6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-15 20:29:48 +00:00
Code Issues Projects Releases Wiki Activity
56f3b7f2fd
llvm-6502/test/DebugInfo/X86/recursive_inlining.ll

276 lines
13 KiB
LLVM
Raw Normal View History

DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
; REQUIRES: object-emission
Fix llvm/test/DebugInfo/X86/recursive_inlining.ll to use %llc_dwarf. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215181 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-08 02:24:05 +00:00
; RUN: %llc_dwarf -filetype=obj -O0 < %s | llvm-dwarfdump -debug-dump=info - | FileCheck %s
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
; This isn't a very pretty test case - I imagine there might be other ways to
; tickle the optimizers into producing the desired code, but I haven't found
; them.
; The issue is when a function is inlined into itself, the inlined argument
; accidentally overwrote the concrete argument and was lost.
; IR generated from the following source compiled with clang -g:
; void fn1(void *);
; void fn2(int, int, int, int);
; void fn3();
; void fn8();
; struct C {
; int b;
; void m_fn2() {
; fn8();
; if (b) fn2(0, 0, 0, 0);
; fn3();
; }
; };
; C *x;
; inline void fn7() {}
; void fn6() {
; fn8();
; x->m_fn2();
; fn7();
; }
; void fn3() { fn6(); }
; void fn4() { x->m_fn2(); }
; void fn5() { x->m_fn2(); }
; The definition of C and declaration of C::m_fn2
; CHECK: DW_TAG_structure_type
; CHECK-NOT: {{DW_TAG|NULL}}
; CHECK: DW_TAG_member
; CHECK-NOT: {{DW_TAG|NULL}}
[dwarfdump] Print the name for referenced specification of abstract_origin DIEs. Reviewers: dblaikie, samsonov, echristo, aprantl Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D5466 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219099 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-06 03:36:31 +00:00
; CHECK: DW_TAG_subprogram
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
; CHECK-NOT: DW_TAG
; CHECK: DW_AT_name {{.*}} "m_fn2"
; CHECK-NOT: {{DW_TAG|NULL}}
; CHECK: [[M_FN2_THIS_DECL:.*]]: DW_TAG_formal_parameter
; The abstract definition of C::m_fn2
; CHECK: [[M_FN2_ABS_DEF:.*]]: DW_TAG_subprogram
; CHECK-NOT: DW_TAG
[dwarfdump] Print the name for referenced specification of abstract_origin DIEs. Reviewers: dblaikie, samsonov, echristo, aprantl Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D5466 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219099 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-06 03:36:31 +00:00
; CHECK: DW_AT_specification {{.*}} "_ZN1C5m_fn2Ev"
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
; CHECK-NOT: DW_TAG
; CHECK: DW_AT_inline
; CHECK-NOT: {{DW_TAG|NULL}}
; CHECK: [[M_FN2_THIS_ABS_DEF:.*]]: DW_TAG_formal_parameter
; CHECK-NOT: DW_TAG
; CHECK: DW_AT_name {{.*}} "this"
; Skip some other functions
; CHECK: DW_TAG_subprogram
; CHECK: DW_TAG_subprogram
; CHECK: DW_TAG_subprogram
; The concrete definition of C::m_fn2
; CHECK: DW_TAG_subprogram
; CHECK-NOT: DW_TAG
[dwarfdump] Print the name for referenced specification of abstract_origin DIEs. Reviewers: dblaikie, samsonov, echristo, aprantl Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D5466 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219099 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-06 03:36:31 +00:00
; CHECK: DW_AT_abstract_origin {{.*}} {[[M_FN2_ABS_DEF]]} "_ZN1C5m_fn2Ev"
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
; CHECK-NOT: {{DW_TAG|NULL}}
; CHECK: DW_TAG_formal_parameter
; CHECK-NOT: DW_TAG
; CHECK: DW_AT_abstract_origin {{.*}} {[[M_FN2_THIS_ABS_DEF]]}
; CHECK-NOT: {{DW_TAG|NULL}}
; Inlined fn3:
; CHECK: DW_TAG_inlined_subroutine
; CHECK-NOT: {{DW_TAG|NULL}}
; Inlined fn6:
; CHECK: DW_TAG_inlined_subroutine
; CHECK-NOT: {{DW_TAG|NULL}}
; Inlined C::m_fn2:
; CHECK: DW_TAG_inlined_subroutine
; CHECK-NOT: DW_TAG
