Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / dma / dmaengine.c
diff --git a/kernel/drivers/dma/dmaengine.c b/kernel/drivers/dma/dmaengine.c
new file mode 100644 (file)
index 0000000..3ddfd1f
--- /dev/null
@@ -0,0 +1,1191 @@
+/*
+ * Copyright(c) 2004 - 2006 Intel Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation; either version 2 of the License, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * The full GNU General Public License is included in this distribution in the
+ * file called COPYING.
+ */
+
+/*
+ * This code implements the DMA subsystem. It provides a HW-neutral interface
+ * for other kernel code to use asynchronous memory copy capabilities,
+ * if present, and allows different HW DMA drivers to register as providing
+ * this capability.
+ *
+ * Due to the fact we are accelerating what is already a relatively fast
+ * operation, the code goes to great lengths to avoid additional overhead,
+ * such as locking.
+ *
+ * LOCKING:
+ *
+ * The subsystem keeps a global list of dma_device structs it is protected by a
+ * mutex, dma_list_mutex.
+ *
+ * A subsystem can get access to a channel by calling dmaengine_get() followed
+ * by dma_find_channel(), or if it has need for an exclusive channel it can call
+ * dma_request_channel().  Once a channel is allocated a reference is taken
+ * against its corresponding driver to disable removal.
+ *
+ * Each device has a channels list, which runs unlocked but is never modified
+ * once the device is registered, it's just setup by the driver.
+ *
+ * See Documentation/dmaengine.txt for more details
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/dma-mapping.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/device.h>
+#include <linux/dmaengine.h>
+#include <linux/hardirq.h>
+#include <linux/spinlock.h>
+#include <linux/percpu.h>
+#include <linux/rcupdate.h>
+#include <linux/mutex.h>
+#include <linux/jiffies.h>
+#include <linux/rculist.h>
+#include <linux/idr.h>
+#include <linux/slab.h>
+#include <linux/acpi.h>
+#include <linux/acpi_dma.h>
+#include <linux/of_dma.h>
+#include <linux/mempool.h>
+
+static DEFINE_MUTEX(dma_list_mutex);
+static DEFINE_IDR(dma_idr);
+static LIST_HEAD(dma_device_list);
+static long dmaengine_ref_count;
+
+/* --- sysfs implementation --- */
+
+/**
+ * dev_to_dma_chan - convert a device pointer to the its sysfs container object
+ * @dev - device node
+ *
+ * Must be called under dma_list_mutex
+ */
+static struct dma_chan *dev_to_dma_chan(struct device *dev)
+{
+       struct dma_chan_dev *chan_dev;
+
+       chan_dev = container_of(dev, typeof(*chan_dev), device);
+       return chan_dev->chan;
+}
+
+static ssize_t memcpy_count_show(struct device *dev,
+                                struct device_attribute *attr, char *buf)
+{
+       struct dma_chan *chan;
+       unsigned long count = 0;
+       int i;
+       int err;
+
+       mutex_lock(&dma_list_mutex);
+       chan = dev_to_dma_chan(dev);
+       if (chan) {
+               for_each_possible_cpu(i)
+                       count += per_cpu_ptr(chan->local, i)->memcpy_count;
+               err = sprintf(buf, "%lu\n", count);
+       } else
+               err = -ENODEV;
+       mutex_unlock(&dma_list_mutex);
+
+       return err;
+}
+static DEVICE_ATTR_RO(memcpy_count);
+
+static ssize_t bytes_transferred_show(struct device *dev,
+                                     struct device_attribute *attr, char *buf)
+{
+       struct dma_chan *chan;
+       unsigned long count = 0;
+       int i;
+       int err;
+
+       mutex_lock(&dma_list_mutex);
+       chan = dev_to_dma_chan(dev);
+       if (chan) {
+               for_each_possible_cpu(i)
+                       count += per_cpu_ptr(chan->local, i)->bytes_transferred;
+               err = sprintf(buf, "%lu\n", count);
+       } else
+               err = -ENODEV;
+       mutex_unlock(&dma_list_mutex);
+
+       return err;
+}
