Re: [patch 1/3] slob: rework freelist handling

[Matt Mackall <mpm@selenic.com>]

On Wed, May 23, 2007 at 03:26:05PM -0700, Christoph Lameter wrote:
> On Wed, 23 May 2007, Matt Mackall wrote:
> 
> > On Wed, May 23, 2007 at 01:02:53PM -0700, Christoph Lameter wrote:
> > > On Wed, 23 May 2007, Matt Mackall wrote:
> > > 
> > > > Meanwhile this function is only called from swsusp.c.
> > > 
> > > NR_SLAB_UNRECLAIMABLE is also used in  __vm_enough_memory and 
> > > in zone reclaim (well ok thats only NUMA).
> > 
> > It's NR_SLAB_RECLAIMABLE in __vm_enough_memory. And that is always
> > zero with SLOB. There aren't any reclaimable slab pages.
> 
> All dentries and inodes are reclaimable via the shrinkers in vmscan.c. So 
> you are saying that SLOB does not allow dentry and inode reclaim?

No. I've already pointed out the EXACT CALL CHAIN that leads to dentry
reclaim. And it's independent of NR_SLAB_RECLAIMABLE and independent
of allocator.

> > SLOB does do UNRECLAIMABLE, true. But there aren't any interesting
> > users of SLAB_UNRECLAIMABLE that I can see.
> 
> > That's because there are no slabs! Memory usage shows up here in
> > exactly the same way that memory usage from get_free_page does.
> 
> So there seems to be a basic unresolvable issue with SLOB....

Already responded to this one too.

-- 
Mathematics is the supreme nostalgia of our time.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>