Bug 22171

Summary:

CTI refactoring: remove m_callFrame; batch JIT all static trampolines.

Product:

WebKit

Reporter:

Gavin Barraclough <barraclough>

Component:

JavaScriptCore

Assignee:

Gavin Barraclough <barraclough>

Status:

RESOLVED FIXED

Severity:

Normal

Priority:

Version:

528+ (Nightly build)

Hardware:

Mac

OS:

OS X 10.5

Attachments:

Description	Flags
The patch	zwarich: review+

Gavin Barraclough

Reported 2008-11-10 20:14:24 PST

Batch compile the set of static trampolines at the point Machine is constructed, using a single allocation. Refactor out m_callFrame from CTI, since this is only needed to access the global data (instead store a pointer to the global data directly, since this is available at the point the Machine is constructed). Add a method to align the code buffer, to allow JIT generation for multiple trampolines in one block.

Attachments
The patch (31.47 KB, patch) 2008-11-10 20:15 PST, Gavin Barraclough	zwarich: review+	Details Formatted Diff Diff
View All Add attachment proposed patch, testcase, etc.

Gavin Barraclough

Comment 1 2008-11-10 20:15:35 PST

Created attachment 25042 [details] The patch no performance impact

Cameron Zwarich (cpst)

Comment 2 2008-11-10 20:53:19 PST

Comment on attachment 25042 [details] The patch r=me, but remove the trailing whitespace you added before emitInt3().

Gavin Barraclough

Comment 3 2008-11-10 21:10:11 PST

Sending JavaScriptCore/ChangeLog Sending JavaScriptCore/VM/CTI.cpp Sending JavaScriptCore/VM/CTI.h Sending JavaScriptCore/VM/Machine.cpp Sending JavaScriptCore/VM/Machine.h Sending JavaScriptCore/masm/X86Assembler.h Sending JavaScriptCore/runtime/JSGlobalData.cpp Transmitting file data ....... Committed revision 38286.

Geoffrey Garen

Comment 4 2008-11-10 22:45:59 PST

I think you should fold the initialize function into the Machine constructor. Is there any reason for it to be separate?

Note You need to log in before you can comment on or make changes to this bug.