
Add support for using the existing SCRATCH_STORE_BLOCK and SCRATCH_LOAD_BLOCK instructions for saving and restoring callee-saved VGPRs. This is controlled by a new subtarget feature, block-vgpr-csr. It does not include WWM registers - those will be saved and restored individually, just like before. This patch does not change the ABI. Use of this feature may lead to slightly increased stack usage, because the memory is not compacted if certain registers don't have to be transferred (this will happen in practice for calling conventions where the callee and caller saved registers are interleaved in groups of 8). However, if the registers at the end of the block of 32 don't have to be transferred, we don't need to use a whole 128-byte stack slot - we can trim some space off the end of the range. In order to implement this feature, we need to rely less on the target-independent code in the PrologEpilogInserter, so we override several new methods in SIFrameLowering. We also add new pseudos, SI_BLOCK_SPILL_V1024_SAVE/RESTORE. One peculiarity is that both the SI_BLOCK_V1024_RESTORE pseudo and the SCRATCH_LOAD_BLOCK instructions will have all the registers that are not transferred added as implicit uses. This is done in order to inform LiveRegUnits that those registers are not available before the restore (since we're not really restoring them - so we can't afford to scavenge them). Unfortunately, this trick doesn't work with the save, so before the save all the registers in the block will be unavailable (see the unit test). This was reverted due to failures in the builds with expensive checks on, now fixed by always updating LiveIntervals and SlotIndexes in SILowerSGPRSpills.
295 lines
28 KiB
YAML
295 lines
28 KiB
YAML
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
|
|
# RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -mattr=+block-vgpr-csr,+wavefrontsize32,-wavefrontsize64 -start-before=si-lower-sgpr-spills -stop-after=prologepilog -verify-machineinstrs -o - %s | FileCheck %s --check-prefixes=CHECK,W32
|
|
# RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -mattr=+block-vgpr-csr,-wavefrontsize32,+wavefrontsize64 -start-before=si-lower-sgpr-spills -stop-after=prologepilog -verify-machineinstrs -o - %s | FileCheck %s --check-prefixes=CHECK,W64
|
|
|
|
--- |
|
|
define void @one_block() { ret void }
|
|
define void @one_block_csr_only() { ret void }
|
|
define void @multiple_blocks() { ret void }
|
|
define void @reg_tuples() { ret void }
|
|
define void @locals() { ret void }
|
|
define void @other_regs() { ret void }
|
|
define amdgpu_kernel void @entry_func() { ret void }
|
|
define void @multiple_basic_blocks() { ret void }
|
|
...
|
|
|
|
# Block load/store v42 and v45. The mask should be 0x9.
|
|
|
|
---
|
|
name: one_block
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
stackPtrOffsetReg: $sgpr32
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr30_sgpr31
|
|
; CHECK-LABEL: name: one_block
|
|
; CHECK: liveins: $sgpr30_sgpr31, $vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71_vgpr72_vgpr73
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 9
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71_vgpr72_vgpr73, $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $vgpr42, implicit-def $vgpr45
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 9
|
|
; CHECK-NEXT: $vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71_vgpr72_vgpr73 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr43, implicit $vgpr44, implicit $vgpr46, implicit $vgpr47, implicit $vgpr56, implicit $vgpr57, implicit $vgpr58, implicit $vgpr59, implicit $vgpr60, implicit $vgpr61, implicit $vgpr62, implicit $vgpr63, implicit $vgpr72, implicit $vgpr73 :: (load (s1024) from %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
S_NOP 0, implicit-def $vgpr42, implicit-def $vgpr45
|
|
S_SETPC_B64_return $sgpr30_sgpr31
|
|
...
