I was able to track this down to a bug in Clang. I have a patch up for review.
http://reviews.llvm.org/D13276 I'm not sure what part of the TypeSystem refactor caused this to finally be exposed, but in any case I expect that patch will go through and this bug will be fixed. On Tue, Sep 29, 2015 at 3:40 PM Zachary Turner <ztur...@google.com> wrote: > Resending since the wrong list was CC'ed. > > > On Tue, Sep 29, 2015 at 3:39 PM Zachary Turner <ztur...@google.com> wrote: > >> Hi Greg. I'm responding on this thread since you said lldb-dev is a >> better place for questions. >> >> I had a chance to look at this some more. It looks like the way we're >> constructing the clang::RecordDecl is incorrect on Windows. I'm not sure >> what record layout is like on non-Windows platforms, but if we have this >> code: >> >> class A { >> virtual ~A() {} >> int a; >> }; >> >> class B : public A { >> virtual ~B() {} >> int b; >> }; >> >> A * pa = new B (); >> >> then using the MS ABI, offset of a is 4 and offset of B is 8. I verified >> this by modifying the source of sbvalue-cast.cpp to print the offsets. >> >> d:\src\llvm\tools\lldb\test\lang\cpp\dynamic-value>a.out >> DerivedA::ctor()-> >> m_base_val=20 >> m_a_val=10 >> *offset of DerivedA::m_a_val is 8* >> *offset of DerivedA::m_base_val is 4* >> >> Clang agrees with me as well. Compiling with -Xclang -emit-llvm-only >> -Xclang -fdump-record-layouts gives me the following AST Record Layout >> >> *** Dumping AST Record Layout >> 0 | class DerivedA >> 0 | class Base (primary base) >> 0 | (Base vftable pointer) >> * 4 | int m_base_val* >> 8 | int m_a_val >> | [sizeof=12, align=4 >> | nvsize=12, nvalign=4] >> >> On the LLDB side, we have ClangASTContext::GetChildCompilerTypeAtIndex >> which computes the byte offset, field name, etc for the ValueObject. On >> line 5754 it begins querying the name and byte offset of the field. The >> name returns `m_base_val`, but the field_offset returns 8 bytes, which we >> can see above is wrong. >> >> Any suggestions where I might look to dig further? >> >> On Wed, Sep 23, 2015 at 10:34 AM Greg Clayton <clayb...@gmail.com> wrote: >> >>> I responded in another e-mail with follow stuff you will need to send >>> me. Always send questions to lldb-dev and not lldb-commits as there is so >>> much traffic in lldb-commmits it might get lost. I keep up with lldb-dev >>> daily. >>> >>> > On Sep 22, 2015, at 10:12 AM, Zachary Turner <ztur...@google.com> >>> wrote: >>> > >>> > In regards to this CL: >>> http://llvm.org/viewvc/llvm-project?rev=247953&view=rev >>> > >>> > It has broken TestCppValueCast on Windows and fixed TestCxxWcharT on >>> Windows as well. Although these appear to be platform specific behavioral >>> changes, as I don't see anyone else reporting failures or fixes, the >>> description of the CL leads me to infer that this change was supposed to >>> have no functional change. I haven't gotten any response pinging the >>> thread so I'll pose it to the wider list for visibility in case someone can >>> shed some light on this. In particular, is this supposed to come with a >>> functional change, and if so what? >>> > >>> > Unfortunately I don't have buildbots running tests on Windows yet, so >>> I can't justify reverting the CL (and it's too old to revert now anyway >>> since other things depend on it), but it would greatly appreciate if >>> someone can take a look at this and see if they can identify the source of >>> the behavioral change >>> > >>> > I do plan to have a buildbot running tests within a week or two, so >>> hopefully we can catch this kind of thing much sooner next time. >>> >>>
_______________________________________________ lldb-dev mailing list lldb-dev@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-dev
