Re: apparent 4 GB memory limit for brk() in solaris 8 on some hardware...?
From: Paul Pluzhnikov (ppluzhnikov-nsp_at_charter.net)
Date: Fri, 29 Jul 2005 09:21:16 -0700
> 1/a) I don't want to run truss wide open (too much data), so I guess I
> will specify memory-allocation-related syscalls only. With that in
> mind, I will probably run this truss command:
> truss -t brk,mmap -u malloc -o outfile -f -p pid
Truss is unlikely to reveal anything -- your pmap output (assuming
it was taken just before the failure) shows *nothing* that would
have prevented further brk() growth.
One other suggestion: try running the app with libmapmalloc (via
LD_PRELOAD) -- it may be able to bypass whatever limit you are
bumping into, or at least provide further clues.
-- In order to understand recursion you must first understand recursion. Remove /-nsp/ for email.