[llvm] r220923 - PR21408: Workaround the appearance of duplicate variables due to problems when inlining two calls to the same function from the same call site.
David Blaikie
dblaikie at gmail.com
Fri Oct 31 09:59:07 PDT 2014
On Thu, Oct 30, 2014 at 1:59 PM, Adrian Prantl <aprantl at apple.com> wrote:
>
> > On Oct 30, 2014, at 1:20 PM, David Blaikie <dblaikie at gmail.com> wrote:
> >
> > Author: dblaikie
> > Date: Thu Oct 30 15:20:11 2014
> > New Revision: 220923
> >
> > URL: http://llvm.org/viewvc/llvm-project?rev=220923&view=rev
> > Log:
> > PR21408: Workaround the appearance of duplicate variables due to
> problems when inlining two calls to the same function from the same call
> site.
> >
> > Added:
> > llvm/trunk/test/DebugInfo/duplicate_inline.ll
> > Modified:
> > llvm/trunk/lib/CodeGen/AsmPrinter/DwarfFile.cpp
> >
> > Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfFile.cpp
> > URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfFile.cpp?rev=220923&r1=220922&r2=220923&view=diff
> >
> ==============================================================================
> > --- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfFile.cpp (original)
> > +++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfFile.cpp Thu Oct 30 15:20:11
> 2014
> > @@ -177,7 +177,12 @@ void DwarfFile::addScopeVariable(Lexical
> > // A later indexed parameter has been found, insert immediately
> before it.
> > if (CurNum > ArgNum)
> > break;
> > - assert(CurNum != ArgNum);
> > + // FIXME: There are still some cases where two inlined functions
> are
> > + // conflated together (two calls to the same function at the same
> > + // location (eg: via a macro, or without column info, etc)) and
> then
> > + // their arguments are conflated as well.
> > + assert((LS->getParent() || CurNum != ArgNum) &&
> > + "Missing parameter for top level (non-inlined) function”);
>
> Shouldn’t the message be “duplicate argument top level (non-inlined)
> function”?
>
Right you are. Fixed in 220967.
In theory 'parameter' is probably more correct than 'argument' here (I
think that's how people use the terms - parameters are the generic things
in the function definition/declaration, arguments are the concrete
instances of those passed at call sites), but we've got a lot of 'argument'
related stuff (arg_variable, arg number, etc) here already, so I'll just go
with that.
>
> -- adrian
>
>
> > ++I;
> > }
> > Vars.insert(I, Var);
> >
> > Added: llvm/trunk/test/DebugInfo/duplicate_inline.ll
> > URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/DebugInfo/duplicate_inline.ll?rev=220923&view=auto
> >
> ==============================================================================
> > --- llvm/trunk/test/DebugInfo/duplicate_inline.ll (added)
> > +++ llvm/trunk/test/DebugInfo/duplicate_inline.ll Thu Oct 30 15:20:11
> 2014
> > @@ -0,0 +1,117 @@
> > +; REQUIRES: object-emission
> > +
> > +; RUN: %llc_dwarf < %s -filetype=obj | llvm-dwarfdump -debug-dump=info
> - | FileCheck %s
> > +
> > +; Built with clang from the following source:
> > +; void f1(int);
> > +; __attribute__((always_inline)) inline void f2(int i) { f1(i); }
> > +;
> > +; #define MULTICALL \
> > +; f2(x); \
> > +; f2(y);
> > +;
> > +; void f3(int x, int y) { MULTICALL; }
> > +
> > +; FIXME: This produces only one inlined_subroutine, with two
> formal_parameters
> > +; (both named "this"), one for each of the actual inlined subroutines.
> ;
> > +; Inlined scopes are differentiated by the combination of 'inlined at'
> (call)
> > +; location and the location within the function. If two calls to the
> same
> > +; function occur at the same location the scopes end up conflated and
> there
> > +; appears to be only one inlined function.
