- squirrelfish-dev - lists.macosforge.org

(no subject)
by wingoog moon 07 Jan '12

07 Jan '12

HI!! I know that DFG JIT can do some general optimizatiopns such as CSE,OSR,dead code eliminations. But all these optimizations are machine independent. So what machine dependent optimizations DFG JIT can do???(for inatance for ARM platform)

2 1

SFX JIT
by wingoog moon 29 Dec '11

29 Dec '11

Hi Tring to understand SFX aritecture and I have found these functions JIT::privateCompileMainPass() JIT::privateCompileSlowCases() in http://trac.webkit.org/browser/trunk/Source/JavaScriptCore/jit/JIT.cpp. So as I understand each instruction is executed seperatly from others. It seems that JIT works tha way as interpretator does. What the difference between SFX JIT and SFX interpretor, exept that JIT keep cashed copies of functions. thanks for attention!

2 1

CTI finctions
by wingoog moon 25 Dec '11

25 Dec '11

Hi, In "Selective Just-in-Time Compilation for Client-side Mobile JavaScript Engine" article I read that in JSC there are 109 bytecode isntructions, and 51 of them are handeled by CTI functions. But i could't find anywhere what is the CTI function and why they are used? thanks.

2 1

Build Squirrelfish for NACL
by mohit bansal 28 Nov '11

28 Nov '11

Hii I am new to Webkit-Squirrelfish.. I want to build stand alone Squirrelfish with NACL but I have some query: 1) Is stand alone build of Squirrelfish is feasible as like V8 in chromium OS browser. 2) I found that to build Squirrelfish we need to given toolkit (like gtk or qt for Linux). Is it feasible to build Squirrelfish without these options. 3) Is Squirrelfish have some dependency on Glibc as they need to be checked in configuration file which made at the time of build Squirrelfish . Thank You !!! Regards, Mohit Bansal

2 1

two questions about the parser
by xiaoming gu 30 Aug '11

30 Aug '11

<squirrelfish-dev(a)lists.webkit.org>Hi, all. I'm Xiaoming Gu and a new guy studying SquirrelFish. I got to know that there are three compilation steps from JS source code to machine code: JS source code to AST tree by the parser, AST tree to byte code by the byte code generator, and byte code to machine code by the JIT compiler. Now I'm getting to know the parser. Following are two questions. 1. Is the parser based on an LL(1) grammar? 2. Where are the grammar rules? The annex in ECMAScript Language Specification? Thanks. Xiaoming

2 1

error in building webkit on win7
by Ax Jiang 30 Jun '11

30 Jun '11

Hi All When I tried to build webkit on win7, I ran into the following problems: " ..\..\wtf\FastMalloc.cpp(83) : fatal error C1083: Cannot open include file: 'pthread.h': No such file or directory 2>c:\WebKit\Source\JavaScriptCore\wtf/unicode/icu/UnicodeIcu.h(27) : fatal error C1083: Cannot open include file: 'unicode/uchar.h': No such file or directory 2>MainThreadWin.cpp 2>TCSystemAlloc.cpp 2>RandomNumber.cpp 2>dtoa.cpp 2>c:\WebKit\Source\JavaScriptCore\wtf/unicode/icu/UnicodeIcu.h(27) : fatal error C1083: Cannot open include file: 'unicode/uchar.h': No such file or directory " It seems that pthread is supported by mingw in windows and I have mingw installed when I installed cygwin with default option. Any help? thanks Wu

2 1

debugging GC area
by Zoltan Herczeg 19 May '11

19 May '11

Hi, I am currently debugging a nasty GC bug. (This one: https://bugs.webkit.org/show_bug.cgi?id=60881) In short, inspector/styles/metrics-box-sizing.html fails in the interpreter with ASSERT_GC_OBJECT_LOOKS_VALID(cell) if all inspector/ tests are running by run-webkit-tests (32 bit, debug mode), but it passes if you test it alone. Actually for tracking memory related issues the valgrind tool has an excellent approach: never reuse the same address. This way we can keep track of the freed chunks. Would it be possible to implement something similar in the GC for JavaScriptCore (disabled by default with some directives)? We would still run the GC as usual, but instead of reusing the cells, we would always allocate a new chunk bz mmap. Naturally the freed cells would contain some invalid pointers to cause crashes, and perhaps a descriptor which describes where it was originally allocated. Would it be feasible to add such feature to JavaScriptCore? Besides I saw a lot of improvements in GC. Is it still a mark and don't sweep allocator? Is it still allocates fixed size cells? Does it move cells? Does Weak<T> contains a GC'ed pointer? Regards, Zoltan

