Re: serious performance regression due to NX patch

[Jakub Jelinek <jakub@redhat.com>]

On Sun, Jul 11, 2004 at 05:52:59AM -0400, Ingo Molnar wrote:
> 
> On Sun, 11 Jul 2004, Ingo Molnar wrote:
> 
> > > ok, agreed. I'll check that it still does the right thing on x86.
> > 
> > it doesnt seem to do the right thing for !PT_GNU_STACK applications on 
> > x86:
> 
> how about the patch below? This way we recognize the fact that x86 didnt
> have any executability check previously at the point where we discover
> that it's a 'legacy' binary.
> 
> 	Ingo
> 
> --- linux/fs/binfmt_elf.c.orig3	
> +++ linux/fs/binfmt_elf.c	
> @@ -627,8 +627,14 @@ static int load_elf_binary(struct linux_
>  				executable_stack = EXSTACK_DISABLE_X;
>  			break;
>  		}
> +#ifdef __i386_
> +	/*
> +	 * Legacy x86 binaries have an expectation of executability for
> +	 * virtually all their address-space - turn executability on:
> +	 */
>  	if (i == elf_ex.e_phnum)
>  		def_flags |= VM_EXEC | VM_MAYEXEC;
> +#endif

This looks incorrect.
There are many arches where legacy binaries expect the executability for
virtually all their address-space (my guess is all but x86-64 and ia64),
and even on those two legacy binaries expected at least stack executable.

And on x86-64 and ia64 ia32 binaries (i.e. when binfmt_elf.c is included
in their binfmt_elf32.c or how is it called) should have VM_EXEC and
VM_MAYEXEC in def_flags.

	Jakub