[Webkit-unassigned] [Bug 113862] Negative zero checks cause unnecessary speculation failures on SunSpider on ARMv7

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=113862


Geoffrey Garen <ggaren at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #197263|review?, commit-queue?      |review+, commit-queue-
               Flag|                            |




--- Comment #9 from Geoffrey Garen <ggaren at apple.com>  2013-04-25 12:33:07 PST ---
(From update of attachment 197263)
View in context: https://bugs.webkit.org/attachment.cgi?id=197263&action=review

r=me

cq- because of the layout test issue. Any committer can fix the layout test and then land, or you can post a new patch, and I can cq+ it.

> LayoutTests/fast/js/regress/script-tests/negative-zero-divide.js:6
> +  for (i = 0; i < 100000; i++) {

Please change this to 100. That's our de facto limit for getting code to run in the DFG, without making regression tests run too long.

> LayoutTests/fast/js/regress/script-tests/negative-zero-divide.js:7
> +    if (i == 99999)

And 99 here.

> LayoutTests/fast/js/regress/script-tests/negative-zero-modulo.js:6
> +  for (i = 0; i < 100000; i++) {
> +    if (i == 99999)

Ditto.

> LayoutTests/fast/js/regress/script-tests/negative-zero-negate.js:6
> +  for (i = 0; i < 100000; i++) {
> +    if (i == 99999)

Ditto.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.