[dwarfdump] Print the name for referenced specification of abstract_origin DIEs. Reviewers: dblaikie, samsonov, echristo, aprantl Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D5466 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219099 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-06 03:36:31 +00:00
; CHECK: DW_AT_abstract_origin {{.*}} {[[M_FN2_ABS_DEF]]} "_ZN1C5m_fn2Ev"
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
; CHECK-NOT: {{DW_TAG|NULL}}
; CHECK: DW_TAG_formal_parameter
; CHECK-NOT: DW_TAG
; CHECK: DW_AT_abstract_origin {{.*}} {[[M_FN2_THIS_ABS_DEF]]}
%struct.C = type { i32 }
@x = global %struct.C* null, align 8
; Function Attrs: nounwind
define void @_Z3fn6v() #0 {
entry:
tail call void @_Z3fn8v() #3, !dbg !31
%0 = load %struct.C** @x, align 8, !dbg !32, !tbaa !33
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
tail call void @llvm.dbg.value(metadata %struct.C* %0, i64 0, metadata !37, metadata !{!"0x102"}) #3, !dbg !38
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
tail call void @_Z3fn8v() #3, !dbg !39
%b.i = getelementptr inbounds %struct.C* %0, i64 0, i32 0, !dbg !40
%1 = load i32* %b.i, align 4, !dbg !40, !tbaa !42
%tobool.i = icmp eq i32 %1, 0, !dbg !40
br i1 %tobool.i, label %_ZN1C5m_fn2Ev.exit, label %if.then.i, !dbg !40
if.then.i: ; preds = %entry
tail call void @_Z3fn2iiii(i32 0, i32 0, i32 0, i32 0) #3, !dbg !45
br label %_ZN1C5m_fn2Ev.exit, !dbg !45
_ZN1C5m_fn2Ev.exit: ; preds = %entry, %if.then.i
tail call void @_Z3fn3v() #3, !dbg !47
ret void, !dbg !48
}
declare void @_Z3fn8v() #1
; Function Attrs: nounwind
define linkonce_odr void @_ZN1C5m_fn2Ev(%struct.C* nocapture readonly %this) #0 align 2 {
entry:
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
tail call void @llvm.dbg.value(metadata %struct.C* %this, i64 0, metadata !24, metadata !{!"0x102"}), !dbg !49
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
tail call void @_Z3fn8v() #3, !dbg !50
%b = getelementptr inbounds %struct.C* %this, i64 0, i32 0, !dbg !51
%0 = load i32* %b, align 4, !dbg !51, !tbaa !42
%tobool = icmp eq i32 %0, 0, !dbg !51
br i1 %tobool, label %if.end, label %if.then, !dbg !51
if.then: ; preds = %entry
tail call void @_Z3fn2iiii(i32 0, i32 0, i32 0, i32 0) #3, !dbg !52
br label %if.end, !dbg !52
if.end: ; preds = %entry, %if.then
tail call void @_Z3fn8v() #3, !dbg !53
%1 = load %struct.C** @x, align 8, !dbg !56, !tbaa !33
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
tail call void @llvm.dbg.value(metadata %struct.C* %1, i64 0, metadata !57, metadata !{!"0x102"}) #3, !dbg !58
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
tail call void @_Z3fn8v() #3, !dbg !59
%b.i.i = getelementptr inbounds %struct.C* %1, i64 0, i32 0, !dbg !60
%2 = load i32* %b.i.i, align 4, !dbg !60, !tbaa !42
%tobool.i.i = icmp eq i32 %2, 0, !dbg !60
br i1 %tobool.i.i, label %_Z3fn6v.exit, label %if.then.i.i, !dbg !60
if.then.i.i: ; preds = %if.end
tail call void @_Z3fn2iiii(i32 0, i32 0, i32 0, i32 0) #3, !dbg !61
br label %_Z3fn6v.exit, !dbg !61
_Z3fn6v.exit: ; preds = %if.end, %if.then.i.i
tail call void @_Z3fn3v() #3, !dbg !62
ret void, !dbg !63
}
; Function Attrs: nounwind
define void @_Z3fn3v() #0 {
entry:
br label %tailrecurse
tailrecurse: ; preds = %tailrecurse.backedge, %entry
tail call void @_Z3fn8v() #3, !dbg !64
%0 = load %struct.C** @x, align 8, !dbg !66, !tbaa !33
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
tail call void @llvm.dbg.value(metadata %struct.C* %0, i64 0, metadata !67, metadata !{!"0x102"}) #3, !dbg !68
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
tail call void @_Z3fn8v() #3, !dbg !69
%b.i.i = getelementptr inbounds %struct.C* %0, i64 0, i32 0, !dbg !70
%1 = load i32* %b.i.i, align 4, !dbg !70, !tbaa !42
%tobool.i.i = icmp eq i32 %1, 0, !dbg !70