> > +; To fix this, we'd need to add some kind of unique metadata per call
> site, possibly something like:
> > +;
> > +; !42 = metadata !{i32 1, i32 0, metadata !43, metadata !44}
> > +; !44 = metadata !{i32 2, i32 0, metadata !45, null}
> > +;
> > +; ->
> > +;
> > +; !42 = metadata !{i32 1, i32 0, metadata !43, metadata !44}
> > +; !44 = metadata !{metadata !45, metadata !44}
> > +; !45 = metadata !{i32 2, i32 0, metadata !45, null}
> > +;
> > +; since cycles in metadata are not uniqued, the !44 node would not be
> shared
> > +; between calls to the same function from the same location, ensuring
> separate
> > +; inlined subroutines would be generated.
> > +;
> > +; Once this is done, the (insufficient) hack in clang that adds column
> > +; information to call sites to differentiate inlined callers can be
> removed as it
> > +; will no longer be necessary.
> > +;
> > +; While it might be nice to omit the duplicate parameter in this case
> (while
> > +; we wait/work on the real fix), it's actually better to leave it in
> because it
> > +; allows us to hold the invariant that every DbgVariable has a DIE,
> every time.
> > +; This has proved valuable in finding other bugs, so I want to avoid
> removing the
> > +; invariant/assertion. Besides, we don't know which one's the right one
> anyway...
> > +
> > +; CHECK: DW_TAG_subprogram
> > +; CHECK: DW_TAG_inlined_subroutine
> > +; CHECK: DW_TAG_formal_parameter
> > +; CHECK-NOT: DW_TAG
> > +; CHECK: DW_TAG_formal_parameter
> > +; CHECK-NOT: DW_TAG
> > +; CHECK: NULL
> > +; CHECK-NOT: DW_TAG
> > +; CHECK: NULL
> > +
> > +; Function Attrs: uwtable
> > +define void @_Z2f3ii(i32 %x, i32 %y) #0 {
> > +entry:
> > + %i.addr.i1 = alloca i32, align 4
> > + %i.addr.i = alloca i32, align 4
> > + %x.addr = alloca i32, align 4
> > + %y.addr = alloca i32, align 4
> > + store i32 %x, i32* %x.addr, align 4
> > + call void @llvm.dbg.declare(metadata !{i32* %x.addr}, metadata !15,
> metadata !16), !dbg !17
> > + store i32 %y, i32* %y.addr, align 4
> > + call void @llvm.dbg.declare(metadata !{i32* %y.addr}, metadata !18,
> metadata !16), !dbg !19
> > + %0 = load i32* %x.addr, align 4, !dbg !20
> > + store i32 %0, i32* %i.addr.i, align 4, !dbg !20
> > + call void @llvm.dbg.declare(metadata !{i32* %i.addr.i}, metadata !21,
> metadata !16), !dbg !22
> > + %1 = load i32* %i.addr.i, align 4, !dbg !23
> > + call void @_Z2f1i(i32 %1), !dbg !23
> > + %2 = load i32* %y.addr, align 4, !dbg !20
> > + store i32 %2, i32* %i.addr.i1, align 4, !dbg !20
> > + call void @llvm.dbg.declare(metadata !{i32* %i.addr.i1}, metadata
> !21, metadata !16), !dbg !22
> > + %3 = load i32* %i.addr.i1, align 4, !dbg !23
> > + call void @_Z2f1i(i32 %3), !dbg !23
> > + ret void, !dbg !24
> > +}
> > +
> > +; Function Attrs: nounwind readnone
> > +declare void @llvm.dbg.declare(metadata, metadata, metadata) #1
> > +
> > +declare void @_Z2f1i(i32) #2
> > +
> > +attributes #0 = { uwtable "less-precise-fpmad"="false"
> "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf"
> "no-infs-fp-math"="false" "no-nans-fp-math"="false"
> "stack-protector-buffer-size"="8" "unsafe-fp-math"="false"
> "use-soft-float"="false" }
> > +attributes #1 = { nounwind readnone }
