Re: Misleading OOM messages

[Pavel Machek]

On Fri 2009-05-15 13:59:50, Christoph Lameter wrote:
> On Fri, 15 May 2009, Pavel Machek wrote:
> 
> > Ok, so kernel should be fixed to make limits 30% of non-mlocked
> > memory.
> 
> There is already ulimit.

...which does not work as described in the message you snipped.

> > > folks.  They go sucking up and locking as much memory as they can get
> > > their hands on.  Adding memory never helps them because they'll use up
> > > whatever is there.
> >
> > Well, but it is uncommon everywhere else. If you have desktop system,
> > job size is pretty much constant. If you have too little memory, you
> > OOM.
> 
> Nope. If you have too little memory for your app then the kernel pages
> portions of the app out to disk. Thats is why you have a VM (VIRTUAL
> machine). The app is not running with physical memory.

Try running your machine with mem=8M, then tell me how virtual memory
works.

If you have too little RAM+swap, you OOM. (Adding memory helps).

If you have *way* too little RAM, you OOM. (Kernel data is
unswappable. Task struct is 8KB. At some point it breaks).
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/