br i1 %tobool.i.i, label %tailrecurse.backedge, label %if.then.i.i, !dbg !70
tailrecurse.backedge: ; preds = %tailrecurse, %if.then.i.i
br label %tailrecurse
if.then.i.i: ; preds = %tailrecurse
tail call void @_Z3fn2iiii(i32 0, i32 0, i32 0, i32 0) #3, !dbg !71
br label %tailrecurse.backedge, !dbg !71
}
; Function Attrs: nounwind
define void @_Z3fn4v() #0 {
entry:
%0 = load %struct.C** @x, align 8, !dbg !72, !tbaa !33
tail call void @_ZN1C5m_fn2Ev(%struct.C* %0), !dbg !72
ret void, !dbg !72
}
; Function Attrs: nounwind
define void @_Z3fn5v() #0 {
entry:
%0 = load %struct.C** @x, align 8, !dbg !73, !tbaa !33
tail call void @_ZN1C5m_fn2Ev(%struct.C* %0), !dbg !73
ret void, !dbg !73
}
declare void @_Z3fn2iiii(i32, i32, i32, i32) #1
; Function Attrs: nounwind readnone
Move the complex address expression out of DIVariable and into an extra argument of the llvm.dbg.declare/llvm.dbg.value intrinsics. Previously, DIVariable was a variable-length field that has an optional reference to a Metadata array consisting of a variable number of complex address expressions. In the case of OpPiece expressions this is wasting a lot of storage in IR, because when an aggregate type is, e.g., SROA'd into all of its n individual members, the IR will contain n copies of the DIVariable, all alike, only differing in the complex address reference at the end. By making the complex address into an extra argument of the dbg.value/dbg.declare intrinsics, all of the pieces can reference the same variable and the complex address expressions can be uniqued across the CU, too. Down the road, this will allow us to move other flags, such as "indirection" out of the DIVariable, too. The new intrinsics look like this: declare void @llvm.dbg.declare(metadata %storage, metadata %var, metadata %expr) declare void @llvm.dbg.value(metadata %storage, i64 %offset, metadata %var, metadata %expr) This patch adds a new LLVM-local tag to DIExpressions, so we can detect and pretty-print DIExpression metadata nodes. What this patch doesn't do: This patch does not touch the "Indirect" field in DIVariable; but moving that into the expression would be a natural next step. http://reviews.llvm.org/D4919 rdar://problem/17994491 Thanks to dblaikie and dexonsmith for reviewing this patch! Note: I accidentally committed a bogus older version of this patch previously. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218787 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-01 18:55:02 +00:00
declare void @llvm.dbg.value(metadata, i64, metadata, metadata) #2
DebugInfo: Fix overwriting/loss of inlined arguments to recursively inlined functions. Due to an unnecessary special case, inlined arguments that happened to be from the same function as they were inlined into were misclassified as non-inline arguments and would overwrite the non-inlined arguments. Assert that we never overwrite a function's arguments, and stop misclassifying inlined arguments as non-inline arguments to fix this issue. Excuse the rather crappy test case - handcrafted IR might do better, or someone who understands better how to tickle the inliner to create a recursive inlining situation like this (though it may also be necessary to tickle the variable in a particular way to cause it to be recorded in the MMI side table and go down this particular path for location information). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@215157 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-07 22:22:49 +00:00
attributes #0 = { nounwind "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-realign-stack" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #1 = { "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-realign-stack" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #2 = { nounwind readnone }
attributes #3 = { nounwind }
!llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!28, !29}
