[PATCH v3 7/9] bus: fsl-mc: dpio: add the DPAA2 DPIO service interface

[Stuart Yoder]

From: Roy Pledge <Roy.Pledge@xxxxxxx>

The DPIO service interface handles initialization of DPIO objects
and exports APIs to be used by other DPAA2 object drivers to perform
queuing and buffer management related operations.  The service allows
registration of callbacks when frames or notifications are received.

Signed-off-by: Roy Pledge <roy.pledge@xxxxxxx>
Signed-off-by: Haiying Wang <haiying.wang@xxxxxxx>
Signed-off-by: Stuart Yoder <stuart.yoder@xxxxxxx>
---

Notes:
    -v3
       -zero memory allocated for a dpio store (bug fix suggested
        by Ioana Radulescu)
    -v2
       -use service_select_by_cpu() for re-arming DPIO interrupts
       -replace use of NR_CPUS with num_possible_cpus()

 drivers/bus/fsl-mc/dpio/Makefile       |   2 +-
 drivers/bus/fsl-mc/dpio/dpio-service.c | 614 +++++++++++++++++++++++++++++++++
 include/linux/fsl/dpaa2-io.h           | 138 ++++++++
 3 files changed, 753 insertions(+), 1 deletion(-)
 create mode 100644 drivers/bus/fsl-mc/dpio/dpio-service.c
 create mode 100644 include/linux/fsl/dpaa2-io.h

diff --git a/drivers/bus/fsl-mc/dpio/Makefile b/drivers/bus/fsl-mc/dpio/Makefile
index 6588498..0778da7 100644
--- a/drivers/bus/fsl-mc/dpio/Makefile
+++ b/drivers/bus/fsl-mc/dpio/Makefile
@@ -6,4 +6,4 @@ subdir-ccflags-y := -Werror
 
 obj-$(CONFIG_FSL_MC_DPIO) += fsl-mc-dpio.o
 
-fsl-mc-dpio-objs := dpio.o qbman-portal.o
+fsl-mc-dpio-objs := dpio.o qbman-portal.o dpio-service.o
diff --git a/drivers/bus/fsl-mc/dpio/dpio-service.c b/drivers/bus/fsl-mc/dpio/dpio-service.c
new file mode 100644
index 0000000..01f0293
--- /dev/null
+++ b/drivers/bus/fsl-mc/dpio/dpio-service.c
@@ -0,0 +1,614 @@
+/*
+ * Copyright 2014-2016 Freescale Semiconductor Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *     * Redistributions of source code must retain the above copyright
+ *	 notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above copyright
+ *	 notice, this list of conditions and the following disclaimer in the
+ *	 documentation and/or other materials provided with the distribution.
+ *     * Neither the name of Freescale Semiconductor nor the
+ *	 names of its contributors may be used to endorse or promote products
+ *	 derived from this software without specific prior written permission.
+ *
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Freescale Semiconductor ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL Freescale Semiconductor BE LIABLE FOR ANY
+ * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include <linux/types.h>
+#include <linux/fsl/mc.h>
+#include <linux/fsl/dpaa2-io.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/interrupt.h>
+#include <linux/dma-mapping.h>
+#include <linux/slab.h>
+
+#include "dpio.h"
+#include "qbman-portal.h"
+
+struct dpaa2_io {
+	atomic_t refs;
+	struct dpaa2_io_desc dpio_desc;
+	struct qbman_swp_desc swp_desc;
+	struct qbman_swp *swp;
+	struct list_head node;
+	spinlock_t lock_mgmt_cmd;
+	spinlock_t lock_notifications;
+	struct list_head notifications;
+};
+
+struct dpaa2_io_store {
+	unsigned int max;
+	dma_addr_t paddr;
+	struct dpaa2_dq *vaddr;
+	void *alloced_addr;    /* unaligned value from kmalloc() */
+	unsigned int idx;      /* position of the next-to-be-returned entry */
+	struct qbman_swp *swp; /* portal used to issue VDQCR */
+	struct device *dev;    /* device used for DMA mapping */
+};
+
+/* keep a per cpu array of DPIOs for fast access */
+static struct dpaa2_io *dpio_by_cpu[NR_CPUS];
+static struct list_head dpio_list = LIST_HEAD_INIT(dpio_list);
+static DEFINE_SPINLOCK(dpio_list_lock);
+
+static inline struct dpaa2_io *service_select_by_cpu(struct dpaa2_io *d,
+						     int cpu)
+{
+	if (d)
+		return d;
+
+	if (unlikely(cpu >= num_possible_cpus()))
+		return NULL;
+
+	/*
+	 * If cpu == -1, choose the current cpu, with no guarantees about
+	 * potentially being migrated away.
