Always create deopt data when using full-codegen if the function is optimizable

Always create deopt data when using full-codegen if the function is optimizable
Committed: http://code.google.com/p/v8/source/detail?r=10040

[Kevin Millikin (Chromium), 2011-11-21 13:47:25]

LGTM.  The AllowOptimize/IsOptimizing/IsOptimizable/HasDeoptimizationSupport
seems pretty complicated.  I have the feeling that one of them is now redundant
(perhaps: HasDeoptimizationSupport iff. IsOptimizable).

http://codereview.chromium.org/8611003/diff/1/src/compiler.cc
File src/compiler.cc (right):

http://codereview.chromium.org/8611003/diff/1/src/compiler.cc#newcode248
src/compiler.cc:248: bool should_recompile =
!info->shared_info()->has_deoptimization_support();
So is all this dead code now?  Can we just assert that the unoptimized code has
deoptimization support?

[Rico, 2011-11-21 14:00:12]

I totally agree, the
AllowOptimize/IsOptimizing/IsOptimizable/HasDeoptimizationSupport is really
confusing. I just took the flag that is used to decide if to optimize in the
runtime profiler.

http://codereview.chromium.org/8611003/diff/1/src/compiler.cc
File src/compiler.cc (right):

http://codereview.chromium.org/8611003/diff/1/src/compiler.cc#newcode248
src/compiler.cc:248: bool should_recompile =
!info->shared_info()->has_deoptimization_support();
On 2011/11/21 13:47:25, Kevin Millikin wrote:
> So is all this dead code now?  Can we just assert that the unoptimized code
has
> deoptimization support?
I think that we can get in here if we had a function that had the
CompilingForDebugging set, in that case we only generate deopt data if the
function already has optimized code. I could change that to also always generate
deopt data - in all the cases where we could potentially later optimize.