[lldb-dev] Heads up: LLDB is broken with LLVM at ToT
Zachary Turner
zturner at google.com
Tue Dec 9 14:50:29 PST 2014
Just an additional heads up that there may be some runtime issues with the
patch as well.
Running this command:
d:\src\llvm\build\ninja\bin\clang++.exe -std=c++11 -g -O0 -m32
-Id:/src/llvm/tools/lldb/test/make/../../include -fno-exceptions -include
d:/src/llvm/tools/lldb/test/make/uncaught_exception.h -D_HAS_EXCEPTIONS=0
-O0 -c -o main.o main.cpp
Results in this:
Stack dump:
0. Program arguments: d:\src\llvm\build\ninja\bin\clang++.exe -cc1
-triple i686-pc-windows-msvc -emit-obj -mrelax-all -disable-free
-main-file-name main.cpp -mrelocation-model static -mthread-model posix
-mdisable-fp-elim -fmath-errno -masm-verbose -mconstructor-aliases
-target-cpu pentium4 -g
-dwarf-column-info -coverage-file
d:\src\llvm\tools\lldb\test\functionalities\data-formatter\data-formatter-stl\libcxx\vbool\main.o
-resource-dir d:\src\llvm\build\ninja\bin\..\lib\clang\3.6.0 -include
d:/src/llvm/tools/lldb/test/make/uncaught_exception.h -D _HAS_EXCEPTIONS=0
-I d:/src/llvm/tools/ll
db/test/make/../../include -internal-isystem
d:\src\llvm\build\ninja\bin\..\lib\clang\3.6.0\include -internal-isystem
C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC\include
-internal-isystem C:\Program Files (x86)\Windows Kits\8.1\include\shared
-internal-isystem C:\Program Files (x86)\Windo
ws Kits\8.1\include\um -internal-isystem C:\Program Files (x86)\Windows
Kits\8.1\include\winrt -O0 -std=c++11 -fdeprecated-macro
-fdebug-compilation-dir
d:\src\llvm\tools\lldb\test\functionalities\data-formatter\data-formatter-stl\libcxx\vbool
-ferror-limit 19 -fmessage-length 300 -mstackrealign -fm
s-extensions -fms-compatibility -fms-compatibility-version=17.00
-fdelayed-template-parsing -fobjc-runtime=gcc -fdiagnostics-show-option
-fcolor-diagnostics -o main.o -x c++ main.cpp
1. <eof> parser at end of file
2. Per-file LLVM IR generation
0x578AAA3B (0x07C903C4 0x07C903E4 0x07C903E4 0x05D8E590), ??3 at YAXPAX@Z() +
0xAB bytes(s)
0x01242744 (0x07C903E4 0x07C903E4 0x05D8E5AC 0x01242B8A),
llvm::MDNode::operator delete() + 0x34 bytes(s),
d:\src\llvm\lib\ir\metadata.cpp, line 388 + 0x14 byte(s)
0x0124D367 (0x00000001 0x00000000 0xCCCCCCCC 0x07C903E4),
llvm::MDNodeFwdDecl::`scalar deleting destructor'() + 0x27 bytes(s)
0x01242B8A (0x07C903E4 0x05D8E788 0x05D8E7FC 0xCCCCCCCC),
llvm::MDNode::deleteTemporary() + 0x8A bytes(s),
d:\src\llvm\lib\ir\metadata.cpp, line 591 + 0x25 byte(s)
0x01148933 (0x06150E78 0x07C903E4 0x05D8E7E4 0x05D8E7FC),
llvm::DIDescriptor::replaceAllUsesWith() + 0x123 bytes(s),
d:\src\llvm\lib\ir\debuginfo.cpp, line 342 + 0x9 byte(s)
0x01C824B7 (0x05D8E848 0x05D8E7FC 0xCCCCCCCC 0xCCCCCCCC),
clang::CodeGen::CGDebugInfo::finalize() + 0x2A7 bytes(s),
d:\src\llvm\tools\clang\lib\codegen\cgdebuginfo.cpp, line 3390
0x01C112BB (0x0618BE90 0x05D8E848 0x01BEFB40 0x0617DDE0),
clang::CodeGen::CodeGenModule::Release() + 0x3CB bytes(s),
d:\src\llvm\tools\clang\lib\codegen\codegenmodule.cpp, line 415
0x01C0E576 (0x0617DDE0 0x05D8E8D0 0x05D8E854 0xCCCCCCCC), `anonymous
namespace'::CodeGeneratorImpl::HandleTranslationUnit() + 0x86 bytes(s),
d:\src\llvm\tools\clang\lib\codegen\modulebuilder.cpp, line 198
0x01BEFB40 (0x0617DDE0 0x05D8EAD0 0x05D8E8E0 0x061B8890),