|
|
|
|
# Block load/store v40-47 and v56-63 (v48-55 and v64-71 are caller-saved). The
|
|
# mask should be 0x00FF00FF.
|
|
|
|
---
|
|
name: one_block_csr_only
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
stackPtrOffsetReg: $sgpr32
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr30_sgpr31
|
|
; CHECK-LABEL: name: one_block_csr_only
|
|
; CHECK: liveins: $sgpr30_sgpr31, $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 16711935
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71, $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $vgpr40, implicit-def $vgpr41, implicit-def $vgpr42, implicit-def $vgpr43, implicit-def $vgpr44, implicit-def $vgpr45, implicit-def $vgpr46, implicit-def $vgpr47, implicit-def $vgpr48, implicit-def $vgpr49, implicit-def $vgpr50, implicit-def $vgpr51, implicit-def $vgpr52, implicit-def $vgpr53, implicit-def $vgpr54, implicit-def $vgpr55, implicit-def $vgpr56, implicit-def $vgpr57, implicit-def $vgpr58, implicit-def $vgpr59, implicit-def $vgpr60, implicit-def $vgpr61, implicit-def $vgpr62, implicit-def $vgpr63, implicit-def $vgpr64, implicit-def $vgpr65, implicit-def $vgpr66
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 16711935
|
|
; CHECK-NEXT: $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (load (s1024) from %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
S_NOP 0, implicit-def $vgpr40, implicit-def $vgpr41, implicit-def $vgpr42, implicit-def $vgpr43, implicit-def $vgpr44, implicit-def $vgpr45, implicit-def $vgpr46, implicit-def $vgpr47, implicit-def $vgpr48, implicit-def $vgpr49, implicit-def $vgpr50, implicit-def $vgpr51, implicit-def $vgpr52, implicit-def $vgpr53, implicit-def $vgpr54, implicit-def $vgpr55, implicit-def $vgpr56, implicit-def $vgpr57, implicit-def $vgpr58, implicit-def $vgpr59, implicit-def $vgpr60, implicit-def $vgpr61, implicit-def $vgpr62, implicit-def $vgpr63, implicit-def $vgpr64, implicit-def $vgpr65, implicit-def $vgpr66
|
|
S_SETPC_B64_return $sgpr30_sgpr31
|
|
...
|
|
|
|
# Block load/store to/from different blocks.
|
|
# Note the mask for storing v232, which is 0x100 because we have to start the
|
|
# block at v224 (since the upper limit is 255). For the same reason, the first
|
|
# stack slot will be 36 bytes long (the first 32 will be empty, since the memory
|
|
# will not get compacted). The second slot, which will hold registers v104 and
|
|
# v110, will be 28 bytes long, and finally the third, holding registers v40 and
|
|
# v41, will be 8 bytes long.
|
|
---
|
|
name: multiple_blocks
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
stackPtrOffsetReg: $sgpr32
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr30_sgpr31
|
|
; CHECK-LABEL: name: multiple_blocks
|
|
; CHECK: liveins: $sgpr30_sgpr31, $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71, $vgpr104_vgpr105_vgpr106_vgpr107_vgpr108_vgpr109_vgpr110_vgpr111_vgpr112_vgpr113_vgpr114_vgpr115_vgpr116_vgpr117_vgpr118_vgpr119_vgpr120_vgpr121_vgpr122_vgpr123_vgpr124_vgpr125_vgpr126_vgpr127_vgpr128_vgpr129_vgpr130_vgpr131_vgpr132_vgpr133_vgpr134_vgpr135, $vgpr224_vgpr225_vgpr226_vgpr227_vgpr228_vgpr229_vgpr230_vgpr231_vgpr232_vgpr233_vgpr234_vgpr235_vgpr236_vgpr237_vgpr238_vgpr239_vgpr240_vgpr241_vgpr242_vgpr243_vgpr244_vgpr245_vgpr246_vgpr247_vgpr248_vgpr249_vgpr250_vgpr251_vgpr252_vgpr253_vgpr254_vgpr255
