Experimental integration with binaryen

[nth10sd]

Here's a first take on the integration with binaryen for experimental deployment.

It uses stdout output from jsfunfuzz, e.g. from w1-out.txt to act as the seed for binaryen. If an issue (e.g. a crash) is found, the wrapper and wasm files are zipped up to be submitted to FuzzManager.

I ran some crashing testcases in another branch to see that this experimentally seems to work.

Caveats:

• Due to a bug in binaryen, the version is restricted to binaryen v52 for now
• Note that after a bug involving a wasm testcase is found, pinpoint and hence autobisectjs auto-integration do not understand multiple testcase input, e.g. ./js wrapper.js wasm.file, so all of such reports to FuzzManager will have quality 10 for now.

I would like to get this in quickly to see if:

• wasm bugs can be reported via the harness
• see if this tweaks our coverage story.

@jschwartzentruber just asking for a cursory glance here, how does this sound?

[codecov-io]

Codecov Report

Merging #219 into master will increase coverage by 1.93%.
The diff coverage is 55.25%.

@@            Coverage Diff             @@
##           master     #219      +/-   ##
==========================================
+ Coverage   44.51%   46.45%   +1.93%     
==========================================
  Files          39       43       +4     
  Lines        2736     2878     +142     
==========================================
+ Hits         1218     1337     +119     
- Misses       1518     1541      +23

Impacted Files	Coverage Δ
src/funfuzz/js/js_interesting.py	17.85% <0%> (+0.38%)	⬆️
tests/js/test_with_binaryen.py	100% <100%> (ø)
src/funfuzz/util/file_system_helpers.py	100% <100%> (ø)
tests/util/test_file_system_helpers.py	100% <100%> (ø)
src/funfuzz/js/compare_jit.py	15.84% <14.28%> (+0.46%)	⬆️
src/funfuzz/js/loop.py	15.78% <6.59%> (-0.04%)	⬇️
src/funfuzz/js/with_binaryen.py	82.75% <82.75%> (ø)
... and 1 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 240a346...29244f7. Read the comment docs.

[nth10sd]

==> forkjoin-10-out.txt <==
/tmp/tmp8lwivi5iloop10/w1-cj-initial-r4 |   0.2s | 0 | fine
/tmp/tmp8lwivi5iloop10/w1-cj-initial-r5 |   0.2s | 0 | fine
/tmp/tmp8lwivi5iloop10/w1-cj-initial-r6 |   0.2s | 0 | fine
/tmp/tmp8lwivi5iloop10/w1-cj-initial-r7 |   0.2s | 0 | fine
/tmp/tmp8lwivi5iloop10/w1-cj-initial-r8 |   0.2s | 0 | fine
/tmp/tmp8lwivi5iloop10/w1-cj-initial-r9 |   0.2s | 0 | fine
/tmp/tmp8lwivi5iloop10/w2 |  17.8s | 0 | fine
/tmp/tmp8lwivi5iloop10/w2-wasm |   0.3s | 0 | fine
/tmp/tmp8lwivi5iloop10/w2-cj-initial-r0 |   0.4s | 0 | fine
/tmp/tmp8lwivi5iloop10/w2-cj-initial-r1 |   0.6s | 0 | fine

==> forkjoin-11-out.txt <==
/tmp/tmpmuxmvk7jloop11/w1-cj-initial-r4 |   0.2s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w1-cj-initial-r5 |   0.2s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w1-cj-initial-r6 |   0.2s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w1-cj-initial-r7 |   0.2s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w1-cj-initial-r8 |   0.6s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w1-cj-initial-r9 |   0.6s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w2 |  13.5s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w2-wasm |   0.3s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w2-cj-initial-r0 |   0.3s | 0 | fine
/tmp/tmpmuxmvk7jloop11/w2-cj-initial-r1 |   3.7s | 0 | fine

Sample running logs.

I'll just land this as it is merely an if statement that activates the integration once tests pass.

[nth10sd]

I'll land this first. The experiment has already proved a success with the discovery of bug 1516720.