[MemDepAnalysis] Don't reuse NonLocalPointerDeps cache if memory location size differs (#116936)

As seen in #111585, we can end up using a previous cache entry where the
size was too large and was UB.

Compile time impact:
https://llvm-compile-time-tracker.com/compare.php?from=6a863f7e2679a60f2f38ae6a920d0b6e1a2c1690&to=faccf4e1f47fcd5360a438de2a56d02b770ad498&stat=instructions:u.

Fixes #111585.

Author: aeubanks

llvm/lib/Analysis/MemoryDependenceAnalysis.cpp

@@ -1066,40 +1066,18 @@ bool MemoryDependenceResults::getNonLocalPointerDepFromBB(
   // Invariant loads don't participate in caching. Thus no need to reconcile.
   if (!isInvariantLoad && !Pair.second) {
     if (CacheInfo->Size != Loc.Size) {
-      bool ThrowOutEverything;
-      if (CacheInfo->Size.hasValue() && Loc.Size.hasValue()) {
-        // FIXME: We may be able to do better in the face of results with mixed
-        // precision. We don't appear to get them in practice, though, so just
-        // be conservative.
-        ThrowOutEverything =
-            CacheInfo->Size.isPrecise() != Loc.Size.isPrecise() ||
-            !TypeSize::isKnownGE(CacheInfo->Size.getValue(),
-                                 Loc.Size.getValue());
-      } else {
-        // For our purposes, unknown size > all others.
-        ThrowOutEverything = !Loc.Size.hasValue();
-      }
-
-      if (ThrowOutEverything) {
-        // The query's Size is greater than the cached one. Throw out the
-        // cached data and proceed with the query at the greater size.
-        CacheInfo->Pair = BBSkipFirstBlockPair();
-        CacheInfo->Size = Loc.Size;
-        for (auto &Entry : CacheInfo->NonLocalDeps)
-          if (Instruction *Inst = Entry.getResult().getInst())
-            RemoveFromReverseMap(ReverseNonLocalPtrDeps, Inst, CacheKey);
-        CacheInfo->NonLocalDeps.clear();
-        // The cache is cleared (in the above line) so we will have lost
-        // information about blocks we have already visited. We therefore must
-        // assume that the cache information is incomplete.
-        IsIncomplete = true;
-      } else {
-        // This query's Size is less than the cached one. Conservatively restart
-        // the query using the greater size.
-        return getNonLocalPointerDepFromBB(
-            QueryInst, Pointer, Loc.getWithNewSize(CacheInfo->Size), isLoad,
-            StartBB, Result, Visited, SkipFirstBlock, IsIncomplete);
-      }
+      // The query's Size is not equal to the cached one. Throw out the cached
+      // data and proceed with the query with the new size.
+      CacheInfo->Pair = BBSkipFirstBlockPair();
+      CacheInfo->Size = Loc.Size;
+      for (auto &Entry : CacheInfo->NonLocalDeps)
+        if (Instruction *Inst = Entry.getResult().getInst())
+          RemoveFromReverseMap(ReverseNonLocalPtrDeps, Inst, CacheKey);
+      CacheInfo->NonLocalDeps.clear();
+      // The cache is cleared (in the above line) so we will have lost
+      // information about blocks we have already visited. We therefore must
+      // assume that the cache information is incomplete.
+      IsIncomplete = true;
     }
 
     // If the query's AATags are inconsistent with the cached one,

llvm/test/Analysis/MemoryDependenceAnalysis/load-size-cache.ll

@@ -22,9 +22,11 @@ define i8 @f(i1 %arg0, i1 %arg1, i1 %arg2) {
 ; CHECK-NEXT:    call void @use(i64 undef)
 ; CHECK-NEXT:    br label %[[BB9:.*]]
 ; CHECK:       [[BB7]]:
+; CHECK-NEXT:    [[LOAD8:%.*]] = load i8, ptr [[CALL]], align 4
 ; CHECK-NEXT:    br label %[[BB9]]
 ; CHECK:       [[BB9]]:
-; CHECK-NEXT:    ret i8 4
+; CHECK-NEXT:    [[PHI10:%.*]] = phi i8 [ [[LOAD8]], %[[BB7]] ], [ 4, %[[BB6]] ]
+; CHECK-NEXT:    ret i8 [[PHI10]]
 ;
 bb:
   br i1 %arg2, label %bb2, label %bb11

