http://bugzilla.novell.com/show_bug.cgi?id=405246
User bart.vanassche@gmail.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=405246#c10
--- Comment #10 from Bart Van Assche 2009-04-06 10:54:11 MDT ---
Until now no one asked why I started looking at the procps internals, so let me
explain this first. Last year I stumbled upon a memory leak in the PAE
implementation in the Linux kernel (http://lkml.org/lkml/2008/3/9/19) which was
resolved quickly. In order to detect the presence or absence of a memory leak
in the kernel, one needs a tool to measure how much memory is in use and how
much memory can still be reclaimed. The patch I posted on linux-mm makes the
procps tools (top, free, vmstat) behave such that "in use" represents the
memory in use and "free" represents the memory that still can be reclaimed such
that it can be allocated by the kernel for other purposes.
Dave, what you are asking is to let the procps tools print some values that
might by consistent with previous versions of the procps tools but that do not
make sense and that are absolutely not easy to understand, to explain or to
document.
Regarding the single line of code that changes the behavior of the procps tools
(kb_main_cached += kb_slab_reclaimable + kb_swap_cached + kb_nfs_unstable): you
are focusing on whether or not to correct the kb_main_cached value with
kb_swap_cached but did not comment on the correction with kb_slab_reclaimable
or kb_nfs_unstable. Even if the correction with kb_swap_cached is not retained
because of historical reasons, the correction with kb_slab_reclaimable and
kb_nfs_unstable should be retained.
--
Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.