From 6df8f27c95a514b19476a0e6f66107575281f158 Mon Sep 17 00:00:00 2001 From: James Molloy Date: Mon, 14 Nov 2016 11:14:41 +0000 Subject: [PATCH] [InlineCost] Remove skew when calculating call costs When calculating the cost of a call instruction we were applying a heuristic penalty as well as the cost of the instruction itself. However, when calculating the benefit from inlining we weren't discounting the equivalent penalty for the call instruction that would be removed! This caused skew in the calculation and meant we wouldn't inline in the following, trivial case: int g() { h(); } int f() { g(); } llvm-svn: 286814 --- llvm/lib/Analysis/InlineCost.cpp | 4 +++- llvm/test/Transforms/Inline/alloca-bonus.ll | 6 ++++++ llvm/test/Transforms/Inline/inline-cold-callee.ll | 6 ++++-- llvm/test/Transforms/Inline/inline-cold.ll | 4 ++++ llvm/test/Transforms/Inline/inline-hot-callee.ll | 6 ++++-- llvm/test/Transforms/Inline/inline-hot-callsite.ll | 6 ++++-- llvm/test/Transforms/Inline/inline-optsize.ll | 3 +++ llvm/test/Transforms/Inline/inline_unreachable-2.ll | 3 +++ .../Inline/optimization-remarks-passed-yaml.ll | 4 ++-- llvm/test/Transforms/Inline/ptr-diff.ll | 11 ++++++++--- 10 files changed, 41 insertions(+), 12 deletions(-) diff --git a/llvm/lib/Analysis/InlineCost.cpp b/llvm/lib/Analysis/InlineCost.cpp index d820069e6ce3..6c3525fad4e1 100644 --- a/llvm/lib/Analysis/InlineCost.cpp +++ b/llvm/lib/Analysis/InlineCost.cpp @@ -1255,7 +1255,9 @@ bool CallAnalyzer::analyzeCall(CallSite CS) { Cost -= InlineConstants::InstrCost; } } - + // The call instruction also disappears after inlining. + Cost -= InlineConstants::InstrCost + InlineConstants::CallPenalty; + // If there is only one call of the function, and it has internal linkage, // the cost of inlining it drops dramatically. bool OnlyOneCallAndLocalLinkage = diff --git a/llvm/test/Transforms/Inline/alloca-bonus.ll b/llvm/test/Transforms/Inline/alloca-bonus.ll index 649fac909fba..76a59ff6915c 100644 --- a/llvm/test/Transforms/Inline/alloca-bonus.ll +++ b/llvm/test/Transforms/Inline/alloca-bonus.ll @@ -22,6 +22,7 @@ define void @inner1(i32 *%ptr) { %E = bitcast i32* %ptr to i8* %F = select i1 false, i32* %ptr, i32* @glbl call void @llvm.lifetime.start(i64 0, i8* %E) + call void @extern() ret void } @@ -42,6 +43,7 @@ define void @inner2(i32 *%ptr) { %E = bitcast i32* %ptr to i8* %F = select i1 false, i32* %ptr, i32* @glbl call void @llvm.lifetime.start(i64 0, i8* %E) + call void @extern() ret void } @@ -56,6 +58,7 @@ define void @outer3() { define void @inner3(i32 *%ptr, i1 %x) { %A = icmp eq i32* %ptr, null %B = and i1 %x, %A + call void @extern() br i1 %A, label %bb.true, label %bb.false bb.true: ; This block musn't be counted in the inline cost. @@ -97,6 +100,7 @@ define void @outer4(i32 %A) { define void @inner4(i32 *%ptr, i32 %A) { %B = getelementptr inbounds i32, i32* %ptr, i32 %A %C = icmp eq i32* %ptr, null + call void @extern() br i1 %C, label %bb.true, label %bb.false bb.true: ; This block musn't be counted in the inline cost. @@ -139,6 +143,7 @@ define void @outer5() { define void @inner5(i1 %flag, i32 *%ptr) { %A = load i32, i32* %ptr store i32 0, i32* %ptr + call void @extern() %C = getelementptr inbounds i32, i32* %ptr, i32 0 br i1 %flag, label %if.then, label %exit @@ -153,3 +158,4 @@ exit: ret void } +declare void @extern() diff --git a/llvm/test/Transforms/Inline/inline-cold-callee.ll b/llvm/test/Transforms/Inline/inline-cold-callee.ll index 8e202e91ffab..cfd56c281730 100644 --- a/llvm/test/Transforms/Inline/inline-cold-callee.ll +++ b/llvm/test/Transforms/Inline/inline-cold-callee.ll @@ -9,7 +9,7 @@ define i32 @callee1(i32 %x) !prof !21 { %x1 = add i32 %x, 1 %x2 = add i32 %x1, 1 %x3 = add i32 %x2, 1 - + call void @extern() ret i32 %x3 } @@ -18,7 +18,7 @@ define i32 @callee2(i32 %x) !prof !22 { %x1 = add i32 %x, 1 %x2 = add i32 %x1, 1 %x3 = add i32 %x2, 1 - + call void @extern() ret i32 %x3 } @@ -32,6 +32,8 @@ define i32 @caller2(i32 %y1) !prof !22 { ret i32 %y3 } +declare void @extern() + !llvm.module.flags = !{!1} !21 = !{!"function_entry_count", i64 100} !22 = !{!"function_entry_count", i64 1} diff --git a/llvm/test/Transforms/Inline/inline-cold.ll b/llvm/test/Transforms/Inline/inline-cold.ll index b3c134538a45..93d2569d87ad 100644 --- a/llvm/test/Transforms/Inline/inline-cold.ll +++ b/llvm/test/Transforms/Inline/inline-cold.ll @@ -17,6 +17,7 @@ ; Function Attrs: nounwind readnone uwtable define i32 @simpleFunction(i32 %a) #0 { entry: + call void @extern() %a1 = load volatile i32, i32* @a %x1 = add i32 %a1, %a1 %a2 = load volatile i32, i32* @a @@ -54,6 +55,7 @@ define i32 @ColdFunction(i32 %a) #1 { ; DEFAULT-LABEL: @ColdFunction ; DEFAULT: ret entry: + call void @extern() %a1 = load volatile i32, i32* @a %x1 = add i32 %a1, %a1 %a2 = load volatile i32, i32* @a @@ -91,6 +93,7 @@ define i32 @ColdFunction2(i32 %a) #1 { ; DEFAULT-LABEL: @ColdFunction2 ; DEFAULT: ret entry: + call void @extern() %a1 = load volatile i32, i32* @a %x1 = add i32 %a1, %a1 %a2 = load volatile i32, i32* @a @@ -196,5 +199,6 @@ entry: ret i32 %add } +declare void @extern() attributes #0 = { nounwind readnone uwtable } attributes #1 = { nounwind cold readnone uwtable } diff --git a/llvm/test/Transforms/Inline/inline-hot-callee.ll b/llvm/test/Transforms/Inline/inline-hot-callee.ll index 7a04b517bf60..59d44bbb501f 100644 --- a/llvm/test/Transforms/Inline/inline-hot-callee.ll +++ b/llvm/test/Transforms/Inline/inline-hot-callee.ll @@ -9,7 +9,7 @@ define i32 @callee1(i32 %x) !prof !21 { %x1 = add i32 %x, 1 %x2 = add i32 %x1, 1 %x3 = add i32 %x2, 1 - + call void @extern() ret i32 %x3 } @@ -18,7 +18,7 @@ define i32 @callee2(i32 %x) !prof !22 { %x1 = add i32 %x, 1 %x2 = add i32 %x1, 1 %x3 = add i32 %x2, 1 - + call void @extern() ret i32 %x3 } @@ -32,6 +32,8 @@ define i32 @caller2(i32 %y1) !prof !22 { ret i32 %y3 } +declare void @extern() + !llvm.module.flags = !