[LLVM] Uninitialized member in copy constructor in DataLayout.h

[David Blaikie]

On Tue, Feb 26, 2013 at 8:37 PM, Peng Cheng <gm4cheng at gmail.com> wrote:
> The usual llvm regression tests for a standard constructor bug seems like an
> overkill, leading to many wasteful code exercises, such as ir reading,
> optimization, code generation, ...

While I'm the last to advocate for indirect testing, this is generally
the attitude in the LLVM project - to exercise from the actual
user-programs. This ensures better coverage (though not necessarily
actually /testing/ all of that, of course) in some ways, worse
coverage in others.

> Are there any other places to place the unit tests for classes in llvm code
> base?  Simply to test whether all the necessary fields are initialized after
> the copy constructor.

We do have true API-level unit tests as well - just not often written,
usually only for core/generic data structures & the like.

>
>
> On Tue, Feb 26, 2013 at 10:51 PM, David Blaikie <dblaikie at gmail.com> wrote:
>>
>> On Tue, Feb 26, 2013 at 7:34 PM, Peng Cheng <gm4cheng at gmail.com> wrote:
>> > It did not fail on any valgrind bots.  But when I used valgrind to check
>> > my
>> > jit engine based on llvm, I found the bug.
>> >
>> > Could you point to me an example of test case for the valgrind bot?  I
>> > need
>> > to see how the test infrastructure is setup in llvm.
>>
>> The current public valgrind bot (
>> http://lab.llvm.org:8011/builders/clang-x86_64-linux-vg - there may be
>> others) just runs the usual LLVM/Clang regression tests. So if you can
>> construct a nice small/simple regression test you can add that (&
>> limit the test to only running under valgrind, potentially)
>>
>> > On Tue, Feb 26, 2013 at 10:28 PM, David Blaikie <dblaikie at gmail.com>
>> > wrote:
>> >>
>> >> On Tue, Feb 26, 2013 at 7:15 PM, Peng Cheng <gm4cheng at gmail.com> wrote:
>> >> > The copy constructor of class DataLayout fails to initialize
>> >> > StackNaturalAlign.  It could lead to conditional jump on
>> >> > uninitialized
>> >> > variables in member function exceedsNaturalStackAlignment.
>> >>
>> >> Does this fail any valgrind bots? Could you write a test case that
>> >> would fail under valgrind?
>> >>
>> >> > See fix in the patch attached.
>> >> >
>> >> > _______________________________________________
>> >> > llvm-commits mailing list
>> >> > llvm-commits at cs.uiuc.edu
>> >> > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>> >> >
>> >
>> >
>
>