[DebugInfo] Handle call site values for instructions before call bundle

[dstenb]

Summary

If a call is bundled then the code that looks for instructions that
produce parameter values would break when reaching the call's bundle
header, due to the ifCall(/*AnyInBundle*/) invocation returning true.

It is not enough to simply ignore bundle headers in the isCall()
invocation, as the bundle header may have defines of parameter registers
due to the call, meaning that such registers would incorrectly be
removed from the worklist. Therefore, do not look at bundle headers at
all.

[aprantl]

llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

621

is there a higher-level iterator that does this skipping automatically?

[vsk]

llvm/test/DebugInfo/MIR/Hexagon/dbgcall-site-instr-before-bundled-call.mir

35

Are the dbg.value/DBG_VALUE's necessary?

[dstenb]

llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

621

AFAICT there is no such iterator.

llvm/test/DebugInfo/MIR/Hexagon/dbgcall-site-instr-before-bundled-call.mir

35

No, I'll clean them up!

[dstenb]

Remove unnecessary debug information from test case.

[djtodoro]

Looks good! Thanks!

llvm/test/DebugInfo/MIR/Hexagon/dbgcall-site-instr-before-bundled-call.mir

175

Cool, thanks! :)

llvm/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

auto finishCallSiteParam = [&](DbgValueLoc DbgLocVal, unsigned Reg) {
 
     DbgCallSiteParam CSParm(FwdReg, DbgLocVal);
     Params.push_back(CSParm);
     ++NumCSParams;
   };
 
   // Search for a loading value in forwarding registers.
   for (; I != MBB->rend(); ++I) {
+    // Skip bundle headers.
+    if (I->isBundle())
+      continue;
+
     // If the next instruction is a call we can not interpret parameter's
     // forwarding registers or we finished the interpretation of all parameters.
     if (I->isCall())
       return;
 
     if (ForwardedRegWorklist.empty())
       return;
 

