Re: [RFC PATCH] binfmt_elf: Fix PIE execution with randomization disabled

[Josh Boyer <jwboyer@redhat.com>]

On Mon, Oct 03, 2011 at 02:10:26PM +0200, Jiri Kosina wrote:
> On Thu, 29 Sep 2011, Josh Boyer wrote:
> 
> > We've had a bug report[1] of some PIE programs getting a SIGKILL upon exec
> > if you disable address randomization with:
> > 
> > 	echo 0 > /proc/sys/kernel/randomize_va_space
> > 
> > I tracked this down to get_unmapped_area_prot returning -ENOMEM because
> > the address being passed in is larger than TASK_SIZE - len for the bss
> > section of the test executable.  That filters back to set_brk returning
> > an error to load_elf_binary and the SIGKILL being sent around line 872
> > of binfmt_elf.c.
> > 
> > H.J. submitted an upstream bug report [2] as well, but got no feedback
> > and we can't view it with kernel.org being down anyway.  He came up with
> > the patch below as well, which is what I'm sending on for comments.  The
> > changelog is my addition, so if that is wrong yell at me.
> > 
> > I wanted to get some more eyes on this, because the current code sets
> > load_bias to 0 unconditionally on CONFIG_X86 or CONFIG_ARM.  I have no
> > idea why that is.  The original execshield patches had an #ifdef on
> > __i386__ but the patch that was commited to add PIE support has the
> > CONFIG_X86 setting.
> > 
> > Thoughts welcome.
> > 
> > [1] https://bugzilla.redhat.com/show_bug.cgi?id=708563
> > [2] http://bugzilla.kernel.org/show_bug.cgi?id=36372
> > 
> > josh
> > 
> > ---
> > 
> > From: H.J. Lu <hongjiu.lu@intel.com>
> > 
> > Set the load_bias for PIE executables to a non-zero address if no virtual
> > address is specified.  This prevents us from running out of room for all
> > the various loadable segments when ASLR is disabled.
> > 
> > Signed-off-by: H.J. Lu <hongjiu.lu@intel.com>
> > Signed-off-by: Josh Boyer <jwboyer@redhat.com>
> > 
> > ---
> > 
> > diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
> > index 303983f..069ee29 100644
> > --- a/fs/binfmt_elf.c
> > +++ b/fs/binfmt_elf.c
> > @@ -794,9 +794,14 @@ static int load_elf_binary(struct linux_binprm *bprm, struct pt_regs *regs)
> >  			/* Try and get dynamic programs out of the way of the
> >  			 * default mmap base, as well as whatever program they
> >  			 * might try to exec.  This is because the brk will
> > -			 * follow the loader, and is not movable.  */
> > +			 * follow the loader, and is not movable.  Don't use
> > +			 * 0 load address since we may not have room for
> > +			 * all loadable segements.  */
> >  #if defined(CONFIG_X86) || defined(CONFIG_ARM)
> > -			load_bias = 0;
> > +			if (vaddr)
> > +				load_bias = 0;
> > +			else
> > +				load_bias = ELF_PAGESTART(ELF_ET_DYN_BASE);
> 
> If you do this unconditionally, I can't see how ASLR could be working at 
> all -- this patch causes that elf_addr() is always called with non-zero 
> 'addr' argument, right? (haven't verified by actually testing the patch 
> yet, but I fail to see how come this simply doesn't break ASLR 
> completely).
> 
> Checking for PF_RANDOMIZE flag should be the way to go here.

I sent out a pseudo-patch that did that (I think).  Could you take a
look and let me know if that seems more appropriate?

josh