Re: [PATCHv2] backing-dev: fix wakeup timer races with bdi_unregister()

[Rabin Vincent <rabin@rab.in>]

On Fri, Jan 20, 2012 at 08:18:31PM +0900, Namjae Jeon wrote:
> The following case is not possible with racing of __mark_inode_dirty  ?
> --------------------------------------------------------------------------------------------------
> spin_lock
> bdi->dev = NULL;
> spin_unlock
>                               spin_lock
>                                ......
>                                ......
>                                trace_writeback_wake_forker_thread(bdi);
>                                spin_unlock
> -----------------------------------------------------------------------------

After this patch, this is what wakeup_timer_fn looks like:

static void wakeup_timer_fn(unsigned long data)
{
        struct backing_dev_info *bdi = (struct backing_dev_info *)data;

        spin_lock_bh(&bdi->wb_lock);
        if (bdi->wb.task) {
                trace_writeback_wake_thread(bdi);
                wake_up_process(bdi->wb.task);
        } else if (bdi->dev) {
                /*
                 * When bdi tasks are inactive for long time, they are killed.
                 * In this case we have to wake-up the forker thread which
                 * should create and run the bdi thread.
                 */
                trace_writeback_wake_forker_thread(bdi);
                wake_up_process(default_backing_dev_info.wb.task);
        }
        spin_unlock_bh(&bdi->wb_lock);
}

So how will trace_writeback_wake_forker_thread() be called if bdi->dev is NULL?

This patch added the if (bdi->dev) check, perhaps you overlooked that?