[llvm] [MachineOutliner] Sort by Benefit to Cost Ratio (PR #90264)
via llvm-commits
llvm-commits at lists.llvm.org
Mon Jun 3 08:10:27 PDT 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-backend-aarch64
Author: Xuan Zhang (xuanzh-meta)
<details>
<summary>Changes</summary>
This PR depends on https://github.com/llvm/llvm-project/pull/90260
We changed the order in which functions are outlined in Machine Outliner.
The formula for priority is found via a black-box Bayesian optimization toolbox. Using this formula for sorting consistently reduces the uncompressed size of large real-world mobile apps. We also ran a few benchmarks using LLVM test suites, and showed that sorting by priority consistently reduces the text segment size.
|run (CTMark/) |baseline (1)|priority (2)|diff (1 -> 2)|
|----------------|------------|------------|-------------|
|lencod |349624 |349264 |-0.1030% |
|SPASS |219672 |219480 |-0.0874% |
|kc |271956 |251200 |-7.6321% |
|sqlite3 |223920 |223708 |-0.0947% |
|7zip-benchmark |405364 |402624 |-0.6759% |
|bullet |139820 |139500 |-0.2289% |
|consumer-typeset|295684 |290196 |-1.8560% |
|pairlocalalign |72236 |72092 |-0.1993% |
|tramp3d-v4 |189572 |189292 |-0.1477% |
This is part of an enhanced version of machine outliner -- see [RFC](https://discourse.llvm.org/t/rfc-enhanced-machine-outliner-part-1-fulllto-part-2-thinlto-nolto-to-come/78732).
---
Patch is 29.07 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/90264.diff
6 Files Affected:
- (modified) llvm/lib/CodeGen/MachineOutliner.cpp (+4-2)
- (added) llvm/test/CodeGen/AArch64/machine-outliner-sort-per-priority.ll (+96)
- (modified) llvm/test/CodeGen/ARM/machine-outliner-calls.mir (+28-52)
- (modified) llvm/test/CodeGen/ARM/machine-outliner-default.mir (+15-18)
- (modified) llvm/test/CodeGen/ARM/machine-outliner-stack-fixup-arm.mir (+24-32)
- (modified) llvm/test/CodeGen/ARM/machine-outliner-stack-fixup-thumb.mir (+27-47)
``````````diff
diff --git a/llvm/lib/CodeGen/MachineOutliner.cpp b/llvm/lib/CodeGen/MachineOutliner.cpp
index 626e577a30bf3..9e2e6316dc6d1 100644
--- a/llvm/lib/CodeGen/MachineOutliner.cpp
+++ b/llvm/lib/CodeGen/MachineOutliner.cpp
@@ -828,10 +828,12 @@ bool MachineOutliner::outline(Module &M,
<< "\n");
bool OutlinedSomething = false;
- // Sort by benefit. The most beneficial functions should be outlined first.
+ // Sort by priority where priority := getNotOutlinedCost / getOutliningCost.
+ // The function with highest priority should be outlined first.
stable_sort(FunctionList,
[](const OutlinedFunction &LHS, const OutlinedFunction &RHS) {
- return LHS.getBenefit() > RHS.getBenefit();
+ return LHS.getNotOutlinedCost() * RHS.getOutliningCost() >
+ RHS.getNotOutlinedCost() * LHS.getOutliningCost();
});
// Walk over each function, outlining them as we go along. Functions are
diff --git a/llvm/test/CodeGen/AArch64/machine-outliner-sort-per-priority.ll b/llvm/test/CodeGen/AArch64/machine-outliner-sort-per-priority.ll
new file mode 100644
index 0000000000000..00efc3c6e71c8
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/machine-outliner-sort-per-priority.ll
@@ -0,0 +1,96 @@
+; This tests the order in which functions are outlined in MachineOutliner
+; There are TWO key OutlinedFunction in FunctionList
+;
+; ===================== First One =====================
+; ```
+; mov w0, #1
+; mov w1, #2
+; mov w2, #3
+; mov w3, #4
+; mov w4, #5
+; ```
+; It has:
+; - `SequenceSize=20` and `OccurrenceCount=6`
+; - each Candidate has `CallOverhead=12` and `FrameOverhead=4`
+; - `NotOutlinedCost=20*6=120` and `OutliningCost=12*6+20+4=96`
+; - `Benefit=120-96=24` and `Priority=120/96=1.25`
+;
+; ===================== Second One =====================
+; ```
+; mov w6, #6
+; mov w7, #7
+; b
+; ```
+; It has:
+; - `SequenceSize=12` and `OccurrenceCount=4`
+; - each Candidate has `CallOverhead=4` and `FrameOverhead=0`
+; - `NotOutlinedCost=12*4=48` and `OutliningCost=4*4+12+0=28`
+; - `Benefit=120-96=20` and `Priority=48/28=1.71`
+;
+; Note that the first one has higher benefit, but lower priority.