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 3
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71, $sgpr32, 64, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 65
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr104_vgpr105_vgpr106_vgpr107_vgpr108_vgpr109_vgpr110_vgpr111_vgpr112_vgpr113_vgpr114_vgpr115_vgpr116_vgpr117_vgpr118_vgpr119_vgpr120_vgpr121_vgpr122_vgpr123_vgpr124_vgpr125_vgpr126_vgpr127_vgpr128_vgpr129_vgpr130_vgpr131_vgpr132_vgpr133_vgpr134_vgpr135, $sgpr32, 36, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.1, align 4, addrspace 5)
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 256
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr224_vgpr225_vgpr226_vgpr227_vgpr228_vgpr229_vgpr230_vgpr231_vgpr232_vgpr233_vgpr234_vgpr235_vgpr236_vgpr237_vgpr238_vgpr239_vgpr240_vgpr241_vgpr242_vgpr243_vgpr244_vgpr245_vgpr246_vgpr247_vgpr248_vgpr249_vgpr250_vgpr251_vgpr252_vgpr253_vgpr254_vgpr255, $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.2, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $vgpr40, implicit-def $vgpr41, implicit-def $vgpr104, implicit-def $vgpr110, implicit-def $vgpr232
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 256
|
|
; CHECK-NEXT: $vgpr224_vgpr225_vgpr226_vgpr227_vgpr228_vgpr229_vgpr230_vgpr231_vgpr232_vgpr233_vgpr234_vgpr235_vgpr236_vgpr237_vgpr238_vgpr239_vgpr240_vgpr241_vgpr242_vgpr243_vgpr244_vgpr245_vgpr246_vgpr247_vgpr248_vgpr249_vgpr250_vgpr251_vgpr252_vgpr253_vgpr254_vgpr255 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr233, implicit $vgpr234, implicit $vgpr235, implicit $vgpr236, implicit $vgpr237, implicit $vgpr238, implicit $vgpr239, implicit $vgpr248, implicit $vgpr249, implicit $vgpr250, implicit $vgpr251, implicit $vgpr252, implicit $vgpr253, implicit $vgpr254, implicit $vgpr255 :: (load (s1024) from %stack.2, align 4, addrspace 5)
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 65
|
|
; CHECK-NEXT: $vgpr104_vgpr105_vgpr106_vgpr107_vgpr108_vgpr109_vgpr110_vgpr111_vgpr112_vgpr113_vgpr114_vgpr115_vgpr116_vgpr117_vgpr118_vgpr119_vgpr120_vgpr121_vgpr122_vgpr123_vgpr124_vgpr125_vgpr126_vgpr127_vgpr128_vgpr129_vgpr130_vgpr131_vgpr132_vgpr133_vgpr134_vgpr135 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 36, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr105, implicit $vgpr106, implicit $vgpr107, implicit $vgpr108, implicit $vgpr109, implicit $vgpr111, implicit $vgpr120, implicit $vgpr121, implicit $vgpr122, implicit $vgpr123, implicit $vgpr124, implicit $vgpr125, implicit $vgpr126, implicit $vgpr127 :: (load (s1024) from %stack.1, align 4, addrspace 5)
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 3
|
|
; CHECK-NEXT: $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 64, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr42, implicit $vgpr43, implicit $vgpr44, implicit $vgpr45, implicit $vgpr46, implicit $vgpr47, implicit $vgpr56, implicit $vgpr57, implicit $vgpr58, implicit $vgpr59, implicit $vgpr60, implicit $vgpr61, implicit $vgpr62, implicit $vgpr63 :: (load (s1024) from %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
S_NOP 0, implicit-def $vgpr40, implicit-def $vgpr41, implicit-def $vgpr104, implicit-def $vgpr110, implicit-def $vgpr232
|
|
S_SETPC_B64_return $sgpr30_sgpr31
|
|
...
