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: -r=%t.o,main,plx \ 40; RUN: -r=%t.o,_ZdaPv, \ 41; RUN: -r=%t.o,sleep, \ 42; RUN: -r=%t.o,_Znam, \ 43; RUN: -memprof-verify-ccg -memprof-verify-nodes -memprof-dump-ccg \ 44; RUN: -memprof-export-to-dot -memprof-dot-file-path-prefix=%t. \ 45; RUN: -stats -pass-remarks=memprof-context-disambiguation -save-temps \ 46; RUN: -o %t.out 2>&1 | FileCheck %s --check-prefix=DUMP \ 47; RUN: --check-prefix=STATS --check-prefix=STATS-BE --check-prefix=REMARKS 48 49; RUN: cat %t.ccg.postbuild.dot | FileCheck %s --check-prefix=DOT 50;; We should have cloned bar, baz, and foo, for the cold memory allocation. 51; RUN: cat %t.ccg.cloned.dot | FileCheck %s --check-prefix=DOTCLONED 52 53; RUN: llvm-dis %t.out.1.4.opt.bc -o - | FileCheck %s --check-prefix=IR 54 55 56;; Try again but with distributed ThinLTO 57; RUN: llvm-lto2 run %t.o -enable-memprof-context-disambiguation \ 58; RUN: -thinlto-distributed-indexes \ 59; RUN: -r=%t.o,main,plx \ 60; RUN: -r=%t.o,_ZdaPv, \ 61; RUN: -r=%t.o,sleep, \ 62; RUN: -r=%t.o,_Znam, \ 63; RUN: -memprof-verify-ccg -memprof-verify-nodes -memprof-dump-ccg \ 64; RUN: -memprof-export-to-dot -memprof-dot-file-path-prefix=%t2. \ 65; RUN: -stats -pass-remarks=memprof-context-disambiguation \ 66; RUN: -o %t2.out 2>&1 | FileCheck %s --check-prefix=DUMP \ 67; RUN: --check-prefix=STATS 68 69; RUN: cat %t2.ccg.postbuild.dot | FileCheck %s --check-prefix=DOT 70;; We should have cloned bar, baz, and foo, for the cold memory allocation. 71; RUN: cat %t2.ccg.cloned.dot | FileCheck %s --check-prefix=DOTCLONED 72 73;; Check distributed index 74; RUN: llvm-dis %t.o.thinlto.bc -o - | FileCheck %s --check-prefix=DISTRIB 75 76;; Run ThinLTO backend 77; RUN: opt -passes=memprof-context-disambiguation \ 78; RUN: -memprof-import-summary=%t.o.thinlto.bc \ 79; RUN: -stats -pass-remarks=memprof-context-disambiguation \ 80; RUN: %t.o -S 2>&1 | FileCheck %s --check-prefix=IR \ 81; RUN: --check-prefix=STATS-BE --check-prefix=REMARKS 82 83source_filename = "memprof-basic.ll" 84target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128" 85target triple = "x86_64-unknown-linux-gnu" 86 87define i32 @main() #0 { 88entry: 89 %call = call ptr @_Z3foov(), !callsite !0 90 %call1 = call ptr @_Z3foov(), !callsite !1 91 ret i32 0 92} 93 94declare void @_ZdaPv() 95 96declare i32 @sleep() 97 98define internal ptr @_Z3barv() #0 { 99entry: 100 %call = call ptr @_Znam(i64 0), !memprof !2, !callsite !7 101 ret ptr null 102} 103 104declare ptr @_Znam(i64) 105 106define internal ptr @_Z3bazv() #0 { 107entry: 108 %call = call ptr @_Z3barv(), !callsite !8 109 ret ptr null 110} 111 112define internal ptr @_Z3foov() #0 { 113entry: 114 %call = call ptr @_Z3bazv(), !callsite !9 115 ret ptr null 116} 117 118; uselistorder directives 119uselistorder ptr @_Z3foov, { 1, 0 } 120 121attributes #0 = { noinline optnone } 122 123!0 = !{i64 8632435727821051414} 124!1 = !{i64 -3421689549917153178} 125!2 = !{!3, !5} 126!3 = !{!4, !"notcold"} 127!4 = !{i64 9086428284934609951, i64 -5964873800580613432, i64 2732490490862098848, i64 8632435727821051414} 128!5 = !{!6, !"cold"} 129!6 = !{i64 9086428284934609951, i64 -5964873800580613432, i64 2732490490862098848, i64 -3421689549917153178} 130!7 = !{i64 9086428284934609951} 131!8 = !{i64 -5964873800580613432} 132!9 = !