[JumpThread] Add test for SimplifyPartiallyRedundantLoad(), NFC

[junbuml]

Summary

Add an additional test to show the case caught by r293896 in O3. Note that the IR was the input of the jump thread when compiling the C code in the comment with O3.

This change also have a minor vairable name change, NFC.

[junbuml]

Just minor update in comments.

include/llvm/Analysis/Loads.h

 /// \param [in,out] ScanFrom The location to start scanning from. When this
 /// function returns, it points at the last instruction scanned.
 /// \param MaxInstsToScan The maximum number of instructions to scan. If this
 /// is zero, the whole block will be scanned.
 /// \param AA Optional pointer to alias analysis, to make the scan more
 /// precise.
 /// \param [out] IsLoadCSE Whether the returned value is a load from the same
 /// location in memory, as opposed to the value operand of a store.
+/// \param [out] NumScannedInst The number of instructions scanned.
 ///
 /// \returns The found value, or nullptr if no value is found.
 Value *FindAvailableLoadedValue(LoadInst *Load,
                                 BasicBlock *ScanBB,
                                 BasicBlock::iterator &ScanFrom,
                                 unsigned MaxInstsToScan = DefMaxInstsToScan,
                                 AliasAnalysis *AA = nullptr,
                                 bool *IsLoadCSE = nullptr,
-                                unsigned *NumScanedInst = nullptr);
+                                unsigned *NumScannedInst = nullptr);
 
 }
 
 #endif

lib/Analysis/Loads.cpp

cl::desc("Use this to specify the default maximum number of instructions "
            "to scan backward from a given instruction, when searching for "
            "available loaded value"));
 
 Value *llvm::FindAvailableLoadedValue(LoadInst *Load,
                                       BasicBlock *ScanBB,
                                       BasicBlock::iterator &ScanFrom,
                                       unsigned MaxInstsToScan,
                                       AliasAnalysis *AA, bool *IsLoadCSE,
-                                      unsigned *NumScanedInst) {
+                                      unsigned *NumScannedInst) {
   if (MaxInstsToScan == 0)
     MaxInstsToScan = ~0U;
 
   Value *Ptr = Load->getPointerOperand();
   Type *AccessTy = Load->getType();
 
   // We can never remove a volatile load
   if (Load->isVolatile())
while (ScanFrom != ScanBB->begin()) {
     // would affect codegen).
     Instruction *Inst = &*--ScanFrom;
     if (isa<DbgInfoIntrinsic>(Inst))
       continue;
 
     // Restore ScanFrom to expected value in case next test succeeds
     ScanFrom++;
 
-    if (NumScanedInst)
-      ++(*NumScanedInst);
+    if (NumScannedInst)
+      ++(*NumScannedInst);
 
     // Don't scan huge blocks.
     if (MaxInstsToScan-- == 0)
       return nullptr;
 
     --ScanFrom;
     // If this is a load of Ptr, the loaded value is available.
     // (This is true even if the load is volatile or atomic, although

lib/Transforms/Scalar/JumpThreading.cpp

bool JumpThreadingPass::SimplifyPartiallyRedundantLoad(LoadInst *LI) {
   // block.  Check to see if it is available in any of the predecessor blocks.
   for (BasicBlock *PredBB : predecessors(LoadBB)) {
     // If we already scanned this predecessor, skip it.
     if (!PredsScanned.insert(PredBB).second)
       continue;
 
     // Scan the predecessor to see if the value is available in the pred.
     BBIt = PredBB->end();
-    unsigned NumScanedInst = 0;
+    unsigned NumScannedInst = 0;
     Value *PredAvailable =
         FindAvailableLoadedValue(LI, PredBB, BBIt, DefMaxInstsToScan, nullptr,
-                                 &IsLoadCSE, &NumScanedInst);
-
+                                 &IsLoadCSE, &NumScannedInst);
     // If PredBB has a single predecessor, continue scanning through the single
     // precessor.
     BasicBlock *SinglePredBB = PredBB;
     while (!PredAvailable && SinglePredBB && BBIt == SinglePredBB->begin() &&
-           NumScanedInst < DefMaxInstsToScan) {
+           NumScannedInst < DefMaxInstsToScan) {
       SinglePredBB = SinglePredBB->getSinglePredecessor();
       if (SinglePredBB) {
         BBIt = SinglePredBB->end();
         PredAvailable = FindAvailableLoadedValue(
-            LI, SinglePredBB, BBIt, (DefMaxInstsToScan - NumScanedInst),
-            nullptr, &IsLoadCSE, &NumScanedInst);
+            LI, SinglePredBB, BBIt, (DefMaxInstsToScan - NumScannedInst),
+            nullptr, &IsLoadCSE, &NumScannedInst);
       }
     }
 