+; Hence, when outlining per priority, the second one will be outlined first.
+
+; RUN: llc %s -enable-machine-outliner=always -filetype=obj -o %t
+; RUN: llvm-objdump -d %t | FileCheck %s --check-prefix=CHECK-SORT-BY-PRIORITY
+
+; RUN: llc %s -enable-machine-outliner=always -outliner-benefit-threshold=22 -filetype=obj -o %t
+; RUN: llvm-objdump -d %t | FileCheck %s --check-prefix=CHECK-THRESHOLD
+
+
+target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
+target triple = "arm64-apple-macosx14.0.0"
+
+declare i32 @_Z3fooiiii(i32 noundef, i32 noundef, i32 noundef, i32 noundef, i32 noundef, i32 noundef, i32 noundef, i32 noundef)
+
+define i32 @_Z2f1v() minsize {
+ %1 = tail call i32 @_Z3fooiiii(i32 noundef 1, i32 noundef 2, i32 noundef 3, i32 noundef 4, i32 noundef 5, i32 noundef 11, i32 noundef 6, i32 noundef 7)
+ ret i32 %1
+}
+
+define i32 @_Z2f2v() minsize {
+ %1 = tail call i32 @_Z3fooiiii(i32 noundef 1, i32 noundef 2, i32 noundef 3, i32 noundef 4, i32 noundef 5, i32 noundef 12, i32 noundef 6, i32 noundef 7)
+ ret i32 %1
+}
+
+define i32 @_Z2f3v() minsize {
+ %1 = tail call i32 @_Z3fooiiii(i32 noundef 1, i32 noundef 2, i32 noundef 3, i32 noundef 4, i32 noundef 5, i32 noundef 13, i32 noundef 6, i32 noundef 7)
+ ret i32 %1
+}
+
+define i32 @_Z2f4v() minsize {
+ %1 = tail call i32 @_Z3fooiiii(i32 noundef 1, i32 noundef 2, i32 noundef 3, i32 noundef 4, i32 noundef 5, i32 noundef 14, i32 noundef 6, i32 noundef 7)
+ ret i32 %1
+}
+
+define i32 @_Z2f5v() minsize {
+ %1 = tail call i32 @_Z3fooiiii(i32 noundef 1, i32 noundef 2, i32 noundef 3, i32 noundef 4, i32 noundef 5, i32 noundef 15, i32 noundef 8, i32 noundef 9)
+ ret i32 %1
+}
+
+define i32 @_Z2f6v() minsize {
+ %1 = tail call i32 @_Z3fooiiii(i32 noundef 1, i32 noundef 2, i32 noundef 3, i32 noundef 4, i32 noundef 5, i32 noundef 16, i32 noundef 9, i32 noundef 8)
+ ret i32 %1
+}
+
+; CHECK-SORT-BY-PRIORITY: <_OUTLINED_FUNCTION_0>:
+; CHECK-SORT-BY-PRIORITY-NEXT: mov w6, #0x6
+; CHECK-SORT-BY-PRIORITY-NEXT: mov w7, #0x7
+; CHECK-SORT-BY-PRIORITY-NEXT: b
+
+; CHECK-SORT-BY-PRIORITY: <_OUTLINED_FUNCTION_1>:
+; CHECK-SORT-BY-PRIORITY-NEXT: mov w0, #0x1
+; CHECK-SORT-BY-PRIORITY-NEXT: mov w1, #0x2
+; CHECK-SORT-BY-PRIORITY-NEXT: mov w2, #0x3
+; CHECK-SORT-BY-PRIORITY-NEXT: mov w3, #0x4
+; CHECK-SORT-BY-PRIORITY-NEXT: mov w4, #0x5
+; CHECK-SORT-BY-PRIORITY-NEXT: ret
+
+; CHECK-THRESHOLD: <_OUTLINED_FUNCTION_0>:
+; CHECK-THRESHOLD-NEXT: mov w0, #0x1
+; CHECK-THRESHOLD-NEXT: mov w1, #0x2
+; CHECK-THRESHOLD-NEXT: mov w2, #0x3
+; CHECK-THRESHOLD-NEXT: mov w3, #0x4
+; CHECK-THRESHOLD-NEXT: mov w4, #0x5
+; CHECK-THRESHOLD-NEXT: ret
+
+; CHECK-THRESHOLD-NOT: <_OUTLINED_FUNCTION_1>:
diff --git a/llvm/test/CodeGen/ARM/machine-outliner-calls.mir b/llvm/test/CodeGen/ARM/machine-outliner-calls.mir
index a92c9dd28be5a..7634ecd6e863a 100644
--- a/llvm/test/CodeGen/ARM/machine-outliner-calls.mir