|
|
|
|
# Make sure we handle register tuples correctly, even when they're straddling
|
|
# the boundary between blocks. The first mask should be 0x00000007 (the bottom
|
|
# 2 registers from the second tuple are not callee saves), the second
|
|
# 0x00000003.
|
|
|
|
---
|
|
name: reg_tuples
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
stackPtrOffsetReg: $sgpr32
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr30_sgpr31
|
|
; CHECK-LABEL: name: reg_tuples
|
|
; CHECK: liveins: $sgpr30_sgpr31, $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71, $vgpr72_vgpr73_vgpr74_vgpr75_vgpr76_vgpr77_vgpr78_vgpr79_vgpr80_vgpr81_vgpr82_vgpr83_vgpr84_vgpr85_vgpr86_vgpr87_vgpr88_vgpr89_vgpr90_vgpr91_vgpr92_vgpr93_vgpr94_vgpr95_vgpr96_vgpr97_vgpr98_vgpr99_vgpr100_vgpr101_vgpr102_vgpr103
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 7
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71, $sgpr32, 8, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 3
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr72_vgpr73_vgpr74_vgpr75_vgpr76_vgpr77_vgpr78_vgpr79_vgpr80_vgpr81_vgpr82_vgpr83_vgpr84_vgpr85_vgpr86_vgpr87_vgpr88_vgpr89_vgpr90_vgpr91_vgpr92_vgpr93_vgpr94_vgpr95_vgpr96_vgpr97_vgpr98_vgpr99_vgpr100_vgpr101_vgpr102_vgpr103, $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.1, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $vgpr40_vgpr41_vgpr42, implicit-def $vgpr70_vgpr71_vgpr72_vgpr73
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 3
|
|
; CHECK-NEXT: $vgpr72_vgpr73_vgpr74_vgpr75_vgpr76_vgpr77_vgpr78_vgpr79_vgpr80_vgpr81_vgpr82_vgpr83_vgpr84_vgpr85_vgpr86_vgpr87_vgpr88_vgpr89_vgpr90_vgpr91_vgpr92_vgpr93_vgpr94_vgpr95_vgpr96_vgpr97_vgpr98_vgpr99_vgpr100_vgpr101_vgpr102_vgpr103 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr74, implicit $vgpr75, implicit $vgpr76, implicit $vgpr77, implicit $vgpr78, implicit $vgpr79, implicit $vgpr88, implicit $vgpr89, implicit $vgpr90, implicit $vgpr91, implicit $vgpr92, implicit $vgpr93, implicit $vgpr94, implicit $vgpr95 :: (load (s1024) from %stack.1, align 4, addrspace 5)
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 7
|
|
; CHECK-NEXT: $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 8, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr43, implicit $vgpr44, implicit $vgpr45, implicit $vgpr46, implicit $vgpr47, implicit $vgpr56, implicit $vgpr57, implicit $vgpr58, implicit $vgpr59, implicit $vgpr60, implicit $vgpr61, implicit $vgpr62, implicit $vgpr63 :: (load (s1024) from %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
S_NOP 0, implicit-def $vgpr40_vgpr41_vgpr42, implicit-def $vgpr70_vgpr71_vgpr72_vgpr73
|
|
S_SETPC_B64_return $sgpr30_sgpr31
|
|
...
|
|
|
|
# Make sure we don't overwrite any stack variables.
|
|
|
|
---
|
|
name: locals
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
stackPtrOffsetReg: $sgpr32
|
|
stack:
|
|
- { id: 0, type: default, offset: 0, size: 12, alignment: 4,
|
|
stack-id: default, callee-saved-register: '', callee-saved-restored: true,
|
|
local-offset: 0, debug-info-variable: '', debug-info-expression: '',
|
|
debug-info-location: '' }
|
|
- { id: 1, type: default, offset: 12, size: 20, alignment: 4,
|
|
stack-id: default, callee-saved-register: '', callee-saved-restored: true,
|
|
local-offset: 0, debug-info-variable: '', debug-info-expression: '',
|
|
debug-info-location: '' }
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr30_sgpr31, $vgpr48
|
|
; CHECK-LABEL: name: locals
|
|
; CHECK: liveins: $vgpr48, $sgpr30_sgpr31, $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 1
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71, $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.2, align 4, addrspace 5)
|
|
; CHECK-NEXT: SCRATCH_STORE_DWORD_SADDR $vgpr48, $sgpr32, 4, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.0, addrspace 5)
|
|
; CHECK-NEXT: SCRATCH_STORE_DWORD_SADDR $vgpr48, $sgpr32, 20, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.1, addrspace 5)
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $vgpr40
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 1
|
|
; CHECK-NEXT: $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr41, implicit $vgpr42, implicit $vgpr43, implicit $vgpr44, implicit $vgpr45, implicit $vgpr46, implicit $vgpr47, implicit $vgpr56, implicit $vgpr57, implicit $vgpr58, implicit $vgpr59, implicit $vgpr60, implicit $vgpr61, implicit $vgpr62, implicit $vgpr63 :: (load (s1024) from %stack.2, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
SCRATCH_STORE_DWORD_SADDR $vgpr48, %stack.0, 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.0, addrspace 5)
|
|
SCRATCH_STORE_DWORD_SADDR $vgpr48, %stack.1, 4, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.1, addrspace 5)
|
|
S_NOP 0, implicit-def $vgpr40
|
|
S_SETPC_B64_return $sgpr30_sgpr31
|
|
...
