[cfe-dev] Static analysis of Hyperion (llvm/clang)

[Maarten Hoes via cfe-dev]

Hi,


On Thu, Nov 12, 2015 at 10:26 PM, Anna Zaks <ganna at apple.com> wrote:
>
>> On Nov 12, 2015, at 1:17 PM, Eric Christopher <echristo at gmail.com> wrote:
>>
>> Only sort-of :)
>>
>> We would be interested in hearing about your results as you go through
them!
>
>
> Yes!
>


Well in that case, here we go...

Im not really a developer myself, but I *am* a user of that codebase and as
I like the code I use to be as bug-free as possible, I was hoping to
trigger the developers into looking at their code more closely. ;-)

Their first impression seem to be - and I guess this doesnt come as a
surprise - that there are false positives, but also some real issues that
need to be looked at. Also, there seems to be a template in use that has a
variable that doesnt always get used/referenced, resulting in a spurious
assignment. The reason it was written this way seems to be that it was
assumed that the compiler would optimize the assignment out. So although
those reports *are* valid, the code is not likely to get changed I guess.
(please forgive me for perhaps not making sense in the above, Im merely
trying to summarize the responses on the right mailing list)

What I *have* personal experience with though, is that I ran into this bug
during ./configure :
https://llvm.org/bugs/show_bug.cgi?id=25502
Perhaps that can be looked at ?


- Maarten
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20151112/4f60ed8e/attachment.html>