1;; Test callsite context graph generation for simple call graph with 2;; two memprof contexts and no inlining. 3;; 4;; Original code looks like: 5;; 6;; char *bar() { 7;; return new char[10]; 8;; } 9;; 10;; char *baz() { 11;; return bar(); 12;; } 13;; 14;; char *foo() { 15;; return baz(); 16;; } 17;; 18;; int main(int argc, char **argv) { 19;; char *x = foo(); 20;; char *y = foo(); 21;; memset(x, 0, 10); 22;; memset(y, 0, 10); 23;; delete[] x; 24;; sleep(10); 25;; delete[] y; 26;; return 0; 27;; } 28;; 29;; Code compiled with -mllvm -memprof-ave-lifetime-cold-threshold=5 so that the 30;; memory freed after sleep(10) results in cold lifetimes. 31;; 32;; The IR was then reduced using llvm-reduce with the expected FileCheck input. 33 34;; -stats requires asserts 35; REQUIRES: asserts 36 37; RUN: opt -thinlto-bc %s >%t.o 38; RUN: llvm-lto2 run %t.o -enable-memprof-context-disambiguation \ 39; RUN: -supports-hot-cold-new \ 40; RUN: -r=%t.o,main,plx \ 41; RUN: -r=%t.o,_ZdaPv, \ 42; RUN: -r=%t.o,sleep, \ 43; RUN: -r=%t.o,_Znam, \ 44; RUN: -memprof-verify-ccg -memprof-verify-nodes -memprof-dump-ccg \ 45; RUN: -memprof-export-to-dot -memprof-dot-file-path-prefix=%t. \ 46; RUN: -stats -pass-remarks=memprof-context-disambiguation -save-temps \ 47; RUN: -o %t.out 2>&1 | FileCheck %s --check-prefix=DUMP \ 48; RUN: --check-prefix=STATS --check-prefix=STATS-BE --check-prefix=REMARKS 49 50; RUN: cat %t.ccg.postbuild.dot | FileCheck %s --check-prefix=DOT 51;; We should have cloned bar, baz, and foo, for the cold memory allocation. 52; RUN: cat %t.ccg.cloned.dot | FileCheck %s --check-prefix=DOTCLONED 53 54; RUN: llvm-dis %t.out.1.4.opt.bc -o - | FileCheck %s --check-prefix=IR 55 56 57;; Try again but with distributed ThinLTO 58; RUN: llvm-lto2 run %t.o -enable-memprof-context-disambiguation \ 59; RUN: -supports-hot-cold-new \ 60; RUN: -thinlto-distributed-indexes \ 61; RUN: -r=%t.o,main,plx \ 62; RUN: -r=%t.o,_ZdaPv, \ 63; RUN: -r=%t.o,sleep, \ 64; RUN: -r=%t.o,_Znam, \ 65; RUN: -memprof-verify-ccg -memprof-verify-nodes -memprof-dump-ccg \ 66; RUN: -memprof-export-to-dot -memprof-dot-file-path-prefix=%t2. \ 67; RUN: -stats -pass-remarks=memprof-context-disambiguation \ 68; RUN: -o %t2.out 2>&1 | FileCheck %s --check-prefix=DUMP \ 69; RUN: --check-prefix=STATS 70 71; RUN: cat %t2.ccg.postbuild.dot | FileCheck %s --check-prefix=DOT 72;; We should have cloned bar, baz, and foo, for the cold memory allocation. 