{!1} !21 = !{!"function_entry_count", i64 300} !22 = !{!"function_entry_count", i64 1} diff --git a/llvm/test/Transforms/Inline/inline-hot-callsite.ll b/llvm/test/Transforms/Inline/inline-hot-callsite.ll index 46993f49cc64..cc18fe6099cf 100644 --- a/llvm/test/Transforms/Inline/inline-hot-callsite.ll +++ b/llvm/test/Transforms/Inline/inline-hot-callsite.ll @@ -9,7 +9,7 @@ define i32 @callee1(i32 %x) { %x1 = add i32 %x, 1 %x2 = add i32 %x1, 1 %x3 = add i32 %x2, 1 - + call void @extern() ret i32 %x3 } @@ -18,7 +18,7 @@ define i32 @callee2(i32 %x) { %x1 = add i32 %x, 1 %x2 = add i32 %x1, 1 %x3 = add i32 %x2, 1 - + call void @extern() ret i32 %x3 } @@ -32,6 +32,8 @@ define i32 @caller2(i32 %y1) { ret i32 %y3 } +declare void @extern() + !llvm.module.flags = !{!1} !21 = !{!"branch_weights", i64 300} !22 = !{!"branch_weights", i64 1} diff --git a/llvm/test/Transforms/Inline/inline-optsize.ll b/llvm/test/Transforms/Inline/inline-optsize.ll index 7e62245fd3f0..c7cd9b3189d1 100644 --- a/llvm/test/Transforms/Inline/inline-optsize.ll +++ b/llvm/test/Transforms/Inline/inline-optsize.ll @@ -12,6 +12,7 @@ ; This function should be larger than the inline threshold for -Oz (25), but ; smaller than the inline threshold for optsize (75). define i32 @inner() { + call void @extern() %a1 = load volatile i32, i32* @a %x1 = add i32 %a1, %a1 %a2 = load volatile i32, i32* @a @@ -42,3 +43,5 @@ define i32 @outer2() minsize { %r = call i32 @inner() ret i32 %r } + +declare void @extern() \ No newline at end of file diff --git a/llvm/test/Transforms/Inline/inline_unreachable-2.ll b/llvm/test/Transforms/Inline/inline_unreachable-2.ll index 57f090effc5b..cc038c9b7208 100644 --- a/llvm/test/Transforms/Inline/inline_unreachable-2.ll +++ b/llvm/test/Transforms/Inline/inline_unreachable-2.ll @@ -8,6 +8,7 @@ define void @caller(i32 %a, i1 %b) #0 { } define void @callee(i32 %a, i1 %b) { + call void @extern() call void asm sideeffect "", ""() br i1 %b, label %bb1, label %bb2 bb1: @@ -17,3 +18,5 @@ bb2: call void asm sideeffect "", ""() ret void } + +declare void @extern() \ No newline at end of file diff --git a/llvm/test/Transforms/Inline/optimization-remarks-passed-yaml.ll b/llvm/test/Transforms/Inline/optimization-remarks-passed-yaml.ll index 1e173342d8be..f532a1648576 100644 --- a/llvm/test/Transforms/Inline/optimization-remarks-passed-yaml.ll +++ b/llvm/test/Transforms/Inline/optimization-remarks-passed-yaml.ll @@ -12,7 +12,7 @@ ; 4 return foo(); ; 5 } -; CHECK: remark: /tmp/s.c:4:10: foo can be inlined into bar with cost={{[0-9]+}} (threshold={{[0-9]+}}) (hotness: 30) +; CHECK: remark: /tmp/s.c:4:10: foo can be inlined into bar with cost={{[0-9\-]+}} (threshold={{[0-9]+}}) (hotness: 30) ; CHECK-NEXT: remark: /tmp/s.c:4:10: foo inlined into bar (hotness: 30) ; YAML: --- !Analysis @@ -28,7 +28,7 @@ ; YAML-NEXT: - Caller: bar ; YAML-NEXT: DebugLoc: { File: /tmp/s.c, Line: 3, Column: 0 } ; YAML-NEXT: - String: ' with cost=' -; YAML-NEXT: - Cost: '{{[0-9]+}}' +; YAML-NEXT: - Cost: '{{[0-9\-]+}}' ; YAML-NEXT: - String: ' (threshold=' ; YAML-NEXT: - Threshold: '{{[0-9]+}}' ; YAML-NEXT: - String: ')' diff --git a/llvm/test/Transforms/Inline/ptr-diff.ll b/llvm/test/Transforms/Inline/ptr-diff.ll index 728f3793ee2d..d4647b7891b1 100644 --- a/llvm/test/Transforms/Inline/ptr-diff.ll +++ b/llvm/test/Transforms/Inline/ptr-diff.ll @@ -4,7 +4,7 @@ target datalayout = "p:32:32-p1:64:64-p2:16:16-n16:32:64" define i32 @outer1() { ; CHECK-LABEL: @outer1( -; CHECK-NOT: call +; CHECK-NOT: call i32 ; CHECK: ret i32 %ptr = alloca i32 @@ -15,6 +15,7 @@ define i32 @outer1() { } define i32 @inner1(i32* %begin, i32* %end) { + call void @extern() %begin.i = ptrtoint i32* %begin to i32 %end.i = ptrtoint i32* %end to i32 %distance = sub i32 %end.i, %begin.i @@ -43,6 +44,7 @@ define i32 @outer2(i32* %ptr) { } define i32 @inner2(i32* %begin, i32* %end) { + call void @extern() %begin.i = ptrtoint i32* %begin to i32 %end.i = ptrtoint i32* %end to i32 %distance = sub i32 %end.i, %begin.i @@ -60,6 +62,7 @@ else: ; The inttoptrs are free since it is a smaller integer to a larger ; pointer size define i32 @inttoptr_free_cost(i32 %a, i32 %b, i32 %c) { + call void @extern() %p1 = inttoptr i32 %a to i32 addrspace(1)* %p2 = inttoptr i32 %b to i32 addrspace(1)* %p3 = inttoptr i32 %c to i32 addrspace(1)* @@ -73,7 +76,7 @@ define i32 @inttoptr_free_cost(i32 %a, i32 %b, i32 %c) { define i32 @inttoptr_free_cost_user(i32 %begin, i32 %end) { ; CHECK-LABEL: @inttoptr_free_cost_user( -; CHECK-NOT: call +; CHECK-NOT: call i32 %x = call i32 @inttoptr_free_cost(i32 %begin, i32 %end, i32 9) ret i32 %x } @@ -81,6 +84,7 @@ define i32 @inttoptr_free_cost_user(i32 %begin, i32 %end) { ; The inttoptrs have a cost since it is a larger integer to a smaller ; pointer size define i32 @inttoptr_cost_smaller_ptr(i32 %a, i32 %b, i32 %c) { + call void @extern() %p1 = inttoptr i32 %a to i32 addrspace(2)* %p2 = inttoptr i32 %b to i32 addrspace(2)* %p3 = inttoptr i32 %c to i32 addrspace(2)* @@ -94,8 +98,9 @@ define i32 @inttoptr_cost_smaller_ptr(i32 %a, i32 %b, i32 %c) { define i32 @inttoptr_cost_smaller_ptr_user(i32 %begin, i32 %end) { ; CHECK-LABEL: @inttoptr_cost_smaller_ptr_user( -; CHECK: call +; CHECK: call i32 %x = call i32 @inttoptr_cost_smaller_ptr(i32 %begin, i32 %end, i32 9) ret i32 %x } +declare void @extern() \ No newline at end of file