|
|
|
|
# Make sure we don't break SGPR or WWM handling, and also that we don't
|
|
# block-spill WWM VGPRs that have already been spilled (the mask for the block
|
|
# load/store should be 0x9 because we don't want to include v41 or v42).
|
|
# Use all VGPRs up to v40, so the WWM registers v41 and v42 and the VGPR used
|
|
# for SGPR spills remain within the block.
|
|
|
|
---
|
|
name: other_regs
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
stackPtrOffsetReg: $sgpr32
|
|
wwmReservedRegs:
|
|
- '$vgpr41'
|
|
- '$vgpr42'
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr30_sgpr31, $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39_vgpr40
|
|
; W32-LABEL: name: other_regs
|
|
; W32: liveins: $sgpr48, $sgpr30_sgpr31, $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39_vgpr40, $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71
|
|
; W32-NEXT: {{ $}}
|
|
; W32-NEXT: $sgpr0 = S_OR_SAVEEXEC_B32 -1, implicit-def $exec, implicit-def dead $scc, implicit $exec
|
|
; W32-NEXT: SCRATCH_STORE_DWORD_SADDR killed $vgpr41, $sgpr32, 16, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.1, addrspace 5)
|
|
; W32-NEXT: SCRATCH_STORE_DWORD_SADDR killed $vgpr42, $sgpr32, 20, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.2, addrspace 5)
|
|
; W32-NEXT: SCRATCH_STORE_DWORD_SADDR killed $vgpr44, $sgpr32, 24, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.3, addrspace 5)
|
|
; W32-NEXT: $exec_lo = S_MOV_B32 killed $sgpr0
|
|
; W32-NEXT: $m0 = S_MOV_B32 9
|
|
; W32-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71, $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.4, align 4, addrspace 5)
|
|
; W32-NEXT: $vgpr44 = SI_SPILL_S32_TO_VGPR $sgpr48, 0, $vgpr44
|
|
; W32-NEXT: S_NOP 0, implicit-def $vgpr40, implicit-def $vgpr41, implicit-def $vgpr43, implicit-def $sgpr22, implicit-def $sgpr48, implicit-def $m0, implicit-def $exec
|
|
; W32-NEXT: S_NOP 0, implicit $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, implicit $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, implicit $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, implicit $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, implicit $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39_vgpr40
|
|
; W32-NEXT: $sgpr48 = SI_RESTORE_S32_FROM_VGPR $vgpr44, 0
|
|
; W32-NEXT: $m0 = S_MOV_B32 9
|
|
; W32-NEXT: $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr41, implicit $vgpr42, implicit $vgpr44, implicit $vgpr45, implicit $vgpr46, implicit $vgpr47, implicit $vgpr56, implicit $vgpr57, implicit $vgpr58, implicit $vgpr59, implicit $vgpr60, implicit $vgpr61, implicit $vgpr62, implicit $vgpr63 :: (load (s1024) from %stack.4, align 4, addrspace 5)
|
|
; W32-NEXT: $sgpr0 = S_OR_SAVEEXEC_B32 -1, implicit-def $exec, implicit-def dead $scc, implicit $exec
|
|
; W32-NEXT: $vgpr41 = SCRATCH_LOAD_DWORD_SADDR $sgpr32, 16, 0, implicit $exec, implicit $flat_scr :: (load (s32) from %stack.1, addrspace 5)
|
|
; W32-NEXT: $vgpr42 = SCRATCH_LOAD_DWORD_SADDR $sgpr32, 20, 0, implicit $exec, implicit $flat_scr :: (load (s32) from %stack.2, addrspace 5)
|
|
; W32-NEXT: $vgpr44 = SCRATCH_LOAD_DWORD_SADDR $sgpr32, 24, 0, implicit $exec, implicit $flat_scr :: (load (s32) from %stack.3, addrspace 5)
|
|
; W32-NEXT: $exec_lo = S_MOV_B32 killed $sgpr0
|
|