llvm/test/DebugInfo/MIR/Hexagon/dbgcall-site-instr-before-bundled-call.mir

+# RUN: llc -mtriple hexagon -debug-entry-values -start-after=machineverifier -filetype=obj %s -o - | llvm-dwarfdump - | FileCheck %s
+
+# Based on the following C reproducer:
+#
+# int ga, gb, gc;
+#
+# extern void callee(int, int, int);
+#
+# void caller() {
+#   int a = ga;
+#   int b = gb;
+#   int c = gc;
+#
+#   // Clobber all integer registers.
+#   __asm("" : : :
+#         "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7", "r8", "r9", "r10",
+#         "r11", "r12", "r13", "r14", "r15", "r16", "r17", "r18", "r19", "r20",
+#         "r21", "r22", "r23", "r24", "r25", "r26", "r27", "r28");
+#
+#   callee(a, b, c);
+# }
+
+--- |
+  target datalayout = "e-m:e-p:32:32:32-a:0-n16:32-i64:64:64-i32:32:32-i16:16:16-i1:8:8-f32:32:32-f64:64:64-v32:32:32-v64:64:64-v512:512:512-v1024:1024:1024-v2048:2048:2048"
+  target triple = "hexagon"
+
+  @ga = common global i32 0, align 4, !dbg !0
+  @gb = common global i32 0, align 4, !dbg !11
+  @gc = common global i32 0, align 4, !dbg !13
+
+  ; Function Attrs: nounwind
+  define void @caller() #0 !dbg !19 {
+  entry:
+    %0 = load i32, i32* @ga, align 4, !dbg !26
+    call void @llvm.dbg.value(metadata i32 %0, metadata !23, metadata !DIExpression()), !dbg !27
+    %1 = load i32, i32* @gb, align 4, !dbg !28
+    call void @llvm.dbg.value(metadata i32 %1, metadata !24, metadata !DIExpression()), !dbg !27
+    %2 = load i32, i32* @gc, align 4, !dbg !29
+    call void @llvm.dbg.value(metadata i32 %2, metadata !25, metadata !DIExpression()), !dbg !27
+    call void asm sideeffect "", "~{r0},~{r1},~{r2},~{r3},~{r4},~{r5},~{r6},~{r7},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15},~{r16},~{r17},~{r18},~{r19},~{r20},~{r21},~{r22},~{r23},~{r24},~{r25},~{r26},~{r27},~{r28}"() #3, !dbg !30, !srcloc !31
+    call void @callee(i32 %0, i32 %1, i32 %2) #3, !dbg !32
+    ret void, !dbg !33
+  }
+
+  declare !dbg !6 void @callee(i32, i32, i32)
+
+  ; Function Attrs: nounwind readnone speculatable willreturn
+  declare void @llvm.dbg.value(metadata, metadata, metadata) #1
+
+  attributes #0 = { nounwind }
+  attributes #1 = { nounwind readnone speculatable willreturn }
+
+  !llvm.dbg.cu = !{!2}
+  !llvm.module.flags = !{!15, !16, !17}
+  !llvm.ident = !{!18}
+
+  !0 = !DIGlobalVariableExpression(var: !1, expr: !DIExpression())
+  !1 = distinct !DIGlobalVariable(name: "ga", scope: !2, file: !3, line: 1, type: !9, isLocal: false, isDefinition: true)
+  !2 = distinct !DICompileUnit(language: DW_LANG_C99, file: !3, producer: "clang version 10.0.0", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug, enums: !4, retainedTypes: !5, globals: !10, nameTableKind: None)
+  !3 = !DIFile(filename: "h.c", directory: "/")
+  !4 = !{}
+  !5 = !{!6}
+  !6 = !DISubprogram(name: "callee", scope: !3, file: !3, line: 3, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
+  !7 = !DISubroutineType(types: !8)
+  !8 = !{null, !9, !9, !9}
+  !9 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
+  !10 = !{!0, !11, !13}
+  !11 = !DIGlobalVariableExpression(var: !12, expr: !DIExpression())
+  !12 = distinct !DIGlobalVariable(name: "gb", scope: !2, file: !3, line: 1, type: !9, isLocal: false, isDefinition: true)
+  !13 = !DIGlobalVariableExpression(var: !14, expr: !DIExpression())
+  !14 = distinct !DIGlobalVariable(name: "gc", scope: !2, file: !3, line: 1, type: !9, isLocal: false, isDefinition: true)
+  !15 = !{i32 7, !"Dwarf Version", i32 4}
+  !16 = !{i32 2, !"Debug Info Version", i32 3}
+  !17 = !{i32 1, !"wchar_size", i32 4}
+  !18 = !{!"clang version 10.0.0"}
+  !19 = distinct !DISubprogram(name: "caller", scope: !3, file: !3, line: 5, type: !20, scopeLine: 5, flags: DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2, retainedNodes: !22)
+  !20 = !DISubroutineType(types: !21)
+  !21 = !{null}
+  !22 = !{!23, !24, !25}
+  !23 = !DILocalVariable(name: "a", scope: !19, file: !3, line: 6, type: !9)
+  !24 = !DILocalVariable(name: "b", scope: !19, file: !3, line: 7, type: !9)
+  !25 = !DILocalVariable(name: "c", scope: !19, file: !3, line: 8, type: !9)
+  !26 = !DILocation(line: 6, scope: !19)
+  !27 = !DILocation(line: 0, scope: !19)
+  !28 = !DILocation(line: 7, scope: !19)
+  !29 = !DILocation(line: 8, scope: !19)
+  !30 = !DILocation(line: 11, scope: !19)
+  !31 = !{i32 158}
+  !32 = !DILocation(line: 16, scope: !19)
