[Devel] [PATCH rh7 1/2] ploop: Stop fast path reqs living their own lives
Konstantin Khorenko
khorenko at virtuozzo.com
Thu Mar 5 19:31:14 MSK 2020
On 03/03/2020 06:13 PM, Kirill Tkhai wrote:
> Doing ploop_quiesce() we want any IO becomes stopped.
> But fast path requests live their own lives, and this
> completelly ignores quiesce state. This may be a reason
> of unpredictable behaviour.
>
> This patch fixes the problem.
>
> Signed-off-by: Kirill Tkhai <ktkhai at virtuozzo.com>
> ---
> drivers/block/ploop/dev.c | 30 ++++++++++++++++++++++++++++--
> include/linux/ploop/ploop.h | 2 ++
> 2 files changed, 30 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/block/ploop/dev.c b/drivers/block/ploop/dev.c
> index 3f70f674e112..9e01866860bd 100644
> --- a/drivers/block/ploop/dev.c
> +++ b/drivers/block/ploop/dev.c
> @@ -164,6 +164,22 @@ static void ploop_uncongest(struct ploop_device *plo)
> }
> }
>
> +static int fast_path_reqs_count(struct ploop_device *plo)
> +{
> + int ret;
> +
> + spin_lock_irq(&plo->lock);
> + ret = plo->fastpath_reqs;
> + spin_unlock_irq(&plo->lock);
Do we really need to take a lock here?
> +
> + return ret;
> +}
> +
> +static void wait_fast_path_reqs(struct ploop_device *plo)
> +{
> + wait_event(plo->fast_path_waitq, fast_path_reqs_count(plo) == 0);
> +}
> +
> static void ploop_init_request(struct ploop_request *preq)
> {
> preq->eng_state = PLOOP_E_ENTRY;
> @@ -656,6 +672,8 @@ DEFINE_BIO_CB(ploop_fast_end_io)
> (test_bit(PLOOP_S_EXITING, &plo->state) ||
> !list_empty(&plo->entry_queue)))
> wake_up_interruptible(&plo->waitq);
> + if (plo->fast_path_disabled_count && !plo->fastpath_reqs)
> + wake_up(&plo->fast_path_waitq);
> spin_unlock_irqrestore(&plo->lock, flags);
>
> BIO_ENDIO(plo->queue, orig, err);
> @@ -969,7 +987,8 @@ static void ploop_make_request(struct request_queue *q, struct bio *bio)
>
> /* No fast path, when maintenance is in progress.
> * (PLOOP_S_TRACK was checked immediately above) */
> - if (FAST_PATH_DISABLED(plo->maintenance_type))
> + if (FAST_PATH_DISABLED(plo->maintenance_type) ||
> + plo->fast_path_disabled_count)
> goto queue;
>
> /* Attention state, always queue */
> @@ -3442,7 +3461,6 @@ static int ploop_replace_delta(struct ploop_device * plo, unsigned long arg)
> return err;
> }
>
> -
> void ploop_quiesce(struct ploop_device * plo)
> {
> struct completion qcomp;
> @@ -3452,6 +3470,8 @@ void ploop_quiesce(struct ploop_device * plo)
> return;
>
> spin_lock_irq(&plo->lock);
> + plo->fast_path_disabled_count++;
> +
> preq = ploop_alloc_request(plo, true);
> preq->req_rw = 0;
> preq->state = (1 << PLOOP_REQ_SYNC) | (1 << PLOOP_REQ_BARRIER);
> @@ -3469,6 +3489,7 @@ void ploop_quiesce(struct ploop_device * plo)
> wake_up_interruptible(&plo->waitq);
> spin_unlock_irq(&plo->lock);
>
> + wait_fast_path_reqs(plo);
> wait_for_completion(&qcomp);
> plo->quiesce_comp = NULL;
> }
> @@ -3476,6 +3497,10 @@ EXPORT_SYMBOL(ploop_quiesce);
>
> void ploop_relax(struct ploop_device * plo)
> {
> + spin_lock_irq(&plo->lock);
> + plo->fast_path_disabled_count--;
> + spin_unlock_irq(&plo->lock);
> +
> if (!test_bit(PLOOP_S_RUNNING, &plo->state))
> return;
>
> @@ -5456,6 +5481,7 @@ static struct ploop_device *__ploop_dev_alloc(int index)
> init_waitqueue_head(&plo->req_waitq);
> init_waitqueue_head(&plo->freeze_waitq);
> init_waitqueue_head(&plo->event_waitq);
> + init_waitqueue_head(&plo->fast_path_waitq);
> plo->tune = DEFAULT_PLOOP_TUNE;
> map_init(plo, &plo->map);
> track_init(plo);
> diff --git a/include/linux/ploop/ploop.h b/include/linux/ploop/ploop.h
> index 66591623f79d..8495ca2145fd 100644
> --- a/include/linux/ploop/ploop.h
> +++ b/include/linux/ploop/ploop.h
> @@ -407,6 +407,7 @@ struct ploop_device
> wait_queue_head_t req_waitq;
> wait_queue_head_t freeze_waitq;
> wait_queue_head_t event_waitq;
> + wait_queue_head_t fast_path_waitq;
>
> struct ploop_map map;
> struct ploop_map *trans_map;
> @@ -473,6 +474,7 @@ struct ploop_device
> struct block_device *dm_crypt_bdev;
>
> unsigned long locking_state; /* plo locked by userspace */
> + unsigned int fast_path_disabled_count;
> };
>
> enum
>
>
> .
>
More information about the Devel
mailing list