{i64 2732490490862098848} 133 134 135; DUMP: CCG before cloning: 136; DUMP: Callsite Context Graph: 137; DUMP: Node [[BAR:0x[a-z0-9]+]] 138; DUMP: Versions: 1 MIB: 139; DUMP: AllocType 1 StackIds: 2, 3, 0 140; DUMP: AllocType 2 StackIds: 2, 3, 1 141; DUMP: (clone 0) 142; DUMP: AllocTypes: NotColdCold 143; DUMP: ContextIds: 1 2 144; DUMP: CalleeEdges: 145; DUMP: CallerEdges: 146; DUMP: Edge from Callee [[BAR]] to Caller: [[BAZ:0x[a-z0-9]+]] AllocTypes: NotColdCold ContextIds: 1 2 147 148; DUMP: Node [[BAZ]] 149; DUMP: Callee: 9832687305761716512 (_Z3barv) Clones: 0 StackIds: 2 (clone 0) 150; DUMP: AllocTypes: NotColdCold 151; DUMP: ContextIds: 1 2 152; DUMP: CalleeEdges: 153; DUMP: Edge from Callee [[BAR]] to Caller: [[BAZ]] AllocTypes: NotColdCold ContextIds: 1 2 154; DUMP: CallerEdges: 155; DUMP: Edge from Callee [[BAZ]] to Caller: [[FOO:0x[a-z0-9]+]] AllocTypes: NotColdCold ContextIds: 1 2 156 157; DUMP: Node [[FOO]] 158; DUMP: Callee: 5878270615442837395 (_Z3bazv) Clones: 0 StackIds: 3 (clone 0) 159; DUMP: AllocTypes: NotColdCold 160; DUMP: ContextIds: 1 2 161; DUMP: CalleeEdges: 162; DUMP: Edge from Callee [[BAZ]] to Caller: [[FOO]] AllocTypes: NotColdCold ContextIds: 1 2 163; DUMP: CallerEdges: 164; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN1:0x[a-z0-9]+]] AllocTypes: NotCold ContextIds: 1 165; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN2:0x[a-z0-9]+]] AllocTypes: Cold ContextIds: 2 166 167; DUMP: Node [[MAIN1]] 168; DUMP: Callee: 6731117468105397038 (_Z3foov) Clones: 0 StackIds: 0 (clone 0) 169; DUMP: AllocTypes: NotCold 170; DUMP: ContextIds: 1 171; DUMP: CalleeEdges: 172; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN1]] AllocTypes: NotCold ContextIds: 1 173; DUMP: CallerEdges: 174 175; DUMP: Node [[MAIN2]] 176; DUMP: Callee: 6731117468105397038 (_Z3foov) Clones: 0 StackIds: 1 (clone 0) 177; DUMP: AllocTypes: Cold 178; DUMP: ContextIds: 2 179; DUMP: CalleeEdges: 180; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN2]] AllocTypes: Cold ContextIds: 2 181; DUMP: CallerEdges: 182 183; DUMP: CCG after cloning: 184; DUMP: Callsite Context Graph: 185; DUMP: Node [[BAR]] 186; DUMP: Versions: 1 MIB: 187; DUMP: AllocType 1 StackIds: 2, 3, 0 188; DUMP: AllocType 2 StackIds: 2, 3, 1 189; DUMP: (clone 0) 190; DUMP: AllocTypes: NotCold 191; DUMP: ContextIds: 1 192; DUMP: CalleeEdges: 193; DUMP: CallerEdges: 194; DUMP: Edge from Callee [[BAR]] to Caller: [[BAZ]] AllocTypes: NotCold ContextIds: 1 195; DUMP: Clones: [[BAR2:0x[a-z0-9]+]] 196 197; DUMP: Node [[BAZ]] 198; DUMP: Callee: 9832687305761716512 (_Z3barv) Clones: 0 StackIds: 2 (clone 0) 199; DUMP: AllocTypes: NotCold 200; DUMP: ContextIds: 1 201; DUMP: CalleeEdges: 202; DUMP: Edge from Callee [[BAR]] to Caller: [[BAZ]] AllocTypes: NotCold ContextIds: 1 203; DUMP: CallerEdges: 204; DUMP: Edge from Callee [[BAZ]] to Caller: [[FOO]] AllocTypes: NotCold ContextIds: 1 205; DUMP: Clones: [[BAZ2:0x[a-z0-9]+]] 206 207; DUMP: Node [[FOO]] 208; DUMP: Callee: 5878270615442837395 (_Z3bazv) Clones: 0 StackIds: 3 (clone 0) 209; DUMP: AllocTypes: NotCold 210; DUMP: ContextIds: 1 211; DUMP: CalleeEdges: 212; DUMP: Edge from Callee [[BAZ]] to Caller: [[FOO]] AllocTypes: NotCold ContextIds: 1 213; DUMP: CallerEdges: 214; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN1]] AllocTypes: NotCold ContextIds: 1 215; DUMP: Clones: [[FOO2:0x[a-z0-9]+]] 216 217; DUMP: Node [[MAIN1]] 218; DUMP: Callee: 6731117468105397038 (_Z3foov) Clones: 0 StackIds: 0 (clone 0) 219; DUMP: AllocTypes: NotCold 220; DUMP: ContextIds: 1 221; DUMP: CalleeEdges: 222; DUMP: Edge from Callee [[FOO]] to