+static DEVICE_ATTR_RO(bytes_transferred);
+
+static ssize_t in_use_show(struct device *dev, struct device_attribute *attr,
+                          char *buf)
+{
+       struct dma_chan *chan;
+       int err;
+
+       mutex_lock(&dma_list_mutex);
+       chan = dev_to_dma_chan(dev);
+       if (chan)
+               err = sprintf(buf, "%d\n", chan->client_count);
+       else
+               err = -ENODEV;
+       mutex_unlock(&dma_list_mutex);
+
+       return err;
+}
+static DEVICE_ATTR_RO(in_use);
+
+static struct attribute *dma_dev_attrs[] = {
+       &dev_attr_memcpy_count.attr,
+       &dev_attr_bytes_transferred.attr,
+       &dev_attr_in_use.attr,
+       NULL,
+};
+ATTRIBUTE_GROUPS(dma_dev);
+
+static void chan_dev_release(struct device *dev)
+{
+       struct dma_chan_dev *chan_dev;
+
+       chan_dev = container_of(dev, typeof(*chan_dev), device);
+       if (atomic_dec_and_test(chan_dev->idr_ref)) {
+               mutex_lock(&dma_list_mutex);
+               idr_remove(&dma_idr, chan_dev->dev_id);
+               mutex_unlock(&dma_list_mutex);
+               kfree(chan_dev->idr_ref);
+       }
+       kfree(chan_dev);
+}
+
+static struct class dma_devclass = {
+       .name           = "dma",
+       .dev_groups     = dma_dev_groups,
+       .dev_release    = chan_dev_release,
+};
+
+/* --- client and device registration --- */
+
+#define dma_device_satisfies_mask(device, mask) \
+       __dma_device_satisfies_mask((device), &(mask))
+static int
+__dma_device_satisfies_mask(struct dma_device *device,
+                           const dma_cap_mask_t *want)
+{
+       dma_cap_mask_t has;
+
+       bitmap_and(has.bits, want->bits, device->cap_mask.bits,
+               DMA_TX_TYPE_END);
+       return bitmap_equal(want->bits, has.bits, DMA_TX_TYPE_END);
+}
+
+static struct module *dma_chan_to_owner(struct dma_chan *chan)
+{
+       return chan->device->dev->driver->owner;
+}
+
+/**
+ * balance_ref_count - catch up the channel reference count
+ * @chan - channel to balance ->client_count versus dmaengine_ref_count
+ *
+ * balance_ref_count must be called under dma_list_mutex
+ */
+static void balance_ref_count(struct dma_chan *chan)
+{
+       struct module *owner = dma_chan_to_owner(chan);
+
+       while (chan->client_count < dmaengine_ref_count) {
+               __module_get(owner);
+               chan->client_count++;
+       }
+}
+
+/**
+ * dma_chan_get - try to grab a dma channel's parent driver module
+ * @chan - channel to grab
+ *
+ * Must be called under dma_list_mutex
+ */
+static int dma_chan_get(struct dma_chan *chan)
+{
+       struct module *owner = dma_chan_to_owner(chan);
+       int ret;
+
+       /* The channel is already in use, update client count */
+       if (chan->client_count) {
+               __module_get(owner);
+               goto out;
+       }
+
+       if (!try_module_get(owner))
+               return -ENODEV;
+
+       /* allocate upon first client reference */
+       if (chan->device->device_alloc_chan_resources) {
+               ret = chan->device->device_alloc_chan_resources(chan);
+               if (ret < 0)
+                       goto err_out;
+       }
+
+       if (!dma_has_cap(DMA_PRIVATE, chan->device->cap_mask))
+               balance_ref_count(chan);
+
+out:
+       chan->client_count++;
+       return 0;
+
+err_out:
+       module_put(owner);
+       return ret;
+}
+
+/**
+ * dma_chan_put - drop a reference to a dma channel's parent driver module
+ * @chan - channel to release
+ *
+ * Must be called under dma_list_mutex
+ */
+static void dma_chan_put(struct dma_chan *chan)
+{
+       /* This channel is not in use, bail out */
+       if (!chan->client_count)
+               return;
+
+       chan->client_count--;
+       module_put(dma_chan_to_owner(chan));
+
+       /* This channel is not in use anymore, free it */
+       if (!chan->client_count && chan->device->device_free_chan_resources)