73; RUN: cat %t2.ccg.cloned.dot | FileCheck %s --check-prefix=DOTCLONED 74 75;; Check distributed index 76; RUN: llvm-dis %t.o.thinlto.bc -o - | FileCheck %s --check-prefix=DISTRIB 77 78;; Run ThinLTO backend 79; RUN: opt -passes=memprof-context-disambiguation \ 80; RUN: -memprof-import-summary=%t.o.thinlto.bc \ 81; RUN: -stats -pass-remarks=memprof-context-disambiguation \ 82; RUN: %t.o -S 2>&1 | FileCheck %s --check-prefix=IR \ 83; RUN: --check-prefix=STATS-BE --check-prefix=REMARKS 84 85source_filename = "memprof-basic.ll" 86target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" 87target triple = "x86_64-unknown-linux-gnu" 88 89define i32 @main() #0 { 90entry: 91 %call = call ptr @_Z3foov(), !callsite !0 92 %call1 = call ptr @_Z3foov(), !callsite !1 93 ret i32 0 94} 95 96declare void @_ZdaPv() 97 98declare i32 @sleep() 99 100define internal ptr @_Z3barv() #0 { 101entry: 102 %call = call ptr @_Znam(i64 0), !memprof !2, !callsite !7 103 ret ptr null 104} 105 106declare ptr @_Znam(i64) 107 108define internal ptr @_Z3bazv() #0 { 109entry: 110 %call = call ptr @_Z3barv(), !callsite !8 111 ret ptr null 112} 113 114define internal ptr @_Z3foov() #0 { 115entry: 116 %call = call ptr @_Z3bazv(), !callsite !9 117 ret ptr null 118} 119 120; uselistorder directives 121uselistorder ptr @_Z3foov, { 1, 0 } 122 123attributes #0 = { noinline optnone } 124 125!0 = !{i64 8632435727821051414} 126!1 = !{i64 -3421689549917153178} 127!2 = !{!3, !5} 128!3 = !{!4, !"notcold"} 129!4 = !{i64 9086428284934609951, i64 -5964873800580613432, i64 2732490490862098848, i64 8632435727821051414} 130!5 = !{!6, !"cold"} 131!6 = !{i64 9086428284934609951, i64 -5964873800580613432, i64 2732490490862098848, i64 -3421689549917153178} 132!7 = !{i64 9086428284934609951} 133!8 = !{i64 -5964873800580613432} 134!9 = !{i64 2732490490862098848} 135 136 137; DUMP: CCG before cloning: 138; DUMP: Callsite Context Graph: 139; DUMP: Node [[BAR:0x[a-z0-9]+]] 140; DUMP: Versions: 1 MIB: 141; DUMP: AllocType 1 StackIds: 2, 3, 0 142; DUMP: AllocType 2 StackIds: 2, 3, 1 143; DUMP: (clone 0) 144; DUMP: AllocTypes: NotColdCold 145; DUMP: ContextIds: 1 2 146; DUMP: CalleeEdges: 147; DUMP: CallerEdges: 148; DUMP: Edge from Callee [[BAR]] to Caller: [[BAZ:0x[a-z0-9]+]] AllocTypes: NotColdCold ContextIds: 1 2 149 150; DUMP: Node [[BAZ]] 151; DUMP: Callee: 11481133863268513686 (_Z3barv) Clones: 0 StackIds: 2 (clone 0) 152; DUMP: AllocTypes: NotColdCold 153; DUMP: ContextIds: 1 2 154; DUMP: CalleeEdges: 155; DUMP: Edge from Callee [[BAR]] to Caller: [[BAZ]] AllocTypes: NotColdCold ContextIds: 1 2 156; DUMP: CallerEdges: 157; DUMP: Edge from Callee [[BAZ]] to Caller: [[FOO:0x[a-z0-9]+]] AllocTypes: NotColdCold ContextIds: 1 2 158 159; DUMP: Node [[FOO]] 160; DUMP: Callee: 1807954217441101578 (_Z3bazv) Clones: 0 StackIds: 3 (clone 0) 161; DUMP: AllocTypes: NotColdCold 162; DUMP: ContextIds: 1 2 163; DUMP: CalleeEdges: 164; DUMP: Edge from Callee [[BAZ]] to Caller: [[FOO]] AllocTypes: NotColdCold ContextIds: 1 2 165; DUMP: CallerEdges: 166; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN1:0x[a-z0-9]+]] AllocTypes: NotCold ContextIds: 1 167; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN2:0x[a-z0-9]+]] AllocTypes: Cold ContextIds: 2 168 169; DUMP: Node [[MAIN1]] 170; DUMP: Callee: 8107868197919466657 (_Z3foov) Clones: 0 StackIds: 0 (clone 0) 171; DUMP: AllocTypes: NotCold 172; DUMP: ContextIds: 1 173; DUMP: CalleeEdges: 174; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN1]] AllocTypes: NotCold ContextIds: 1 175; DUMP: CallerEdges: 176 177; DUMP: Node [[MAIN2]] 178; DUMP: Callee: 8107868197919466657 (_Z3foov) Clones: 0 StackIds: 1 (clone 0) 179; DUMP: AllocTypes: Cold 180; DUMP: ContextIds: 2 181; DUMP: CalleeEdges: 182; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN2]] AllocTypes: Cold ContextIds: 2 183; DUMP: CallerEdges: 184 185; DUMP: CCG after cloning: 186; DUMP: Callsite Context Graph: 187; DUMP: Node [[BAR]] 188; DUMP: Versions: 1 MIB: 189; DUMP: AllocType 1 StackIds: 2, 3, 0 190; DUMP: AllocType 2 StackIds: 2, 3, 1 191; DUMP: (clone 0) 192; DUMP: AllocTypes: NotCold 193; DUMP: ContextIds: 1 194; DUMP: CalleeEdges: 195; DUMP: CallerEdges: 196; DUMP: Edge from Callee [[BAR]] to Caller: [[BAZ]] AllocTypes: NotCold ContextIds: 1 197; DUMP: Clones: [[BAR2:0x[a-z0-9]+]] 198 199; DUMP: Node [[BAZ]] 200; DUMP: Callee: 11481133863268513686 (_Z3barv) Clones: 0 StackIds: 2 (clone 0) 201; DUMP: AllocTypes: NotCold 202; DUMP: ContextIds: 1 203; DUMP: CalleeEdges: 204; DUMP: Edge from Callee [[BAR]] to Caller: [[BAZ]] AllocTypes: NotCold ContextIds: 1 205; DUMP: CallerEdges: 206; DUMP: Edge from Callee [[BAZ]] to Caller: [[FOO]] AllocTypes: NotCold ContextIds: 1 207; DUMP: Clones: [[BAZ2:0x[a-z0-9]+]] 208 209; DUMP: Node [[FOO]] 210; DUMP: Callee: 1807954217441101578 (_Z3bazv) Clones: 0 StackIds: 3 (clone 0) 211; DUMP: AllocTypes: NotCold 212; DUMP: ContextIds: 1 213; DUMP: CalleeEdges: 214; DUMP: Edge from Callee [[BAZ]] to Caller: [[FOO]] AllocTypes: NotCold ContextIds: 1 215; DUMP: CallerEdges: 216; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN1]] AllocTypes: NotCold ContextIds: 1 217; DUMP: Clones: [[FOO2:0x[a-z0-9]+]] 218 219; DUMP: Node [[MAIN1]] 220; DUMP: Callee: 8107868197919466657 (_Z3foov) Clones: 0 StackIds: 0 (clone 0) 221; DUMP: AllocTypes: NotCold 222; DUMP: ContextIds: 1 223; DUMP: CalleeEdges: 224; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN1]] AllocTypes: NotCold ContextIds: 1 225; DUMP: CallerEdges: 226 227; DUMP: Node [[MAIN2]] 228; DUMP: Callee: 8107868197919466657 (_Z3foov) Clones: 0 StackIds: 1 (clone 0) 229; DUMP: AllocTypes: Cold 230; DUMP: ContextIds: 2 231; DUMP: CalleeEdges: 232; DUMP: Edge from Callee [[FOO2]] to Caller: [[MAIN2]] AllocTypes: Cold ContextIds: 2 233; DUMP: CallerEdges: 234 235; DUMP: