On Tue 11-03-14 11:23:55, Andrew Morton wrote: > On Wed, 19 Feb 2014 14:01:39 -0500 Tejun Heo <tj@xxxxxxxxxx> wrote: > > > Hello, Jan. > > > > On Wed, Feb 19, 2014 at 10:27:31AM +0100, Jan Kara wrote: > > > You are the workqueue expert so you may know better ;) But the way I > > > understand it is that queue_delayed_work() does nothing if the timer is > > > already running. Since we queue flusher work to run either immediately or > > > after dirty_writeback_interval we are safe to run queue_delayed_work() > > > whenever we want it to run after dirty_writeback_interval and > > > mod_delayed_work() whenever we want to run it immediately. > > > > Ah, okay, so it's always mod on immediate and queue on delayed. Yeah, > > that should work. > > > > > But it's subtle and some interface where we could say queue delayed work > > > after no later than X would be easier to grasp. > > > > Yeah, I think it'd be better if we had something like > > mod_delayed_work_if_later(). Hmm... > > The code comments which you asked for were not forthcoming. > > Are you otherwise OK with merging this into 3.14 and -stable? I have actually added the comment when nothing happened and have a patch with Tejun's Reviewed-by. Just noone picked it up. I'll send it your way together with another fix in flush work handling. Honza > From: Derek Basehore <dbasehore@xxxxxxxxxxxx> > Subject: backing_dev: fix hung task on sync > > bdi_wakeup_thread_delayed() used mod_delayed_work() to schedule work to > writeback dirty inodes. The problem with this is that it can delay work > that is scheduled for immediate execution, such as the work from > sync_inodes_sb(). This can happen since mod_delayed_work can now steal > work from a work_queue. This fixes the problem by using > queue_delayed_work instead. This is a regression from the move to the bdi > workqueue design. > > The reason that this causes a problem is that laptop-mode will change the > delay, dirty_writeback_centisecs, to 60000 (10 minutes) by default. In > the case that bdi_wakeup_thread_delayed races with sync_inodes_sb, sync > will be stopped for 10 minutes and trigger a hung task. Even if > dirty_writeback_centisecs is not long enough to cause a hung task, we > still don't want to delay sync for that long. > > For the same reason, this also changes bdi_writeback_workfn to immediately > queue the work again in the case that the work_list is not empty. The > same problem can happen if the sync work is run on the rescue worker. > > Signed-off-by: Derek Basehore <dbasehore@xxxxxxxxxxxx> > Reviewed-by: Jan Kara <jack@xxxxxxx> > Cc: Alexander Viro <viro@xxxxxxxxxxxxxxxxxx> > Cc: Tejun Heo <tj@xxxxxxxxxx> > Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> > Cc: "Darrick J. Wong" <darrick.wong@xxxxxxxxxx> > Cc: Derek Basehore <dbasehore@xxxxxxxxxxxx> > Cc: Kees Cook <keescook@xxxxxxxxxxxx> > Cc: Benson Leung <bleung@xxxxxxxxxxxx> > Cc: Sonny Rao <sonnyrao@xxxxxxxxxxxx> > Cc: Luigi Semenzato <semenzato@xxxxxxxxxxxx> > Cc: Jens Axboe <axboe@xxxxxxxxx> > Cc: Dave Chinner <david@xxxxxxxxxxxxx> > Cc: <stable@xxxxxxxxxxxxxxx> > Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> > --- > > fs/fs-writeback.c | 5 +++-- > mm/backing-dev.c | 2 +- > 2 files changed, 4 insertions(+), 3 deletions(-) > > diff -puN fs/fs-writeback.c~backing_dev-fix-hung-task-on-sync fs/fs-writeback.c > --- a/fs/fs-writeback.c~backing_dev-fix-hung-task-on-sync > +++ a/fs/fs-writeback.c > @@ -1039,8 +1039,9 @@ void bdi_writeback_workfn(struct work_st > trace_writeback_pages_written(pages_written); > } > > - if (!list_empty(&bdi->work_list) || > - (wb_has_dirty_io(wb) && dirty_writeback_interval)) > + if (!list_empty(&bdi->work_list)) > + mod_delayed_work(bdi_wq, &wb->dwork, 0); > + else if (wb_has_dirty_io(wb) && dirty_writeback_interval) > queue_delayed_work(bdi_wq, &wb->dwork, > msecs_to_jiffies(dirty_writeback_interval * 10)); > > diff -puN mm/backing-dev.c~backing_dev-fix-hung-task-on-sync mm/backing-dev.c > --- a/mm/backing-dev.c~backing_dev-fix-hung-task-on-sync > +++ a/mm/backing-dev.c > @@ -294,7 +294,7 @@ void bdi_wakeup_thread_delayed(struct ba > unsigned long timeout; > > timeout = msecs_to_jiffies(dirty_writeback_interval * 10); > - mod_delayed_work(bdi_wq, &bdi->wb.dwork, timeout); > + queue_delayed_work(bdi_wq, &bdi->wb.dwork, timeout); > } > > /* > _ > -- Jan Kara <jack@xxxxxxx> SUSE Labs, CR -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
