1137610 - Differential Testing: Different output message involving startgc

Status: RESOLVED INVALID

(Core :: JavaScript: GC, defect)

Description

[Gary Kwong [:gkw] [:nth10sd] (NOT official MoCo now)]

try {
    (function() {
        x.valueOf = (function() {});
    })()
} catch (e) {}
try {
    startgc();
    print("FOO");
} catch (e) {}


$ ./js-dbg-64-dm-nsprBuild-darwin-0a8b3b67715a --fuzzing-safe --no-threads --ion-eager testcase.js
FOO

$ ./js-dbg-64-dm-nsprBuild-darwin-0a8b3b67715a --fuzzing-safe --no-threads --ion-eager --gc-zeal=8 testcase.js
$

Tested this on m-c rev 0a8b3b67715a.

My configure flags are:

CC="clang -Qunused-arguments" CXX="clang++ -Qunused-arguments" AR=ar AUTOCONF=/usr/local/Cellar/autoconf213/2.13/bin/autoconf213 sh ~/trees/mozilla-central/js/src/configure --target=x86_64-apple-darwin12.5.0 --enable-debug --enable-nspr-build --enable-more-deterministic --with-ccache --enable-gczeal --enable-debug-symbols --disable-tests

python -u ~/fuzzing/js/compileShell.py -b "--enable-debug --enable-more-deterministic --enable-nspr-build -R ~/trees/mozilla-central" -r 0a8b3b67715a

autoBisect shows this is probably related to the following changeset:

The first bad revision is:
changeset:   https://hg.mozilla.org/mozilla-central/rev/2c46028281dd
user:        Jon Coppeard
date:        Mon Jan 12 10:29:38 2015 +0000
summary:     Bug 1119759 - Add gcstart() test function to start an incremental GC r=terrence

Jon, is bug 1119759 a likely regressor?

Comment 1

[Gary Kwong [:gkw] [:nth10sd] (NOT official MoCo now)]

Not a SM issue, rather, an issue with the fuzzer. -> INVALID

Comment 2

[Jesse Ruderman]

I fixed the problem in the fuzzer.

Well, I hope I did. I said the same thing last time, and I was wrong.

changeset:   160d7f36a1d5
user:        Jesse Ruderman <[e-mail]>
date:        Sat Feb 28 12:27:58 2015 -0800
summary:     Bug 1137610 - Fix another differential-testing issue with eval. Turns out that when something throws, eval returns the value of the previous expression-statement.

Comment 3

[Jon Coppeard (:jonco)]

The problem is that startgc() will fail if a GC is already running, and running with GC zeal will cause that to happen.  So yes I'd say this expected behaviour.