Re: [PATCH] stronger ELF sanity checks v2

[Jesse Pollard <jesse@cats-chateau.net>]

On Friday 16 January 2004 10:08, Pavel Machek wrote:
> Hi!
>
> > On Tue, Jan 13, 2004 at 02:55:07AM +0100, Jesper Juhl wrote:
> > > Here's the second version of my patch to add better sanity checks for
> > > binfmt_elf
> >
> > I assume this breaks Brian Raiter's tiny ELF executables[1]. Even
> > though these binaries are evil hacks that don't comply to standards
> > and serve no serious purpose, I'm not sure what the purpose of the
> > sanity checks is. Are there any risks associated with running
> > non-compliant ELF executables? (Now that I mention it, the
>
> You get vy ugly behaviour. If you compile executable with huge static
> data, it will compile okay, link okay, *launch okay* and die on
> segfault. That's wrong, it should have died on -ENOMEM during exec.
> 								Pave

Wouldn't that depend on the overcommit options?

With permitted overcommit -
	compile/link ok
	launch - segfault/-ENOMEM if heap/stack + static data UPDATES exceed
		 system capacity
Without overcommit:
	-ENOMEM if the heap/stack can't be initialized; as in even the
		first page of the heap/stack fails - and before actual
		launch completes, as the situation you describe.

If static data is just referenced, it should page in, and get dropped.