> > +attributes #2 = { "less-precise-fpmad"="false"
> "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf"
> "no-infs-fp-math"="false" "no-nans-fp-math"="false"
> "stack-protector-buffer-size"="8" "unsafe-fp-math"="false"
> "use-soft-float"="false" }
> > +
> > +!llvm.dbg.cu = !{!0}
> > +!llvm.module.flags = !{!12, !13}
> > +!llvm.ident = !{!14}
> > +
> > +!0 = metadata !{metadata !"0x11\004\00clang version 3.6.0
> \000\00\000\00\001", metadata !1, metadata !2, metadata !2, metadata !3,
> metadata !2, metadata !2} ; [ DW_TAG_compile_unit ]
> [/tmp/dbginfo/duplicate_inline.cpp] [DW_LANG_C_plus_plus]
> > +!1 = metadata !{metadata !"duplicate_inline.cpp", metadata
> !"/tmp/dbginfo"}
> > +!2 = metadata !{}
> > +!3 = metadata !{metadata !4, metadata !9}
> > +!4 = metadata !{metadata
> !"0x2e\00f3\00f3\00_Z2f3ii\008\000\001\000\000\00256\000\008", metadata !1,
> metadata !5, metadata !6, null, void (i32, i32)* @_Z2f3ii, null, null,
> metadata !2} ; [ DW_TAG_subprogram ] [line 8] [def] [f3]
> > +!5 = metadata !{metadata !"0x29", metadata !1} ; [ DW_TAG_file_type
> ] [/tmp/dbginfo/duplicate_inline.cpp]
> > +!6 = metadata !{metadata !"0x15\00\000\000\000\000\000\000", null,
> null, null, metadata !7, null, null, null} ; [ DW_TAG_subroutine_type ]
> [line 0, size 0, align 0, offset 0] [from ]
> > +!7 = metadata !{null, metadata !8, metadata !8}
> > +!8 = metadata !{metadata !"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 = metadata !{metadata
> !"0x2e\00f2\00f2\00_Z2f2i\002\000\001\000\000\00256\000\002", metadata !1,
> metadata !5, metadata !10, null, null, null, null, metadata !2} ; [
> DW_TAG_subprogram ] [line 2] [def] [f2]
> > +!10 = metadata !{metadata !"0x15\00\000\000\000\000\000\000", null,
> null, null, metadata !11, null, null, null} ; [ DW_TAG_subroutine_type ]
> [line 0, size 0, align 0, offset 0] [from ]
> > +!11 = metadata !{null, metadata !8}
> > +!12 = metadata !{i32 2, metadata !"Dwarf Version", i32 4}
> > +!13 = metadata !{i32 2, metadata !"Debug Info Version", i32 2}
> > +!14 = metadata !{metadata !"clang version 3.6.0 "}
> > +!15 = metadata !{metadata !"0x101\00x\0016777224\000", metadata !4,
> metadata !5, metadata !8} ; [ DW_TAG_arg_variable ] [x] [line 8]
> > +!16 = metadata !{metadata !"0x102"} ; [ DW_TAG_expression
> ]
> > +!17 = metadata !{i32 8, i32 13, metadata !4, null}
> > +!18 = metadata !{metadata !"0x101\00y\0033554440\000", metadata !4,
> metadata !5, metadata !8} ; [ DW_TAG_arg_variable ] [y] [line 8]
> > +!19 = metadata !{i32 8, i32 20, metadata !4, null}
> > +!20 = metadata !{i32 8, i32 25, metadata !4, null}
> > +!21 = metadata !{metadata !"0x101\00i\0016777218\000", metadata !9,
> metadata !5, metadata !8} ; [ DW_TAG_arg_variable ] [i] [line 2]
> > +!22 = metadata !{i32 2, i32 51, metadata !9, metadata !20}
> > +!23 = metadata !{i32 2, i32 56, metadata !9, metadata !20}
> > +!24 = metadata !{i32 8, i32 36, metadata !4, null}
> >
> >
> > _______________________________________________
> > llvm-commits mailing list
> > llvm-commits at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141031/e6e6035a/attachment.html>
More information about the llvm-commits
mailing list