+               chan->device->device_free_chan_resources(chan);
+}
+
+enum dma_status dma_sync_wait(struct dma_chan *chan, dma_cookie_t cookie)
+{
+       enum dma_status status;
+       unsigned long dma_sync_wait_timeout = jiffies + msecs_to_jiffies(5000);
+
+       dma_async_issue_pending(chan);
+       do {
+               status = dma_async_is_tx_complete(chan, cookie, NULL, NULL);
+               if (time_after_eq(jiffies, dma_sync_wait_timeout)) {
+                       pr_err("%s: timeout!\n", __func__);
+                       return DMA_ERROR;
+               }
+               if (status != DMA_IN_PROGRESS)
+                       break;
+               cpu_relax();
+       } while (1);
+
+       return status;
+}
+EXPORT_SYMBOL(dma_sync_wait);
+
+/**
+ * dma_cap_mask_all - enable iteration over all operation types
+ */
+static dma_cap_mask_t dma_cap_mask_all;
+
+/**
+ * dma_chan_tbl_ent - tracks channel allocations per core/operation
+ * @chan - associated channel for this entry
+ */
+struct dma_chan_tbl_ent {
+       struct dma_chan *chan;
+};
+
+/**
+ * channel_table - percpu lookup table for memory-to-memory offload providers
+ */
+static struct dma_chan_tbl_ent __percpu *channel_table[DMA_TX_TYPE_END];
+
+static int __init dma_channel_table_init(void)
+{
+       enum dma_transaction_type cap;
+       int err = 0;
+
+       bitmap_fill(dma_cap_mask_all.bits, DMA_TX_TYPE_END);
+
+       /* 'interrupt', 'private', and 'slave' are channel capabilities,
+        * but are not associated with an operation so they do not need
+        * an entry in the channel_table
+        */
+       clear_bit(DMA_INTERRUPT, dma_cap_mask_all.bits);
+       clear_bit(DMA_PRIVATE, dma_cap_mask_all.bits);
+       clear_bit(DMA_SLAVE, dma_cap_mask_all.bits);
+
+       for_each_dma_cap_mask(cap, dma_cap_mask_all) {
+               channel_table[cap] = alloc_percpu(struct dma_chan_tbl_ent);
+               if (!channel_table[cap]) {
+                       err = -ENOMEM;
+                       break;
+               }
+       }
+
+       if (err) {
+               pr_err("initialization failure\n");
+               for_each_dma_cap_mask(cap, dma_cap_mask_all)
+                       free_percpu(channel_table[cap]);
+       }
+
+       return err;
+}
+arch_initcall(dma_channel_table_init);
+
+/**
+ * dma_find_channel - find a channel to carry out the operation
+ * @tx_type: transaction type
+ */
+struct dma_chan *dma_find_channel(enum dma_transaction_type tx_type)
+{
+       return this_cpu_read(channel_table[tx_type]->chan);
+}
+EXPORT_SYMBOL(dma_find_channel);
+
+/**
+ * dma_issue_pending_all - flush all pending operations across all channels
+ */
+void dma_issue_pending_all(void)
+{
+       struct dma_device *device;
+       struct dma_chan *chan;
+
+       rcu_read_lock();
+       list_for_each_entry_rcu(device, &dma_device_list, global_node) {
+               if (dma_has_cap(DMA_PRIVATE, device->cap_mask))
+                       continue;
+               list_for_each_entry(chan, &device->channels, device_node)
+                       if (chan->client_count)
+                               device->device_issue_pending(chan);
+       }
+       rcu_read_unlock();
+}
+EXPORT_SYMBOL(dma_issue_pending_all);
+
+/**
+ * dma_chan_is_local - returns true if the channel is in the same numa-node as the cpu
+ */
+static bool dma_chan_is_local(struct dma_chan *chan, int cpu)
+{
+       int node = dev_to_node(chan->device->dev);
+       return node == -1 || cpumask_test_cpu(cpu, cpumask_of_node(node));
+}
+
+/**
+ * min_chan - returns the channel with min count and in the same numa-node as the cpu
+ * @cap: capability to match
+ * @cpu: cpu index which the channel should be close to
+ *
+ * If some channels are close to the given cpu, the one with the lowest
+ * reference count is returned. Otherwise, cpu is ignored and only the
+ * reference count is taken into account.
