[llvm-dev] -sanitizer-coverage-prune-blocks=true and LibFuzzer

[Kostya Serebryany via llvm-dev]

Exciting!

(btw, I'd prefer libfuzzer at googlegroups.com for such discussions, please
start new topics there)

I can reproduce this too, but if i either increase FUZZER_TESTING_SECONDS
to 600 or change seed=1 to seed=2 the problem is gone.
Looks like one of the binaries got simply unlucky with a particular seed.
You can observe it like this:

for S in 1 2 3 4 5 6; do ./target-asan-8bit-prune-build/fuzzer
-seed=$S -runs=10000000 2>&1 | grep DONE &  done

#10000000       DONE   cov: 60 bits: 91 indir: 1 units: 59 exec/s: 625000
#10000000       DONE   cov: 60 bits: 91 indir: 1 units: 57 exec/s: 588235
#10000000       DONE   cov: 253 bits: 901 indir: 12 units: 467 exec/s: 526315
#10000000       DONE   cov: 63 bits: 95 indir: 1 units: 64 exec/s: 476190
#10000000       DONE   cov: 252 bits: 923 indir: 12 units: 491 exec/s: 454545
#10000000       DONE   cov: 253 bits: 880 indir: 12 units: 471 exec/s: 384615


Similar things happen with other binaries:

for S in 1 2 3 4 5 6; do ./target-asan-8bit-nopru-build/fuzzer
-seed=$S -runs=10000000 2>&1 | grep DONE &  done

#10000000       DONE   cov: 103 bits: 190 indir: 1 units: 62 exec/s: 526315
#10000000       DONE   cov: 443 bits: 1730 indir: 12 units: 529 exec/s: 357142
#10000000       DONE   cov: 443 bits: 1695 indir: 12 units: 509 exec/s: 344827
#10000000       DONE   cov: 443 bits: 1682 indir: 12 units: 500 exec/s: 333333
#10000000       DONE   cov: 444 bits: 1675 indir: 12 units: 501 exec/s: 277777
#10000000       DONE   cov: 401 bits: 1443 indir: 12 units: 341 exec/s: 263157


I've also tried building with trace-pc-guard (the new thing) and results
are similar.

name    cov     bits    execs   execs_per_sec   units   actual_cov
 actual_bits
asan-8bit-nopru 401     1443    19790806        324439  340     401     1441
asan-8bit-prune 256     897     26528866        434899  485     447     1651
asan-edge-nopru 447     0       35589496        583434  137     447     719
asan-edge-prune 256     0       37576436        616007  137     447     719
asan-trac-nopru 401     1443    12566606        206009  340     401     1441
asan-trac-prune 256     891     16295346        267136  480     447     1640


Conclusions:
* testing a fuzzing engine is not trivial :(
* testing it on a very short run with a single seed may be misleading


BTW, I am also looking into more automation of libFuzzer testing.
With trace-pc-guard we now have libFuzzer's flag -print_coverage=1 that
will print all the covered lines.
My hope is that this feature can be used for more detailed analysis of
coverage differences.

--kcc


On Wed, Sep 21, 2016 at 6:00 AM, Jonas Wagner <jonas.wagner at epfl.ch> wrote:

> Hello,
>
> Is this reproducible?
>> Fuzzing is a probabilistic business and one or even two runs don't prove
>> much.
>>
>
> I've reproduced the behavior on two different machines. Attached is a
> script to do so. To use the script,
>
> - create an empty folder and copy both prune-blocks.sh and
> ff-http-parser.sh in there
> - ensure clang and clang++ are in your $PATH
> - cd /path/to/prune-blocks.sh
> - ./prune-blocks.sh
>
> Let me know how it goes.
>
>
>
>> Note that I am going to change all of these coverage options soon.
>> The new thing will be http://clang.llvm.org/docs/SanitizerCoverage.html#
>> tracing-pcs-with-guards
>> It will replace regular (boolean) and 8-bit-counters coverage.
>>
>
> Yay, sounds exciting! I've done a couple experiments to measure the
> performance and effect of the different coverage options in the recent
> past. If you're interested, I'd be happy to discuss off-list; simply send
> me an email.
>
> Best,
> Jonas
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20160921/acaffbd1/attachment.html>