; W32-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
;
|
|
; W64-LABEL: name: other_regs
|
|
; W64: liveins: $sgpr48, $sgpr30_sgpr31, $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39_vgpr40, $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71
|
|
; W64-NEXT: {{ $}}
|
|
; W64-NEXT: $sgpr0_sgpr1 = S_OR_SAVEEXEC_B64 -1, implicit-def $exec, implicit-def dead $scc, implicit $exec
|
|
; W64-NEXT: SCRATCH_STORE_DWORD_SADDR killed $vgpr41, $sgpr32, 16, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.1, addrspace 5)
|
|
; W64-NEXT: SCRATCH_STORE_DWORD_SADDR killed $vgpr42, $sgpr32, 20, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.2, addrspace 5)
|
|
; W64-NEXT: SCRATCH_STORE_DWORD_SADDR killed $vgpr44, $sgpr32, 24, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %stack.3, addrspace 5)
|
|
; W64-NEXT: $exec = S_MOV_B64 killed $sgpr0_sgpr1
|
|
; W64-NEXT: $m0 = S_MOV_B32 9
|
|
; W64-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71, $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.4, align 4, addrspace 5)
|
|
; W64-NEXT: $vgpr44 = SI_SPILL_S32_TO_VGPR $sgpr48, 0, $vgpr44
|
|
; W64-NEXT: S_NOP 0, implicit-def $vgpr40, implicit-def $vgpr41, implicit-def $vgpr43, implicit-def $sgpr22, implicit-def $sgpr48, implicit-def $m0, implicit-def $exec
|
|
; W64-NEXT: S_NOP 0, implicit $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, implicit $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, implicit $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, implicit $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, implicit $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39_vgpr40
|
|
; W64-NEXT: $sgpr48 = SI_RESTORE_S32_FROM_VGPR $vgpr44, 0
|
|
; W64-NEXT: $m0 = S_MOV_B32 9
|
|
; W64-NEXT: $vgpr40_vgpr41_vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr41, implicit $vgpr42, implicit $vgpr44, implicit $vgpr45, implicit $vgpr46, implicit $vgpr47, implicit $vgpr56, implicit $vgpr57, implicit $vgpr58, implicit $vgpr59, implicit $vgpr60, implicit $vgpr61, implicit $vgpr62, implicit $vgpr63 :: (load (s1024) from %stack.4, align 4, addrspace 5)
|
|
; W64-NEXT: $sgpr0_sgpr1 = S_OR_SAVEEXEC_B64 -1, implicit-def $exec, implicit-def dead $scc, implicit $exec
|
|
; W64-NEXT: $vgpr41 = SCRATCH_LOAD_DWORD_SADDR $sgpr32, 16, 0, implicit $exec, implicit $flat_scr :: (load (s32) from %stack.1, addrspace 5)
|
|
; W64-NEXT: $vgpr42 = SCRATCH_LOAD_DWORD_SADDR $sgpr32, 20, 0, implicit $exec, implicit $flat_scr :: (load (s32) from %stack.2, addrspace 5)
|
|
; W64-NEXT: $vgpr44 = SCRATCH_LOAD_DWORD_SADDR $sgpr32, 24, 0, implicit $exec, implicit $flat_scr :: (load (s32) from %stack.3, addrspace 5)
|
|
; W64-NEXT: $exec = S_MOV_B64 killed $sgpr0_sgpr1
|
|
; W64-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
S_NOP 0, implicit-def $vgpr40, implicit-def $vgpr41, implicit-def $vgpr43, implicit-def $sgpr22, implicit-def $sgpr48, implicit-def $m0, implicit-def $exec
|
|
S_NOP 0, implicit $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, implicit $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, implicit $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, implicit $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, implicit $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39_vgpr40
|
|
|
|
S_SETPC_B64_return $sgpr30_sgpr31
|
|
...