2 3

tar1583@gmail.com is inviting you to freecharge.in
by tar1583＠gmail.com 11 Mar '11

11 Mar '11

1 0

true as 1 (int32)
by Zoltan Herczeg 04 Mar '11

04 Mar '11

Hi, I found something interesting in the v8 source code. They store the true and false as integers: 1 and 0. Actually it was a surprise to me, but "true + 1" is "2" in JavaScript, so this kind of object handling seems ok. Advantage of their approach: they don't need to care about boolean values for all arithmetic and conditional operations in JIT. This could reduce the source code (maintainability), and probably faster, since you don't need to check boolean values. Disadvantage: type resolving is a bit more difficult, but since that is a rare operation, I think the gain could be bigger. What is your opinion about it? Would it be possible to implement something like this for JSValue32_64? Would it worth it? Regards, Zoltan

3 5

Re: [squirrelfish] true as 1 (int32)
by Zoltan Herczeg 04 Mar '11

04 Mar '11

This is their initialization: { MaybeObject* maybe_obj = CreateOddball("true", Smi::FromInt(1), Oddball::kTrue); if (!maybe_obj->ToObject(&obj)) return false; } set_true_value(obj); { MaybeObject* maybe_obj = CreateOddball("false", Smi::FromInt(0), Oddball::kFalse); if (!maybe_obj->ToObject(&obj)) return false; } set_false_value(obj); Of course, you need to know in several places that it IS a boolean, but for arithmetic operations (most common in JIT), it seems an int is more than enough. (Smi thing is similar to the old JSValue32 approach) Regards, Zoltan > Hello, > > This is not the case, at least in the full-codegen, where v8 uses specific > heap values for the boolean True and False. > I am not sure what Crankshaft is able to optimize, but it does also use > these values. > > See src/heap.h line 56 > #define UNCONDITIONAL_STRONG_ROOT_LIST(V) > ... > V(Object, true_value, TrueValue) > \ > V(Object, false_value, FalseValue) > > For example ia32 then uses Factory::true_value() to access it. > (full-codegen-ia32.cc) > > 527 // Emit the inlined tests assumed by the stub. > 528 __ cmp(result_register(), Factory::undefined_value()); > 529 __ j(equal, if_false); > 530 __ cmp(result_register(), Factory::true_value()); > 531 __ j(equal, if_true); > 532 __ cmp(result_register(), Factory::false_value()); > 533 __ j(equal, if_false); > > > Alexandre > > > On Fri, Mar 4, 2011 at 1:37 PM, Zoltan Herczeg > <zherczeg(a)inf.u-szeged.hu>wrote: > >> Hi, >> >> I found something interesting in the v8 source code. They store the true >> and false as integers: 1 and 0. Actually it was a surprise to me, but >> "true + 1" is "2" in JavaScript, so this kind of object handling seems >> ok. >> >> Advantage of their approach: they don't need to care about boolean >> values >> for all arithmetic and conditional operations in JIT. This could reduce >> the source code (maintainability), and probably faster, since you don't >> need to check boolean values. >> >> Disadvantage: type resolving is a bit more difficult, but since that is >> a >> rare operation, I think the gain could be bigger. >> >> What is your opinion about it? Would it be possible to implement >> something >> like this for JSValue32_64? Would it worth it? >> >> Regards, >> Zoltan >> >> >> _______________________________________________ >> squirrelfish-dev mailing list >> squirrelfish-dev(a)lists.webkit.org >> http://lists.webkit.org/mailman/listinfo.cgi/squirrelfish-dev >> >

1 0