Re: [PATCH 5/5] x86,mm: make pagefault killable

[Oleg Nesterov <oleg@redhat.com>]

On 03/22, KOSAKI Motohiro wrote:
>
> This patch makes pagefault interruptible by SIGKILL.

Not a comment, but the question...

> --- a/arch/x86/mm/fault.c
> +++ b/arch/x86/mm/fault.c
> @@ -1035,6 +1035,7 @@ do_page_fault(struct pt_regs *regs, unsigned long error_code)
>  	if (user_mode_vm(regs)) {
>  		local_irq_enable();
>  		error_code |= PF_USER;
> +		flags |= FAULT_FLAG_KILLABLE;

OK, this is clear.

I am wondering, can't we set FAULT_FLAG_KILLABLE unconditionally
but check PF_USER when we get VM_FAULT_RETRY? I mean,

	if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current)) {
		if (!(error_code & PF_USER))
			no_context(...);
		return;
	}


Probably not... but I can't find any example of in-kernel fault which
can be broken by -EFAULT if current was killed.

mm_release()->put_user(clear_child_tid) should be fine...

Just curious, I feel I missed something obvious.

Oleg.