I have some comments on changes to the platform-independent files.

I haven't looked at the SH4-specific stuff.


https://codereview.chromium.org/11275184/diff/1/SConstruct
File SConstruct (right):

https://codereview.chromium.org/11275184/diff/1/SConstruct#newcode183
SConstruct:183: 'arch:sh4': {
Building V8 with SCons is deprecated (and already deactivated by means
of a flag). I'm planning to delete this file and its friends before the
end of the year. Please update the GYP based build system instead.

https://codereview.chromium.org/11275184/diff/1/src/full-codegen.h
File src/full-codegen.h (right):

https://codereview.chromium.org/11275184/diff/1/src/full-codegen.h#newcode131
src/full-codegen.h:131: // TODO(STM): what is this for ?
For estimating which functions are hot, the profiler divides back edge
jump distances of loops by this value to get a normalized estimate of
the amount of code executed in the loop. The value for ia32 has been
chosen arbitrarily, the values for other platforms are chosen to match
the resulting behavior for ia32 (specifically, that a loop generated
from the same JS source code is deemed "hot" after roughly the same
number of iterations). As long as you don't support Crankshaft, you can
afford to not care. When you do want optimizations, it should be safe to
pick a value that's roughly in line with the other architectures, until
you get around to brute-force-benchmarking what works best.

https://codereview.chromium.org/11275184/diff/1/src/globals.h
File src/globals.h (right):

https://codereview.chromium.org/11275184/diff/1/src/globals.h#newcode133
src/globals.h:133: #error Target architecture sh4 is only supported on
sh4 host
How about adding a simulator to allow development/testing on x86
hardware? :-)

https://codereview.chromium.org/11275184/diff/1/src/jsregexp.cc
File src/jsregexp.cc (right):

https://codereview.chromium.org/11275184/diff/1/src/jsregexp.cc#newcode6138
src/jsregexp.cc:6138: #elif V8_TARGET_ARCH_SH4
nit: let's keep the whitespace consistent

https://codereview.chromium.org/11275184/diff/1/src/platform-posix.cc
File src/platform-posix.cc (right):

https://codereview.chromium.org/11275184/diff/1/src/platform-posix.cc#newcode154
src/platform-posix.cc:154: // return a qNaN as exepcted by the v8 core.
In sh4 tool chain, the NAN macro
nit: "Return", "expected"

https://codereview.chromium.org/11275184/

--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev

Reply via email to