[Webkit-unassigned] [Bug 96916] Make the old JIT not rely on the timeout check register

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Mon Nov 12 20:54:06 PST 2012


--- Comment #12 from Geoffrey Garen <ggaren at apple.com>  2012-11-12 20:55:50 PST ---
> > So your plan is to do timeout checking by polling an in-memory counter at loop boundaries in the interpreter, the baseline JIT, and the DFG?
> Yes. That is a consistent approach. Or perhaps do you have any better idea?

That depends on the performance of polling an in-memory counter. For example, we could use a flag and an alarm to avoid having to read-modify-write the counter. Or we could poll in the interpreter and baseline JIT, where performance matters less, and use some other solution in the DFG.

If there's no performance cost to a read-modify-write counter in all engines, that seems like a fine approach. But I suspect there will be a performance cost. Have you tested this?

Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

More information about the webkit-unassigned mailing list