I think the first patch set makes more sense. In many of the bailout cases
that
jump to "slow", we can't use handlers anyway. The stub cache should only be
polled for sane receivers and keys that are unique names.
This also ties in nicely with the next step we discussed, which is to use
the
stub cache to handle the "check_name" label; GenerateKeyNameCheck and
GenerateKeyedLoadReceiverCheck would still happen and after that the stub
cache
can be probed (instead of the KLC).
https://codereview.chromium.org/878263002/
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