+	 */
+	if (unlikely(cpu < 0))
+		cpu = smp_processor_id();
+
+	/* If a specific cpu was requested, pick it up immediately */
+	return dpio_by_cpu[cpu];
+}
+
+static inline struct dpaa2_io *service_select(struct dpaa2_io *d)
+{
+	if (d)
+		return d;
+
+	spin_lock(&dpio_list_lock);
+	d = list_entry(dpio_list.next, struct dpaa2_io, node);
+	list_del(&d->node);
+	list_add_tail(&d->node, &dpio_list);
+	spin_unlock(&dpio_list_lock);
+
+	return d;
+}
+
+/**
+ * dpaa2_io_create() - create a dpaa2_io object.
+ * @desc: the dpaa2_io descriptor
+ *
+ * Activates a "struct dpaa2_io" corresponding to the given config of an actual
+ * DPIO object.
+ *
+ * Return a valid dpaa2_io object for success, or NULL for failure.
+ */
+struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc)
+{
+	struct dpaa2_io *obj = kmalloc(sizeof(*obj), GFP_KERNEL);
+
+	if (!obj)
+		return NULL;
+
+	/* check if CPU is out of range (-1 means any cpu) */
+	if (desc->cpu >= num_possible_cpus())
+		return NULL;
+
+	atomic_set(&obj->refs, 1);
+	obj->dpio_desc = *desc;
+	obj->swp_desc.cena_bar = obj->dpio_desc.regs_cena;
+	obj->swp_desc.cinh_bar = obj->dpio_desc.regs_cinh;
+	obj->swp_desc.qman_version = obj->dpio_desc.qman_version;
+	obj->swp = qbman_swp_init(&obj->swp_desc);
+
+	if (!obj->swp) {
+		kfree(obj);
+		return NULL;
+	}
+
+	INIT_LIST_HEAD(&obj->node);
+	spin_lock_init(&obj->lock_mgmt_cmd);
+	spin_lock_init(&obj->lock_notifications);
+	INIT_LIST_HEAD(&obj->notifications);
+
+	/* For now only enable DQRR interrupts */
+	qbman_swp_interrupt_set_trigger(obj->swp,
+					QBMAN_SWP_INTERRUPT_DQRI);
+	qbman_swp_interrupt_clear_status(obj->swp, 0xffffffff);
+	if (obj->dpio_desc.receives_notifications)
+		qbman_swp_push_set(obj->swp, 0, 1);
+
+	spin_lock(&dpio_list_lock);
+	list_add_tail(&obj->node, &dpio_list);
+	if (desc->cpu >= 0 && !dpio_by_cpu[desc->cpu])
+		dpio_by_cpu[desc->cpu] = obj;
+	spin_unlock(&dpio_list_lock);
+
+	return obj;
+}
+EXPORT_SYMBOL(dpaa2_io_create);
+
+/**
+ * dpaa2_io_down() - release the dpaa2_io object.
+ * @d: the dpaa2_io object to be released.
+ *
+ * The "struct dpaa2_io" type can represent an individual DPIO object (as
+ * described by "struct dpaa2_io_desc") or an instance of a "DPIO service",
+ * which can be used to group/encapsulate multiple DPIO objects. In all cases,
+ * each handle obtained should be released using this function.
+ */
+void dpaa2_io_down(struct dpaa2_io *d)
+{
+	if (!atomic_dec_and_test(&d->refs))
+		return;
+	kfree(d);
+}
+EXPORT_SYMBOL(dpaa2_io_down);
+
+#define DPAA_POLL_MAX 32
+
+/**
+ * dpaa2_io_irq() - ISR for DPIO interrupts
+ *
+ * @obj: the given DPIO object.
+ *
+ * Return IRQ_HANDLED for success or IRQ_NONE if there
+ * were no pending interrupts.