Caller: [[MAIN1]] AllocTypes: NotCold ContextIds: 1 223; DUMP: CallerEdges: 224 225; DUMP: Node [[MAIN2]] 226; DUMP: Callee: 6731117468105397038 (_Z3foov) Clones: 0 StackIds: 1 (clone 0) 227; DUMP: AllocTypes: Cold 228; DUMP: ContextIds: 2 229; DUMP: CalleeEdges: 230; DUMP: Edge from Callee [[FOO2]] to Caller: [[MAIN2]] AllocTypes: Cold ContextIds: 2 231; DUMP: CallerEdges: 232 233; DUMP: Node [[FOO2]] 234; DUMP: Callee: 5878270615442837395 (_Z3bazv) Clones: 0 StackIds: 3 (clone 0) 235; DUMP: AllocTypes: Cold 236; DUMP: ContextIds: 2 237; DUMP: CalleeEdges: 238; DUMP: Edge from Callee [[BAZ2]] to Caller: [[FOO2]] AllocTypes: Cold ContextIds: 2 239; DUMP: CallerEdges: 240; DUMP: Edge from Callee [[FOO2]] to Caller: [[MAIN2]] AllocTypes: Cold ContextIds: 2 241; DUMP: Clone of [[FOO]] 242 243; DUMP: Node [[BAZ2]] 244; DUMP: Callee: 9832687305761716512 (_Z3barv) Clones: 0 StackIds: 2 (clone 0) 245; DUMP: AllocTypes: Cold 246; DUMP: ContextIds: 2 247; DUMP: CalleeEdges: 248; DUMP: Edge from Callee [[BAR2]] to Caller: [[BAZ2]] AllocTypes: Cold ContextIds: 2 249; DUMP: CallerEdges: 250; DUMP: Edge from Callee [[BAZ2]] to Caller: [[FOO2]] AllocTypes: Cold ContextIds: 2 251; DUMP: Clone of [[BAZ]] 252 253; DUMP: Node [[BAR2]] 254; DUMP: Versions: 1 MIB: 255; DUMP: AllocType 1 StackIds: 2, 3, 0 256; DUMP: AllocType 2 StackIds: 2, 3, 1 257; DUMP: (clone 0) 258; DUMP: AllocTypes: Cold 259; DUMP: ContextIds: 2 260; DUMP: CalleeEdges: 261; DUMP: CallerEdges: 262; DUMP: Edge from Callee [[BAR2]] to Caller: [[BAZ2]] AllocTypes: Cold ContextIds: 2 263; DUMP: Clone of [[BAR]] 264 265 266; REMARKS: call in clone main assigned to call function clone _Z3foov.memprof.1 267; REMARKS: created clone _Z3barv.memprof.1 268; REMARKS: call in clone _Z3barv marked with memprof allocation attribute notcold 269; REMARKS: call in clone _Z3barv.memprof.1 marked with memprof allocation attribute cold 270; REMARKS: created clone _Z3bazv.memprof.1 271; REMARKS: call in clone _Z3bazv.memprof.1 assigned to call function clone _Z3barv.memprof.1 272; REMARKS: created clone _Z3foov.memprof.1 273; REMARKS: call in clone _Z3foov.memprof.1 assigned to call function clone _Z3bazv.memprof.1 274 275 276; IR: define {{.*}} @main 277;; The first call to foo does not allocate cold memory. It should call the 278;; original functions, which ultimately call the original allocation decorated 279;; with a "notcold" attribute. 280; IR: call {{.*}} @_Z3foov() 281;; The second call to foo allocates cold memory. It should call cloned functions 282;; which ultimately call a cloned allocation decorated with a "cold" attribute. 283; IR: call {{.*}} @_Z3foov.memprof.1() 284; IR: define internal {{.*}} @_Z3barv() 285; IR: call {{.*}} @_Znam(i64 0) #[[NOTCOLD:[0-9]+]] 286; IR: define internal {{.*}} @_Z3bazv() 287; IR: call {{.*}} @_Z3barv() 288; IR: define internal {{.*}} @_Z3foov() 289; IR: call {{.*}} @_Z3bazv() 290; IR: define internal {{.*}} @_Z3barv.memprof.1() 291; IR: call {{.*}} @_Znam(i64 0) #[[COLD:[0-9]+]] 292; IR: define internal {{.*}} @_Z3bazv.memprof.1() 293; IR: call {{.*}} @_Z3barv.memprof.1() 294; IR: define internal {{.*}} @_Z3foov.memprof.1() 295; IR: call {{.