     if (!PredAvailable) {
       OneUnavailablePred = PredBB;
       continue;
     }
 

test/Transforms/JumpThreading/thread-loads.ll

 
 end:
   ret i32 0
 }
 
 declare void @fn2(i64)
 declare void @fn3(i64)
 
+; This test if SimplifyPartiallyRedundantLoad() can detect %1 as a partically
+; redundant load for %2 through lor.lhs.false -> while.end -> land.rhs.
+; By detecting it, we can thread from while.end to return across lor.lhs.false.
+; This IR is the input of the jump thread pass in O3 for the below C code which
+; is modified from dict_match() in spec2000/parser.
+;
+;int dict_match(char * s, char * t) {
+;  while((*s != '\0') && (*t != '\0')) {
+;    s++;
+;    t++;
+;  }
+;
+;  if ((*s == '*') || (*t == '\0'))
+;    return 0;
+;
+;  return *s - *t;
+;}
+;
+; CHECK-LABEL: @dict_match
+; CHECK-LABEL: while.end:
+; CHECK: br i1 %cmp7, label %return, label %return
+;
+define i32 @dict_match(i8* nocapture readonly %s, i8* nocapture readonly %t) local_unnamed_addr #0 {
+entry:
+  br label %while.cond
+
+while.cond:                                       ; preds = %while.body, %entry
+  %s.addr.0 = phi i8* [ %s, %entry ], [ %incdec.ptr, %while.body ]
+  %t.addr.0 = phi i8* [ %t, %entry ], [ %incdec.ptr5, %while.body ]
+  %0 = load i8, i8* %s.addr.0, align 1
+  %conv = zext i8 %0 to i32
+  %cond = icmp eq i8 %0, 0
+  br i1 %cond, label %lor.lhs.false, label %land.rhs
+
+land.rhs:                                         ; preds = %while.cond
+  %1 = load i8, i8* %t.addr.0, align 1
+  %cmp3 = icmp ne i8 %1, 0
+  br i1 %cmp3, label %while.body, label %while.end
+
+while.body:                                       ; preds = %land.rhs
+  %incdec.ptr = getelementptr inbounds i8, i8* %s.addr.0, i64 1
+  %incdec.ptr5 = getelementptr inbounds i8, i8* %t.addr.0, i64 1
+  br label %while.cond
+
+while.end:                                        ; preds = %land.rhs
+  %cmp7 = icmp eq i8 %0, 42
+  br i1 %cmp7, label %return, label %lor.lhs.false
+
+lor.lhs.false:                                    ; preds = %while.cond, %while.end
+  %2 = load i8, i8* %t.addr.0, align 1
+  %cmp10 = icmp eq i8 %2, 0
+  br i1 %cmp10, label %return, label %if.end
+
+if.end:                                           ; preds = %lor.lhs.false
+  %conv9 = zext i8 %2 to i32
+  %sub = sub nsw i32 %conv, %conv9
+  br label %return
+
+return:                                           ; preds = %while.end, %lor.lhs.false, %if.end
+  %retval.0 = phi i32 [ %sub, %if.end ], [ 0, %lor.lhs.false ], [ 0, %while.end ]
+  ret i32 %retval.0
+}
+
 
 !0 = !{!3, !3, i64 0}
 !1 = !{!"omnipotent char", !2}
 !2 = !{!"Simple C/C++ TBAA"}
 !3 = !{!"int", !1}
 !4 = !{ i32 0, i32 1 }
 !5 = !{ i32 8, i32 10 }
 !6 = !{!6}
 !7 = !{!7, !6}
 !8 = !{!8, !6}
 !9 = !{!7}
 !10 = !{!8}