Node [[FOO2]] 236; DUMP: Callee: 1807954217441101578 (_Z3bazv) Clones: 0 StackIds: 3 (clone 0) 237; DUMP: AllocTypes: Cold 238; DUMP: ContextIds: 2 239; DUMP: CalleeEdges: 240; DUMP: Edge from Callee [[BAZ2]] to Caller: [[FOO2]] AllocTypes: Cold ContextIds: 2 241; DUMP: CallerEdges: 242; DUMP: Edge from Callee [[FOO2]] to Caller: [[MAIN2]] AllocTypes: Cold ContextIds: 2 243; DUMP: Clone of [[FOO]] 244 245; DUMP: Node [[BAZ2]] 246; DUMP: Callee: 11481133863268513686 (_Z3barv) Clones: 0 StackIds: 2 (clone 0) 247; DUMP: AllocTypes: Cold 248; DUMP: ContextIds: 2 249; DUMP: CalleeEdges: 250; DUMP: Edge from Callee [[BAR2]] to Caller: [[BAZ2]] AllocTypes: Cold ContextIds: 2 251; DUMP: CallerEdges: 252; DUMP: Edge from Callee [[BAZ2]] to Caller: [[FOO2]] AllocTypes: Cold ContextIds: 2 253; DUMP: Clone of [[BAZ]] 254 255; DUMP: Node [[BAR2]] 256; DUMP: Versions: 1 MIB: 257; DUMP: AllocType 1 StackIds: 2, 3, 0 258; DUMP: AllocType 2 StackIds: 2, 3, 1 259; DUMP: (clone 0) 260; DUMP: AllocTypes: Cold 261; DUMP: ContextIds: 2 262; DUMP: CalleeEdges: 263; DUMP: CallerEdges: 264; DUMP: Edge from Callee [[BAR2]] to Caller: [[BAZ2]] AllocTypes: Cold ContextIds: 2 265; DUMP: Clone of [[BAR]] 266 267 268; REMARKS: call in clone main assigned to call function clone _Z3foov.memprof.1 269; REMARKS: created clone _Z3barv.memprof.1 270; REMARKS: call in clone _Z3barv marked with memprof allocation attribute notcold 271; REMARKS: call in clone _Z3barv.memprof.1 marked with memprof allocation attribute cold 272; REMARKS: created clone _Z3bazv.memprof.1 273; REMARKS: call in clone _Z3bazv.memprof.1 assigned to call function clone _Z3barv.memprof.1 274; REMARKS: created clone _Z3foov.memprof.1 275; REMARKS: call in clone _Z3foov.memprof.1 assigned to call function clone _Z3bazv.memprof.1 276 277 278; IR: define {{.*}} @main 279;; The first call to foo does not allocate cold memory. It should call the 280;; original functions, which ultimately call the original allocation decorated 281;; with a "notcold" attribute. 282; IR: call {{.*}} @_Z3foov() 283;; The second call to foo allocates cold memory. It should call cloned functions 284;; which ultimately call a cloned allocation decorated with a "cold" attribute. 285; IR: call {{.*}} @_Z3foov.memprof.1() 286; IR: define internal {{.*}} @_Z3barv() 287; IR: call {{.*}} @_Znam(i64 0) #[[NOTCOLD:[0-9]+]] 288; IR: define internal {{.*}} @_Z3bazv() 289; IR: call {{.*}} @_Z3barv() 290; IR: define internal {{.*}} @_Z3foov() 291; IR: call {{.*}} @_Z3bazv() 292; IR: define internal {{.*}} @_Z3barv.memprof.1() 293; IR: call {{.*}} @_Znam(i64 0) #[[COLD:[0-9]+]] 294; IR: define internal {{.*}} @_Z3bazv.memprof.1() 295; IR: call {{.*}} @_Z3barv.memprof.1() 296; IR: define internal {{.