Skip to content

[AIEX][NFC] Fix 6 failing lit tests for macOS compatibility and updated codegen #836

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
F-Stuckmann wants to merge 1 commit into
base: aie-public
Choose a base branch
from
Draft

[AIEX][NFC] Fix 6 failing lit tests for macOS compatibility and updated codegen #836

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
2 changes: 1 addition & 1 deletion llvm/test/CodeGen/AIE/GlobalISel/xfail-irtranslator-formal-arguments-vect.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@

; REQUIRES: asserts

; CHECK: Assertion `LocVT == MVT::i32 && "VT should be split in 32-bits chunks"' failed.
; CHECK: VT should be split in 32-bits chunks
define <16 x i64> @arg_v16int64(<16 x i64> %a) {
ret <16 x i64> %a
}
2 changes: 1 addition & 1 deletion llvm/test/CodeGen/AIE/aie2/hardware-loops/multi-use.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@
# RUN: not --crash llc -O2 -mtriple=aie2 -run-pass=aie-hardware-loops %s -o - 2>&1 | FileCheck %s
# requires: asserts

# CHECK: Assertion `0 && "AIE Loops: Wrong/multiple use of LoopDec result!.\n"' failed.
# CHECK: AIE Loops: Wrong/multiple use of LoopDec result!
---
name: multi_use
alignment: 16
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/CodeGen/AIE/aie2/hardware-loops/overwritten-use.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@
# RUN: not --crash llc -O2 -mtriple=aie2 -run-pass=aie-hardware-loops %s -o - 2>&1 | FileCheck %s
# requires: asserts

# CHECK: Assertion `0 && "AIE Loops: Wrong/multiple use of LoopDec result!.\n"' failed.
# CHECK: AIE Loops: Wrong/multiple use of LoopDec result!
---
name: overwritten
alignment: 16
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/CodeGen/AIE/aie2/schedule/interblock/order.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@
#
# (c) Copyright 2023-2024 Advanced Micro Devices, Inc. or its affiliates
# RUN: llc --mtriple=aie2 --start-before=postmisched --debug-only=sched-blocks \
# RUN: --aie-loop-aware=1 %s -o /dev/null |& FileCheck %s
# RUN: --aie-loop-aware=1 %s -o /dev/null 2>&1 | FileCheck %s
# REQUIRES: asserts

# Check post-order scheduling of blocks. This mainly follows the flow
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/CodeGen/AIE/aie2/schedule/loopaware/loop-classify.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@
# (c) Copyright 2023-2024 Advanced Micro Devices, Inc. or its affiliates
#
# RUN: llc --mtriple=aie2 --start-before=postmisched --debug-only=sched-blocks \
# RUN: --aie-loop-aware=1 %s -o /dev/null |& FileCheck %s
# RUN: --aie-loop-aware=1 %s -o /dev/null 2>&1 | FileCheck %s
# REQUIRES: asserts

# Check that a loop that has a loop as epilogue isn't considered a loop
Expand Down
44 changes: 22 additions & 22 deletions llvm/test/CodeGen/AIE/aie2p/ra/waw_reg_renaming_swpaware.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,28 +55,28 @@ body: |
; CHECK-NEXT: successors: %bb.2(0x7c000000), %bb.3(0x04000000)
; CHECK-NEXT: liveins: $dc0, $dc1, $dc4, $dj0, $dj1, $dj4, $dm2:0x000000001800000C, $dm4:0x000000001800000C, $dn0, $dn1, $dn4, $m0, $m1, $p0, $p1, $p2, $r0, $r2, $r3, $r4, $r5, $x0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $x2, $p2, $dc1 = VLD_2D_x_pseudo_split killed $p2, $m1, $dn1, $dj1, killed $dc1 :: (load (<16 x s32>), addrspace 6)
; CHECK-NEXT: renamable $x4, renamable $p1 = VLD_x_pstm_nrm_imm_pseudo killed renamable $p1, 64 :: (load (<32 x s16>), addrspace 5)
; CHECK-NEXT: renamable $x6, renamable $p1 = VLD_x_pstm_nrm_imm_pseudo killed renamable $p1, 64 :: (load (<32 x s16>), addrspace 5)
; CHECK-NEXT: $x8, $p1, $dc0, $dc4 = VLD_3D_x_pseudo_split killed $p1, $m0, $dn0, $dj0, killed $dc0, undef $m4, $dn4, $dj4, killed $dc4 :: (load (<32 x s16>), addrspace 5)
; CHECK-NEXT: renamable $x10 = VEXTBCST_128_vec_extract_broadcast_imm renamable $x2, 0
; CHECK-NEXT: renamable $x1 = VSHIFT renamable $x4, renamable $x6, renamable $r0
; CHECK-NEXT: renamable $dm3 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm2, renamable $x4, renamable $x10, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x3 = VEXTBCST_128_vec_extract_broadcast_imm renamable $x2, 1
; CHECK-NEXT: renamable $dm4 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm4, renamable $x6, killed renamable $x10, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x7 = VSHIFT renamable $x6, renamable $x8, renamable $r0
; CHECK-NEXT: renamable $x5 = VSHIFT renamable $x4, renamable $x6, renamable $r3
; CHECK-NEXT: renamable $dm3 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm3, killed renamable $x1, renamable $x3, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x9 = VEXTBCST_128_vec_extract_broadcast_imm renamable $x2, 2
; CHECK-NEXT: renamable $dm1 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm4, killed renamable $x7, killed renamable $x3, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x10 = VSHIFT renamable $x6, renamable $x8, renamable $r3
; CHECK-NEXT: renamable $x4 = VSHIFT killed renamable $x4, renamable $x6, renamable $r4
; CHECK-NEXT: renamable $dm0 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm3, killed renamable $x5, renamable $x9, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x11 = VEXTBCST_128_vec_extract_broadcast_imm killed renamable $x2, 3
; CHECK-NEXT: renamable $dm1 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm1, killed renamable $x10, killed renamable $x9, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x6 = VSHIFT killed renamable $x6, killed renamable $x8, renamable $r4
; CHECK-NEXT: renamable $dm2 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm0, killed renamable $x4, renamable $x11, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $dm4 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm1, killed renamable $x6, killed renamable $x11, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: $x8, $p2, $dc1 = VLD_2D_x_pseudo_split killed $p2, $m1, $dn1, $dj1, killed $dc1 :: (load (<16 x s32>), addrspace 6)
Copy link
Copy Markdown
Collaborator Author