+  !33 = !DILocation(line: 17, scope: !19)
+
+...
+---
+name:            caller
+tracksRegLiveness: true
+frameInfo:
+  stackSize:       64
+  maxAlignment:    4
+  adjustsStack:    true
+  hasCalls:        true
+  maxCallFrameSize: 0
+fixedStack:
+  - { id: 0, type: spill-slot, offset: -48, size: 8, alignment: 8, callee-saved-register: '$d13' }
+  - { id: 1, type: spill-slot, offset: -40, size: 8, alignment: 8, callee-saved-register: '$d12' }
+  - { id: 2, type: spill-slot, offset: -32, size: 8, alignment: 8, callee-saved-register: '$d11' }
+  - { id: 3, type: spill-slot, offset: -24, size: 8, alignment: 8, callee-saved-register: '$d10' }
+  - { id: 4, type: spill-slot, offset: -16, size: 8, alignment: 8, callee-saved-register: '$d9' }
+  - { id: 5, type: spill-slot, offset: -8, size: 8, alignment: 8, callee-saved-register: '$d8' }
+stack:
+  - { id: 0, type: spill-slot, offset: -52, size: 4, alignment: 4 }
+  - { id: 1, type: spill-slot, offset: -56, size: 4, alignment: 4 }
+  - { id: 2, type: spill-slot, offset: -60, size: 4, alignment: 4 }
+callSites:
+  - { bb: 0, offset: 46, fwdArgRegs:
+      - { arg: 0, reg: '$r0' }
+      - { arg: 1, reg: '$r1' }
+      - { arg: 2, reg: '$r2' } }
+body:             |
+  bb.0.entry:
+    liveins: $d8, $d9, $d10, $d11, $d12, $d13, $d8, $d9, $d10, $d11, $d12, $d13
+
+    BUNDLE implicit-def $r29, implicit-def $r30, implicit $r29, implicit killed $framekey, implicit killed $framelimit, implicit killed $r30, implicit killed $r31, implicit killed $d8, debug-location !26 {
+      $r29 = S2_allocframe $r29, 64, implicit-def $r30, implicit killed $framekey, implicit killed $framelimit, implicit killed $r30, implicit killed $r31, debug-location !26 :: (store 4 into stack)
+      S2_storerd_io internal $r29, -16, killed $d8, debug-location !26 :: (store 8 into %fixed-stack.5)
+    }
+    CFI_INSTRUCTION def_cfa $r30, 8
+    CFI_INSTRUCTION offset $r31, -4
+    CFI_INSTRUCTION offset $r30, -8
+    CFI_INSTRUCTION offset $r17, -12
+    CFI_INSTRUCTION offset $r16, -16
+    CFI_INSTRUCTION offset $r19, -20
+    CFI_INSTRUCTION offset $r18, -24
+    CFI_INSTRUCTION offset $r21, -28
+    CFI_INSTRUCTION offset $r20, -32
+    CFI_INSTRUCTION offset $r23, -36
+    CFI_INSTRUCTION offset $r22, -40
+    CFI_INSTRUCTION offset $r25, -44
+    CFI_INSTRUCTION offset $r24, -48
+    CFI_INSTRUCTION offset $r27, -52
+    CFI_INSTRUCTION offset $r26, -56
+    BUNDLE implicit $r29, implicit killed $d9, implicit killed $d10, debug-location !26 {
+      S2_storerd_io $r29, 48, killed $d9, debug-location !26 :: (store 8 into %fixed-stack.4)
+      S2_storerd_io $r29, 40, killed $d10, debug-location !26 :: (store 8 into %fixed-stack.3)
+    }
+    BUNDLE implicit $r29, implicit killed $d11, implicit killed $d12, debug-location !26 {
+      S2_storerd_io $r29, 32, killed $d11, debug-location !26 :: (store 8 into %fixed-stack.2)
+      S2_storerd_io $r29, 24, killed $d12, debug-location !26 :: (store 8 into %fixed-stack.1)
+    }
+    DBG_VALUE internal $r0, $noreg, !23, !DIExpression(), debug-location !27
+    BUNDLE implicit-def $r0, implicit $r29, implicit killed $d13, implicit $gp, debug-location !26 {
+      S2_storerd_io $r29, 16, killed $d13, debug-location !26 :: (store 8 into %fixed-stack.0)
+      renamable $r0 = L2_loadrigp @ga, implicit $gp, debug-location !26 :: (dereferenceable load 4 from @ga)
+    }
+    DBG_VALUE $r29, 0, !23, !DIExpression(DW_OP_plus_uconst, 12), debug-location !27
+    DBG_VALUE internal $r0, $noreg, !24, !DIExpression(), debug-location !27
+    BUNDLE implicit-def $r0, implicit $r29, implicit killed $r0, implicit $gp, debug-location !28 {
+      S2_storeri_io $r29, 12, killed renamable $r0, debug-location !28 :: (store 4 into %stack.0)
+      renamable $r0 = L2_loadrigp @gb, implicit $gp, debug-location !28 :: (dereferenceable load 4 from @gb)
+    }
+    DBG_VALUE $r29, 0, !24, !DIExpression(DW_OP_plus_uconst, 8), debug-location !27
+    DBG_VALUE internal $r0, $noreg, !25, !DIExpression(), debug-location !27
+    BUNDLE implicit-def $r0, implicit $r29, implicit killed $r0, implicit killed $gp, debug-location !29 {
+      S2_storeri_io $r29, 8, killed renamable $r0, debug-location !29 :: (store 4 into %stack.1)
+      renamable $r0 = L2_loadrigp @gc, implicit killed $gp, debug-location !29 :: (dereferenceable load 4 from @gc)
+    }