+ */
+irqreturn_t dpaa2_io_irq(struct dpaa2_io *obj)
+{
+	const struct dpaa2_dq *dq;
+	int max = 0;
+	struct qbman_swp *swp;
+	u32 status;
+
+	swp = obj->swp;
+	status = qbman_swp_interrupt_read_status(swp);
+	if (!status)
+		return IRQ_NONE;
+
+	dq = qbman_swp_dqrr_next(swp);
+	while (dq) {
+		if (qbman_result_is_SCN(dq)) {
+			struct dpaa2_io_notification_ctx *ctx;
+			u64 q64;
+
+			q64 = qbman_result_SCN_ctx(dq);
+			ctx = (void *)q64;
+			ctx->cb(ctx);
+		} else {
+			pr_crit("fsl-mc-dpio: Unrecognised/ignored DQRR entry\n");
+		}
+		qbman_swp_dqrr_consume(swp, dq);
+		++max;
+		if (max > DPAA_POLL_MAX)
+			goto done;
+		dq = qbman_swp_dqrr_next(swp);
+	}
+done:
+	qbman_swp_interrupt_clear_status(swp, status);
+	qbman_swp_interrupt_set_inhibit(swp, 0);
+	return IRQ_HANDLED;
+}
+EXPORT_SYMBOL(dpaa2_io_irq);
+
+/**
+ * dpaa2_io_service_register() - Prepare for servicing of FQDAN or CDAN
+ *                               notifications on the given DPIO service.
+ * @d:   the given DPIO service.
+ * @ctx: the notification context.
+ *
+ * The caller should make the MC command to attach a DPAA2 object to
+ * a DPIO after this function completes successfully.  In that way:
+ *    (a) The DPIO service is "ready" to handle a notification arrival
+ *        (which might happen before the "attach" command to MC has
+ *        returned control of execution back to the caller)
+ *    (b) The DPIO service can provide back to the caller the 'dpio_id' and
+ *        'qman64' parameters that it should pass along in the MC command
+ *        in order for the object to be configured to produce the right
+ *        notification fields to the DPIO service.
+ *
+ * Return 0 for success, or -ENODEV for failure.
+ */
+int dpaa2_io_service_register(struct dpaa2_io *d,
+			      struct dpaa2_io_notification_ctx *ctx)
+{
+	unsigned long irqflags;
+
+	d = service_select_by_cpu(d, ctx->desired_cpu);
+	if (!d)
+		return -ENODEV;
+
+	ctx->dpio_id = d->dpio_desc.dpio_id;
+	ctx->qman64 = (u64)ctx;
+	ctx->dpio_private = d;
+	spin_lock_irqsave(&d->lock_notifications, irqflags);
+	list_add(&ctx->node, &d->notifications);
+	spin_unlock_irqrestore(&d->lock_notifications, irqflags);
+
+	/* Enable the generation of CDAN notifications */
+	if (ctx->is_cdan)
+		qbman_swp_CDAN_set_context_enable(d->swp,
+						  (u16)ctx->id,
+						  ctx->qman64);
+	return 0;
+}
+EXPORT_SYMBOL(dpaa2_io_service_register);
+
+/**
+ * dpaa2_io_service_deregister - The opposite of 'register'.
+ * @service: the given DPIO service.
+ * @ctx: the notification context.
+ *
+ * This function should be called only after sending the MC command to
+ * to detach the notification-producing device from the DPIO.
+ */
+void dpaa2_io_service_deregister(struct dpaa2_io *service,
+				struct dpaa2_io_notification_ctx *ctx)
+{
+	struct dpaa2_io *d = ctx->dpio_private;
+	unsigned long irqflags;
+
+	if (ctx->is_cdan)
+		qbman_swp_CDAN_disable(d->swp, (u16)ctx->id);
+
+	spin_lock_irqsave(&d->lock_notifications, irqflags);
+	list_del(&ctx->node);
+	spin_unlock_irqrestore(&d->lock_notifications, irqflags);
+}
+EXPORT_SYMBOL(dpaa2_io_service_deregister);
+
+/**
+ * dpaa2_io_service_rearm() - Rearm the notification for the given DPIO service.
+ * @service: the given DPIO service.
+ * @ctx: the notification context.
+ *
+ * Once a FQDAN/CDAN has been produced, the corresponding FQ/channel is
+ * considered "disarmed". Ie. the user can issue pull dequeue operations on that
+ * traffic source for as long as it likes. Eventually it may wish to "rearm"
+ * that source to allow it to produce another FQDAN/CDAN, that's what this
+ * function achieves.
