https://codereview.chromium.org/24596002/diff/11001/src/arm/lithium-arm.cc
File src/arm/lithium-arm.cc (right):
https://codereview.chromium.org/24596002/diff/11001/src/arm/lithium-arm.cc#newcode1214
src/arm/lithium-arm.cc:1214: : UseFixed(instr->context(), cp);
On 2013/09/27 12:08:00, ulan wrote:
I think we can use UseAny here because the context is used in deferred
code.
If we use UseAny, the context will we copied into cp if necessary.
However, as at the end we need the context in cp, I think it's better to
let the allocator moving the context into cp (in some case, it should
avoid some mov).
In fact, it's just a question of probability, we should use the deferred
code most of the time because it's not used only when we have a tagged
value which is a smi (untagged cases don't need the context). I think
that there are more probabilities that a tagged value hold a number or
an integer than a smi because the cases untagged number, untagged
integer or smi are already managed by a different path.
If you think these assertions are false, I will put UseAny.
https://codereview.chromium.org/24596002/diff/11001/src/arm/lithium-codegen-arm.cc
File src/arm/lithium-codegen-arm.cc (right):
https://codereview.chromium.org/24596002/diff/11001/src/arm/lithium-codegen-arm.cc#newcode3683
src/arm/lithium-codegen-arm.cc:3683:
ASSERT(ToRegister(instr->context()).is(cp));
On 2013/09/27 12:08:00, ulan wrote:
I think we don't need context in cp here since it is loaded in call to
runtime
from deferred.
It is just to be consistent with LChunkBuilder::DoMathAbs. If we use
UseAny, these assert will be removed.
https://codereview.chromium.org/24596002/
--
--
v8-dev mailing list
v8-dev@googlegroups.com
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to v8-dev+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