llvm/test/Transforms/GVN/PRE/rle.ll

@@ -673,15 +673,15 @@ define i8 @phi_trans4(ptr %p) {
 ; CHECK-NEXT:    [[X3:%.*]] = getelementptr i8, ptr [[P:%.*]], i32 192
 ; CHECK-NEXT:    store i8 -64, ptr [[X3]], align 1
 ; CHECK-NEXT:    [[X:%.*]] = getelementptr i8, ptr [[P]], i32 4
-; CHECK-NEXT:    [[Y:%.*]] = load i8, ptr [[X]], align 1
+; CHECK-NEXT:    [[Y2_PRE:%.*]] = load i8, ptr [[X]], align 1
 ; CHECK-NEXT:    br label [[LOOP:%.*]]
 ; CHECK:       loop:
-; CHECK-NEXT:    [[Y2:%.*]] = phi i8 [ [[Y]], [[ENTRY:%.*]] ], [ 0, [[LOOP]] ]
+; CHECK-NEXT:    [[Y2:%.*]] = phi i8 [ [[Y2_PRE]], [[ENTRY:%.*]] ], [ 0, [[LOOP]] ]
 ; CHECK-NEXT:    [[COND:%.*]] = call i1 @cond2()
 ; CHECK-NEXT:    store i32 0, ptr [[X3]], align 4
 ; CHECK-NEXT:    br i1 [[COND]], label [[LOOP]], label [[OUT:%.*]]
 ; CHECK:       out:
-; CHECK-NEXT:    [[R:%.*]] = add i8 [[Y]], [[Y2]]
+; CHECK-NEXT:    [[R:%.*]] = add i8 [[Y2_PRE]], [[Y2]]
 ; CHECK-NEXT:    ret i8 [[R]]
 ;
 entry:
@@ -772,18 +772,18 @@ define i32 @phi_trans6(ptr noalias nocapture readonly %x, i1 %cond) {
 ; CHECK-NEXT:    call void @use_i32(i32 [[L0]])
 ; CHECK-NEXT:    br label [[HEADER:%.*]]
 ; CHECK:       header:
-; CHECK-NEXT:    [[L1:%.*]] = phi i32 [ [[L0]], [[ENTRY:%.*]] ], [ [[L1_PRE:%.*]], [[LATCH_HEADER_CRIT_EDGE:%.*]] ]
+; CHECK-NEXT:    [[L1_PRE:%.*]] = phi i32 [ [[L0]], [[ENTRY:%.*]] ], [ [[L1_PRE1:%.*]], [[LATCH_HEADER_CRIT_EDGE:%.*]] ]
 ; CHECK-NEXT:    [[IV:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[IV_NEXT:%.*]], [[LATCH_HEADER_CRIT_EDGE]] ]
 ; CHECK-NEXT:    indirectbr ptr blockaddress(@phi_trans6, [[LATCH:%.*]]), [label %latch]
 ; CHECK:       latch:
 ; CHECK-NEXT:    [[IV_NEXT]] = add i32 [[IV]], 1
 ; CHECK-NEXT:    br i1 [[COND:%.*]], label [[EXIT:%.*]], label [[LATCH_HEADER_CRIT_EDGE]]
 ; CHECK:       latch.header_crit_edge:
 ; CHECK-NEXT:    [[GEP_1_PHI_TRANS_INSERT_PHI_TRANS_INSERT:%.*]] = getelementptr i32, ptr [[X]], i32 [[IV_NEXT]]
-; CHECK-NEXT:    [[L1_PRE]] = load i32, ptr [[GEP_1_PHI_TRANS_INSERT_PHI_TRANS_INSERT]], align 4
+; CHECK-NEXT:    [[L1_PRE1]] = load i32, ptr [[GEP_1_PHI_TRANS_INSERT_PHI_TRANS_INSERT]], align 4
 ; CHECK-NEXT:    br label [[HEADER]]
 ; CHECK:       exit:
-; CHECK-NEXT:    ret i32 [[L1]]
+; CHECK-NEXT:    ret i32 [[L1_PRE]]
 ;
 entry:
   %l0 = load i32, ptr %x
@@ -1057,18 +1057,18 @@ define void @load_load_partial_alias_loop(ptr %P) {
 ; LE-NEXT:    [[TMP0:%.*]] = trunc i32 [[V_1_32]] to i8