+ *
+ * Return 0 for success.
+ */
+int dpaa2_io_service_rearm(struct dpaa2_io *d,
+			   struct dpaa2_io_notification_ctx *ctx)
+{
+	unsigned long irqflags;
+	int err;
+
+	d = service_select_by_cpu(d, ctx->desired_cpu);
+	if (!unlikely(d))
+		return -ENODEV;
+
+	spin_lock_irqsave(&d->lock_mgmt_cmd, irqflags);
+	if (ctx->is_cdan)
+		err = qbman_swp_CDAN_enable(d->swp, (u16)ctx->id);
+	else
+		err = qbman_swp_fq_schedule(d->swp, ctx->id);
+	spin_unlock_irqrestore(&d->lock_mgmt_cmd, irqflags);
+
+	return err;
+}
+EXPORT_SYMBOL(dpaa2_io_service_rearm);
+
+/**
+ * dpaa2_io_service_pull_fq() - pull dequeue functions from a fq.
+ * @d: the given DPIO service.
+ * @fqid: the given frame queue id.
+ * @s: the dpaa2_io_store object for the result.
+ *
+ * Return 0 for success, or error code for failure.
+ */
+int dpaa2_io_service_pull_fq(struct dpaa2_io *d, u32 fqid,
+			     struct dpaa2_io_store *s)
+{
+	struct qbman_pull_desc pd;
+	int err;
+
+	qbman_pull_desc_clear(&pd);
+	qbman_pull_desc_set_storage(&pd, s->vaddr, s->paddr, 1);
+	qbman_pull_desc_set_numframes(&pd, (u8)s->max);
+	qbman_pull_desc_set_fq(&pd, fqid);
+
+	d = service_select(d);
+	if (!d)
+		return -ENODEV;
+	s->swp = d->swp;
+	err = qbman_swp_pull(d->swp, &pd);
+	if (err)
+		s->swp = NULL;
+
+	return err;
+}
+EXPORT_SYMBOL(dpaa2_io_service_pull_fq);
+
+/**
+ * dpaa2_io_service_pull_channel() - pull dequeue functions from a channel.
+ * @d: the given DPIO service.
+ * @channelid: the given channel id.
+ * @s: the dpaa2_io_store object for the result.
+ *
+ * Return 0 for success, or error code for failure.
+ */
+int dpaa2_io_service_pull_channel(struct dpaa2_io *d, u32 channelid,
+				  struct dpaa2_io_store *s)
+{
+	struct qbman_pull_desc pd;
+	int err;
+
+	qbman_pull_desc_clear(&pd);
+	qbman_pull_desc_set_storage(&pd, s->vaddr, s->paddr, 1);
+	qbman_pull_desc_set_numframes(&pd, (u8)s->max);
+	qbman_pull_desc_set_channel(&pd, channelid, qbman_pull_type_prio);
+
+	d = service_select(d);
+	if (!d)
+		return -ENODEV;
+
+	s->swp = d->swp;
+	err = qbman_swp_pull(d->swp, &pd);
+	if (err)
+		s->swp = NULL;
+
+	return err;
+}
+EXPORT_SYMBOL(dpaa2_io_service_pull_channel);
+
+/**
+ * dpaa2_io_service_enqueue_fq() - Enqueue a frame to a frame queue.
+ * @d: the given DPIO service.
+ * @fqid: the given frame queue id.
+ * @fd: the frame descriptor which is enqueued.
+ *
+ * Return 0 for successful enqueue, -EBUSY if the enqueue ring is not ready,
+ * or -ENODEV if there is no dpio service.
+ */
+int dpaa2_io_service_enqueue_fq(struct dpaa2_io *d,
+				u32 fqid,
+				const struct dpaa2_fd *fd)
+{
+	struct qbman_eq_desc ed;
+
+	d = service_select(d);
+	if (!d)
+		return -ENODEV;
+
+	qbman_eq_desc_clear(&ed);
+	qbman_eq_desc_set_no_orp(&ed, 0);
+	qbman_eq_desc_set_fq(&ed, fqid);
+
+	return qbman_swp_enqueue(d->swp, &ed, fd);
+}
+EXPORT_SYMBOL(dpaa2_io_service_enqueue_fq);
+
+/**
+ * dpaa2_io_service_enqueue_qd() - Enqueue a frame to a QD.