!llvm.ident = !{!30}
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
!0 = !{!"0x11\004\00clang version 3.6.0 \001\00\000\00\001", !1, !2, !3, !13, !26, !2} ; [ DW_TAG_compile_unit ] [/usr/local/google/home/blaikie/dev/scratch/missing_concrete_variable_on_darwin/reduce/<stdin>] [DW_LANG_C_plus_plus]
!1 = !{!"<stdin>", !"/usr/local/google/home/blaikie/dev/scratch/missing_concrete_variable_on_darwin/reduce"}
!2 = !{}
!3 = !{!4}
!4 = !{!"0x13\00C\005\0032\0032\000\000\000", !5, null, null, !6, null, null, !"_ZTS1C"} ; [ DW_TAG_structure_type ] [C] [line 5, size 32, align 32, offset 0] [def] [from ]
!5 = !{!"recursive_inlining.cpp", !"/usr/local/google/home/blaikie/dev/scratch/missing_concrete_variable_on_darwin/reduce"}
!6 = !{!7, !9}
!7 = !{!"0xd\00b\006\0032\0032\000\000", !5, !"_ZTS1C", !8} ; [ DW_TAG_member ] [b] [line 6, size 32, align 32, offset 0] [from int]
!8 = !{!"0x24\00int\000\0032\0032\000\000\005", null, null} ; [ DW_TAG_base_type ] [int] [line 0, size 32, align 32, offset 0, enc DW_ATE_signed]
!9 = !{!"0x2e\00m_fn2\00m_fn2\00_ZN1C5m_fn2Ev\007\000\000\000\006\00256\001\007", !5, !"_ZTS1C", !10, null, null, null, i32 0, null} ; [ DW_TAG_subprogram ] [line 7] [m_fn2]
!10 = !{!"0x15\00\000\000\000\000\000\000", i32 0, null, null, !11, null, null, null} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ]
!11 = !{null, !12}
!12 = !{!"0xf\00\000\0064\0064\000\001088", null, null, !"_ZTS1C"} ; [ DW_TAG_pointer_type ] [line 0, size 64, align 64, offset 0] [artificial] [from _ZTS1C]
!13 = !{!14, !18, !19, !20, !21, !22}
!14 = !{!"0x2e\00fn6\00fn6\00_Z3fn6v\0015\000\001\000\006\00256\001\0015", !5, !15, !16, null, void ()* @_Z3fn6v, null, null, !2} ; [ DW_TAG_subprogram ] [line 15] [def] [fn6]
!15 = !{!"0x29", !5} ; [ DW_TAG_file_type ] [/usr/local/google/home/blaikie/dev/scratch/missing_concrete_variable_on_darwin/reduce/recursive_inlining.cpp]
!16 = !{!"0x15\00\000\000\000\000\000\000", i32 0, null, null, !17, null, null, null} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ]
!17 = !{null}
!18 = !{!"0x2e\00fn3\00fn3\00_Z3fn3v\0020\000\001\000\006\00256\001\0020", !5, !15, !16, null, void ()* @_Z3fn3v, null, null, !2} ; [ DW_TAG_subprogram ] [line 20] [def] [fn3]
!19 = !{!"0x2e\00fn4\00fn4\00_Z3fn4v\0021\000\001\000\006\00256\001\0021", !5, !15, !16, null, void ()* @_Z3fn4v, null, null, !2} ; [ DW_TAG_subprogram ] [line 21] [def] [fn4]
!20 = !{!"0x2e\00fn5\00fn5\00_Z3fn5v\0022\000\001\000\006\00256\001\0022", !5, !15, !16, null, void ()* @_Z3fn5v, null, null, !2} ; [ DW_TAG_subprogram ] [line 22] [def] [fn5]
!21 = !{!"0x2e\00fn7\00fn7\00_Z3fn7v\0014\000\001\000\006\00256\001\0014", !5, !15, !16, null, null, null, null, !2} ; [ DW_TAG_subprogram ] [line 14] [def] [fn7]
!22 = !{!"0x2e\00m_fn2\00m_fn2\00_ZN1C5m_fn2Ev\007\000\001\000\006\00256\001\007", !5, !"_ZTS1C", !10, null, void (%struct.C*)* @_ZN1C5m_fn2Ev, null, !9, !23} ; [ DW_TAG_subprogram ] [line 7] [def] [m_fn2]
!23 = !{!24}
!24 = !{!"0x101\00this\0016777216\001088", !22, null, !25} ; [ DW_TAG_arg_variable ] [this] [line 0]
!25 = !{!"0xf\00\000\0064\0064\000\000", null, null, !"_ZTS1C"} ; [ DW_TAG_pointer_type ] [line 0, size 64, align 64, offset 0] [from _ZTS1C]
!26 = !{!27}
!27 = !{!"0x34\00x\00x\00\0013\000\001", null, !15, !25, %struct.C** @x, null} ; [ DW_TAG_variable ] [x] [line 13] [def]
!28 = !{i32 2, !"Dwarf Version", i32 4}
!29 = !{i32 2, !"Debug Info Version", i32 2}
!30 = !{!"clang version 3.6.0 "}
IR: Move MDLocation into place This commit moves `MDLocation`, finishing off PR21433. There's an accompanying clang commit for frontend testcases. I'll attach the testcase upgrade script I used to PR21433 to help out-of-tree frontends/backends. This changes the schema for `DebugLoc` and `DILocation` from: !{i32 3, i32 7, !7, !8} to: !MDLocation(line: 3, column: 7, scope: !7, inlinedAt: !8) Note that empty fields (line/column: 0 and inlinedAt: null) don't get printed by the assembly writer. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@226048 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-14 22:27:36 +00:00