+ * Must be called under dma_list_mutex.
+ */
+static struct dma_chan *min_chan(enum dma_transaction_type cap, int cpu)
+{
+       struct dma_device *device;
+       struct dma_chan *chan;
+       struct dma_chan *min = NULL;
+       struct dma_chan *localmin = NULL;
+
+       list_for_each_entry(device, &dma_device_list, global_node) {
+               if (!dma_has_cap(cap, device->cap_mask) ||
+                   dma_has_cap(DMA_PRIVATE, device->cap_mask))
+                       continue;
+               list_for_each_entry(chan, &device->channels, device_node) {
+                       if (!chan->client_count)
+                               continue;
+                       if (!min || chan->table_count < min->table_count)
+                               min = chan;
+
+                       if (dma_chan_is_local(chan, cpu))
+                               if (!localmin ||
+                                   chan->table_count < localmin->table_count)
+                                       localmin = chan;
+               }
+       }
+
+       chan = localmin ? localmin : min;
+
+       if (chan)
+               chan->table_count++;
+
+       return chan;
+}
+
+/**
+ * dma_channel_rebalance - redistribute the available channels
+ *
+ * Optimize for cpu isolation (each cpu gets a dedicated channel for an
+ * operation type) in the SMP case,  and operation isolation (avoid
+ * multi-tasking channels) in the non-SMP case.  Must be called under
+ * dma_list_mutex.
+ */
+static void dma_channel_rebalance(void)
+{
+       struct dma_chan *chan;
+       struct dma_device *device;
+       int cpu;
+       int cap;
+
+       /* undo the last distribution */
+       for_each_dma_cap_mask(cap, dma_cap_mask_all)
+               for_each_possible_cpu(cpu)
+                       per_cpu_ptr(channel_table[cap], cpu)->chan = NULL;
+
+       list_for_each_entry(device, &dma_device_list, global_node) {
+               if (dma_has_cap(DMA_PRIVATE, device->cap_mask))
+                       continue;
+               list_for_each_entry(chan, &device->channels, device_node)
+                       chan->table_count = 0;
+       }
+
+       /* don't populate the channel_table if no clients are available */
+       if (!dmaengine_ref_count)
+               return;
+
+       /* redistribute available channels */
+       for_each_dma_cap_mask(cap, dma_cap_mask_all)
+               for_each_online_cpu(cpu) {
+                       chan = min_chan(cap, cpu);
+                       per_cpu_ptr(channel_table[cap], cpu)->chan = chan;
+               }
+}
+
+int dma_get_slave_caps(struct dma_chan *chan, struct dma_slave_caps *caps)
+{
+       struct dma_device *device;
+
+       if (!chan || !caps)
+               return -EINVAL;
+
+       device = chan->device;
+
+       /* check if the channel supports slave transactions */
+       if (!test_bit(DMA_SLAVE, device->cap_mask.bits))
+               return -ENXIO;
+
+       /*
+        * Check whether it reports it uses the generic slave
+        * capabilities, if not, that means it doesn't support any
+        * kind of slave capabilities reporting.