+ * @d: the given DPIO service.
+ * @qdid: the given queuing destination id.
+ * @prio: the given queuing priority.
+ * @qdbin: the given queuing destination bin.
+ * @fd: the frame descriptor which is enqueued.
+ *
+ * Return 0 for successful enqueue, or -EBUSY if the enqueue ring is not ready,
+ * or -ENODEV if there is no dpio service.
+ */
+int dpaa2_io_service_enqueue_qd(struct dpaa2_io *d,
+				u32 qdid, u8 prio, u16 qdbin,
+				const struct dpaa2_fd *fd)
+{
+	struct qbman_eq_desc ed;
+
+	d = service_select(d);
+	if (!d)
+		return -ENODEV;
+
+	qbman_eq_desc_clear(&ed);
+	qbman_eq_desc_set_no_orp(&ed, 0);
+	qbman_eq_desc_set_qd(&ed, qdid, qdbin, prio);
+
+	return qbman_swp_enqueue(d->swp, &ed, fd);
+}
+EXPORT_SYMBOL(dpaa2_io_service_enqueue_qd);
+
+/**
+ * dpaa2_io_service_release() - Release buffers to a buffer pool.
+ * @d: the given DPIO object.
+ * @bpid: the buffer pool id.
+ * @buffers: the buffers to be released.
+ * @num_buffers: the number of the buffers to be released.
+ *
+ * Return 0 for success, and negative error code for failure.
+ */
+int dpaa2_io_service_release(struct dpaa2_io *d,
+			     u32 bpid,
+			     const u64 *buffers,
+			     unsigned int num_buffers)
+{
+	struct qbman_release_desc rd;
+
+	d = service_select(d);
+	if (!d)
+		return -ENODEV;
+
+	qbman_release_desc_clear(&rd);
+	qbman_release_desc_set_bpid(&rd, bpid);
+
+	return qbman_swp_release(d->swp, &rd, buffers, num_buffers);
+}
+EXPORT_SYMBOL(dpaa2_io_service_release);
+
+/**
+ * dpaa2_io_service_acquire() - Acquire buffers from a buffer pool.
+ * @d: the given DPIO object.
+ * @bpid: the buffer pool id.
+ * @buffers: the buffer addresses for acquired buffers.
+ * @num_buffers: the expected number of the buffers to acquire.
+ *
+ * Return a negative error code if the command failed, otherwise it returns
+ * the number of buffers acquired, which may be less than the number requested.
+ * Eg. if the buffer pool is empty, this will return zero.
+ */
+int dpaa2_io_service_acquire(struct dpaa2_io *d,
+			     u32 bpid,
+			     u64 *buffers,
+			     unsigned int num_buffers)
+{
+	unsigned long irqflags;
+	int err;
+
+	d = service_select(d);
+	if (!d)
+		return -ENODEV;
+
+	spin_lock_irqsave(&d->lock_mgmt_cmd, irqflags);
+	err = qbman_swp_acquire(d->swp, bpid, buffers, num_buffers);
+	spin_unlock_irqrestore(&d->lock_mgmt_cmd, irqflags);
+
+	return err;
+}
+EXPORT_SYMBOL(dpaa2_io_service_acquire);
+
+/*
+ * 'Stores' are reusable memory blocks for holding dequeue results, and to
+ * assist with parsing those results.
+ */
+
+/**
+ * dpaa2_io_store_create() - Create the dma memory storage for dequeue result.
+ * @max_frames: the maximum number of dequeued result for frames, must be <= 16.
+ * @dev:        the device to allow mapping/unmapping the DMAable region.
+ *
+ * The size of the storage is "max_frames*sizeof(struct dpaa2_dq)".
+ * The 'dpaa2_io_store' returned is a DPIO service managed object.
+ *
+ * Return pointer to dpaa2_io_store struct for successfuly created storage
+ * memory, or NULL on error.