+++ b/llvm/test/CodeGen/ARM/machine-outliner-calls.mir
@@ -26,15 +26,15 @@ body: |
; CHECK: frame-setup CFI_INSTRUCTION def_cfa_offset 8
; CHECK: frame-setup CFI_INSTRUCTION offset $lr, -4
; CHECK: frame-setup CFI_INSTRUCTION offset $r4, -8
- ; CHECK: BL @OUTLINED_FUNCTION_0
+ ; CHECK: BL @OUTLINED_FUNCTION_2
; CHECK: bb.1:
- ; CHECK: BL @OUTLINED_FUNCTION_0
+ ; CHECK: BL @OUTLINED_FUNCTION_2
; CHECK: bb.2:
- ; CHECK: BL @OUTLINED_FUNCTION_0
+ ; CHECK: BL @OUTLINED_FUNCTION_2
; CHECK: bb.3:
- ; CHECK: BL @OUTLINED_FUNCTION_0
+ ; CHECK: BL @OUTLINED_FUNCTION_2
; CHECK: bb.4:
- ; CHECK: BL @OUTLINED_FUNCTION_0
+ ; CHECK: BL @OUTLINED_FUNCTION_2
; CHECK: bb.5:
; CHECK: $sp = frame-destroy LDMIA_UPD $sp, 14 /* CC::al */, $noreg, def $r4, def $lr
; CHECK: BX_RET 14 /* CC::al */, $noreg
@@ -139,13 +139,13 @@ body: |
; CHECK: frame-setup CFI_INSTRUCTION def_cfa_offset 8
; CHECK: frame-setup CFI_INSTRUCTION offset $lr, -4
; CHECK: frame-setup CFI_INSTRUCTION offset $r4, -8
- ; CHECK: BL @OUTLINED_FUNCTION_1
+ ; CHECK: BL @OUTLINED_FUNCTION_0
; CHECK: bb.1:
- ; CHECK: BL @OUTLINED_FUNCTION_1
+ ; CHECK: BL @OUTLINED_FUNCTION_0
; CHECK: bb.2:
- ; CHECK: BL @OUTLINED_FUNCTION_1
+ ; CHECK: BL @OUTLINED_FUNCTION_0
; CHECK: bb.3:
- ; CHECK: BL @OUTLINED_FUNCTION_1
+ ; CHECK: BL @OUTLINED_FUNCTION_0
; CHECK: bb.4:
; CHECK: $sp = frame-destroy LDMIA_UPD $sp, 14 /* CC::al */, $noreg, def $r4, def $lr
; CHECK: BX_RET 14 /* CC::al */, $noreg
@@ -245,19 +245,19 @@ body: |
; CHECK: frame-setup CFI_INSTRUCTION offset $lr, -4
; CHECK: frame-setup CFI_INSTRUCTION offset $r4, -8
; CHECK: BL @"\01mcount", csr_aapcs, implicit-def dead $lr, implicit $sp
- ; CHECK: BL @OUTLINED_FUNCTION_2
+ ; CHECK: BL @OUTLINED_FUNCTION_1
; CHECK: bb.1:
; CHECK: BL @"\01mcount", csr_aapcs, implicit-def dead $lr, implicit $sp
- ; CHECK: BL @OUTLINED_FUNCTION_2
+ ; CHECK: BL @OUTLINED_FUNCTION_1
; CHECK: bb.2:
; CHECK: BL @"\01mcount", csr_aapcs, implicit-def dead $lr, implicit $sp
- ; CHECK: BL @OUTLINED_FUNCTION_2
+ ; CHECK: BL @OUTLINED_FUNCTION_1
; CHECK: bb.3:
; CHECK: BL @"\01mcount", csr_aapcs, implicit-def dead $lr, implicit $sp
- ; CHECK: BL @OUTLINED_FUNCTION_2
+ ; CHECK: BL @OUTLINED_FUNCTION_1
; CHECK: bb.4:
; CHECK: BL @"\01mcount", csr_aapcs, implicit-def dead $lr, implicit $sp
- ; CHECK: BL @OUTLINED_FUNCTION_2
+ ; CHECK: BL @OUTLINED_FUNCTION_1
; CHECK: bb.5:
; CHECK: $sp = frame-destroy LDMIA_UPD $sp, 14 /* CC::al */, $noreg, def $r4, def $lr
; CHECK: BX_RET 14 /* CC::al */, $noreg
@@ -307,38 +307,17 @@ body: |
bb.0:
BX_RET 14, $noreg
-
; CHECK-LABEL: name: OUTLINED_FUNCTION_0
; CHECK: bb.0:
- ; CHECK: liveins: $r11, $r10, $r9, $r8, $r7, $r6, $r5, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8, $lr
- ; CHECK: early-clobber $sp = frame-setup STR_PRE_IMM killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: frame-setup CFI_INSTRUCTION def_cfa_offset 8
