Bug 137285 – CSS JIT: add the initial implementation of :nth-child(An+B of selector)

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Benjamin Poulain]

CSS JIT: add the initial implementation of :nth-child(An+B of selector)

[Benjamin Poulain]

Created attachment 238996 [details]
Patch

[Yusuke Suzuki]

(In reply to comment #0)
> CSS JIT: add the initial implementation of :nth-child(An+B of selector)

I've landed the StackAllocator.h part in the separate patch[1] as reviewed by you[2].
So now, we can use `allocateUninitialized(int)` :)

[1]: http://trac.webkit.org/changeset/174142
[2]: https://bugs.webkit.org/show_bug.cgi?id=135293

[Benjamin Poulain]

Comment on attachment 238996 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=238996&action=review

> Source/WebCore/cssjit/SelectorCompiler.cpp:1509
> +    if (unsigned stackRequirementCount = backtrackingMemoryRequirements.stackCount) {
> +        for (unsigned i = 0; i < stackRequirementCount; ++i)
> +            m_backtrackingStack.append(m_stackAllocator.allocateUninitialized());
> +        if (!temporaryStackBase.isValid())
> +            temporaryStackBase = m_backtrackingStack.first();
>      }

This is terrible, I did not get a chance to fix it tonight.

I'll fix it before landing by using http://trac.webkit.org/changeset/174142

[Benjamin Poulain]

(In reply to comment #2)
> (In reply to comment #0)
> > CSS JIT: add the initial implementation of :nth-child(An+B of selector)
> 
> I've landed the StackAllocator.h part in the separate patch[1] as reviewed by you[2].
> So now, we can use `allocateUninitialized(int)` :)
> 
> [1]: http://trac.webkit.org/changeset/174142
> [2]: https://bugs.webkit.org/show_bug.cgi?id=135293

Thanks! I was just adding the comment about that. :)

[Benjamin Poulain]

Comment on attachment 238996 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=238996&action=review

> Source/WebCore/cssjit/SelectorCompiler.cpp:1593
> +    // FIXME: ASSERT clean state.

Oops

[Andreas Kling]

Comment on attachment 238996 [details]
Patch

r=me.

[Benjamin Poulain]

Created attachment 239146 [details]
Patch

[Benjamin Poulain]

Comment on attachment 239146 [details]
Patch

Clearing flags on attachment: 239146

Committed r174245: <http://trac.webkit.org/changeset/174245>

[Benjamin Poulain]

All reviewed patches have been landed.  Closing bug.

[Darin Adler]

I think this patch caused some leaks in the CSS parser because the new rules don’t all delete nth_selector_ending.

[Benjamin Poulain]

(In reply to comment #10)
> I think this patch caused some leaks in the CSS parser because the new rules
> don’t all delete nth_selector_ending.

I guess you meant to comment on https://bugs.webkit.org/show_bug.cgi?id=136845 ?

I think you are right. I'll check tomorrow.