+ */
+struct dpaa2_io_store *dpaa2_io_store_create(unsigned int max_frames,
+					     struct device *dev)
+{
+	struct dpaa2_io_store *ret;
+	size_t size;
+
+	if (!max_frames || (max_frames > 16))
+		return NULL;
+
+	ret = kmalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return NULL;
+
+	ret->max = max_frames;
+	size = max_frames * sizeof(struct dpaa2_dq) + 64;
+	ret->alloced_addr = kzalloc(size, GFP_KERNEL);
+	if (!ret->alloced_addr) {
+		kfree(ret);
+		return NULL;
+	}
+
+	ret->vaddr = PTR_ALIGN(ret->alloced_addr, 64);
+	ret->paddr = dma_map_single(dev, ret->vaddr,
+				    sizeof(struct dpaa2_dq) * max_frames,
+				    DMA_FROM_DEVICE);
+	if (dma_mapping_error(dev, ret->paddr)) {
+		kfree(ret->alloced_addr);
+		kfree(ret);
+		return NULL;
+	}
+
+	ret->idx = 0;
+	ret->dev = dev;
+
+	return ret;
+}
+EXPORT_SYMBOL(dpaa2_io_store_create);
+
+/**
+ * dpaa2_io_store_destroy() - Frees the dma memory storage for dequeue
+ *                            result.
+ * @s: the storage memory to be destroyed.
+ */
+void dpaa2_io_store_destroy(struct dpaa2_io_store *s)
+{
+	dma_unmap_single(s->dev, s->paddr, sizeof(struct dpaa2_dq) * s->max,
+			 DMA_FROM_DEVICE);
+	kfree(s->alloced_addr);
+	kfree(s);
+}
+EXPORT_SYMBOL(dpaa2_io_store_destroy);
+
+/**
+ * dpaa2_io_store_next() - Determine when the next dequeue result is available.
+ * @s: the dpaa2_io_store object.
+ * @is_last: indicate whether this is the last frame in the pull command.
+ *
+ * When an object driver performs dequeues to a dpaa2_io_store, this function
+ * can be used to determine when the next frame result is available. Once
+ * this function returns non-NULL, a subsequent call to it will try to find
+ * the next dequeue result.
+ *
+ * Note that if a pull-dequeue has a NULL result because the target FQ/channel
+ * was empty, then this function will also return NULL (rather than expecting
+ * the caller to always check for this. As such, "is_last" can be used to
+ * differentiate between "end-of-empty-dequeue" and "still-waiting".
+ *
+ * Return dequeue result for a valid dequeue result, or NULL for empty dequeue.
+ */
+struct dpaa2_dq *dpaa2_io_store_next(struct dpaa2_io_store *s, int *is_last)
+{
+	int match;
+	struct dpaa2_dq *ret = &s->vaddr[s->idx];
+
+	match = qbman_result_has_new_result(s->swp, ret);
+	if (!match) {
+		*is_last = 0;
+		return NULL;
+	}
+
+	s->idx++;
+
+	if (dpaa2_dq_is_pull_complete(ret)) {
+		*is_last = 1;
+		s->idx = 0;
+		/*
+		 * If we get an empty dequeue result to terminate a zero-results
+		 * vdqcr, return NULL to the caller rather than expecting him to
+		 * check non-NULL results every time.
+		 */
+		if (!(dpaa2_dq_flags(ret) & DPAA2_DQ_STAT_VALIDFRAME))
+			ret = NULL;
+	} else {
+		*is_last = 0;
+	}
+
+	return ret;
+}
+EXPORT_SYMBOL(dpaa2_io_store_next);
diff --git a/include/linux/fsl/dpaa2-io.h b/include/linux/fsl/dpaa2-io.h
new file mode 100644
index 0000000..12e7ad8
--- /dev/null
+++ b/include/linux/fsl/dpaa2-io.h
@@ -0,0 +1,138 @@
+/*
+ * Copyright 2014-2016 Freescale Semiconductor Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in the
+ *       documentation and/or other materials provided with the distribution.
+ *     * Neither the name of Freescale Semiconductor nor the
+ *       names of its contributors may be used to endorse or promote products
+ *       derived from this software without specific prior written permission.
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Freescale Semiconductor ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL Freescale Semiconductor BE LIABLE FOR ANY
+ * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef __FSL_DPAA2_IO_H
+#define __FSL_DPAA2_IO_H
+
+#include <linux/types.h>
+#include <linux/cpumask.h>
+
+#include <linux/fsl/dpaa2-fd.h>
+#include <linux/fsl/dpaa2-global.h>
+
+struct dpaa2_io;
+struct dpaa2_io_store;
+struct device;
+
+/**
+ * DOC: DPIO Service
+ *
+ * The DPIO service provides APIs for users to interact with the datapath
+ * by enqueueing and dequeing frame descriptors.