- ; CHECK: frame-setup CFI_INSTRUCTION offset $lr, -8
- ; CHECK: BL @bar, implicit-def dead $lr, implicit $sp
- ; CHECK: $r0 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $r1 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $r2 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $r3 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $r4 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $lr, $sp = frame-destroy LDR_POST_IMM $sp, $noreg, 8, 14 /* CC::al */, $noreg
- ; CHECK: MOVPCLR 14 /* CC::al */, $noreg
-
- ; CHECK-LABEL: name: OUTLINED_FUNCTION_1
- ; CHECK: bb.0:
- ; CHECK: liveins: $r11, $r10, $r9, $r8, $r7, $r6, $r5, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8, $lr
- ; CHECK: early-clobber $sp = frame-setup STR_PRE_IMM killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: frame-setup CFI_INSTRUCTION def_cfa_offset 8
- ; CHECK: frame-setup CFI_INSTRUCTION offset $lr, -8
- ; CHECK: BL @bar, implicit-def dead $lr, implicit $sp
+ ; CHECK: liveins: $r11, $r10, $r9, $r8, $r7, $r6, $r5, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8
; CHECK: $r0 = MOVi 2, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r1 = MOVi 2, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r2 = MOVi 2, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r3 = MOVi 2, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r4 = MOVi 2, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $lr, $sp = frame-destroy LDR_POST_IMM $sp, $noreg, 8, 14 /* CC::al */, $noreg
; CHECK: TAILJMPd @bar, implicit $sp
- ; CHECK-LABEL: name: OUTLINED_FUNCTION_2
+ ; CHECK-LABEL: name: OUTLINED_FUNCTION_1
; CHECK: bb.0:
; CHECK: liveins: $r11, $r10, $r9, $r8, $r7, $r6, $r5, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8
; CHECK: $r0 = MOVi 3, 14 /* CC::al */, $noreg, $noreg
@@ -348,31 +327,28 @@ body: |
; CHECK: $r4 = MOVi 3, 14 /* CC::al */, $noreg, $noreg
; CHECK: MOVPCLR 14 /* CC::al */, $noreg
+ ; CHECK-LABEL: name: OUTLINED_FUNCTION_2
+ ; CHECK: bb.0:
+ ; CHECK: liveins: $r11, $r10, $r9, $r8, $r7, $r6, $r5, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8
+ ; CHECK: $r0 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: $r1 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: $r2 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: $r3 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: $r4 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: MOVPCLR 14 /* CC::al */, $noreg
+
; CHECK-LABEL: name: OUTLINED_FUNCTION_3
; CHECK: bb.0:
- ; CHECK: liveins: $r11, $r10, $r9, $r8, $r6, $r5, $r4, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8, $lr
- ; CHECK: early-clobber $sp = frame-setup t2STR_PRE killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: frame-setup CFI_INSTRUCTION def_cfa_offset 8
- ; CHECK: frame-setup CFI_INSTRUCTION offset $lr, -8
- ; CHECK: tBL 14 /* CC::al */, $noreg, @bar, implicit-def dead $lr, implicit $sp
