There is an unusual case that the card is busy when trying to send a
command, and we can not polling the card status in interrupt context
by using request_atomic() to dispatch requests.
Thus we should queue a work to try again in the non-atomic context
in case the host releases the busy signal later.
Signed-off-by: Baolin Wang <[email protected]>
---
drivers/mmc/host/mmc_hsq.c | 37 ++++++++++++++++++++++++++++++++++++-
drivers/mmc/host/mmc_hsq.h | 1 +
2 files changed, 37 insertions(+), 1 deletion(-)
diff --git a/drivers/mmc/host/mmc_hsq.c b/drivers/mmc/host/mmc_hsq.c
index fdbaa98..3edad11 100644
--- a/drivers/mmc/host/mmc_hsq.c
+++ b/drivers/mmc/host/mmc_hsq.c
@@ -15,11 +15,33 @@
#define HSQ_NUM_SLOTS 64
#define HSQ_INVALID_TAG HSQ_NUM_SLOTS
+static void mmc_hsq_retry_handler(struct work_struct *work)
+{
+ struct mmc_hsq *hsq = container_of(work, struct mmc_hsq, retry_work);
+ struct mmc_host *mmc = hsq->mmc;
+ struct mmc_request *mrq = hsq->mrq;
+ struct mmc_data *data = mrq->data;
+
+ if (mmc->ops->request) {
+ mmc->ops->request(mmc, mrq);
+ return;
+ }
+
+ /*
+ * If host does not supply the callback in normal context to
+ * handle request, just finish this request.
+ */
+ data->error = -EBUSY;
+ data->bytes_xfered = 0;
+ mmc_hsq_finalize_request(mmc, mrq);
+}
+
static void mmc_hsq_pump_requests(struct mmc_hsq *hsq)
{
struct mmc_host *mmc = hsq->mmc;
struct hsq_slot *slot;
unsigned long flags;
+ int ret = 0;
spin_lock_irqsave(&hsq->lock, flags);
@@ -42,9 +64,21 @@ static void mmc_hsq_pump_requests(struct mmc_hsq *hsq)
spin_unlock_irqrestore(&hsq->lock, flags);
if (mmc->ops->request_atomic)
- mmc->ops->request_atomic(mmc, hsq->mrq);
+ ret = mmc->ops->request_atomic(mmc, hsq->mrq);
else
mmc->ops->request(mmc, hsq->mrq);
+
+ /*
+ * If returning BUSY from request_atomic(), which means the card
+ * may be busy now, and we should change to non-atomic context to
+ * try again for this unusual case, to avoid time-consuming operations
+ * in the atomic context.
+ *
+ * Note: we can ignore other error cases, since the host driver
+ * will handle them.
+ */
+ if (ret == -EBUSY)
+ schedule_work(&hsq->retry_work);
}
static void mmc_hsq_update_next_tag(struct mmc_hsq *hsq, int remains)
@@ -327,6 +361,7 @@ int mmc_hsq_init(struct mmc_hsq *hsq, struct mmc_host *mmc)
hsq->mmc->cqe_private = hsq;
mmc->cqe_ops = &mmc_hsq_ops;
+ INIT_WORK(&hsq->retry_work, mmc_hsq_retry_handler);
spin_lock_init(&hsq->lock);
init_waitqueue_head(&hsq->wait_queue);
diff --git a/drivers/mmc/host/mmc_hsq.h b/drivers/mmc/host/mmc_hsq.h
index d51beb7..81f6c4f 100644
--- a/drivers/mmc/host/mmc_hsq.h
+++ b/drivers/mmc/host/mmc_hsq.h
@@ -12,6 +12,7 @@ struct mmc_hsq {
wait_queue_head_t wait_queue;
struct hsq_slot *slot;
spinlock_t lock;
+ struct work_struct retry_work;
int next_tag;
int num_slots;
--
1.9.1
On 19/03/20 12:54 pm, Baolin Wang wrote:
> There is an unusual case that the card is busy when trying to send a
> command, and we can not polling the card status in interrupt context
> by using request_atomic() to dispatch requests.
>
> Thus we should queue a work to try again in the non-atomic context
> in case the host releases the busy signal later.
I think this should be part of patch 1
>
> Signed-off-by: Baolin Wang <[email protected]>
Sorry for the slow reply.
