicit $m0, implicit $exec
|
|
|
|
...
|
|
|
|
# Test that a reload into a copy of exec is not folded to be a reload of exec directly.
|
|
|
|
---
|
|
|
|
name: reload_sgpr_spill_into_copy_to_exec_lo
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; CHECK-LABEL: name: reload_sgpr_spill_into_copy_to_exec_lo
|
|
; CHECK: renamable $vgpr0 = IMPLICIT_DEF
|
|
; CHECK-NEXT: S_NOP 0, implicit-def renamable $sgpr0, implicit-def dead renamable $sgpr1, implicit-def $exec_lo
|
|
; CHECK-NEXT: renamable $vgpr0 = V_WRITELANE_B32 killed $sgpr0, 0, killed $vgpr0
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 $vgpr0, 0
|
|
; CHECK-NEXT: S_NOP 0, implicit killed renamable $sgpr0, implicit-def dead renamable $sgpr1, implicit-def dead renamable $sgpr0
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 killed $vgpr0, 0
|
|
; CHECK-NEXT: $exec_lo = S_MOV_B32 killed $sgpr0
|
|
; CHECK-NEXT: S_SENDMSG 0, implicit $m0, implicit $exec
|
|
S_NOP 0, implicit-def %0:sreg_32, implicit-def %1:sreg_32, implicit-def $exec_lo
|
|
S_NOP 0, implicit %0, implicit-def %3:sreg_32, implicit-def %4:sreg_32
|
|
$exec_lo = COPY %0
|
|
S_SENDMSG 0, implicit $m0, implicit $exec
|
|
|
|
...
|
|
---
|
|
|
|
name: reload_sgpr_spill_into_copy_to_exec_hi
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; CHECK-LABEL: name: reload_sgpr_spill_into_copy_to_exec_hi
|
|
; CHECK: renamable $vgpr0 = IMPLICIT_DEF
|
|
; CHECK-NEXT: S_NOP 0, implicit-def renamable $sgpr0, implicit-def dead renamable $sgpr1, implicit-def $exec_hi
|
|
; CHECK-NEXT: renamable $vgpr0 = V_WRITELANE_B32 killed $sgpr0, 0, killed $vgpr0
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 $vgpr0, 0
|
|
; CHECK-NEXT: S_NOP 0, implicit killed renamable $sgpr0, implicit-def dead renamable $sgpr1, implicit-def dead renamable $sgpr0
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 killed $vgpr0, 0
|
|
; CHECK-NEXT: $exec_hi = S_MOV_B32 killed $sgpr0
|
|
; CHECK-NEXT: S_SENDMSG 0, implicit $m0, implicit $exec
|
|
S_NOP 0, implicit-def %0:sreg_32, implicit-def %1:sreg_32, implicit-def $exec_hi
|
|
S_NOP 0, implicit %0, implicit-def %3:sreg_32, implicit-def %4:sreg_32
|
|
$exec_hi = COPY %0
|
|
S_SENDMSG 0, implicit $m0, implicit $exec
|
|
|
|
...
|
|
---
|
|
|
|
name: reload_sgpr_spill_into_copy_to_exec
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; CHECK-LABEL: name: reload_sgpr_spill_into_copy_to_exec
|
|
; CHECK: renamable $vgpr0 = IMPLICIT_DEF
|
|
; CHECK-NEXT: S_NOP 0, implicit-def renamable $sgpr0_sgpr1, implicit-def dead renamable $sgpr2_sgpr3, implicit-def $exec
|
|
; CHECK-NEXT: renamable $vgpr0 = V_WRITELANE_B32 killed $sgpr0, 0, killed $vgpr0, implicit-def $sgpr0_sgpr1, implicit $sgpr0_sgpr1
|
|
; CHECK-NEXT: renamable $vgpr0 = V_WRITELANE_B32 killed $sgpr1, 1, killed $vgpr0, implicit $sgpr0_sgpr1
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 $vgpr0, 0, implicit-def $sgpr0_sgpr1
|
|
; CHECK-NEXT: $sgpr1 = V_READLANE_B32 $vgpr0, 1
|
|
; CHECK-NEXT: S_NOP 0, implicit killed renamable $sgpr0_sgpr1, implicit-def dead renamable $sgpr2_sgpr3, implicit-def dead renamable $sgpr0_sgpr1
|
|
; CHECK-NEXT: $sgpr0 = V_READLANE_B32 $vgpr0, 0, implicit-def $sgpr0_sgpr1
|
|
; CHECK-NEXT: $sgpr1 = V_READLANE_B32 killed $vgpr0, 1
|
|
; CHECK-NEXT: $exec = S_MOV_B64 killed $sgpr0_sgpr1
|
|
; CHECK-NEXT: S_SENDMSG 0, implicit $m0, implicit $exec
|
|
S_NOP 0, implicit-def %0:sreg_64, implicit-def %1:sreg_64, implicit-def $exec
|
|
S_NOP 0, implicit %0, implicit-def %3:sreg_64, implicit-def %4:sreg_64
|
|
$exec = COPY %0
|
|
S_SENDMSG 0, implicit $m0, implicit $exec
|
|
|
|
...
|