clang::BackendConsumer::HandleTranslationUnit() + 0x60 bytes(s),
d:\src\llvm\tools\clang\lib\codegen\codegenaction.cpp, line 131 + 0x21
byte(s)
0x025B1717 (0x061A2460 0x00000000 0x00000000 0x05D8EAD8), clang::ParseAST()
+ 0x237 bytes(s), d:\src\llvm\tools\clang\lib\parse\parseast.cpp, line 154
+ 0x18 byte(s)
0x018BF641 (0x05D8EB54 0x05D8EAD8 0xCCCCCCCC 0xCCCCCCCC),
clang::ASTFrontendAction::ExecuteAction() + 0x101 bytes(s),
d:\src\llvm\tools\clang\lib\frontend\frontendaction.cpp, line 527 + 0x30
byte(s)
0x01BEC33F (0x05D8EB00 0xCCCCCCCC 0xCCCCCCCC 0xCCCCCCCC),
clang::CodeGenAction::ExecuteAction() + 0x3CF bytes(s),
d:\src\llvm\tools\clang\lib\codegen\codegenaction.cpp, line 729
0x018BF22E (0x05D8EBF8 0x00000000 0xCCCCCCCC 0xCCCCCCCC),
clang::FrontendAction::Execute() + 0x7E bytes(s),
d:\src\llvm\tools\clang\lib\frontend\frontendaction.cpp, line 428 + 0xF
byte(s)
0x0187A821 (0x06158390 0x05D8F100 0x00000000 0xCCCCCCCC),
clang::CompilerInstance::ExecuteAction() + 0x2A1 bytes(s),
d:\src\llvm\tools\clang\lib\frontend\compilerinstance.cpp, line 812
0x019CDDBE (0x06152A98 0x05D8FEC8 0xCCCCCCCC 0xCCCCCCCC),
clang::ExecuteCompilerInvocation() + 0x30E bytes(s),
d:\src\llvm\tools\clang\lib\frontendtool\executecompilerinvocation.cpp,
line 222 + 0x11 byte(s)
0x0041461F (0x05D8FAB0 0x0000003D 0x06155AD0 0x0035139D), cc1_main() +
0x2FF bytes(s), d:\src\llvm\tools\clang\tools\driver\cc1_main.cpp, line 110
+ 0xE byte(s)
0x003FFC18 (0x05D8FAA8 0x0000003F 0x06155AFC 0x00000000), ExecuteCC1Tool()
+ 0x78 bytes(s), d:\src\llvm\tools\clang\tools\driver\driver.cpp, line 369
+ 0x2B byte(s)
0x003FFFB5 (0x0000003F 0x061519E8 0x06145120 0x5DB515C0), main() + 0x2F5
bytes(s), d:\src\llvm\tools\clang\tools\driver\driver.cpp, line 415 + 0x33
byte(s)
0x034BDA89 (0x05D8FF2C 0x75CC919F 0x7EDD4000 0x05D8FF70),
__tmainCRTStartup() + 0x199 bytes(s),
f:\dd\vctools\crt\crtw32\dllstuff\crtexe.c, line 626 + 0x19 byte(s)
0x034BDBCD (0x7EDD4000 0x05D8FF70 0x77A40BBB 0x7EDD4000), mainCRTStartup()
+ 0xD bytes(s), f:\dd\vctools\crt\crtw32\dllstuff\crtexe.c, line 466
0x75CC919F (0x7EDD4000 0x2F016EAA 0x00000000 0x00000000),
BaseThreadInitThunk() + 0xE bytes(s)
0x77A40BBB (0xFFFFFFFF 0x77A2C9D1 0x00000000 0x00000000),
RtlInitializeExceptionChain() + 0x84 bytes(s)
0x77A40B91 (0x034BDBC0 0x7EDD4000 0x00000000 0x00000000),
RtlInitializeExceptionChain() + 0x5A bytes(s)
^C
d:\src\llvm\tools\lldb\test\functionalities\data-formatter\data-formatter-stl\libcxx\vbool>
On Tue Dec 09 2014 at 2:41:09 PM Duncan P. N. Exon Smith <
dexonsmith at apple.com> wrote:
> Awesome, thanks for the help!
>
> > On 2014 Dec 9, at 14:28, Zachary Turner <zturner at google.com> wrote:
> >
> > This fixes the issue. I will check in your patch. Thanks!
> >
> > On Tue Dec 09 2014 at 2:20:02 PM Zachary Turner <zturner at google.com>
> wrote:
> > Testing now. I notice a few of the changes aren't actually required in
> order to fix compiler errors. For example, this line:
> >
> > - ConstantInt *constant_int = dyn_cast<ConstantInt>(alloc_
> md->getOperand(0));
> > + ConstantInt *constant_int = mdconst::dyn_extract<
> ConstantInt>(alloc_md->getOperand(0));
> >
> > wasn't actually failing. Not sure if that's important, but I'm using
> your fixed version anyway.