+ *
+ * The following set of APIs can be used to enqueue and dequeue frames
+ * as well as producing notification callbacks when data is available
+ * for dequeue.
+ */
+
+/**
+ * struct dpaa2_io_desc - The DPIO descriptor
+ * @receives_notifications: Use notificaton mode. Non-zero if the DPIO
+ *                  has a channel.
+ * @has_8prio:      Set to non-zero for channel with 8 priority WQs.  Ignored
+ *                  unless receives_notification is TRUE.
+ * @cpu:            The cpu index that at least interrupt handlers will
+ *                  execute on.
+ * @stash_affinity: The stash affinity for this portal favour 'cpu'
+ * @regs_cena:      The cache enabled regs.
+ * @regs_cinh:      The cache inhibited regs
+ * @dpio_id:        The dpio index
+ * @qman_version:   The qman version
+ *
+ * Describes the attributes and features of the DPIO object.
+ */
+struct dpaa2_io_desc {
+	int receives_notifications;
+	int has_8prio;
+	int cpu;
+	void *regs_cena;
+	void *regs_cinh;
+	int dpio_id;
+	u32 qman_version;
+};
+
+struct dpaa2_io *dpaa2_io_create(const struct dpaa2_io_desc *desc);
+
+void dpaa2_io_down(struct dpaa2_io *d);
+
+irqreturn_t dpaa2_io_irq(struct dpaa2_io *obj);
+
+/**
+ * struct dpaa2_io_notification_ctx - The DPIO notification context structure
+ * @cb:           The callback to be invoked when the notification arrives
+ * @is_cdan:      Zero for FQDAN, non-zero for CDAN
+ * @id:           FQID or channel ID, needed for rearm
+ * @desired_cpu:  The cpu on which the notifications will show up. -1 means
+ *                any CPU.
+ * @dpio_id:      The dpio index
+ * @qman64:       The 64-bit context value shows up in the FQDAN/CDAN.
+ * @node:         The list node
+ * @dpio_private: The dpio object internal to dpio_service
+ *
+ * Used when a FQDAN/CDAN registration is made by drivers.
+ */
+struct dpaa2_io_notification_ctx {
+	void (*cb)(struct dpaa2_io_notification_ctx *);
+	int is_cdan;
+	u32 id;
+	int desired_cpu;
+	int dpio_id;
+	u64 qman64;
+	struct list_head node;
+	void *dpio_private;
+};
+
+int dpaa2_io_service_register(struct dpaa2_io *service,
+			      struct dpaa2_io_notification_ctx *ctx);
+void dpaa2_io_service_deregister(struct dpaa2_io *service,
+				 struct dpaa2_io_notification_ctx *ctx);
+int dpaa2_io_service_rearm(struct dpaa2_io *service,
+			   struct dpaa2_io_notification_ctx *ctx);
+
+int dpaa2_io_service_pull_fq(struct dpaa2_io *d, u32 fqid,
+			     struct dpaa2_io_store *s);
+int dpaa2_io_service_pull_channel(struct dpaa2_io *d, u32 channelid,
+				  struct dpaa2_io_store *s);
+
+int dpaa2_io_service_enqueue_fq(struct dpaa2_io *d, u32 fqid,
+				const struct dpaa2_fd *fd);
+int dpaa2_io_service_enqueue_qd(struct dpaa2_io *d, u32 qdid, u8 prio,
+				u16 qdbin, const struct dpaa2_fd *fd);
+int dpaa2_io_service_release(struct dpaa2_io *d, u32 bpid,
+			     const u64 *buffers, unsigned int num_buffers);
+int dpaa2_io_service_acquire(struct dpaa2_io *d, u32 bpid,
+			     u64 *buffers, unsigned int num_buffers);
+
+struct dpaa2_io_store *dpaa2_io_store_create(unsigned int max_frames,
+					     struct device *dev);
+void dpaa2_io_store_destroy(struct dpaa2_io_store *s);
+struct dpaa2_dq *dpaa2_io_store_next(struct dpaa2_io_store *s, int *is_last);
+
+#endif /* __FSL_DPAA2_IO_H */
-- 
1.9.0