The following testcase crashes on mozilla-central revision b8141448e0ba (build with --enable-debug, run with --fuzzing-safe --no-blinterp):

quit();

Backtrace:

#0  js::jit::Compile (cx=<optimized out>, script=..., osrFrame=<optimized out>, osrPc=0x0, forceRecompile=<optimized out>) at js/src/jit/Ion.cpp:2116
#1  0x000055e1db25c159 in js::jit::CanEnterIon (cx=0x7f1148d19000, state=...) at js/src/jit/Ion.cpp:2268
#2  0x000055e1db305f64 in js::jit::MaybeEnterJit (cx=0x7f1148d19000, state=...) at js/src/jit/Jit.cpp:156
#3  0x000055e1da650931 in js::RunScript (cx=0x7f1148d19000, state=...) at js/src/vm/Interpreter.cpp:410
#4  0x000055e1da666d70 in js::ExecuteKernel (cx=0x7f1148d19000, script=..., envChainArg=..., newTargetValue=..., evalInFrame=..., result=0x7fff161972b8) at js/src/vm/Interpreter.cpp:787
/snip

For detailed crash information, see attachment.

Setting [fuzzblocker] due to the simplicity of the testcase.

autobisectjs shows this is probably related to the following changeset:

The first bad revision is:
changeset: https://hg.mozilla.org/mozilla-central/rev/56b33927fd49
user: Jan de Mooij
date: Sat Jul 20 08:56:36 2019 +0000
summary: Bug 1566332 part 4 - Make IsBaselineJitEnabled imply IsBaselineInterpreterEnabled. r=tcampbell

Jan, is bug 1566332 a likely regressor?

Also moves these functions to JitOptions.h because Ion.h cannot include BaselineJIT.h
due to include dependencies. Ion.h should be split up eventually but JitOptions.h isn't
unreasonable for now.

https://hg.mozilla.org/mozilla-central/rev/78a0b312e621