LGTM for what I have commented on.

http://codereview.chromium.org/6170001/diff/125002/src/assembler.h
File src/assembler.h (right):

http://codereview.chromium.org/6170001/diff/125002/src/assembler.h#newcode474
src/assembler.h:474: // FP_CALL - builtin/runtime call that returns
floating point.
On 2011/02/03 07:27:31, zaheer wrote:
On 2011/02/02 13:24:38, Søren Gjesse wrote:
> Please change "builtin/runtime call that returns floating point." to
something
> like "direct call to a C-function which will never cause a GC". It
is not
always
> a function with the signature double f(double, double), e.g.
> "native_compare_doubles".
if iam not mistaken, native_compare_doubles uses the BUILTIN_CALL case
and not
FP_RETURN case. In that sense isnt the current comment accurate?



You are right it does, and of cause returns an int. However it is rather
misleading to use BUILTIN_CALL, as the function is actually expecting
two doubles on the stack, but lets leave it like this for now.

One small thing though, the function power_double_double in assembler.cc
should be marked with FP_RETURN_CALL - it is currently wrongly missing
the true argument. It is not currently used on ARM, but probably will.

http://codereview.chromium.org/6170001/

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

Reply via email to