 ; LE-NEXT:    br label [[LOOP:%.*]]
 ; LE:       loop:
-; LE-NEXT:    [[V_I:%.*]] = phi i8 [ [[TMP0]], [[ENTRY:%.*]] ], [ [[V_I_PRE:%.*]], [[LOOP_LOOP_CRIT_EDGE:%.*]] ]
+; LE-NEXT:    [[V_I:%.*]] = phi i8 [ [[TMP0]], [[ENTRY:%.*]] ], [ [[TMP2:%.*]], [[LOOP_LOOP_CRIT_EDGE:%.*]] ]
 ; LE-NEXT:    [[I:%.*]] = phi i64 [ 1, [[ENTRY]] ], [ [[I_INC:%.*]], [[LOOP_LOOP_CRIT_EDGE]] ]
 ; LE-NEXT:    [[P_I:%.*]] = getelementptr i8, ptr [[P]], i64 [[I]]
 ; LE-NEXT:    call void @use.i8(i8 [[V_I]])
 ; LE-NEXT:    [[V_I_32:%.*]] = load i32, ptr [[P_I]], align 4
 ; LE-NEXT:    call void @use.i32(i32 [[V_I_32]])
 ; LE-NEXT:    [[I_INC]] = add i64 [[I]], 1
 ; LE-NEXT:    [[CMP:%.*]] = icmp ne i64 [[I_INC]], 64
+; LE-NEXT:    [[TMP1:%.*]] = lshr i32 [[V_I_32]], 8
+; LE-NEXT:    [[TMP2]] = trunc i32 [[TMP1]] to i8
 ; LE-NEXT:    br i1 [[CMP]], label [[LOOP_LOOP_CRIT_EDGE]], label [[EXIT:%.*]]
 ; LE:       loop.loop_crit_edge:
-; LE-NEXT:    [[P_I_PHI_TRANS_INSERT:%.*]] = getelementptr i8, ptr [[P]], i64 [[I_INC]]
-; LE-NEXT:    [[V_I_PRE]] = load i8, ptr [[P_I_PHI_TRANS_INSERT]], align 1
 ; LE-NEXT:    br label [[LOOP]]
 ; LE:       exit:
 ; LE-NEXT:    ret void
@@ -1084,18 +1084,18 @@ define void @load_load_partial_alias_loop(ptr %P) {
 ; BE-NEXT:    [[TMP1:%.*]] = trunc i32 [[TMP0]] to i8
 ; BE-NEXT:    br label [[LOOP:%.*]]
 ; BE:       loop:
-; BE-NEXT:    [[V_I:%.*]] = phi i8 [ [[TMP1]], [[ENTRY:%.*]] ], [ [[V_I_PRE:%.*]], [[LOOP_LOOP_CRIT_EDGE:%.*]] ]
+; BE-NEXT:    [[V_I:%.*]] = phi i8 [ [[TMP1]], [[ENTRY:%.*]] ], [ [[TMP3:%.*]], [[LOOP_LOOP_CRIT_EDGE:%.*]] ]
 ; BE-NEXT:    [[I:%.*]] = phi i64 [ 1, [[ENTRY]] ], [ [[I_INC:%.*]], [[LOOP_LOOP_CRIT_EDGE]] ]
 ; BE-NEXT:    [[P_I:%.*]] = getelementptr i8, ptr [[P]], i64 [[I]]
 ; BE-NEXT:    call void @use.i8(i8 [[V_I]])
 ; BE-NEXT:    [[V_I_32:%.*]] = load i32, ptr [[P_I]], align 4
 ; BE-NEXT:    call void @use.i32(i32 [[V_I_32]])
 ; BE-NEXT:    [[I_INC]] = add i64 [[I]], 1
 ; BE-NEXT:    [[CMP:%.*]] = icmp ne i64 [[I_INC]], 64
+; BE-NEXT:    [[TMP2:%.*]] = lshr i32 [[V_I_32]], 16
+; BE-NEXT:    [[TMP3]] = trunc i32 [[TMP2]] to i8
 ; BE-NEXT:    br i1 [[CMP]], label [[LOOP_LOOP_CRIT_EDGE]], label [[EXIT:%.*]]
 ; BE:       loop.loop_crit_edge:
-; BE-NEXT:    [[P_I_PHI_TRANS_INSERT:%.*]] = getelementptr i8, ptr [[P]], i64 [[I_INC]]
-; BE-NEXT:    [[V_I_PRE]] = load i8, ptr [[P_I_PHI_TRANS_INSERT]], align 1
 ; BE-NEXT:    br label [[LOOP]]
 ; BE:       exit:
 ; BE-NEXT:    ret void