MathICs should be able to emit only a jump along the inline path when they don't have any type data

• Component: JavaScriptCore
• Status: RESOLVED FIXED

[Saam Barati]

And we should repatch them on first execution

[Saam Barati]

Created attachment 284532 [details]
WIP

almsot done

[WebKit Commit Bot]

Attachment 284532 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/jit/JITMathIC.h:112:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
ERROR: Source/JavaScriptCore/jit/JITMathIC.h:124:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
Total errors found: 2 in 8 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Saam Barati]

Created attachment 284656 [details]
patch

[WebKit Commit Bot]

Attachment 284656 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/jit/JITMathIC.h:122:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
ERROR: Source/JavaScriptCore/jit/JITMathIC.h:134:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
Total errors found: 2 in 6 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Mark Lam]

Comment on attachment 284656 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=284656&action=review

r=me with fixes.

> Source/JavaScriptCore/jit/JITMathIC.h:74
> +                // It looks like the MathIC has yet to execute. We don't want to emit code in this
> +                // case for a couple reasons. First, the operation may never execute, so if we don't emit
> +                // code, it's a win. Second, if the operation does execute, we can emit better code
> +                // once we have an idea about the types of lhs and rhs.

nice!

> Source/JavaScriptCore/jit/JITMathIC.h:81
> +                ASSERT(static_cast<ptrdiff_t>(inlineSize) <= MacroAssembler::maxJumpReplacementSize());

I think you should use ASSERT_UNUSED(inlineSize, ...) here to fix the build breakage.

> Source/JavaScriptCore/jit/JITMathICInlineResult.h:34
> +    DelayFastPath,

I don't see this used anywhere.  Can remove?

[Saam Barati]

Comment on attachment 284656 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=284656&action=review

>> Source/JavaScriptCore/jit/JITMathICInlineResult.h:34
>> +    DelayFastPath,
> 
> I don't see this used anywhere.  Can remove?

Yeah. I forgot to remove this. I had this in the original patch but forgot to remove it after I refactored.

[Saam Barati]

Created attachment 284712 [details]
patch for landing

[WebKit Commit Bot]

Attachment 284712 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/jit/JITMathIC.h:122:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
ERROR: Source/JavaScriptCore/jit/JITMathIC.h:134:  Place brace on its own line for function definitions.  [whitespace/braces] [4]
Total errors found: 2 in 5 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[WebKit Commit Bot]

Comment on attachment 284712 [details]
patch for landing

Clearing flags on attachment: 284712

Committed r203786: <http://trac.webkit.org/changeset/203786>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.