+ ; CHECK: liveins: $r11, $r10, $r9, $r8, $r6, $r5, $r4, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8
; CHECK: $r0 = t2MOVi 2, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r1 = t2MOVi 2, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r2 = t2MOVi 2, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $lr, $sp = frame-destroy t2LDR_POST $sp, 8, 14 /* CC::al */, $noreg
; CHECK: tTAILJMPdND @bar, 14 /* CC::al */, $noreg, implicit $sp
; CHECK-LABEL: name: OUTLINED_FUNCTION_4
; CHECK: bb.0:
- ; CHECK: liveins: $r11, $r10, $r9, $r8, $r6, $r5, $r4, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8, $lr
- ; CHECK: early-clobber $sp = frame-setup t2STR_PRE killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: frame-setup CFI_INSTRUCTION def_cfa_offset 8
- ; CHECK: frame-setup CFI_INSTRUCTION offset $lr, -8
- ; CHECK: tBL 14 /* CC::al */, $noreg, @bar, implicit-def dead $lr, implicit $sp
+ ; CHECK: liveins: $r11, $r10, $r9, $r8, $r6, $r5, $r4, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8
; CHECK: $r0 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r1 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r2 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $lr, $sp = frame-destroy t2LDR_POST $sp, 8, 14 /* CC::al */, $noreg
; CHECK: tBX_RET 14 /* CC::al */, $noreg
-
-
-
diff --git a/llvm/test/CodeGen/ARM/machine-outliner-default.mir b/llvm/test/CodeGen/ARM/machine-outliner-default.mir
index 6d0218dbfe636..de2b8f5596976 100644
--- a/llvm/test/CodeGen/ARM/machine-outliner-default.mir
+++ b/llvm/test/CodeGen/ARM/machine-outliner-default.mir
@@ -19,17 +19,17 @@ body: |
; CHECK: bb.0:
; CHECK: liveins: $lr
; CHECK: early-clobber $sp = frame-setup STR_PRE_IMM killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: BL @OUTLINED_FUNCTION_0
+ ; CHECK: BL @OUTLINED_FUNCTION_1
; CHECK: $lr, $sp = frame-destroy LDR_POST_IMM $sp, $noreg, 8, 14 /* CC::al */, $noreg
; CHECK: bb.1:
; CHECK: liveins: $lr, $r6, $r7, $r8, $r9, $r10, $r11
; CHECK: early-clobber $sp = frame-setup STR_PRE_IMM killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: BL @OUTLINED_FUNCTION_0
+ ; CHECK: BL @OUTLINED_FUNCTION_1
; CHECK: $lr, $sp = frame-destroy LDR_POST_IMM $sp, $noreg, 8, 14 /* CC::al */, $noreg
; CHECK: bb.2:
; CHECK: liveins: $lr, $r6, $r7, $r8, $r9, $r10, $r11
; CHECK: early-clobber $sp = frame-setup STR_PRE_IMM killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: BL @OUTLINED_FUNCTION_0
+ ; CHECK: BL @OUTLINED_FUNCTION_1
; CHECK: $lr, $sp = frame-destroy LDR_POST_IMM $sp, $noreg, 8, 14 /* CC::al */, $noreg
; CHECK: bb.3:
; CHECK: liveins: $lr, $r6, $r7, $r8, $r9, $r10, $r11
@@ -73,17 +73,17 @@ body: |
; CHECK: bb.0:
; CHECK: liveins: $lr
; CHECK: early-clobber $sp = frame-setup t2STR_PRE killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: tBL 14 /* CC::al */, $noreg, @OUTLINED_FUNCTION_1
+ ; CHECK: tBL 14 /* CC::al */, $noreg, @OUTLINED_FUNCTION_0
