Re: [PATCH] Asan static optimization (draft)

[Konstantin Serebryany <konstantin dot s dot serebryany at gmail dot com>]

On Thu, Aug 14, 2014 at 11:55 PM, Yuri Gribov <tetra2005@gmail.com> wrote:
> On Thu, Aug 14, 2014 at 8:53 PM, Konstantin Serebryany
> <konstantin.s.serebryany@gmail.com> wrote:
>> In order for your work to be generally useful, I'd ask several things:
>> - Update https://code.google.com/p/address-sanitizer/wiki/CompileTimeOptimizations
>> with examples that will be handled
>
> Done (to be honest I only plan to do full redundancy elimination for
> now, hoisting would hopefully follow later). Note I'm still
> experimenting so there may be some changes in actual implementation.

Thanks.
if we are running -O0, we should not care about optimizing asan
instrumentation.
If this is -O1 or higher,  then most (but not all) of your cases
*should* be optimized by the compiler before asan kicks in.
(This may be different for GCC-asan because GCC-asan runs a bit too
early, AFAICT. Maybe this *is* the problem we need to fix).
If there is a case where the regular optimizer can optimize the code
before asan but it doesn't --
we should fix the general optimizer or the phase ordering instead of
enhancing asan opt phase.
I am mainly interested in cases where the general optimizer can not
possibly improve the code,
but asan opt can eliminate redundant checks.


>
>> - Create small standalone test cases in C
>> - Don't put the tests under GPL (otherwise we'll not be able to reuse
>> them in LLVM)
>
> I already have a bunch of tests (which I plan to extend further). How
> should I submit them s.t. they could be reused by LLVM?

Maybe just start accumulating tests on the CompileTimeOptimizations wiki
(as full functions that one can copy-paste to a .c file and build)?
Once some new optimization is implemented in a compiler X,
we'll copy the test with proper harness code (FileCheck/dejagnu/etc)
to the X's repository

>

>>>> make sure that sanopt performs conservative optimization
>> Yes. I don't know a good solution to this problem, which is why we did
>> not attack it before.
>> Increasing asan speed will save lots of CPU time, but missing a single software
>> bug due to an overly aggressive optimization may cost much more.
>
> Yeah. I thought about manually inspecting optimizations that are
> performed for some large files (e.g. GCC's asan.c) and maybe doing
> some  random verifications of Asan trophies
> (http://code.google.com/p/address-sanitizer/wiki/FoundBugs). Ideally
> we'd have some test generator but making a reasonable one for C sounds
> laughable. Perhaps there is some prior work on verification of Java
> range checks optimizers?

There is ton of work for range check optimizers, but none of that
fully applies to asan,
since asan also checks use-after-free and init-order-fiasco.


>
> -Y