[llvm] 29b2c1c - [Peeling] Extend the scope of peeling a bit

[Serguei Katkov via llvm-commits]

Author: Serguei Katkov
Date: 2020-06-22T12:17:44+07:00
New Revision: 29b2c1ca72096ca06415b5e626e6728c42ef1e74

URL: https://github.com/llvm/llvm-project/commit/29b2c1ca72096ca06415b5e626e6728c42ef1e74
DIFF: https://github.com/llvm/llvm-project/commit/29b2c1ca72096ca06415b5e626e6728c42ef1e74.diff

LOG: [Peeling] Extend the scope of peeling a bit

Currently we allow peeling of the loops if there is a exiting latch block
and all other exits are blocks ending with deopt.

Actually we want that exit would end up with deopt unconditionally but
it is not required that exit itself ends with deopt.

Reviewers: reames, ashlykov, fhahn, apilipenko, fedor.sergeev
Reviewed By: apilipenko
Subscribers: hiraditya, zzheng, dantrushin, llvm-commits
Differential Revision: https://reviews.llvm.org/D81140

Added: 
    llvm/test/Transforms/LoopUnroll/peel-loop-pgo-deopt-pd.ll

Modified: 
    llvm/lib/Transforms/Utils/LoopUnrollPeel.cpp
    llvm/lib/Transforms/Utils/LoopUtils.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Transforms/Utils/LoopUnrollPeel.cpp b/llvm/lib/Transforms/Utils/LoopUnrollPeel.cpp
index 43dfaf3e50dc..23fe5330babc 100644
--- a/llvm/lib/Transforms/Utils/LoopUnrollPeel.cpp
+++ b/llvm/lib/Transforms/Utils/LoopUnrollPeel.cpp
@@ -90,7 +90,7 @@ bool llvm::canPeel(Loop *L) {
       const BranchInst *T = dyn_cast<BranchInst>(Latch->getTerminator());
       return T && T->isConditional() && L->isLoopExiting(Latch) &&
              all_of(Exits, [](const BasicBlock *BB) {
-               return BB->getTerminatingDeoptimizeCall();
+               return BB->getPostdominatingDeoptimizeCall();
              });
     }
   }

diff  --git a/llvm/lib/Transforms/Utils/LoopUtils.cpp b/llvm/lib/Transforms/Utils/LoopUtils.cpp
index 9241377012a4..bd4fb4bedc17 100644
--- a/llvm/lib/Transforms/Utils/LoopUtils.cpp
+++ b/llvm/lib/Transforms/Utils/LoopUtils.cpp
@@ -744,7 +744,7 @@ static BranchInst *getExpectedExitLoopLatchBranch(Loop *L) {
   SmallVector<BasicBlock *, 4> ExitBlocks;
   L->getUniqueNonLatchExitBlocks(ExitBlocks);
   if (any_of(ExitBlocks, [](const BasicBlock *EB) {
-        return !EB->getTerminatingDeoptimizeCall();
+        return !EB->getPostdominatingDeoptimizeCall();
       }))
     return nullptr;
 

diff  --git a/llvm/test/Transforms/LoopUnroll/peel-loop-pgo-deopt-pd.ll b/llvm/test/Transforms/LoopUnroll/peel-loop-pgo-deopt-pd.ll
new file mode 100644
index 000000000000..06a69e17b2fd
--- /dev/null
+++ b/llvm/test/Transforms/LoopUnroll/peel-loop-pgo-deopt-pd.ll
@@ -0,0 +1,47 @@
+; RUN: opt < %s -S -loop-unroll -unroll-runtime -unroll-peel-multi-deopt-exit 2>&1 | FileCheck %s
+
+; Make sure that we can peel even if deopt block is not in immidiate exit.
+
+; CHECK-LABEL: @basic
+; CHECK: br i1 %c, label %{{.*}}, label %side_exit
+; CHECK: br i1 %{{.*}}, label %[[NEXT0:.*]], label %exit
+; CHECK: [[NEXT0]]:
+; CHECK: br i1 %c, label %{{.*}}, label %side_exit
+; CHECK: br i1 %{{.*}}, label %[[NEXT1:.*]], label %exit
+
+define i32 @basic(i32* %p, i32 %k, i1 %c) #0 !prof !1 {
+entry:
+  %cmp3 = icmp slt i32 0, %k
+  br i1 %cmp3, label %header, label %deopt
+
+header:
+  br label %body
+
+body:
+  %i = phi i32 [ 0, %header ], [ %inc, %backedge ]
+  %addr = getelementptr inbounds i32, i32* %p, i32 %i
+  store i32 %i, i32* %addr, align 4
+  %inc = add nsw i32 %i, 1
+  %cmp = icmp slt i32 %inc, %k
+  br i1 %c, label %backedge, label %side_exit, !prof !3
+
+backedge:
+  br i1 %cmp, label %body, label %exit, !prof !2
+
+exit:
+  ret i32 %i
+
+side_exit:
+  br label %deopt
+
+deopt:
+  %deopt_kind = phi i32 [0, %entry], [1, %side_exit]
+  %rval = call i32(...) @llvm.experimental.deoptimize.i32() [ "deopt"(i32 %deopt_kind) ]
+  ret i32 %rval
+}
+
+declare i32 @llvm.experimental.deoptimize.i32(...)
+
+!1 = !{!"function_entry_count", i64 1}
+!2 = !{!"branch_weights", i32 0, i32 1}
+!3 = !{!"branch_weights", i32 1, i32 0}