[llvm] 7401672 - [DSE] Update dereferenceable attributes when adjusting memintrinsic ptr (#125073)

[via llvm-commits]

Author: Björn Pettersson
Date: 2025-02-18T17:51:14+01:00
New Revision: 74016728e34ff4fcc6f0e0c412415cf70c2584e5

URL: https://github.com/llvm/llvm-project/commit/74016728e34ff4fcc6f0e0c412415cf70c2584e5
DIFF: https://github.com/llvm/llvm-project/commit/74016728e34ff4fcc6f0e0c412415cf70c2584e5.diff

LOG: [DSE] Update dereferenceable attributes when adjusting memintrinsic ptr (#125073)

Consider IR like this
call void @llvm.memset.p0.i64(ptr dereferenceable(28) %p, i8 0, i64 28,
i1 false)
  store i32 1, ptr %p

In the past it has been optimized like this:
  %p2 = getelementptr inbounds i8, ptr %p, i64 4
call void @llvm.memset.p0.i64(ptr dereferenceable(28) %p2, i8 0, i64 24,
i1 false)
  store i32 1, ptr %p

As the input IR doesn't guarantee that it is OK to deref 28 bytes
starting at the adjusted pointer %p2 the transformation has been a bit
flawed.

With this patch we make sure to drop any
dereferenceable/dereferenceable_or_null attributes when doing such
transforms. An alternative would have been to adjust the amount of
dereferenceable bytes, but since a memset with a constant length already
implies dereferenceability by itself it is simpler to just drop the
attributes.

The new filtering of attributes is done using a helper that only keep
attributes that we explicitly handle. For the adjusted mem instrinsic
pointers that currently involve "NonNull", "NoUndef" and "Alignment"
(when the alignment is known to be fulfilled also after offsetting the
pointer).

Fixes #115976

Added: 
    

Modified: 
    llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp
    llvm/test/Transforms/DeadStoreElimination/OverwriteStoreBegin.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp b/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp
index 38454053b039e..a1649c276de83 100644
--- a/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp
+++ b/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp
@@ -50,6 +50,7 @@
 #include "llvm/Analysis/TargetLibraryInfo.h"
 #include "llvm/Analysis/ValueTracking.h"
 #include "llvm/IR/Argument.h"
+#include "llvm/IR/AttributeMask.h"
 #include "llvm/IR/BasicBlock.h"
 #include "llvm/IR/Constant.h"
 #include "llvm/IR/ConstantRangeList.h"
@@ -563,6 +564,43 @@ static void shortenAssignment(Instruction *Inst, Value *OriginalDest,
   for_each(LinkedDVRAssigns, InsertAssignForOverlap);
 }
 
+/// Update the attributes given that a memory access is updated (the
+/// dereferenced pointer could be moved forward when shortening a
+/// mem intrinsic).
+static void adjustArgAttributes(AnyMemIntrinsic *Intrinsic, unsigned ArgNo,
+                                uint64_t PtrOffset) {
+  // Remember old attributes.
+  AttributeSet OldAttrs = Intrinsic->getParamAttributes(ArgNo);
+
+  // Find attributes that should be kept, and remove the rest.
+  AttributeMask AttrsToRemove;
+  for (auto &Attr : OldAttrs) {
+    if (Attr.hasKindAsEnum()) {
+      switch (Attr.getKindAsEnum()) {
+      default:
+        break;
+      case Attribute::Alignment:
+        // Only keep alignment if PtrOffset satisfy the alignment.
+        if (isAligned(Attr.getAlignment().valueOrOne(), PtrOffset))
+          continue;
+        break;
+      case Attribute::Dereferenceable:
+      case Attribute::DereferenceableOrNull:
+        // We could reduce the size of these attributes according to
+        // PtrOffset. But we simply drop these for now.
+        break;
+      case Attribute::NonNull:
+      case Attribute::NoUndef:
+        continue;
+      }
+    }
+    AttrsToRemove.addAttribute(Attr);
+  }
+
+  // Remove the attributes that should be dropped.
+  Intrinsic->removeParamAttrs(ArgNo, AttrsToRemove);
+}
+
 static bool tryToShorten(Instruction *DeadI, int64_t &DeadStart,
                          uint64_t &DeadSize, int64_t KillingStart,
                          uint64_t KillingSize, bool IsOverwriteEnd) {
@@ -644,6 +682,7 @@ static bool tryToShorten(Instruction *DeadI, int64_t &DeadStart,
         DeadI->getIterator());
     NewDestGEP->setDebugLoc(DeadIntrinsic->getDebugLoc());
     DeadIntrinsic->setDest(NewDestGEP);
+    adjustArgAttributes(DeadIntrinsic, 0, ToRemoveSize);
   }
 
   // Update attached dbg.assign intrinsics. Assume 8-bit byte.

diff  --git a/llvm/test/Transforms/DeadStoreElimination/OverwriteStoreBegin.ll b/llvm/test/Transforms/DeadStoreElimination/OverwriteStoreBegin.ll
index bc1756f6ca9d1..43fbcfcc600c6 100644
--- a/llvm/test/Transforms/DeadStoreElimination/OverwriteStoreBegin.ll
+++ b/llvm/test/Transforms/DeadStoreElimination/OverwriteStoreBegin.ll
@@ -402,3 +402,33 @@ entry:
   store i64 1, ptr %p, align 1
   ret void
 }
+
+; Verify that we adjust/drop the dereferenceable attribute.
+define void @dereferenceable(ptr nocapture %p) {
+; CHECK-LABEL: @dereferenceable(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 4
+; CHECK-NEXT:    call void @llvm.memset.p0.i64(ptr align 4 [[TMP0]], i8 0, i64 24, i1 false)
+; CHECK-NEXT:    store i32 1, ptr [[P]], align 4
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.memset.p0.i64(ptr dereferenceable(28) align 4 %p, i8 0, i64 28, i1 false)
+  store i32 1, ptr %p, align 4
+  ret void
+}
+
+; Verify that we adjust/drop the dereferenceable_or_null attribute.
+define void @dereferenceable_or_null(ptr nocapture %p) {
+; CHECK-LABEL: @dereferenceable_or_null(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 8
+; CHECK-NEXT:    call void @llvm.memset.p0.i64(ptr align 4 [[TMP0]], i8 0, i64 20, i1 false)
+; CHECK-NEXT:    store i64 1, ptr [[P]], align 4
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.memset.p0.i64(ptr dereferenceable_or_null(28) align 4 %p, i8 0, i64 28, i1 false)
+  store i64 1, ptr %p, align 4
+  ret void
+}