http://codereview.chromium.org/2843010/diff/1/2 File src/arm/codegen-arm.cc (right):
http://codereview.chromium.org/2843010/diff/1/2#newcode7368 src/arm/codegen-arm.cc:7368: ASSERT(heap_number_map.is(r6)); // This is a register that we don't use. On 2010/06/17 14:15:25, Søren Gjesse wrote:
don't use -> don't otherwise use in this stub?
I delayed the allocation of r6 to avoid penalizing those programs that avoid non-Smis. And in the process, the need for this comment evaporated. http://codereview.chromium.org/2843010/diff/1/2#newcode7762 src/arm/codegen-arm.cc:7762: ASSERT(heap_number_map.is(r6)); // This is a register we don't use. On 2010/06/17 14:15:25, Søren Gjesse wrote:
don't use -> don't otherwise use in this stub?
Ditto. http://codereview.chromium.org/2843010/diff/1/2#newcode8012 src/arm/codegen-arm.cc:8012: Register num_map = r6; On 2010/06/17 14:15:25, Søren Gjesse wrote:
num_map -> heap_number_map for consistency.
This variable went away! http://codereview.chromium.org/2843010/diff/1/6 File src/arm/macro-assembler-arm.cc (right): http://codereview.chromium.org/2843010/diff/1/6#newcode1670 src/arm/macro-assembler-arm.cc:1670: // Store heap number map in the allocated object. On 2010/06/17 14:15:25, Søren Gjesse wrote:
Why not also use AssertRegisterIsRoot here?
Done. http://codereview.chromium.org/2843010/show -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
