789111 - Ion: Inline path for LIntToString

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Tom Schuster [:evilpie]]

Attachment: v1

This can't hurt and is a small win a micro-benchmark. Also makes me wonder why we don't handle some other easy stuff in ToString.

Comment 1

[Tom Schuster [:evilpie]]

Comment on attachment 658922 [details] [diff] [review]
v1

Actually there is some failing cross compartment check I need to look at. Sorry.

Comment 2

[Tom Schuster [:evilpie]]

Attachment: v2

Ups, Assembler::Below inverted is not Assembler::Above.

Comment 3

[David Anderson [:dvander] - inactive, e-mail if emergency]

What benchmark is this for?

Comment 4

[Jan de Mooij [:jandem]]

Comment on attachment 658930 [details] [diff] [review]
v2

Review of attachment 658930 [details] [diff] [review]:
-----------------------------------------------------------------

Tom, sorry for the delay here! This looks great, r=me with the comment below addressed.

::: js/src/ion/CodeGenerator.cpp
@@ +207,5 @@
> +    if (!ool)
> +        return false;
> +
> +    masm.cmpPtr(input, ImmWord(StaticStrings::INT_STATIC_LIMIT));
> +    masm.j(Assembler::AboveOrEqual, ool->entry());

This should be (maybe wrap if it's too long):

masm.branch32(Assembler::AboveOrEqual, input, Imm32(StaticStrings::INT_STATIC_LIMIT), ool->entry());

Comment 5

[Jan de Mooij [:jandem]]

I need this for bug 794475 so I pushed this for you with nit addressed

https://hg.mozilla.org/integration/mozilla-inbound/rev/978385fba015

Comment 6

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/978385fba015

Comment 7

[Tom Schuster [:evilpie]]

Hehe nice, thank you!