> >
> > On Tue Dec 09 2014 at 2:11:08 PM Duncan P. N. Exon Smith <
> dexonsmith at apple.com> wrote:
> > Attaching a patch that should fix it. Can you try compiling?
> >
> >
> >
> >
> > > On 2014 Dec 9, at 14:03, Zachary Turner <zturner at google.com> wrote:
> > >
> > > Here's the 3 errors.
> > >
> > > ..\..\tools\lldb\source\Expression\IRForTarget.cpp(413) : error
> C2678: binary '!=' : no operator found which takes a left-hand operand of
> type 'const llvm::MDOperand' (or there is no acceptable conversion)
> > > d:\src\llvm\include\llvm/ADT/APSInt.h(291): could be 'bool
> llvm::operator !=(int64_t,const llvm::APSInt &)'
> > > d:\src\llvm\include\llvm/ADT/APInt.h(1724): or 'bool
> llvm::operator !=(uint64_t,const llvm::APInt &)'
> > > d:\src\llvm\include\llvm/ADT/StringRef.h(527): or 'bool
> llvm::operator !=(llvm::StringRef,llvm::StringRef)'
> > > while trying to match the argument list '(const
> llvm::MDOperand, const llvm::GlobalValue *)'
> > >
> > >
> > >
> > > ..\..\tools\lldb\source\Expression\IRForTarget.cpp(647) : error
> C2664: 'llvm::MDNode *llvm::MDNode::get(llvm::LLVMContext
> &,llvm::ArrayRef<llvm::Metadata *>)' : cannot convert argument 2 from
> 'llvm::ArrayRef<llvm::Value *>' to 'llvm::ArrayRef<llvm::Metadata *>'
> > > No user-defined-conversion operator available that can perform
> this conversion, or the operator cannot be called
> > >
> > >
> > > ..\..\tools\lldb\source\Expression\IRForTarget.cpp(1254) : error
> C2664: 'llvm::MDNode *llvm::MDNode::get(llvm::LLVMContext
> &,llvm::ArrayRef<llvm::Metadata *>)' : cannot convert argument 2 from
> 'llvm::ArrayRef<llvm::Value *>' to 'llvm::ArrayRef<llvm::Metadata *>'
> > > No user-defined-conversion operator available that can perform
> this conversion, or the operator cannot be called
> > >
> > >
> > > On Tue Dec 09 2014 at 1:56:03 PM Zachary Turner <zturner at google.com>
> wrote:
> > > I'm not sure what the status of building LLDB with CMake is, but the
> normal way apple people build LLDB is to structure the directory like this:
> > >
> > > lldb
> > > --llvm
> > > ----tools
> > > ------clang
> > >
> > > Then open the xcode workspace and build the lldb-tool target.
> > >
> > > I'll sync to ToT again though so that I can get you a dump of the
> compiler errors I'm seeing.
> > >
> > > On Tue Dec 09 2014 at 1:52:40 PM Duncan P. N. Exon Smith <
> dexonsmith at apple.com> wrote:
> > > I'm trying to build lldb now, but it's hung trying to build LLVM and
> > > clang... but I already built these? I don't really understand.
> > >
> > > Is there any way to make lldb point to my LLVM checkout? I just want
> to
> > > see the compiler errors so I can fix them :(.
> > >
> > > Alternatively, if some can email me a compile log with all the errors
> > > (can you do a `make -k` or something?) I can send you a patch... this
> is
> > > mechanical to update, but I need the compiler to tell me what's wrong.
> > >
> > > I expect the fixes to all be along these lines:
> > >
> > > - Value *Ops[] => Metadata *Ops[]
> > >
> > > - cast<ConstantInt>(N->getOperand(0)) =>
> > > mdconst::extract<ConstantInt>(N->getOperand(0))
> > >
> > > - Value *Op = ConstantInt::get(...) =>
> > > Metadata *Op = ConstantAsMetadata::get(ConstantInt::get(...))
> > >
> > > Sorry for the breakage :(.
> > >
> > > > On 2014 Dec 9, at 13:41, Zachary Turner <zturner at google.com> wrote:
> > > >
> > > > If you sync llvm past git revision db7b69e3a634c5fdff0eceeee2a41ee49c3270a2
> (svn r223802), your LLDB build will be broken.
> > > >
> > > > I was able to fix one of the errors, but this isn't really my area
> of expertise, so I stopped when I realized I didn't know what I was doing.
> > > >
> > > > Just a warning not to sync LLVM until this is fixed. I've CC'ed
> Sean and Duncan as they are probably the best people to fix this, but if
> anyone else knows how to fix it, feel free.
> > >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-dev/attachments/20141209/f857da15/attachment.html>
More information about the lldb-dev
mailing list