+        */
+       if (!device->directions)
+               return -ENXIO;
+
+       caps->src_addr_widths = device->src_addr_widths;
+       caps->dst_addr_widths = device->dst_addr_widths;
+       caps->directions = device->directions;
+       caps->residue_granularity = device->residue_granularity;
+
+       /*
+        * Some devices implement only pause (e.g. to get residuum) but no
+        * resume. However cmd_pause is advertised as pause AND resume.
+        */
+       caps->cmd_pause = !!(device->device_pause && device->device_resume);
+       caps->cmd_terminate = !!device->device_terminate_all;
+
+       return 0;
+}
+EXPORT_SYMBOL_GPL(dma_get_slave_caps);
+
+static struct dma_chan *private_candidate(const dma_cap_mask_t *mask,
+                                         struct dma_device *dev,
+                                         dma_filter_fn fn, void *fn_param)
+{
+       struct dma_chan *chan;
+
+       if (!__dma_device_satisfies_mask(dev, mask)) {
+               pr_debug("%s: wrong capabilities\n", __func__);
+               return NULL;
+       }
+       /* devices with multiple channels need special handling as we need to
+        * ensure that all channels are either private or public.
+        */
+       if (dev->chancnt > 1 && !dma_has_cap(DMA_PRIVATE, dev->cap_mask))
+               list_for_each_entry(chan, &dev->channels, device_node) {
+                       /* some channels are already publicly allocated */
+                       if (chan->client_count)
+                               return NULL;
+               }
+
+       list_for_each_entry(chan, &dev->channels, device_node) {
+               if (chan->client_count) {
+                       pr_debug("%s: %s busy\n",
+                                __func__, dma_chan_name(chan));
+                       continue;
+               }
+               if (fn && !fn(chan, fn_param)) {
+                       pr_debug("%s: %s filter said false\n",
+                                __func__, dma_chan_name(chan));
+                       continue;
+               }
+               return chan;
+       }
+
+       return NULL;
+}
+
+/**
+ * dma_request_slave_channel - try to get specific channel exclusively
+ * @chan: target channel
+ */
+struct dma_chan *dma_get_slave_channel(struct dma_chan *chan)
+{
+       int err = -EBUSY;
+
+       /* lock against __dma_request_channel */
+       mutex_lock(&dma_list_mutex);
+
+       if (chan->client_count == 0) {
+               err = dma_chan_get(chan);
+               if (err)
+                       pr_debug("%s: failed to get %s: (%d)\n",
+                               __func__, dma_chan_name(chan), err);
+       } else
+               chan = NULL;
+
+       mutex_unlock(&dma_list_mutex);
+
+
+       return chan;
+}
+EXPORT_SYMBOL_GPL(dma_get_slave_channel);
+
+struct dma_chan *dma_get_any_slave_channel(struct dma_device *device)
+{
+       dma_cap_mask_t mask;
+       struct dma_chan *chan;
+       int err;
+
+       dma_cap_zero(mask);
+       dma_cap_set(DMA_SLAVE, mask);
+
+       /* lock against __dma_request_channel */
+       mutex_lock(&dma_list_mutex);
+
+       chan = private_candidate(&mask, device, NULL, NULL);
+       if (chan) {
+               dma_cap_set(DMA_PRIVATE, device->cap_mask);
+               device->privatecnt++;
+               err = dma_chan_get(chan);
+               if (err) {
+                       pr_debug("%s: failed to get %s: (%d)\n",
+                               __func__, dma_chan_name(chan), err);
+                       chan = NULL;
+                       if (--device->privatecnt == 0)
+                               dma_cap_clear(DMA_PRIVATE, device->cap_mask);
+               }
+       }
+
+       mutex_unlock(&dma_list_mutex);
+
+       return chan;
+}
+EXPORT_SYMBOL_GPL(dma_get_any_slave_channel);
+
+/**
+ * __dma_request_channel - try to allocate an exclusive channel
+ * @mask: capabilities that the channel must satisfy
+ * @fn: optional callback to disposition available channels
+ * @fn_param: opaque parameter to pass to dma_filter_fn
+ *
+ * Returns pointer to appropriate DMA channel on success or NULL.