!31 = !MDLocation(line: 16, scope: !14)
!32 = !MDLocation(line: 17, scope: !14)
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
!33 = !{!34, !34, i64 0}
!34 = !{!"any pointer", !35, i64 0}
!35 = !{!"omnipotent char", !36, i64 0}
!36 = !{!"Simple C/C++ TBAA"}
!37 = !{!"0x101\00this\0016777216\001088", !22, null, !25, !32} ; [ DW_TAG_arg_variable ] [this] [line 0]
IR: Move MDLocation into place This commit moves `MDLocation`, finishing off PR21433. There's an accompanying clang commit for frontend testcases. I'll attach the testcase upgrade script I used to PR21433 to help out-of-tree frontends/backends. This changes the schema for `DebugLoc` and `DILocation` from: !{i32 3, i32 7, !7, !8} to: !MDLocation(line: 3, column: 7, scope: !7, inlinedAt: !8) Note that empty fields (line/column: 0 and inlinedAt: null) don't get printed by the assembly writer. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@226048 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-14 22:27:36 +00:00
!38 = !MDLocation(line: 0, scope: !22, inlinedAt: !32)
!39 = !MDLocation(line: 8, scope: !22, inlinedAt: !32)
!40 = !MDLocation(line: 9, scope: !41, inlinedAt: !32)
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
!41 = !{!"0xb\009\000\000", !5, !22} ; [ DW_TAG_lexical_block ] [/usr/local/google/home/blaikie/dev/scratch/missing_concrete_variable_on_darwin/reduce/recursive_inlining.cpp]
!42 = !{!43, !44, i64 0}
!43 = !{!"_ZTS1C", !44, i64 0}
!44 = !{!"int", !35, i64 0}
IR: Move MDLocation into place This commit moves `MDLocation`, finishing off PR21433. There's an accompanying clang commit for frontend testcases. I'll attach the testcase upgrade script I used to PR21433 to help out-of-tree frontends/backends. This changes the schema for `DebugLoc` and `DILocation` from: !{i32 3, i32 7, !7, !8} to: !MDLocation(line: 3, column: 7, scope: !7, inlinedAt: !8) Note that empty fields (line/column: 0 and inlinedAt: null) don't get printed by the assembly writer. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@226048 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-14 22:27:36 +00:00
!45 = !MDLocation(line: 9, scope: !46, inlinedAt: !32)
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
!46 = !{!"0xb\009\000\001", !5, !41} ; [ DW_TAG_lexical_block ] [/usr/local/google/home/blaikie/dev/scratch/missing_concrete_variable_on_darwin/reduce/recursive_inlining.cpp]
IR: Move MDLocation into place This commit moves `MDLocation`, finishing off PR21433. There's an accompanying clang commit for frontend testcases. I'll attach the testcase upgrade script I used to PR21433 to help out-of-tree frontends/backends. This changes the schema for `DebugLoc` and `DILocation` from: !{i32 3, i32 7, !7, !8} to: !MDLocation(line: 3, column: 7, scope: !7, inlinedAt: !8) Note that empty fields (line/column: 0 and inlinedAt: null) don't get printed by the assembly writer. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@226048 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-14 22:27:36 +00:00
!47 = !MDLocation(line: 10, scope: !22, inlinedAt: !32)
!48 = !MDLocation(line: 19, scope: !14)
!49 = !MDLocation(line: 0, scope: !22)
!50 = !MDLocation(line: 8, scope: !22)
!51 = !MDLocation(line: 9, scope: !41)
!52 = !MDLocation(line: 9, scope: !46)
!53 = !MDLocation(line: 16, scope: !14, inlinedAt: !54)
!54 = !MDLocation(line: 20, scope: !18, inlinedAt: !55)
!55 = !MDLocation(line: 10, scope: !22)
!56 = !MDLocation(line: 17, scope: !14, inlinedAt: !54)
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