@F-Stuckmann F-Stuckmann Mar 10, 2026 •
edited
Loading

Choose a reason for hiding this comment

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

wait why do registers change?

; CHECK-NEXT: renamable $x10, renamable $p1 = VLD_x_pstm_nrm_imm_pseudo killed renamable $p1, 64 :: (load (<32 x s16>), addrspace 5)
; CHECK-NEXT: renamable $x1, renamable $p1 = VLD_x_pstm_nrm_imm_pseudo killed renamable $p1, 64 :: (load (<32 x s16>), addrspace 5)
; CHECK-NEXT: $x6, $p1, $dc0, $dc4 = VLD_3D_x_pseudo_split killed $p1, $m0, $dn0, $dj0, killed $dc0, undef $m4, $dn4, $dj4, killed $dc4 :: (load (<32 x s16>), addrspace 5)
; CHECK-NEXT: renamable $x9 = VEXTBCST_128_vec_extract_broadcast_imm renamable $x8, 0
; CHECK-NEXT: renamable $x5 = VSHIFT renamable $x10, renamable $x1, renamable $r0
; CHECK-NEXT: renamable $dm3 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm2, renamable $x10, renamable $x9, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x11 = VEXTBCST_128_vec_extract_broadcast_imm renamable $x8, 1
; CHECK-NEXT: renamable $dm4 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm4, renamable $x1, killed renamable $x9, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x3 = VSHIFT renamable $x1, renamable $x6, renamable $r0
; CHECK-NEXT: renamable $x7 = VSHIFT renamable $x10, renamable $x1, renamable $r3
; CHECK-NEXT: renamable $dm3 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm3, killed renamable $x5, renamable $x11, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x2 = VEXTBCST_128_vec_extract_broadcast_imm renamable $x8, 2
; CHECK-NEXT: renamable $dm1 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm4, killed renamable $x3, killed renamable $x11, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x5 = VSHIFT renamable $x1, renamable $x6, renamable $r3
; CHECK-NEXT: renamable $x10 = VSHIFT killed renamable $x10, renamable $x1, renamable $r4
; CHECK-NEXT: renamable $dm0 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm3, killed renamable $x7, renamable $x2, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x4 = VEXTBCST_128_vec_extract_broadcast_imm killed renamable $x8, 3
; CHECK-NEXT: renamable $dm1 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm1, killed renamable $x5, killed renamable $x2, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $x7 = VSHIFT killed renamable $x1, killed renamable $x6, renamable $r4
; CHECK-NEXT: renamable $dm2 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm0, killed renamable $x10, renamable $x4, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: renamable $dm4 = VMAC_f_vmac_bf_vmul_bf_core_X_X killed renamable $dm1, killed renamable $x7, killed renamable $x4, renamable $r5, implicit-def dead $srfpflags, implicit $crfpmask
; CHECK-NEXT: PseudoLoopEnd <mcsymbol .L_LEnd0>, %bb.2
; CHECK-NEXT: PseudoJ_jump_imm %bb.3
; CHECK-NEXT: {{ $}}
Expand Down
Loading