[Webkit-unassigned] [Bug 106740] Fix the atomicIncrement implementation for Windows

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=106740





--- Comment #6 from Zoltan Arvai <zarvai at inf.u-szeged.hu>  2013-01-17 08:06:47 PST ---
I tried a dirty implementation in Atomics.h, based on the linked blog content.
It builds fine, QtTestBrowser seems ok, MiniBrowser runs with strange duplicated output - but this can be caused by remote desktop connection that I use.


inline int64_t atomicIncrement(int64_t volatile* addend)
{ 
    // return InterlockedIncrement64(reinterpret_cast<long volatile*>(addend)); 
    __asm
    {
        mov edi, addend
        mov eax, [edi]      //read current value non-atomically here...
        mov edx, [edi+4]    //  it's just a guess, if it's wrong we'll try again
    tryAgain:
        mov ebx, 1
        mov ecx, 0
        add ebx, eax
        adc ecx, edx
        lock cmpxchg8b qword ptr [edi]
        jnz tryAgain
    }
    return *addend;
}

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