> ---
> drivers/mmc/host/mmc_hsq.c | 37 ++++++++++++++++++++++++++++++++++++-
> drivers/mmc/host/mmc_hsq.h | 1 +
> 2 files changed, 37 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/host/mmc_hsq.c b/drivers/mmc/host/mmc_hsq.c
> index fdbaa98..3edad11 100644
> --- a/drivers/mmc/host/mmc_hsq.c
> +++ b/drivers/mmc/host/mmc_hsq.c
> @@ -15,11 +15,33 @@
> #define HSQ_NUM_SLOTS 64
> #define HSQ_INVALID_TAG HSQ_NUM_SLOTS
>
> +static void mmc_hsq_retry_handler(struct work_struct *work)
> +{
> + struct mmc_hsq *hsq = container_of(work, struct mmc_hsq, retry_work);
> + struct mmc_host *mmc = hsq->mmc;
> + struct mmc_request *mrq = hsq->mrq;
> + struct mmc_data *data = mrq->data;
> +
> + if (mmc->ops->request) {
->request() is not an optional mmc operation so checking it is not necessary.
> + mmc->ops->request(mmc, mrq);
> + return;
> + }
> +
> + /*
> + * If host does not supply the callback in normal context to
> + * handle request, just finish this request.
> + */
> + data->error = -EBUSY;
> + data->bytes_xfered = 0;
> + mmc_hsq_finalize_request(mmc, mrq);
> +}
> +
> static void mmc_hsq_pump_requests(struct mmc_hsq *hsq)
> {
> struct mmc_host *mmc = hsq->mmc;
> struct hsq_slot *slot;
> unsigned long flags;
> + int ret = 0;
>
> spin_lock_irqsave(&hsq->lock, flags);
>
> @@ -42,9 +64,21 @@ static void mmc_hsq_pump_requests(struct mmc_hsq *hsq)
> spin_unlock_irqrestore(&hsq->lock, flags);
>
> if (mmc->ops->request_atomic)
> - mmc->ops->request_atomic(mmc, hsq->mrq);
> + ret = mmc->ops->request_atomic(mmc, hsq->mrq);
> else
> mmc->ops->request(mmc, hsq->mrq);
> +
> + /*
> + * If returning BUSY from request_atomic(), which means the card
> + * may be busy now, and we should change to non-atomic context to
> + * try again for this unusual case, to avoid time-consuming operations
> + * in the atomic context.
> + *
> + * Note: we can ignore other error cases, since the host driver
> + * will handle them.
> + */
> + if (ret == -EBUSY)
> + schedule_work(&hsq->retry_work);
Let's add a warning for unexpected return values i.e.
WARN_ON_ONCE(ret && ret != -EBUSY);
> }
>
> static void mmc_hsq_update_next_tag(struct mmc_hsq *hsq, int remains)
> @@ -327,6 +361,7 @@ int mmc_hsq_init(struct mmc_hsq *hsq, struct mmc_host *mmc)
> hsq->mmc->cqe_private = hsq;
> mmc->cqe_ops = &mmc_hsq_ops;
>
> + INIT_WORK(&hsq->retry_work, mmc_hsq_retry_handler);
> spin_lock_init(&hsq->lock);
> init_waitqueue_head(&hsq->wait_queue);
>
> diff --git a/drivers/mmc/host/mmc_hsq.h b/drivers/mmc/host/mmc_hsq.h
> index d51beb7..81f6c4f 100644
> --- a/drivers/mmc/host/mmc_hsq.h
> +++ b/drivers/mmc/host/mmc_hsq.h
> @@ -12,6 +12,7 @@ struct mmc_hsq {
> wait_queue_head_t wait_queue;
> struct hsq_slot *slot;
> spinlock_t lock;
> + struct work_struct retry_work;
>
> int next_tag;
> int num_slots;
>
Hi Adrian,
On Thu, Apr 2, 2020 at 6:45 PM Adrian Hunter <[email protected]> wrote:
>
> On 19/03/20 12:54 pm, Baolin Wang wrote:
> > There is an unusual case that the card is busy when trying to send a
> > command, and we can not polling the card status in interrupt context
> > by using request_atomic() to dispatch requests.
> >
> > Thus we should queue a work to try again in the non-atomic context
> > in case the host releases the busy signal later.
>
> I think this should be part of patch 1
OK. Will move these changes into patch 1.