; CHECK: $lr, $sp = frame-destroy t2LDR_POST $sp, 8, 14 /* CC::al */, $noreg
; CHECK: bb.1:
; CHECK: liveins: $lr, $r4, $r5, $r6, $r7, $r8, $r9, $r10, $r11
; CHECK: early-clobber $sp = frame-setup t2STR_PRE killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: tBL 14 /* CC::al */, $noreg, @OUTLINED_FUNCTION_1
+ ; CHECK: tBL 14 /* CC::al */, $noreg, @OUTLINED_FUNCTION_0
; CHECK: $lr, $sp = frame-destroy t2LDR_POST $sp, 8, 14 /* CC::al */, $noreg
; CHECK: bb.2:
; CHECK: liveins: $lr, $r4, $r5, $r6, $r7, $r8, $r9, $r10, $r11
; CHECK: early-clobber $sp = frame-setup t2STR_PRE killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ; CHECK: tBL 14 /* CC::al */, $noreg, @OUTLINED_FUNCTION_1
+ ; CHECK: tBL 14 /* CC::al */, $noreg, @OUTLINED_FUNCTION_0
; CHECK: $lr, $sp = frame-destroy t2LDR_POST $sp, 8, 14 /* CC::al */, $noreg
; CHECK: bb.3:
; CHECK: liveins: $lr, $r4, $r5, $r6, $r7, $r8, $r9, $r10, $r11
@@ -114,6 +114,15 @@ body: |
; CHECK-LABEL: name: OUTLINED_FUNCTION_0
; CHECK: bb.0:
+ ; CHECK: liveins: $lr, $r4, $r5, $r6, $r7, $r8, $r9, $r10, $r11
+ ; CHECK: $r0 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: $r1 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: $r2 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: $r3 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK: tBX_RET 14 /* CC::al */, $noreg
+
+ ; CHECK-LABEL: name: OUTLINED_FUNCTION_1
+ ; CHECK: bb.0:
; CHECK: liveins: $lr, $r6, $r7, $r8, $r9, $r10, $r11
; CHECK: $r0 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r1 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
@@ -122,15 +131,3 @@ body: |
; CHECK: $r4 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
; CHECK: $r5 = MOVi 1, 14 /* CC::al */, $noreg, $noreg
; CHECK: MOVPCLR 14 /* CC::al */, $noreg
-
- ; CHECK-LABEL: name: OUTLINED_FUNCTION_1
- ; CHECK: bb.0:
- ; CHECK: liveins: $lr, $r4, $r5, $r6, $r7, $r8, $r9, $r10, $r11
- ; CHECK: $r0 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $r1 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $r2 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: $r3 = t2MOVi 1, 14 /* CC::al */, $noreg, $noreg
- ; CHECK: tBX_RET 14 /* CC::al */, $noreg
-
-
-
diff --git a/llvm/test/CodeGen/ARM/machine-outliner-stack-fixup-arm.mir b/llvm/test/CodeGen/ARM/machine-outliner-stack-fixup-arm.mir
index ae5caa5b7c06d..e71edc8ceb3f6 100644
--- a/llvm/test/CodeGen/ARM/machine-outliner-stack-fixup-arm.mir
+++ b/llvm/test/CodeGen/ARM/machine-outliner-stack-fixup-arm.mir
@@ -18,6 +18,7 @@ body: |
liveins: $r0
; CHECK-LABEL: name: CheckAddrMode_i12
; CHECK: $r1 = MOVr killed $r0, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK-NEXT: BL @foo, implicit-def dead $lr, implicit $sp
; CHECK-NEXT: BL @OUTLINED_FUNCTION_[[I12:[0-9]+]]
; CHECK-NEXT: $r6 = LDRi12 $sp, 4088, 14 /* CC::al */, $noreg
$r1 = MOVr killed $r0, 14, $noreg, $noreg
