Open Bug 902690 Opened 11 years ago Updated 2 years ago

IonMonkey: Fake parallel compilation scheduling for reducing fuzz-test cases.

Categories

(Core :: JavaScript Engine, defect, P5)

defect

Tracking

()

People

(Reporter: nbp, Unassigned)

References

Details

(Keywords: sec-want)

Attachments

(1 file)

One of the trouble which can be encountered by fuzzers is that running the compilation on a separated thread can cause different kind of issues where we might collect information in baseline or TI which should have invalidated/updated the compiled script

I suggest (based on gary's recommendation to not use a command line argument, but a shell function), to add a shell function such as "jitOptions(…)", which would give the ability to set a variable for the expected number of the Ion calls which should be done before we run the rest of the compilation.

I suggest the name jitOptions, because *later*, I assume that we might also want to handle ion-eager and other flags the same way in addition to the command line arguments.
> I suggest (based on gary's recommendation to not use a command line
> argument, but a shell function), to add a shell function such as
> "jitOptions(…)", which would give the ability to set a variable for the
> expected number of the Ion calls which should be done before we run the rest
> of the compilation.

I mentioned this because I remembered the OOMAfterAllocations function which came about as a result of moving away from -A in bug 872823.
Keywords: sec-want
See Also: → 872823
This patch add a shell function for changing JIT parameters.
Assignee: nicolas.b.pierron → nobody
Status: ASSIGNED → NEW
Priority: -- → P5
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: