[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Bug classpath/29263] Memory leak in VMDouble.toString()
From: |
BenGardiner at nanometrics dot ca |
Subject: |
[Bug classpath/29263] Memory leak in VMDouble.toString() |
Date: |
10 Jul 2009 19:39:25 -0000 |
------- Comment #5 from BenGardiner at nanometrics dot ca 2009-07-10 19:39
-------
Created an attachment (id=18177)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=18177&action=view)
also free any _p5s allocated during the dtoa
I tried the patch in comment #3 but found that there were still some leaks with
the following test program, using the mtrace feature of glibc:
import java.util.Random;
public class DToATester
{
/**
* @param args
*/
public static void main(String[] args)
{
Random random = new Random(1);
while(true)
{
double val = Double.longBitsToDouble(random.nextLong());
System.out.print(val + " ");
}
}
}
turns out that there are some additional _Jv_Bigint's allocated in pow5mult
assigned to the '_p5s' linked-list member of reent. The following patch gets
rid of all allocations made in mprec_calloc -- according the mtrace. It was
created against gcj-4.3.3's import of classpath which I think is 0.92 according
to the classpath README.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29263
- [Bug classpath/29263] Memory leak in VMDouble.toString(),
BenGardiner at nanometrics dot ca <=