|
|
|
|
# Make sure we don't break anything for entry functions.
|
|
|
|
---
|
|
name: entry_func
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
stackPtrOffsetReg: $sgpr32
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr30_sgpr31
|
|
; CHECK-LABEL: name: entry_func
|
|
; CHECK: liveins: $sgpr30_sgpr31
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $vgpr42, implicit-def $vgpr45, implicit-def $vgpr51
|
|
; CHECK-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
S_NOP 0, implicit-def $vgpr42, implicit-def $vgpr45, implicit-def $vgpr51
|
|
S_SETPC_B64_return $sgpr30_sgpr31
|
|
...
|
|
|
|
---
|
|
name: multiple_basic_blocks
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
stackPtrOffsetReg: $sgpr32
|
|
body: |
|
|
; CHECK-LABEL: name: multiple_basic_blocks
|
|
; CHECK: bb.0:
|
|
; CHECK-NEXT: successors: %bb.1(0x80000000)
|
|
; CHECK-NEXT: liveins: $vgpr44, $sgpr30_sgpr31, $vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71_vgpr72_vgpr73
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 11
|
|
; CHECK-NEXT: SCRATCH_STORE_BLOCK_SADDR $vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71_vgpr72_vgpr73, $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0 :: (store (s1024) into %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $vgpr42, implicit-def $vgpr45
|
|
; CHECK-NEXT: S_BRANCH %bb.1
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: bb.1:
|
|
; CHECK-NEXT: successors: %bb.2(0x80000000)
|
|
; CHECK-NEXT: liveins: $vgpr44, $sgpr30_sgpr31
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: S_NOP 0, implicit-def $vgpr43, implicit $vgpr44
|
|
; CHECK-NEXT: S_BRANCH %bb.2
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: bb.2:
|
|
; CHECK-NEXT: liveins: $sgpr30_sgpr31, $vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71_vgpr72_vgpr73
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: $m0 = S_MOV_B32 11
|
|
; CHECK-NEXT: $vgpr42_vgpr43_vgpr44_vgpr45_vgpr46_vgpr47_vgpr48_vgpr49_vgpr50_vgpr51_vgpr52_vgpr53_vgpr54_vgpr55_vgpr56_vgpr57_vgpr58_vgpr59_vgpr60_vgpr61_vgpr62_vgpr63_vgpr64_vgpr65_vgpr66_vgpr67_vgpr68_vgpr69_vgpr70_vgpr71_vgpr72_vgpr73 = SCRATCH_LOAD_BLOCK_SADDR $sgpr32, 0, 0, implicit $exec, implicit $flat_scr, implicit $m0, implicit $vgpr44, implicit $vgpr46, implicit $vgpr47, implicit $vgpr56, implicit $vgpr57, implicit $vgpr58, implicit $vgpr59, implicit $vgpr60, implicit $vgpr61, implicit $vgpr62, implicit $vgpr63, implicit $vgpr72, implicit $vgpr73 :: (load (s1024) from %stack.0, align 4, addrspace 5)
|
|
; CHECK-NEXT: S_SETPC_B64_return $sgpr30_sgpr31
|
|
bb.0:
|
|
liveins: $sgpr30_sgpr31, $vgpr44
|
|
S_NOP 0, implicit-def $vgpr42, implicit-def $vgpr45
|
|
S_BRANCH %bb.1
|
|
|
|
bb.1:
|
|
liveins: $sgpr30_sgpr31, $vgpr44
|
|
S_NOP 0, implicit-def $vgpr43, implicit $vgpr44
|
|
S_BRANCH %bb.2
|
|
|
|
bb.2:
|
|
liveins: $sgpr30_sgpr31
|
|
S_SETPC_B64_return $sgpr30_sgpr31
|
|
...
|