Comment #2 on issue 3168 by [email protected]: A64:
mjsunit/numops-fuzz-part2 fails when checking veneers
http://code.google.com/p/v8/issues/detail?id=3168
This failure is fixed as a side effect of
54332bac7a457007e6d3dfceff90b292f952b368.
Without the patch a lot of debug code is generated to check the stack
consistency (csp/jssp). This causes us to hit - much earlier than expected
- the situation where more code than the veneer margin is generated without
checking for veneers (see last point of comments #2 amd #3 for issue
169893002).
It is good that the CHECK catches this.
I suggest we increase kVeneerDistanceMargin to 4KB until the cleaner check
in CheckBuffer is implemented. This will hopefully give us more room. This
should not penalize much non-TBZ branches (bigger range), and most TBZ/TBNZ
branches to distant targets are likely deoptimization exits for which we
don't care about the veneer performance penalty.
Does that sound ok?
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
--
--
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/groups/opt_out.