*}} @_Z3foov.memprof.1() 297; IR: call {{.*}} @_Z3bazv.memprof.1() 298; IR: attributes #[[NOTCOLD]] = { "memprof"="notcold" } 299; IR: attributes #[[COLD]] = { "memprof"="cold" } 300 301 302; STATS: 1 memprof-context-disambiguation - Number of cold static allocations (possibly cloned) 303; STATS-BE: 1 memprof-context-disambiguation - Number of cold static allocations (possibly cloned) during ThinLTO backend 304; STATS: 1 memprof-context-disambiguation - Number of not cold static allocations (possibly cloned) 305; STATS-BE: 1 memprof-context-disambiguation - Number of not cold static allocations (possibly cloned) during ThinLTO backend 306; STATS-BE: 2 memprof-context-disambiguation - Number of allocation versions (including clones) during ThinLTO backend 307; STATS: 3 memprof-context-disambiguation - Number of function clones created during whole program analysis 308; STATS-BE: 3 memprof-context-disambiguation - Number of function clones created during ThinLTO backend 309; STATS-BE: 3 memprof-context-disambiguation - Number of functions that had clones created during ThinLTO backend 310; STATS-BE: 2 memprof-context-disambiguation - Maximum number of allocation versions created for an original allocation during ThinLTO backend 311; STATS-BE: 1 memprof-context-disambiguation - Number of original (not cloned) allocations with memprof profiles during ThinLTO backend 312 313 314; DOT: digraph "postbuild" { 315; DOT: label="postbuild"; 316; DOT: Node[[BAR:0x[a-z0-9]+]] [shape=record,tooltip="N[[BAR]] ContextIds: 1 2",fillcolor="mediumorchid1",style="filled",style="filled",label="{OrigId: Alloc0\n_Z3barv -\> alloc}"]; 317; DOT: Node[[BAZ:0x[a-z0-9]+]] [shape=record,tooltip="N[[BAZ]] ContextIds: 1 2",fillcolor="mediumorchid1",style="filled",style="filled",label="{OrigId: 12481870273128938184\n_Z3bazv -\> _Z3barv}"]; 318; DOT: Node[[BAZ]] -> Node[[BAR]][tooltip="ContextIds: 1 2",fillcolor="mediumorchid1"]; 319; DOT: Node[[FOO:0x[a-z0-9]+]] [shape=record,tooltip="N[[FOO]] ContextIds: 1 2",fillcolor="mediumorchid1",style="filled",style="filled",label="{OrigId: 2732490490862098848\n_Z3foov -\> _Z3bazv}"]; 320; DOT: Node[[FOO]] -> Node[[BAZ]][tooltip="ContextIds: 1 2",fillcolor="mediumorchid1"]; 321; DOT: Node[[MAIN1:0x[a-z0-9]+]] [shape=record,tooltip="N[[MAIN1]] ContextIds: 1",fillcolor="brown1",style="filled",style="filled",label="{OrigId: 8632435727821051414\nmain -\> _Z3foov}"]; 322; DOT: Node[[MAIN1]] -> Node[[FOO]][tooltip="ContextIds: 1",fillcolor="brown1"]; 323; DOT: Node[[MAIN2:0x[a-z0-9]+]] [shape=record,tooltip="N[[MAIN2]] ContextIds: 2",fillcolor="cyan",style="filled",style="filled",label="{OrigId: 15025054523792398438\nmain -\> _Z3foov}"]; 324; DOT: Node[[MAIN2]] -> Node[[FOO]][tooltip="ContextIds: 