http://codereview.chromium.org/7273066/diff/1/src/arm/full-codegen-arm.cc File src/arm/full-codegen-arm.cc (right):
http://codereview.chromium.org/7273066/diff/1/src/arm/full-codegen-arm.cc#newcode865 src/arm/full-codegen-arm.cc:865: EmitCallIC(ic, RelocInfo::CODE_TARGET, clause->CompareId()); On 2011/06/29 13:00:39, fschneider wrote:
I'd still prefer statically helping to never forget emitting the patch
info.
Dynamically checking with ASSERT only works for debug mode, and only
if our
tests hit all the places. That's why the parameter was introduced.
Plus there is an extra line at each call site.
After some semi-religious offline discussions involving 3 people, I think I'll leave it as it is... ;-) http://codereview.chromium.org/7273066/diff/1/src/arm/full-codegen-arm.cc#newcode4203 src/arm/full-codegen-arm.cc:4203: if (StatsCounter* counter = ic->Counter(isolate())) { On 2011/06/29 13:00:39, fschneider wrote:
Do we really need those counters at all? We don't count loads/stores
in
Crankshaft generated code, so I think they aren't very useful anymore.
Maybe we
should just get rid of them.
OK, I'll nuke those 4 counters then. http://codereview.chromium.org/7273066/diff/1/src/objects-inl.h File src/objects-inl.h (right): http://codereview.chromium.org/7273066/diff/1/src/objects-inl.h#newcode928 src/objects-inl.h:928: (*reinterpret_cast<const int*>(FIELD_ADDR(p, offset))) On 2011/06/29 12:58:42, Mads Ager wrote:
Unrelated change?
Ooops, I'll remove that one... http://codereview.chromium.org/7273066/ -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