*}} @_Z3bazv.memprof.1() 296; IR: attributes #[[NOTCOLD]] = { "memprof"="notcold" } 297; IR: attributes #[[COLD]] = { "memprof"="cold" } 298 299 300; STATS: 1 memprof-context-disambiguation - Number of cold static allocations (possibly cloned) 301; STATS-BE: 1 memprof-context-disambiguation - Number of cold static allocations (possibly cloned) during ThinLTO backend 302; STATS: 1 memprof-context-disambiguation - Number of not cold static allocations (possibly cloned) 303; STATS-BE: 1 memprof-context-disambiguation - Number of not cold static allocations (possibly cloned) during ThinLTO backend 304; STATS-BE: 2 memprof-context-disambiguation - Number of allocation versions (including clones) during ThinLTO backend 305; STATS: 3 memprof-context-disambiguation - Number of function clones created during whole program analysis 306; STATS-BE: 3 memprof-context-disambiguation - Number of function clones created during ThinLTO backend 307; STATS-BE: 3 memprof-context-disambiguation - Number of functions that had clones created during ThinLTO backend 308; STATS-BE: 2 memprof-context-disambiguation - Maximum number of allocation versions created for an original allocation during ThinLTO backend 309; STATS-BE: 1 memprof-context-disambiguation - Number of original (not cloned) allocations with memprof profiles during ThinLTO backend 310 311 312; DOT: digraph "postbuild" { 313; DOT: label="postbuild"; 314; 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}"]; 315; 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}"]; 316; DOT: Node[[BAZ]] -> Node[[BAR]][tooltip="ContextIds: 1 2",fillcolor="mediumorchid1"]; 317; 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}"]; 318; DOT: Node[[FOO]] -> Node[[BAZ]][tooltip="ContextIds: 1 2",fillcolor="mediumorchid1"]; 319; 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}"]; 320; DOT: Node[[MAIN1]] -> Node[[FOO]][tooltip="ContextIds: 1",fillcolor="brown1"]; 321; 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}"]; 322; DOT: Node[[MAIN2]] -> Node[[FOO]][tooltip="ContextIds: 2",fillcolor="cyan"]; 323; DOT: } 324 325 326; DOTCLONED: digraph "cloned" { 327; DOTCLONED: label="cloned"; 328; 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}"]; 329; 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}"]; 330; DOTCLONED: Node[[BAZ]] -> Node[[BAR]][tooltip="ContextIds: 1",fillcolor="brown1"]; 331; 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}"]; 332; DOTCLONED: Node[[FOO]] -> Node[[BAZ]][tooltip="ContextIds: 1",fillcolor="brown1"]; 333; 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}"]; 334; DOTCLONED: Node[[MAIN1]] -> Node[[FOO]][tooltip="ContextIds: 1",fillcolor="brown1"]; 335; 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}"]; 336; DOTCLONED: Node[[MAIN2]] -> Node[[FOO2:0x[a-z0-9]+]][tooltip="ContextIds: 2",fillcolor="cyan"]; 337; 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}"]; 338; DOTCLONED: Node[[FOO2]] -> Node[[BAZ2:0x[a-z0-9]+]][tooltip="ContextIds: 2",fillcolor="cyan"]; 339; 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}"]; 340; DOTCLONED: Node[[BAZ2]] -> Node[[BAR2:0x[a-z0-9]+]][tooltip="ContextIds: 2",fillcolor="cyan"]; 341; 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}"]; 342; DOTCLONED: } 343 344 345; DISTRIB: ^[[BAZ:[0-9]+]] = gv: (guid: 5878270615442837395, {{.*}} callsites: ((callee: ^[[BAR:[0-9]+]], clones: (0, 1) 346; DISTRIB: ^[[FOO:[0-9]+]] = gv: (guid: 6731117468105397038, {{.*}} callsites: ((callee: ^[[BAZ]], clones: (0, 1) 347; DISTRIB: ^[[BAR]] = gv: (guid: 9832687305761716512, {{.*}} allocs: ((versions: (notcold, cold) 348; DISTRIB: ^[[MAIN:[0-9]+]] = gv: (guid: 15822663052811949562, {{.*}} callsites: ((callee: ^[[FOO]], clones: (0), {{.*}} (callee: ^[[FOO]], clones: (1) 349