[Webkit-unassigned] [Bug 123955] New: Avoid calling fastRealloc when it returns the same address

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=123955

           Summary: Avoid calling fastRealloc when it returns the same
                    address
           Product: WebKit
           Version: 528+ (Nightly build)
          Platform: Macintosh
        OS/Version: Mac OS X 10.8
            Status: NEW
          Severity: Enhancement
          Priority: P5
         Component: Web Template Framework
        AssignedTo: webkit-unassigned at lists.webkit.org
        ReportedBy: hbono at chromium.org


Created an attachment (id=216259)
 --> (https://bugs.webkit.org/attachment.cgi?id=216259&action=review)
A patch that uses fastMallocGoodSize

Greetings,

On OS X, malloc() rounds up the specified size to a multiple of malloc_good_size(). (For example, malloc(1) returns a 16-byte block on OS X.) Also, as listed in the source code <http://www.opensource.apple.com/source/Libc/Libc-594.1.4/gen/malloc.c>, realloc() returns the same address when the specified new size is less than the block size of the given memory, i.e. |p0| is equal to |p1| in the following code snippet.

  int main(int argc, char* argv[]) {
    void* p0 = malloc(1);
    void* p1 = realloc(p0, 10);
    printf("p0=%p, p1=%p\n", p0, p1);
    return 0;
  }

To consider this fact, it may be good for the StringBuilder class to use fastMallocGoodSize(length) instead of length to check if it needs to call StringBuilder<T>::reallocateBuffer() as listed in the attached diff. (For your information, the StringBUilder class tried reallocating memory to the same block size >13000 times when I ran 'run-webkit-tests --no-http' on my Mac.)

Regards,

Hironori Bono

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