[StackColoring] Delete dead stack slots #75351
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@llvm/pr-subscribers-backend-x86 @llvm/pr-subscribers-debuginfo Author: None (mohammed-nurulhoque) ChangesThe PR above got merged then reverted because of a test failure. It wasn't rebased before merging and in the meantime 3114bd32e75 made remapInstructions not run when SlotRemap is empty. This fixes the issue. Full diff: https://github.com/llvm/llvm-project/pull/75351.diff 5 Files Affected:
diff --git a/llvm/lib/CodeGen/StackColoring.cpp b/llvm/lib/CodeGen/StackColoring.cpp
index 37f7aa9290054e..0f298f5c9c09fd 100644
--- a/llvm/lib/CodeGen/StackColoring.cpp
+++ b/llvm/lib/CodeGen/StackColoring.cpp
@@ -900,6 +900,15 @@ void StackColoring::remapInstructions(DenseMap<int, int> &SlotRemap) {
unsigned FixedMemOp = 0;
unsigned FixedDbg = 0;
+ // Remove debug information for deleted slots.
+ erase_if(MF->getVariableDbgInfo(), [&](auto &VI) {
+ if (!VI.inStackSlot())
+ return false;
+ int Slot = VI.getStackSlot();
+ return Slot >= 0 && Intervals[Slot]->empty() &&
+ InterestingSlots.test(Slot) && !ConservativeSlots.test(Slot);
+ });
+
// Remap debug information that refers to stack slots.
for (auto &VI : MF->getVariableDbgInfo()) {
if (!VI.Var || !VI.inStackSlot())
@@ -1250,8 +1259,15 @@ bool StackColoring::runOnMachineFunction(MachineFunction &Func) {
// Do not bother looking at empty intervals.
for (unsigned I = 0; I < NumSlots; ++I) {
- if (Intervals[SortedSlots[I]]->empty())
+ int Slot = SortedSlots[I];
+ if (Intervals[Slot]->empty()) {
+ if (InterestingSlots.test(Slot) && !ConservativeSlots.test(Slot)) {
+ RemovedSlots += 1;
+ ReducedSize += MFI->getObjectSize(Slot);
+ MFI->RemoveStackObject(Slot);
+ }
SortedSlots[I] = -1;
+ }
}
// This is a simple greedy algorithm for merging allocas. First, sort the
diff --git a/llvm/test/CodeGen/PowerPC/aix32-cc-abi-vaarg.ll b/llvm/test/CodeGen/PowerPC/aix32-cc-abi-vaarg.ll
index bf66a1ed042d22..1b0a803734ae9f 100644
--- a/llvm/test/CodeGen/PowerPC/aix32-cc-abi-vaarg.ll
+++ b/llvm/test/CodeGen/PowerPC/aix32-cc-abi-vaarg.ll
@@ -347,7 +347,6 @@ entry:
; 32BIT-LABEL: stack:
; 32BIT-DAG: - { id: 0, name: arg1, type: default, offset: 0, size: 4, alignment: 4,
-; 32BIT-DAG: - { id: 1, name: arg2, type: default, offset: 0, size: 4, alignment: 4,
; 32BIT-DAG: - { id: 2, name: '', type: default, offset: 0, size: 8, alignment: 8,
; 32BIT-DAG: - { id: 3, name: '', type: default, offset: 0, size: 8, alignment: 8,
diff --git a/llvm/test/CodeGen/PowerPC/aix64-cc-abi-vaarg.ll b/llvm/test/CodeGen/PowerPC/aix64-cc-abi-vaarg.ll
index ccf89aac2d5408..a8684fdfe1c568 100644
--- a/llvm/test/CodeGen/PowerPC/aix64-cc-abi-vaarg.ll
+++ b/llvm/test/CodeGen/PowerPC/aix64-cc-abi-vaarg.ll
@@ -138,9 +138,7 @@
; 64BIT-LABEL: fixedStack:
; 64BIT-DAG: - { id: 0, type: default, offset: 112, size: 8, alignment: 16, stack-id: default,
-; 64BIT-LABEL: stack:
-; 64BIT-DAG: - { id: 0, name: arg1, type: default, offset: 0, size: 8, alignment: 8,
-; 64BIT-DAG: - { id: 1, name: arg2, type: default, offset: 0, size: 8, alignment: 8,
+; 64BIT-LABEL: stack: []
; 64BIT-LABEL: body: |
; 64BIT-DAG: liveins: $x3, $x4, $x5, $x6, $x7, $x8, $x9, $x10
@@ -305,9 +303,7 @@
; 64BIT-LABEL: fixedStack:
; 64BIT-DAG: - { id: 0, type: default, offset: 152, size: 8
-; 64BIT-LABEL: stack:
-; 64BIT-DAG: - { id: 0, name: arg1, type: default, offset: 0, size: 8
-; 64BIT-DAG: - { id: 1, name: arg2, type: default, offset: 0, size: 8
+; 64BIT-LABEL: stack: []
; 64BIT-LABEL: body: |
; 64BIT-DAG: liveins: $f1, $f2, $f3, $f4, $f5, $f6, $f7, $f8, $f9, $f10, $f11, $f12, $f13
diff --git a/llvm/test/CodeGen/RISCV/dead-stack-slot.ll b/llvm/test/CodeGen/RISCV/dead-stack-slot.ll
new file mode 100644
index 00000000000000..49b0d2ab58c4f6
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/dead-stack-slot.ll
@@ -0,0 +1,25 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -verify-machineinstrs < %s \
+; RUN: | FileCheck %s
+; RUN: llc -mtriple=riscv64 -verify-machineinstrs < %s \
+; RUN: | FileCheck %s
+
+; Remove the lifetime-marked alloca, but not the unmarked one.