+    S2_storeri_io $r29, 4, killed renamable $r0, debug-location !30 :: (store 4 into %stack.2)
+    DBG_VALUE $r29, 0, !25, !DIExpression(DW_OP_plus_uconst, 4), debug-location !27
+    INLINEASM &"", 1, 12, implicit-def dead early-clobber $r0, 12, implicit-def dead early-clobber $r1, 12, implicit-def dead early-clobber $r2, 12, implicit-def dead early-clobber $r3, 12, implicit-def dead early-clobber $r4, 12, implicit-def dead early-clobber $r5, 12, implicit-def dead early-clobber $r6, 12, implicit-def dead early-clobber $r7, 12, implicit-def dead early-clobber $r8, 12, implicit-def dead early-clobber $r9, 12, implicit-def dead early-clobber $r10, 12, implicit-def dead early-clobber $r11, 12, implicit-def dead early-clobber $r12, 12, implicit-def dead early-clobber $r13, 12, implicit-def dead early-clobber $r14, 12, implicit-def dead early-clobber $r15, 12, implicit-def dead early-clobber $r16, 12, implicit-def dead early-clobber $r17, 12, implicit-def dead early-clobber $r18, 12, implicit-def dead early-clobber $r19, 12, implicit-def dead early-clobber $r20, 12, implicit-def dead early-clobber $r21, 12, implicit-def dead early-clobber $r22, 12, implicit-def dead early-clobber $r23, 12, implicit-def dead early-clobber $r24, 12, implicit-def dead early-clobber $r25, 12, implicit-def dead early-clobber $r26, 12, implicit-def dead early-clobber $r27, 12, implicit-def dead early-clobber $r28, !31, debug-location !30
+    BUNDLE implicit-def $r0, implicit-def $r1, implicit $r29, debug-location !32 {
+      $r0 = L2_loadri_io $r29, 12, debug-location !32 :: (load 4 from %stack.0)
+      $r1 = L2_loadri_io $r29, 8, debug-location !32 :: (load 4 from %stack.1)
+    }
+    BUNDLE implicit-def dead $r2, implicit-def dead $pc, implicit-def dead $r31, implicit-def $r29, implicit $r29, implicit killed $r0, implicit killed $r1, debug-location !32 {
+      $r2 = L2_loadri_io $r29, 4, debug-location !32 :: (load 4 from %stack.2)
+      J2_call @callee, hexagoncsr, implicit-def dead $pc, implicit-def dead $r31, implicit $r29, implicit killed $r0, implicit killed $r1, implicit internal killed $r2, implicit-def $r29, debug-location !32
+    }
+    BUNDLE implicit-def $d8, implicit-def $r16, implicit-def $r17, implicit-def $d9, implicit-def $r18, implicit-def $r19, implicit $r29, debug-location !33 {
+      $d8 = L2_loadrd_io $r29, 56, debug-location !33 :: (load 8 from %fixed-stack.5)
+      $d9 = L2_loadrd_io $r29, 48, debug-location !33 :: (load 8 from %fixed-stack.4)
+    }
+    BUNDLE implicit-def $d10, implicit-def $r20, implicit-def $r21, implicit-def $d11, implicit-def $r22, implicit-def $r23, implicit $r29, debug-location !33 {
+      $d10 = L2_loadrd_io $r29, 40, debug-location !33 :: (load 8 from %fixed-stack.3)
+      $d11 = L2_loadrd_io $r29, 32, debug-location !33 :: (load 8 from %fixed-stack.2)
+    }
+    BUNDLE implicit-def $d12, implicit-def $r24, implicit-def $r25, implicit-def $d13, implicit-def $r26, implicit-def $r27, implicit killed $r29, debug-location !33 {
+      $d12 = L2_loadrd_io $r29, 24, debug-location !33 :: (load 8 from %fixed-stack.1)
+      $d13 = L2_loadrd_io killed $r29, 16, debug-location !33 :: (load 8 from %fixed-stack.0)
+    }
+    $d15 = L4_return killed $r30, implicit-def $pc, implicit-def $r29, implicit killed $framekey, implicit-def dead $pc, implicit $d8, implicit $d9, implicit $d10, implicit $d11, implicit $d12, implicit $d13, debug-location !33
+
+...
+
+# Verify that call site entries are emitted for all three parameters.
+# Previously the code that's looking for instructions to describe parameters
+# with would stop when reaching the bundle header for the bundled call,
+# resulting in $r0 and $r1 not being described.
+#
+# Please note that at the time of creating this test the Hexagon target did not
+# support call site information, so the "callSites" array has been manually
+# added.
+
+# CHECK: DW_TAG_GNU_call_site_parameter
+# CHECK-NEXT: DW_AT_location      (DW_OP_reg2 R2)
+# CHECK-NEXT: DW_AT_GNU_call_site_value   (DW_OP_breg29 R29+4, DW_OP_deref_size 0x4)
+
+# CHECK: DW_TAG_GNU_call_site_parameter
+# CHECK-NEXT: DW_AT_location      (DW_OP_reg1 R1)
+# CHECK-NEXT: DW_AT_GNU_call_site_value   (DW_OP_breg29 R29+8, DW_OP_deref_size 0x4)
+
+# CHECK: DW_TAG_GNU_call_site_parameter
+# CHECK-NEXT: DW_AT_location      (DW_OP_reg0 R0)
+# CHECK-NEXT: DW_AT_GNU_call_site_value   (DW_OP_breg29 R29+12, DW_OP_deref_size 0x4)