[Webkit-unassigned] [Bug 94712] High memory usage spike in AssemblerBuffer

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=94712





--- Comment #24 from Yong Li <yoli at rim.com>  2012-08-27 14:07:44 PST ---
(In reply to comment #23)
> > I also think it should be used by all assemblers. but I don't have dev environments for all of them, neither have enough time to do so. So my proposal is make ARMv7 the first step, and it is up to every assembler's maintainer to make them use new ArrayBuffer. Then finally we can get rid of the old one. Would that be OK?
> 
> JSC does not really work like that.
> 
> That would be a mess if every architecture started doing their own thing.
> 
> You do not need to be able to run all the dev environment. The EWS bots can help you to some extent in verifying everything.

I tried making X86Assmebler work with the new AssemblerBuffer. But there are too many dependencies on the flat consecutive buffer. I guess probably more than half of the source code will be changed. And to be honest, I'm not sure using segmented buffer is good for all assemblers. I noticed X86Assmebler makes a lot of calls to putIntegralUnchecked, and this was supposed to give performance boost on date/time tests. 

Speaking of assemblers, there is already another AssemblerBuffer for ARM (traditional): AssemblerBufferWithConstantPool. AssemblerBuffer used by different assemblers are not always same. So to be neat, I'm considering to introduce AssemblerBufferSegmented (a brother of AssemblerBuffer and AssemblerBufferWithConstantPool), and it is only used for ARMv7Assembler and MacroAssemblerARMv7. Objection?

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