[LoopUnroll] Make use of MaxTripCount for loops with "#pragma unroll" (#74703)

Fix loop unroll fail caused by branches folding.

For example:
SimplifyCFG foldloop branches then cause loop unroll failed for "#program unroll" loop.
```
#program unroll
for (int I = 0; I < ConstNum; ++I) { // folding "I < ConstNum" and "Cond2"
  if (Cond2) {
  break;
  }
  xxx loop body;
}
```

The pragma unroll metadata only takes effect if there is an exact trip
count, but not if there is an upper bound trip count. This patch make it
work with an upper bound trip count as well in shouldPragmaUnroll().

Loop unroll is important in stack nervous devices (e.g. GPU, and that is
why a lot of GPU code mark loop with "#program unroll").
It usually much simplify the address (offset) calculations in old
iterations, then we can do a lot of others optimizations, e.g, SROA, for
these simplifed address (escape alloca the whole aggregates).

Author: xiangzh1

llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp

@@ -755,7 +755,7 @@ static unsigned getFullUnrollBoostingFactor(const EstimatedUnrollCost &Cost,
 static std::optional<unsigned>
 shouldPragmaUnroll(Loop *L, const PragmaInfo &PInfo,
                    const unsigned TripMultiple, const unsigned TripCount,
-                   const UnrollCostEstimator UCE,
+                   unsigned MaxTripCount, const UnrollCostEstimator UCE,
                    const TargetTransformInfo::UnrollingPreferences &UP) {
 
   // Using unroll pragma
@@ -776,6 +776,10 @@ shouldPragmaUnroll(Loop *L, const PragmaInfo &PInfo,
   if (PInfo.PragmaFullUnroll && TripCount != 0)
     return TripCount;
 
+  if (PInfo.PragmaEnableUnroll && !TripCount && MaxTripCount &&
+      MaxTripCount <= UnrollMaxUpperBound)
+    return MaxTripCount;
+
   // if didn't return until here, should continue to other priorties
   return std::nullopt;
 }
@@ -902,7 +906,7 @@ bool llvm::computeUnrollCount(
   // 1st priority is unroll count set by "unroll-count" option.
   // 2nd priority is unroll count set by pragma.
   if (auto UnrollFactor = shouldPragmaUnroll(L, PInfo, TripMultiple, TripCount,
-                                             UCE, UP)) {
+                                             MaxTripCount, UCE, UP)) {
     UP.Count = *UnrollFactor;
 
     if (UserUnrollCount || (PragmaCount > 0)) {