!57 = !{!"0x101\00this\0016777216\001088", !22, null, !25, !56} ; [ DW_TAG_arg_variable ] [this] [line 0]
IR: Move MDLocation into place This commit moves `MDLocation`, finishing off PR21433. There's an accompanying clang commit for frontend testcases. I'll attach the testcase upgrade script I used to PR21433 to help out-of-tree frontends/backends. This changes the schema for `DebugLoc` and `DILocation` from: !{i32 3, i32 7, !7, !8} to: !MDLocation(line: 3, column: 7, scope: !7, inlinedAt: !8) Note that empty fields (line/column: 0 and inlinedAt: null) don't get printed by the assembly writer. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@226048 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-14 22:27:36 +00:00
!58 = !MDLocation(line: 0, scope: !22, inlinedAt: !56)
!59 = !MDLocation(line: 8, scope: !22, inlinedAt: !56)
!60 = !MDLocation(line: 9, scope: !41, inlinedAt: !56)
!61 = !MDLocation(line: 9, scope: !46, inlinedAt: !56)
!62 = !MDLocation(line: 10, scope: !22, inlinedAt: !56)
!63 = !MDLocation(line: 11, scope: !22)
!64 = !MDLocation(line: 16, scope: !14, inlinedAt: !65)
!65 = !MDLocation(line: 20, scope: !18)
!66 = !MDLocation(line: 17, scope: !14, inlinedAt: !65)
IR: Make metadata typeless in assembly Now that `Metadata` is typeless, reflect that in the assembly. These are the matching assembly changes for the metadata/value split in r223802. - Only use the `metadata` type when referencing metadata from a call intrinsic -- i.e., only when it's used as a `Value`. - Stop pretending that `ValueAsMetadata` is wrapped in an `MDNode` when referencing it from call intrinsics. So, assembly like this: define @foo(i32 %v) { call void @llvm.foo(metadata !{i32 %v}, metadata !0) call void @llvm.foo(metadata !{i32 7}, metadata !0) call void @llvm.foo(metadata !1, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{metadata !3}, metadata !0) ret void, !bar !2 } !0 = metadata !{metadata !2} !1 = metadata !{i32* @global} !2 = metadata !{metadata !3} !3 = metadata !{} turns into this: define @foo(i32 %v) { call void @llvm.foo(metadata i32 %v, metadata !0) call void @llvm.foo(metadata i32 7, metadata !0) call void @llvm.foo(metadata i32* @global, metadata !0) call void @llvm.foo(metadata !3, metadata !0) call void @llvm.foo(metadata !{!3}, metadata !0) ret void, !bar !2 } !0 = !{!2} !1 = !{i32* @global} !2 = !{!3} !3 = !{} I wrote an upgrade script that handled almost all of the tests in llvm and many of the tests in cfe (even handling many `CHECK` lines). I've attached it (or will attach it in a moment if you're speedy) to PR21532 to help everyone update their out-of-tree testcases. This is part of PR21532. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@224257 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-15 19:07:53 +00:00
!67 = !{!"0x101\00this\0016777216\001088", !22, null, !25, !66} ; [ DW_TAG_arg_variable ] [this] [line 0]
IR: Move MDLocation into place This commit moves `MDLocation`, finishing off PR21433. There's an accompanying clang commit for frontend testcases. I'll attach the testcase upgrade script I used to PR21433 to help out-of-tree frontends/backends. This changes the schema for `DebugLoc` and `DILocation` from: !{i32 3, i32 7, !7, !8} to: !MDLocation(line: 3, column: 7, scope: !7, inlinedAt: !8) Note that empty fields (line/column: 0 and inlinedAt: null) don't get printed by the assembly writer. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@226048 91177308-0d34-0410-b5e6-96231b3b80d8
2015-01-14 22:27:36 +00:00
!68 = !MDLocation(line: 0, scope: !22, inlinedAt: !66)
!69 = !MDLocation(line: 8, scope: !22, inlinedAt: !66)
!70 = !MDLocation(line: 9, scope: !41, inlinedAt: !66)
!71 = !MDLocation(line: 9, scope: !46, inlinedAt: !66)
!72 = !MDLocation(line: 21, scope: !19)
!73 = !MDLocation(line: 22, scope: !20)
Reference in New Issue Copy Permalink