a779af3f88
This PR updates the SGPR layout to a striped caller/callee-saved design, similar to the VGPR layout. To ensure that s30-s31 (return address), s32 (stack pointer), s33 (frame pointer), and s34 (base pointer) remain callee-saved, the striped layout starts from s40, with a stripe width of 8. The last stripe is 10 wide instead of 8 to avoid ending with a 2-wide stripe. Fixes #113782.
126 lines
9.0 KiB
YAML
126 lines
9.0 KiB
YAML
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
|
|
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -verify-machineinstrs -run-pass=prologepilog %s -o - | FileCheck --check-prefix=MUBUF %s
|
|
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -verify-machineinstrs -run-pass=prologepilog -mattr=+enable-flat-scratch %s -o - | FileCheck --check-prefix=FLATSCR %s
|
|
|
|
---
|
|
name: use_restore_frame_reg
|
|
tracksRegLiveness: true
|
|
|
|
stack:
|
|
- { id: 0, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 1, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 2, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 3, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 4, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 5, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 6, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 7, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 8, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 9, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 10, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 11, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 12, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 13, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 14, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 15, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 16, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 17, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
- { id: 18, type: default, offset: 0, size: 4, alignment: 8192 }
|
|
|
|
machineFunctionInfo:
|
|
isEntryFunction: false
|
|
scratchRSrcReg: $sgpr0_sgpr1_sgpr2_sgpr3
|
|
frameOffsetReg: $sgpr33
|
|
stackPtrOffsetReg: $sgpr32
|
|
|
|
body: |
|
|
; MUBUF-LABEL: name: use_restore_frame_reg
|
|
; MUBUF: bb.0:
|
|
; MUBUF-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
|
|
; MUBUF-NEXT: liveins: $sgpr40, $sgpr41, $vgpr1
|
|
; MUBUF-NEXT: {{ $}}
|
|
; MUBUF-NEXT: $sgpr40 = frame-setup COPY $sgpr33
|
|
; MUBUF-NEXT: $sgpr33 = frame-setup S_ADD_I32 $sgpr32, 524224, implicit-def $scc
|
|
; MUBUF-NEXT: $sgpr33 = frame-setup S_AND_B32 killed $sgpr33, 4294443008, implicit-def dead $scc
|
|
; MUBUF-NEXT: $sgpr41 = frame-setup COPY $sgpr34
|
|
; MUBUF-NEXT: $sgpr34 = frame-setup COPY $sgpr32
|
|
; MUBUF-NEXT: $sgpr32 = frame-setup S_ADD_I32 $sgpr32, 11010048, implicit-def dead $scc
|
|
; MUBUF-NEXT: S_CMP_EQ_U32 0, 0, implicit-def $scc
|
|
; MUBUF-NEXT: S_NOP 0, implicit-def $sgpr4, implicit-def $sgpr5, implicit-def $sgpr6, implicit-def $sgpr7, implicit-def $sgpr8, implicit-def $sgpr9, implicit-def $sgpr10, implicit-def $sgpr11, implicit-def $sgpr12, implicit-def $sgpr13, implicit-def $sgpr14, implicit-def $sgpr15, implicit-def $sgpr16, implicit-def $sgpr17, implicit-def $sgpr18, implicit-def $sgpr19, implicit-def $sgpr20, implicit-def $sgpr21, implicit-def $sgpr22, implicit-def $sgpr23, implicit-def $sgpr24, implicit-def $sgpr25, implicit-def $sgpr26, implicit-def $sgpr27, implicit-def $sgpr28, implicit-def $sgpr29, implicit-def $sgpr30, implicit-def $sgpr31, implicit-def $vcc
