Please take another look. That bug you noticed took quite some effort to
clean
up... Now I always move the receiver into r0 on arm, and the method will
end up
in r1. This means that in some cases I'll have to keep the receiver in r0
and
use r1 for the result until the end where I move r1 back over r0.
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/arm/stub-cache-arm.cc
File src/arm/stub-cache-arm.cc (left):
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/arm/stub-cache-arm.cc#oldcode766
src/arm/stub-cache-arm.cc:766:
On 2013/11/25 16:31:53, Igor Sheludko wrote:
Non-equivalent modification for one of the GenerateCallFunction()
usages. See
below.
Done.
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/arm/stub-cache-arm.cc#oldcode1617
src/arm/stub-cache-arm.cc:1617: // -----------------------------------
Yes they are. Look at Load/Store, that's what I want to achieve. They
are platform independent, and all configurations of variables aren't
done anymore in comments but by using abstractions for the variables.
On 2013/11/25 16:31:53, Igor Sheludko wrote:
Are these (and similar) comments really useless here?
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/arm/stub-cache-arm.cc
File src/arm/stub-cache-arm.cc (right):
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/arm/stub-cache-arm.cc#newcode2380
src/arm/stub-cache-arm.cc:2380: // -----------------------------------
I removed the comment altogether. I prefer to add more abstractions over
time (eg, LoadReceiver(reg)) than adding comments.
On 2013/11/25 16:31:53, Igor Sheludko wrote:
What do you think about putting these comments here:
// -- sp[(argc - n - 1) * 4] : arg[n] (zero-based)
// -- ...
// -- sp[argc * 4] : receiver
// -----------------------------------
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/arm/stub-cache-arm.cc#newcode2383
src/arm/stub-cache-arm.cc:2383: Register reg = r1;
On 2013/11/25 16:31:53, Igor Sheludko wrote:
It looks like the other usages of "r1" below should be replaced with
"reg".
Done.
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/arm/stub-cache-arm.cc#newcode2524
src/arm/stub-cache-arm.cc:2524: GenerateCallFunction(masm(), object,
arguments(), &miss, extra_state_);
On 2013/11/25 16:31:53, Igor Sheludko wrote:
Modified GenerateCallFunction() is no longer patches the receiver on
the stack.
Is it true that we no longer need this check here?
Done.
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/stub-cache.h
File src/stub-cache.h (right):
https://chromiumcodereview.appspot.com/83753002/diff/70001/src/stub-cache.h#newcode909
src/stub-cache.h:909: Register HandlerFrontendHeader(Handle<Object>
object,
On 2013/11/25 16:31:53, Igor Sheludko wrote:
Probably is it worth adding a comment about what is returned from the
function.
Done.
https://chromiumcodereview.appspot.com/83753002/
--
--
v8-dev mailing list
[email protected]
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 [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