>
> >
> > Signed-off-by: Baolin Wang <[email protected]>
>
> Sorry for the slow reply.
>
> > ---
> > drivers/mmc/host/mmc_hsq.c | 37 ++++++++++++++++++++++++++++++++++++-
> > drivers/mmc/host/mmc_hsq.h | 1 +
> > 2 files changed, 37 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/mmc/host/mmc_hsq.c b/drivers/mmc/host/mmc_hsq.c
> > index fdbaa98..3edad11 100644
> > --- a/drivers/mmc/host/mmc_hsq.c
> > +++ b/drivers/mmc/host/mmc_hsq.c
> > @@ -15,11 +15,33 @@
> > #define HSQ_NUM_SLOTS 64
> > #define HSQ_INVALID_TAG HSQ_NUM_SLOTS
> >
> > +static void mmc_hsq_retry_handler(struct work_struct *work)
> > +{
> > + struct mmc_hsq *hsq = container_of(work, struct mmc_hsq, retry_work);
> > + struct mmc_host *mmc = hsq->mmc;
> > + struct mmc_request *mrq = hsq->mrq;
> > + struct mmc_data *data = mrq->data;
> > +
> > + if (mmc->ops->request) {
>
> ->request() is not an optional mmc operation so checking it is not necessary.
Yes, will remove the checking.
>
> > + mmc->ops->request(mmc, mrq);
> > + return;
> > + }
> > +
> > + /*
> > + * If host does not supply the callback in normal context to
> > + * handle request, just finish this request.
> > + */
> > + data->error = -EBUSY;
> > + data->bytes_xfered = 0;
> > + mmc_hsq_finalize_request(mmc, mrq);
> > +}
> > +
> > static void mmc_hsq_pump_requests(struct mmc_hsq *hsq)
> > {
> > struct mmc_host *mmc = hsq->mmc;
> > struct hsq_slot *slot;
> > unsigned long flags;
> > + int ret = 0;
> >
> > spin_lock_irqsave(&hsq->lock, flags);
> >
> > @@ -42,9 +64,21 @@ static void mmc_hsq_pump_requests(struct mmc_hsq *hsq)
> > spin_unlock_irqrestore(&hsq->lock, flags);
> >
> > if (mmc->ops->request_atomic)
> > - mmc->ops->request_atomic(mmc, hsq->mrq);
> > + ret = mmc->ops->request_atomic(mmc, hsq->mrq);
> > else
> > mmc->ops->request(mmc, hsq->mrq);
> > +
> > + /*
> > + * If returning BUSY from request_atomic(), which means the card
> > + * may be busy now, and we should change to non-atomic context to
> > + * try again for this unusual case, to avoid time-consuming operations
> > + * in the atomic context.
> > + *
> > + * Note: we can ignore other error cases, since the host driver
> > + * will handle them.
> > + */
> > + if (ret == -EBUSY)
> > + schedule_work(&hsq->retry_work);
>
> Let's add a warning for unexpected return values i.e.
>
> WARN_ON_ONCE(ret && ret != -EBUSY);
Sure. Thanks for your comments.
>
>
> > }
> >
> > static void mmc_hsq_update_next_tag(struct mmc_hsq *hsq, int remains)
> > @@ -327,6 +361,7 @@ int mmc_hsq_init(struct mmc_hsq *hsq, struct mmc_host *mmc)
> > hsq->mmc->cqe_private = hsq;
> > mmc->cqe_ops = &mmc_hsq_ops;
> >
> > + INIT_WORK(&hsq->retry_work, mmc_hsq_retry_handler);
> > spin_lock_init(&hsq->lock);
> > init_waitqueue_head(&hsq->wait_queue);
> >
> > diff --git a/drivers/mmc/host/mmc_hsq.h b/drivers/mmc/host/mmc_hsq.h
> > index d51beb7..81f6c4f 100644
> > --- a/drivers/mmc/host/mmc_hsq.h
> > +++ b/drivers/mmc/host/mmc_hsq.h
> > @@ -12,6 +12,7 @@ struct mmc_hsq {
> > wait_queue_head_t wait_queue;
> > struct hsq_slot *slot;
> > spinlock_t lock;
> > + struct work_struct retry_work;
> >
> > int next_tag;
> > int num_slots;
> >
>
--
Baolin Wang