Oliver, Thanks a great deal for this explanation. I am considering running JS SquirrelFish extreme engine on an ARM-based embedded platform. You are right that posix systems should provide mmap or posix_memalign that will serve the purpose. I am concerned about this issue because in my case, there is no mmap, vm_map etc. to do the job. The only available option is to use standard memory allocation routines. What is the way out for me to deal with such a situation? Obviously, standard malloc() call from within fastMalloc() is not doing anything good for me.
Regards, J R Shah On Thu, Dec 25, 2008 at 3:54 PM, Oliver Hunt <oli...@apple.com> wrote: > > On Dec 25, 2008, at 2:18 AM, Javed Rabbani wrote: > > Oliver, > > <"That call to VirtualAlloc is inside a PLATFORM(WIN_OS) guard -- you > should not be hitting it on a non-windows platform"> > > You are absolutely right about that. The VirtualAlloc() should never be hit > on non-Windows platform. Let me clarify, I am not hitting the VirtualAlloc > call on a non-Windows platform. I was just referring to the point that what > can be a possible alternative to VirtualAlloc on embedded platforms. Looking > through the code, the VirtualAlloc is enabled for WIN_OS only as expected. > For other platforms like SYMBIAN for example, the fastMalloc() is enabled > with a comment that "VirtualAlloc is not available on Symbian, so hack with > fastMalloc". > > > It's perfectly possible it's broken in symbian -- the last symbian release > was from more than two years ago. > > A bit of digging showed that Simon Hausmann added the SYMBIAN fastMalloc > branch, which implies this was for WebKit/Qt which builds with fastMalloc > disabled. So that call to fastMalloc is actually just a call to standard > system malloc, which i would guess, just by pure luck allocates adequately > aligned memory. > > What are specific memory requirements that VirtualAlloc() fulfills and > fastMalloc() does not? What can be the workaround for VirtualAlloc() on > embedded platforms? > > > Alignment -- VirtualAlloc, mmap, vm_map all produce page aligned memory > (which is at least 4k aligned), and the posix_memalign version appears to > actually over align memory (my following of the code makes me thinks it's > requesting 64k alignment :-/ ) > > I hope my point of concern is clearely stated now. > > > I'm still not sure what your problem is -- posix systems should provide > mmap or posix_memalign either of which will achieve correct behaviour > > --Oliver > >
_______________________________________________ webkit-dev mailing list webkit-dev@lists.webkit.org http://lists.webkit.org/mailman/listinfo.cgi/webkit-dev