|
|
; MUBUF-NEXT: $vgpr0 = V_LSHRREV_B32_e64 6, $sgpr33, implicit $exec
|
|
; MUBUF-NEXT: $vgpr0 = V_ADD_U32_e32 8192, killed $vgpr0, implicit $exec
|
|
; MUBUF-NEXT: $vgpr2 = V_LSHRREV_B32_e64 6, $sgpr33, implicit $exec
|
|
; MUBUF-NEXT: $vgpr2 = V_ADD_U32_e32 155648, killed $vgpr2, implicit $exec
|
|
; MUBUF-NEXT: $vgpr0 = V_OR_B32_e32 killed $vgpr2, $vgpr1, implicit $exec, implicit $sgpr4, implicit $sgpr5, implicit $sgpr6, implicit $sgpr7, implicit $sgpr8, implicit $sgpr9, implicit $sgpr10, implicit $sgpr11, implicit $sgpr12, implicit $sgpr13, implicit $sgpr14, implicit $sgpr15, implicit $sgpr16, implicit $sgpr17, implicit $sgpr18, implicit $sgpr19, implicit $sgpr20, implicit $sgpr21, implicit $sgpr22, implicit $sgpr23, implicit $sgpr24, implicit $sgpr25, implicit $sgpr26, implicit $sgpr27, implicit $sgpr28, implicit $sgpr29, implicit $sgpr30, implicit $sgpr31
|
|
; MUBUF-NEXT: S_CBRANCH_VCCNZ %bb.2, implicit $vcc
|
|
; MUBUF-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc
|
|
; MUBUF-NEXT: {{ $}}
|
|
; MUBUF-NEXT: bb.1:
|
|
; MUBUF-NEXT: successors: %bb.2(0x80000000)
|
|
; MUBUF-NEXT: liveins: $sgpr40, $sgpr41
|
|
; MUBUF-NEXT: {{ $}}
|
|
; MUBUF-NEXT: S_NOP 0
|
|
; MUBUF-NEXT: {{ $}}
|
|
; MUBUF-NEXT: bb.2:
|
|
; MUBUF-NEXT: liveins: $sgpr40, $sgpr41
|
|
; MUBUF-NEXT: {{ $}}
|
|
; MUBUF-NEXT: $sgpr32 = frame-destroy COPY $sgpr34
|
|
; MUBUF-NEXT: $sgpr34 = frame-destroy COPY $sgpr41
|
|
; MUBUF-NEXT: $sgpr33 = frame-destroy COPY $sgpr40
|
|
; MUBUF-NEXT: S_ENDPGM 0
|
|
;
|
|
; FLATSCR-LABEL: name: use_restore_frame_reg
|
|
; FLATSCR: bb.0:
|
|
; FLATSCR-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
|
|
; FLATSCR-NEXT: liveins: $sgpr40, $sgpr41, $vgpr1
|
|
; FLATSCR-NEXT: {{ $}}
|
|
; FLATSCR-NEXT: $sgpr40 = frame-setup COPY $sgpr33
|
|
; FLATSCR-NEXT: $sgpr33 = frame-setup S_ADD_I32 $sgpr32, 8191, implicit-def $scc
|
|
; FLATSCR-NEXT: $sgpr33 = frame-setup S_AND_B32 killed $sgpr33, 4294959104, implicit-def dead $scc
|
|
; FLATSCR-NEXT: $sgpr41 = frame-setup COPY $sgpr34
|
|
; FLATSCR-NEXT: $sgpr34 = frame-setup COPY $sgpr32
|
|
; FLATSCR-NEXT: $sgpr32 = frame-setup S_ADD_I32 $sgpr32, 172032, implicit-def dead $scc
|
|
; FLATSCR-NEXT: S_CMP_EQ_U32 0, 0, implicit-def $scc
|
|
; FLATSCR-NEXT: S_NOP 0, implicit-def $sgpr4, implicit-def $sgpr5, implicit-def $sgpr6, implicit-def $sgpr7, implicit-def $sgpr8, implicit-def $sgpr9, implicit-def $sgpr10, implicit-def $sgpr11, implicit-def $sgpr12, implicit-def $sgpr13, implicit-def $sgpr14, implicit-def $sgpr15, implicit-def $sgpr16, implicit-def $sgpr17, implicit-def $sgpr18, implicit-def $sgpr19, implicit-def $sgpr20, implicit-def $sgpr21, implicit-def $sgpr22, implicit-def $sgpr23, implicit-def $sgpr24, implicit-def $sgpr25, implicit-def $sgpr26, implicit-def $sgpr27, implicit-def $sgpr28, implicit-def $sgpr29, implicit-def $sgpr30, implicit-def $sgpr31, implicit-def $vcc