+define signext i32 @f1() nounwind {
+; CHECK-LABEL: f1:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi sp, sp, -32
+; CHECK-NEXT: li a0, 0
+; CHECK-NEXT: addi sp, sp, 32
+; CHECK-NEXT: ret
+ %1 = alloca [32 x i8], align 4
+ %2 = alloca [32 x i8], align 4
+ %3 = getelementptr inbounds [32 x i8], ptr %1, i64 0, i64 0
+ call void @llvm.lifetime.start.p0(i64 32, ptr nonnull %3)
+ call void @llvm.lifetime.end.p0(i64 32, ptr nonnull %3)
+ ret i32 0
+}
+
+declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
+declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
+
diff --git a/llvm/test/DebugInfo/COFF/lexicalblock.ll b/llvm/test/DebugInfo/COFF/lexicalblock.ll
index 40dd8f894252c2..3bfae85f6c9bad 100644
--- a/llvm/test/DebugInfo/COFF/lexicalblock.ll
+++ b/llvm/test/DebugInfo/COFF/lexicalblock.ll
@@ -63,32 +63,12 @@
; CHECK: LocalSym {
; CHECK: VarName: localA
; CHECK: }
-; CHECK: LocalSym {
-; CHECK: VarName: localB
-; CHECK: }
-; CHECK: BlockSym {
-; CHECK: Kind: S_BLOCK32 {{.*}}
-; CHECK: BlockName:
-; CHECK: }
-; CHECK: ScopeEndSym {
-; CHECK: Kind: S_END {{.*}}
-; CHECK: }
-; CHECK: BlockSym {
-; CHECK: Kind: S_BLOCK32 {{.*}}
-; CHECK: BlockName:
-; CHECK: }
-; CHECK: ScopeEndSym {
-; CHECK: Kind: S_END {{.*}}
-; CHECK: }
; CHECK: BlockSym {
; CHECK: Kind: S_BLOCK32 {{.*}}
; CHECK: BlockName:
; CHECK: }
-; CHECK: ScopeEndSym {
-; CHECK: }
-; CHECK: BlockSym {
-; CHECK: Kind: S_BLOCK32 {{.*}}
-; CHECK: BlockName:
+; CHECK: LocalSym {
+; CHECK: VarName: localB
; CHECK: }
; CHECK: ScopeEndSym {
; CHECK: Kind: S_END {{.*}}
|
mohammed-nurulhoque
changed the title
deletes slots that have lifetime markers and the lifetime ranges are empty.
mohammed-nurulhoque
force-pushed
the
ussc
branch
from
cd73bb9
to
88d52fd
Compare
mohammed-nurulhoque
force-pushed
the
ussc
branch
from
88d52fd
to
9afbff9
Compare
|
Hi, our Linux CI is seing the following failure after this patch, and based on the assert, I'm pretty sure its related to this change. If this is going to be hard to fix, can you revert this until its ready to reland? |
|
I'm seeing the same on s390x. Looks like there's still debug metadata referencing the deleted stack slot, and this causes the assert. I've attached a reduced test case. |
ilovepi
added a commit
to ilovepi/llvm-project
that referenced
this pull request
Dec 15, 2023
This reverts commit 08b306d. it causes the following assertion failure: llvm/include/llvm/CodeGen/MachineFrameInfo.h:530: int64_t llvm::MachineFrameInfo::getObjectOffset(int) const: Assertion `!isDeadObjectIndex(ObjectIdx) && "Getting frame offset for a dead object?"' failed.
ilovepi
added a commit
that referenced
this pull request
Dec 15, 2023
This reverts commit 08b306d. it causes the following assertion failure: llvm/include/llvm/CodeGen/MachineFrameInfo.h:530: int64_t llvm::MachineFrameInfo::getObjectOffset(int) const: Assertion `!isDeadObjectIndex(ObjectIdx) && "Getting frame offset for a dead object?"' failed.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #72633 which got merged then reverted because of a test failure. It wasn't rebased before merging and in the meantime 3114bd32e75 made
remapInstructionsnot run when SlotRemap is empty. This fixes the issue.