[cfe-commits] r157970 - in /cfe/trunk: lib/CodeGen/CGDebugInfo.cpp test/CodeGenCXX/debug-info-template-member.cpp test/CodeGenCXX/debug-info-user-def.cpp test/CodeGenCXX/debug-lambda-expressions.cpp

[Nico Weber]

On Tue, Jun 5, 2012 at 11:01 AM, Eric Christopher <echristo at apple.com>wrote:

>
> On Jun 5, 2012, at 12:28 AM, Eric Christopher <echristo at apple.com> wrote:
>
> >
> > On Jun 4, 2012, at 11:12 PM, John McCall wrote:
> >
> >> On Jun 4, 2012, at 9:45 PM, John McCall wrote:
> >>> On Jun 4, 2012, at 5:15 PM, Eric Christopher wrote:
> >>>> Author: echristo
> >>>> Date: Mon Jun  4 19:15:06 2012
> >>>> New Revision: 157970
> >>>>
> >>>> URL: http://llvm.org/viewvc/llvm-project?rev=157970&view=rev
> >>>> Log:
> >>>> Only emit debug information for methods that are user defined, there's
> >>>> not much reason to emit for constructors and destructors that aren't
> >>>> user defined.
> >>>>
> >>>> rdar://11593099
> >>>
> >>> Eric, it looks like this is breaking one of the clang debug info tests:
> >>> Clang :: debuginfo-tests___forward-declare-class.cpp
> >>>
> >>>
> http://smooshlab.apple.com:8013/builders/clang-x86_64-darwin10-nobootstrap-RA/builds/633
> >>
> >> I reverted in r157983.
> >
> > Weird, thanks. I was trying to check the bots this evening but they'd
> frozen for a while. I'll check why this failed in a bit.
>
> Awesome gdb bug:
>
> 0x00000032:     TAG_class_type [4] *
>                 AT_name( "A" )
>                 AT_byte_size( 0x04 )
>                 AT_decl_file(
> "/Volumes/Data/sources/llvm/tools/clang/test/debuginfo/forward-declare-class.cpp"
> )
>                 AT_decl_line( 18 )
>
> 0x0000003a:         TAG_member [5]
>                     AT_name( "MyData" )
>                     AT_type( {0x0000002b} ( int ) )
>                     AT_decl_file(
> "/Volumes/Data/sources/llvm/tools/clang/test/debuginfo/forward-declare-class.cpp"
> )
>                     AT_decl_line( 20 )
>                     AT_data_member_location( +0 )
>                     AT_accessibility( DW_ACCESS_public )
>
> 0x00000049:         NULL
>
> yet gdb doesn't believe the tag:
>
> (gdb) ptype A
> type = struct A {
>    int MyData;
> }
>
> because there aren't any methods/constructors in there.
>
> lldb correctly identifies the type:
>
> (lldb) image lookup -t A
> 1 match found in
> /Volumes/Data/builds/build-llvm/tools/clang/test/debuginfo/Output/forward-declare-class.cpp.tmp.o:
> id = {0x00000032}, name = "A", byte-size = 4, decl =
> forward-declare-class.cpp:18, clang_type = "class A {
>    int MyData;
> }"
>
> I'll fix the test to work around this and recommit. Sorry for the noise.
>

Won't this break gdb for users then?


>
> -eric
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20120605/c1262f6f/attachment.html>