On 04/17, Peter Zijlstra wrote:
>
> On Mon, 2012-04-16 at 17:34 +0200, Oleg Nesterov wrote:
> > On 04/16, Peter Zijlstra wrote:
> > >
> > > Can't we 'optimize' read_opcode() by doing the pagefault_disable() +
> > > __copy_from_user_inatomic() optimistically before going down the whole
> > > gup()+lock+kmap path?
> >
> > Unlikely, the task is not current.
>
> Easy enough to test that though.. and that should make the regular path
> fast enough, no?
>
>
> ---
> kernel/events/uprobes.c | 9 +++++++++
> 1 files changed, 9 insertions(+), 0 deletions(-)
>
> diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
> index 985be4d..7f5d8c5 100644
> --- a/kernel/events/uprobes.c
> +++ b/kernel/events/uprobes.c
> @@ -312,6 +312,15 @@ static int read_opcode(struct mm_struct *mm, unsigned long vaddr, uprobe_opcode_
> void *vaddr_new;
> int ret;
>
> + if (mm == current->mm) {
> + pagefault_disable();
> + ret = __copy_from_user_inatomic(opcode, (void __user *)vaddr,
> + sizeof(*opcode));
> + pagefault_enable();
> + if (!ret)
> + return 0;
> + }
Indeed. And then we do not need is_swbp_at_addr_fast().
This reminds me. Why read_opcode() does lock_page? I was going
to send the cleanup which removes it, but I need to recheck.
Perhaps you can explain the reason?
Oleg.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>