@@ -47,6 +48,7 @@ body: |
liveins: $r1
; CHECK-LABEL: name: CheckAddrMode3
; CHECK: $r0 = MOVr killed $r1, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK-NEXT: BL @foo, implicit-def dead $lr, implicit $sp
; CHECK-NEXT: BL @OUTLINED_FUNCTION_[[I3:[0-9]+]]
; CHECK-NEXT: $r6 = LDRSH $sp, $noreg, 248, 14 /* CC::al */, $noreg
$r0 = MOVr killed $r1, 14, $noreg, $noreg
@@ -76,6 +78,7 @@ body: |
liveins: $r2
; CHECK-LABEL: name: CheckAddrMode5
; CHECK: $r0 = MOVr killed $r2, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK-NEXT: BL @foo, implicit-def dead $lr, implicit $sp
; CHECK-NEXT: BL @OUTLINED_FUNCTION_[[I5:[0-9]+]]
; CHECK-NEXT: $d5 = VLDRD $sp, 254, 14 /* CC::al */, $noreg
$r0 = MOVr killed $r2, 14, $noreg, $noreg
@@ -110,6 +113,7 @@ body: |
liveins: $r3
; CHECK-LABEL: name: CheckAddrMode5FP16
; CHECK: $r0 = MOVr killed $r3, 14 /* CC::al */, $noreg, $noreg
+ ; CHECK-NEXT: BL @foo, implicit-def dead $lr, implicit $sp
; CHECK-NEXT: BL @OUTLINED_FUNCTION_[[I5FP16:[0-9]+]]
; CHECK-NEXT: $s6 = VLDRH $sp, 252, 14, $noreg
$r0 = MOVr killed $r3, 14, $noreg, $noreg
@@ -146,41 +150,29 @@ body: |
BX_RET 14, $noreg
;CHECK: name: OUTLINED_FUNCTION_[[I5]]
- ;CHECK: early-clobber $sp = frame-setup STR_PRE_IMM killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ;CHECK-NEXT: frame-setup CFI_INSTRUCTION def_cfa_offset 8
- ;CHECK-NEXT: frame-setup CFI_INSTRUCTION offset $lr, -8
- ;CHECK-NEXT: BL @foo, implicit-def dead $lr, implicit $sp
- ;CHECK-NEXT: $d0 = VLDRD $sp, 2, 14 /* CC::al */, $noreg
- ;CHECK-NEXT: $d1 = VLDRD $sp, 10, 14 /* CC::al */, $noreg
- ;CHECK-NEXT: $d4 = VLDRD $sp, 255, 14 /* CC::al */, $noreg
- ;CHECK-NEXT: $lr, $sp = frame-destroy LDR_POST_IMM $sp, $noreg, 8, 14 /* CC::al */, $noreg
+ ;CHECK: liveins: $r10, $r9, $r8, $r7, $r6, $r5, $r4, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8
+ ;CHECK: $d0 = VLDRD $sp, 0, 14 /* CC::al */, $noreg
+ ;CHECK-NEXT: $d1 = VLDRD $sp, 8, 14 /* CC::al */, $noreg
+ ;CHECK-NEXT: $d4 = VLDRD $sp, 253, 14 /* CC::al */, $noreg
+ ;CHECK-NEXT: BX_RET 14 /* CC::al */, $noreg
;CHECK: name: OUTLINED_FUNCTION_[[I5FP16]]
- ;CHECK: early-clobber $sp = frame-setup STR_PRE_IMM killed $lr, $sp, -8, 14 /* CC::al */, $noreg
- ;CHECK-NEXT: frame-setup CFI_INSTRUCTION def_cfa_offset 8
- ;CHECK-NEXT: frame-setup CFI_INSTRUCTION offset $lr, -8
- ;CHECK-NEXT: BL @foo, implicit-def dead $lr, implicit $sp
- ;CHECK-NEXT: $s1 = VLDRH $sp, 4, 14, $noreg
- ;CHECK-NEXT: $s2 = VLDRH $sp, 12, 14, $noreg
- ;CHECK-NEXT: $s5 = VLDRH $sp, 244, 14, $noreg
- ;CHECK-NEXT: $lr, $sp = frame-destroy LDR_POST_IMM $sp, $noreg, 8, 14 /* CC::al */, $noreg
+ ;CHECK: liveins: $r10, $r9, $r8, $r7, $r6, $r5, $r4, $d15, $d14, $d13, $d12, $d11, $d10, $d9, $d8
+ ;CHECK: $s1 = VLDRH $sp, 0, 14, $noreg
+ ;CHECK-NEXT: $s2 = VLDRH $...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/90264
More information about the llvm-commits
mailing list