readline = function() {}; // Adapted from randomly chosen test: js/src/jit-test/tests/basic/testGuardCalleeSneakAttack2.js Function.prototype.toString = function() { for (var i = 0; i < 2; i++) { this() } }; oomTest(getBacktrace({ thisprops: true })) asserts js debug shell on m-c changeset d53a52b39a95 with --fuzzing-safe --no-threads --ion-eager at Assertion failure: phaseNestingDepth == 0, at gc/Statistics.cpp Configure options: CC="clang -Qunused-arguments" CXX="clang++ -Qunused-arguments" AR=ar AUTOCONF=/usr/local/Cellar/autoconf213/2.13/bin/autoconf213 sh /Users/skywalker/trees/mozilla-central/js/src/configure --target=x86_64-apple-darwin12.5.0 --enable-debug --disable-threadsafe --enable-more-deterministic --with-ccache --enable-gczeal --enable-debug-symbols --disable-tests python -u ~/funfuzz/js/compileShell.py -b "--enable-debug --enable-more-deterministic" -r d53a52b39a95 autoBisect shows this is probably related to the following changeset: The first bad revision is: changeset: https://hg.mozilla.org/mozilla-central/rev/9c365490d4ce user: Jon Coppeard date: Tue Oct 13 13:37:07 2015 +0100 summary: Bug 1212469 - Make oomTest() into a shell function r=nbp Jon, is bug 1212469 a likely regressor?

(lldb) bt 5 * thread #1: tid = 0x1f0ec4, 0x00000001008e40c4 js-dbg-64-dm-darwin-d53a52b39a95`js::gcstats::Statistics::startTimingMutator(this=<unavailable>) + 132 at Statistics.cpp:1024, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x0) * frame #0: 0x00000001008e40c4 js-dbg-64-dm-darwin-d53a52b39a95`js::gcstats::Statistics::startTimingMutator(this=<unavailable>) + 132 at Statistics.cpp:1024 frame #1: 0x000000010000dbc2 js-dbg-64-dm-darwin-d53a52b39a95`StartTimingMutator(cx=<unavailable>, argc=<unavailable>, vp=0x00007fff5fbfcdc8) + 98 at js.cpp:1627 frame #2: 0x0000000100676cb2 js-dbg-64-dm-darwin-d53a52b39a95`js::Invoke(JSContext*, JS::CallArgs const&, js::MaybeConstruct) [inlined] js::CallJSNative(cx=0x0000000102c45400, native=(js-dbg-64-dm-darwin-d53a52b39a95`StartTimingMutator(JSContext*, unsigned int, JS::Value*) at js.cpp:1619))(JSContext*, unsigned int, JS::Value*), JS::CallArgs const&) + 174 at jscntxtinlines.h:235 frame #3: 0x0000000100676c04 js-dbg-64-dm-darwin-d53a52b39a95`js::Invoke(cx=0x0000000102c45400, args=0x00007fff5fbfcd70, construct=<unavailable>) + 596 at Interpreter.cpp:489 frame #4: 0x000000010069c4fd js-dbg-64-dm-darwin-d53a52b39a95`js::Invoke(cx=0x0000000102c45400, thisv=0x00007fff5fbfcf58, fval=<unavailable>, argc=<unavailable>, argv=<unavailable>, rval=<unavailable>) + 669 at Interpreter.cpp:542 (lldb)

Not related to bug 1212469. I can remove the call to oomTest() and still get the failure. BTW making Function.prototype.toString call |this()| is really evil...

Patch to make Statistics::startTimingMutator() return false if called at the wrong time rather than asserting.

Comment on attachment 8679941 [details] [diff] [review] bug1218900-mutator-timing Review of attachment 8679941 [details] [diff] [review]: ----------------------------------------------------------------- ::: js/src/gc/Statistics.cpp @@ +1021,5 @@ > Statistics::startTimingMutator() > { > + // Should only be called from outside of GC. > + if (phaseNestingDepth != 0) > + return false; This bug shows that the comment is incorrect. The intention was to prevent mistakes where somehow startTimingMutator would be called from within a GC, which would mess up the phase nesting. But it's additionally guarding against nested invocations of itself. As in, this script shows the same crash: startTimingMutator(); startTimingMutator(); So I think the right fix would be to change this to if (phaseNestingDepth != 0) { // Should only be called from outside of GC. MOZ_ASSERT(phaseNestingDepth == 1); MOZ_ASSERT(phaseNesting[0] == PHASE_MUTATOR); return false; }

autoBisect shows this is probably related to the following changeset: The first bad revision is: changeset: https://hg.mozilla.org/mozilla-central/rev/b261745c586a user: Steve Fink date: Tue Nov 18 11:26:11 2014 -0800 summary: Bug 1088831 - Count storebuffer overflows, account for minor GCs, and implement timed regions, r=jonco Yes, bug 1088831 does seem to be the regressor. (I bisected using a testcase without oomTest)

removing the b2g 2.5 flag since this commit has been reverted due to an incorrect merge, sorry for the confusion