|
|
; FLATSCR-NEXT: $sgpr42 = S_ADDC_U32 $sgpr33, 8192, implicit-def $scc, implicit $scc
|
|
; FLATSCR-NEXT: S_BITCMP1_B32 $sgpr42, 0, implicit-def $scc
|
|
; FLATSCR-NEXT: $sgpr42 = S_BITSET0_B32 0, $sgpr42
|
|
; FLATSCR-NEXT: $vgpr0 = V_MOV_B32_e32 killed $sgpr42, implicit $exec
|
|
; FLATSCR-NEXT: $sgpr42 = S_ADDC_U32 $sgpr33, 155648, implicit-def $scc, implicit $scc
|
|
; FLATSCR-NEXT: S_BITCMP1_B32 $sgpr42, 0, implicit-def $scc
|
|
; FLATSCR-NEXT: $sgpr42 = S_BITSET0_B32 0, $sgpr42
|
|
; FLATSCR-NEXT: $vgpr0 = V_OR_B32_e32 killed $sgpr42, $vgpr1, implicit $exec, implicit $sgpr4, implicit $sgpr5, implicit $sgpr6, implicit $sgpr7, implicit $sgpr8, implicit $sgpr9, implicit $sgpr10, implicit $sgpr11, implicit $sgpr12, implicit $sgpr13, implicit $sgpr14, implicit $sgpr15, implicit $sgpr16, implicit $sgpr17, implicit $sgpr18, implicit $sgpr19, implicit $sgpr20, implicit $sgpr21, implicit $sgpr22, implicit $sgpr23, implicit $sgpr24, implicit $sgpr25, implicit $sgpr26, implicit $sgpr27, implicit $sgpr28, implicit $sgpr29, implicit $sgpr30, implicit $sgpr31
|
|
; FLATSCR-NEXT: S_CBRANCH_VCCNZ %bb.2, implicit $vcc
|
|
; FLATSCR-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc
|
|
; FLATSCR-NEXT: {{ $}}
|
|
; FLATSCR-NEXT: bb.1:
|
|
; FLATSCR-NEXT: successors: %bb.2(0x80000000)
|
|
; FLATSCR-NEXT: liveins: $sgpr40, $sgpr41
|
|
; FLATSCR-NEXT: {{ $}}
|
|
; FLATSCR-NEXT: S_NOP 0
|
|
; FLATSCR-NEXT: {{ $}}
|
|
; FLATSCR-NEXT: bb.2:
|
|
; FLATSCR-NEXT: liveins: $sgpr40, $sgpr41
|
|
; FLATSCR-NEXT: {{ $}}
|
|
; FLATSCR-NEXT: $sgpr32 = frame-destroy COPY $sgpr34
|
|
; FLATSCR-NEXT: $sgpr34 = frame-destroy COPY $sgpr41
|
|
; FLATSCR-NEXT: $sgpr33 = frame-destroy COPY $sgpr40
|
|
; FLATSCR-NEXT: S_ENDPGM 0
|
|
bb.0:
|
|
liveins: $vgpr1
|
|
|
|
S_CMP_EQ_U32 0, 0, implicit-def $scc
|
|
S_NOP 0, implicit-def $sgpr4, implicit-def $sgpr5, implicit-def $sgpr6, implicit-def $sgpr7, implicit-def $sgpr8, implicit-def $sgpr9, implicit-def $sgpr10, implicit-def $sgpr11, implicit-def $sgpr12, implicit-def $sgpr13, implicit-def $sgpr14, implicit-def $sgpr15, implicit-def $sgpr16, implicit-def $sgpr17, implicit-def $sgpr18, implicit-def $sgpr19, implicit-def $sgpr20, implicit-def $sgpr21, implicit-def $sgpr22, implicit-def $sgpr23, implicit-def $sgpr24, implicit-def $sgpr25, implicit-def $sgpr26, implicit-def $sgpr27, implicit-def $sgpr28, implicit-def $sgpr29, implicit-def $sgpr30, implicit-def $sgpr31, implicit-def $vcc
|
|
$vgpr0 = V_MOV_B32_e32 %stack.0, implicit $exec
|
|
$vgpr0 = V_OR_B32_e32 %stack.18, $vgpr1, implicit $exec, implicit $sgpr4, implicit $sgpr5, implicit $sgpr6, implicit $sgpr7, implicit $sgpr8, implicit $sgpr9, implicit $sgpr10, implicit $sgpr11, implicit $sgpr12, implicit $sgpr13, implicit $sgpr14, implicit $sgpr15, implicit $sgpr16, implicit $sgpr17, implicit $sgpr18, implicit $sgpr19, implicit $sgpr20, implicit $sgpr21, implicit $sgpr22, implicit $sgpr23, implicit $sgpr24, implicit $sgpr25, implicit $sgpr26, implicit $sgpr27, implicit $sgpr28, implicit $sgpr29, implicit $sgpr30, implicit $sgpr31
|
|
S_CBRANCH_VCCNZ %bb.2, implicit $vcc
|
|
S_CBRANCH_SCC1 %bb.2, implicit $scc
|
|
|
|
bb.1:
|
|
S_NOP 0
|
|
|
|
bb.2:
|
|
S_ENDPGM 0
|
|
|
|
...
|