maximum memory for gcj-compiled executable?

[Boehm, Hans]

Does the real program still appear to leak on Linux?  Or does it also stabilize eventually?

Assuming you're still convinced it leaks:

It looks to me like the only growth is occurring via brk or sbrk.  The brute force way to debug this would be to put breakpoints there (after the process should have reached its final size) and see what's on the call stack.  If the growth continues, and the GC heap size stays fixed, most likely the memory is really obtained through malloc.  But the call stacks at brk or sbrk calls might still give you a hint.  If that doesn't work, I'd suggest a debugging malloc next.

Hans

> -----Original Message-----
> From: Lars Andersen [mailto:lars@rimfaxe.com]
> Sent: Friday, April 04, 2003 2:08 AM
> To: Boehm, Hans
> Cc: java@gcc.gnu.org
> Subject: RE: maximum memory for gcj-compiled executable?
> 
> 
> On Thu, 2003-04-03 at 23:46, Boehm, Hans wrote:
> > On linux, you should easily be able to track this down 
> further by looking at /proc/<pid>/maps and checking which 
> mapping is growing.
> > 
> > Hans
> > 
> 
> 
> I have attached the output from /proc/<pid>/maps with intervals.
> 
> Between the readouts I have performed the stressing that 
> appear to cause
> the growth.
> 
> note that this is not the test-program i posted earlier, but the real
> biggie - the one that could reach 150MB within 24hours. But 
> on linux the
> growth is much, much slower.
> 
> /Lars Andersen
>