2",fillcolor="cyan"]; 325; DOT: } 326 327 328; DOTCLONED: digraph "cloned" { 329; DOTCLONED: label="cloned"; 330; DOTCLONED: Node[[BAR:0x[a-z0-9]+]] [shape=record,tooltip="N[[BAR]] ContextIds: 1",fillcolor="brown1",style="filled",style="filled",label="{OrigId: Alloc0\n_Z3barv -\> alloc}"]; 331; DOTCLONED: Node[[BAZ:0x[a-z0-9]+]] [shape=record,tooltip="N[[BAZ]] ContextIds: 1",fillcolor="brown1",style="filled",style="filled",label="{OrigId: 12481870273128938184\n_Z3bazv -\> _Z3barv}"]; 332; DOTCLONED: Node[[BAZ]] -> Node[[BAR]][tooltip="ContextIds: 1",fillcolor="brown1"]; 333; DOTCLONED: Node[[FOO:0x[a-z0-9]+]] [shape=record,tooltip="N[[FOO]] ContextIds: 1",fillcolor="brown1",style="filled",style="filled",label="{OrigId: 2732490490862098848\n_Z3foov -\> _Z3bazv}"]; 334; DOTCLONED: Node[[FOO]] -> Node[[BAZ]][tooltip="ContextIds: 1",fillcolor="brown1"]; 335; DOTCLONED: Node[[MAIN1:0x[a-z0-9]+]] [shape=record,tooltip="N[[MAIN1]] ContextIds: 1",fillcolor="brown1",style="filled",style="filled",label="{OrigId: 8632435727821051414\nmain -\> _Z3foov}"]; 336; DOTCLONED: Node[[MAIN1]] -> Node[[FOO]][tooltip="ContextIds: 1",fillcolor="brown1"]; 337; DOTCLONED: Node[[MAIN2:0x[a-z0-9]+]] [shape=record,tooltip="N[[MAIN2]] ContextIds: 2",fillcolor="cyan",style="filled",style="filled",label="{OrigId: 15025054523792398438\nmain -\> _Z3foov}"]; 338; DOTCLONED: Node[[MAIN2]] -> Node[[FOO2:0x[a-z0-9]+]][tooltip="ContextIds: 2",fillcolor="cyan"]; 339; DOTCLONED: Node[[FOO2]] [shape=record,tooltip="N[[FOO2]] ContextIds: 2",fillcolor="cyan",style="filled",color="blue",style="filled,bold,dashed",label="{OrigId: 0\n_Z3foov -\> _Z3bazv}"]; 340; DOTCLONED: Node[[FOO2]] -> Node[[BAZ2:0x[a-z0-9]+]][tooltip="ContextIds: 2",fillcolor="cyan"]; 341; DOTCLONED: Node[[BAZ2]] [shape=record,tooltip="N[[BAZ2]] ContextIds: 2",fillcolor="cyan",style="filled",color="blue",style="filled,bold,dashed",label="{OrigId: 0\n_Z3bazv -\> _Z3barv}"]; 342; DOTCLONED: Node[[BAZ2]] -> Node[[BAR2:0x[a-z0-9]+]][tooltip="ContextIds: 2",fillcolor="cyan"]; 343; DOTCLONED: Node[[BAR2]] [shape=record,tooltip="N[[BAR2]] ContextIds: 2",fillcolor="cyan",style="filled",color="blue",style="filled,bold,dashed",label="{OrigId: Alloc0\n_Z3barv -\> alloc}"]; 344; DOTCLONED: } 345 346 347; DISTRIB: ^[[BAZ:[0-9]+]] = gv: (guid: 1807954217441101578, {{.*}} callsites: ((callee: ^[[BAR:[0-9]+]], clones: (0, 1) 348; DISTRIB: ^[[FOO:[0-9]+]] = gv: (guid: 8107868197919466657, {{.*}} callsites: ((callee: ^[[BAZ]], clones: (0, 1) 349; DISTRIB: ^[[BAR]] = gv: (guid: 11481133863268513686, {{.*}} allocs: ((versions: (notcold, cold) 350; DISTRIB: ^[[MAIN:[0-9]+]] = gv: (guid: 15822663052811949562, {{.*}} callsites: ((callee: ^[[FOO]], clones: (0), {{.*}} (callee: ^[[FOO]], clones: (1) 351