What sanitizers preferred?

Joel Sherrill joel at rtems.org
Mon Mar 16 16:55:25 UTC 2020


On Mon, Mar 16, 2020 at 11:05 AM Sebastian Huber <
sebastian.huber at embedded-brains.de> wrote:

> On 15/03/2020 17:52, suyash singh wrote:
>
> Hello,
> Out of
> AddressSanitizer <http://clang.llvm.org/docs/AddressSanitizer.html>,
> ThreadSanitizer <http://clang.llvm.org/docs/ThreadSanitizer.html>,
> MemorySanitizer <http://clang.llvm.org/docs/MemorySanitizer.html>, and
> DataFlowSanitizer <http://clang.llvm.org/docs/DataFlowSanitizer.html>.
>
> Are all of them useful as RTEMS-tools to integrate? Any other sanitizers
> suggested?
>
> Asking as part of my proposed gsoc project,
>
> https://docs.google.com/document/d/1OerM8Iix7PbDUCyb_J_KEzcMgLbvdqxd4FnK_BqI7XI/edit?usp=sharing
>
> I am not sure of these sanitizers can be used in RTEMS at all. We don't
> have virtual memory. Firstly, it would be necessary to check for each
> sanitizer if it uses virtual memory. If yes, then is this absolutely
> necessary or just convenient? Also the sanitizers are not available on all
> LLVM architectures. We have to evaluate if the architectures-specific
> adoptions can be used in RTEMS.
>

+1

Another area is the compiler stack protection techniques. I have no idea if
those are applicable to RTEMS but they would be useful if they can work.

Ignoring other constraints from embedded systems, any possible technique to
use with RTEMS must be evaluated against the constraints imposed by having
a single address space, no VM, and single process model.

You need to check if any of these will work before this project has a
chance.

I would be willing to entertain a project for an appropriate solution that
is limited to say ARM, x86, and RISC-V.

--joel


> _______________________________________________
> devel mailing list
> devel at rtems.org
> http://lists.rtems.org/mailman/listinfo/devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20200316/0c020219/attachment.html>


More information about the devel mailing list