+ */
+struct dma_chan *__dma_request_channel(const dma_cap_mask_t *mask,
+                                      dma_filter_fn fn, void *fn_param)
+{
+       struct dma_device *device, *_d;
+       struct dma_chan *chan = NULL;
+       int err;
+
+       /* Find a channel */
+       mutex_lock(&dma_list_mutex);
+       list_for_each_entry_safe(device, _d, &dma_device_list, global_node) {
+               chan = private_candidate(mask, device, fn, fn_param);
+               if (chan) {
+                       /* Found a suitable channel, try to grab, prep, and
+                        * return it.  We first set DMA_PRIVATE to disable
+                        * balance_ref_count as this channel will not be
+                        * published in the general-purpose allocator
+                        */
+                       dma_cap_set(DMA_PRIVATE, device->cap_mask);
+                       device->privatecnt++;
+                       err = dma_chan_get(chan);
+
+                       if (err == -ENODEV) {
+                               pr_debug("%s: %s module removed\n",
+                                        __func__, dma_chan_name(chan));
+                               list_del_rcu(&device->global_node);
+                       } else if (err)
+                               pr_debug("%s: failed to get %s: (%d)\n",
+                                        __func__, dma_chan_name(chan), err);
+                       else
+                               break;
+                       if (--device->privatecnt == 0)
+                               dma_cap_clear(DMA_PRIVATE, device->cap_mask);
+                       chan = NULL;
+               }
+       }
+       mutex_unlock(&dma_list_mutex);
+
+       pr_debug("%s: %s (%s)\n",
+                __func__,
+                chan ? "success" : "fail",
+                chan ? dma_chan_name(chan) : NULL);
+
+       return chan;
+}
+EXPORT_SYMBOL_GPL(__dma_request_channel);
+
+/**
+ * dma_request_slave_channel - try to allocate an exclusive slave channel
+ * @dev:       pointer to client device structure
+ * @name:      slave channel name
+ *
+ * Returns pointer to appropriate DMA channel on success or an error pointer.
+ */
+struct dma_chan *dma_request_slave_channel_reason(struct device *dev,
+                                                 const char *name)
+{
+       /* If device-tree is present get slave info from here */
+       if (dev->of_node)
+               return of_dma_request_slave_channel(dev->of_node, name);
+
+       /* If device was enumerated by ACPI get slave info from here */
+       if (ACPI_HANDLE(dev))
+               return acpi_dma_request_slave_chan_by_name(dev, name);
+
+       return ERR_PTR(-ENODEV);
+}
+EXPORT_SYMBOL_GPL(dma_request_slave_channel_reason);
+
+/**
+ * dma_request_slave_channel - try to allocate an exclusive slave channel
+ * @dev:       pointer to client device structure
+ * @name:      slave channel name
+ *
+ * Returns pointer to appropriate DMA channel on success or NULL.
+ */
+struct dma_chan *dma_request_slave_channel(struct device *dev,
+                                          const char *name)
+{
+       struct dma_chan *ch = dma_request_slave_channel_reason(dev, name);
+       if (IS_ERR(ch))
+               return NULL;
+       return ch;
+}
+EXPORT_SYMBOL_GPL(dma_request_slave_channel);
+
+void dma_release_channel(struct dma_chan *chan)
+{
+       mutex_lock(&dma_list_mutex);
+       WARN_ONCE(chan->client_count != 1,
+                 "chan reference count %d != 1\n", chan->client_count);
+       dma_chan_put(chan);
+       /* drop PRIVATE cap enabled by __dma_request_channel() */
+       if (--chan->device->privatecnt == 0)
+               dma_cap_clear(DMA_PRIVATE, chan->device->cap_mask);
+       mutex_unlock(&dma_list_mutex);
+}
+EXPORT_SYMBOL_GPL(dma_release_channel);
+
+/**
+ * dmaengine_get - register interest in dma_channels
+ */
+void dmaengine_get(void)
+{
+       struct dma_device *device, *_d;
+       struct dma_chan *chan;
+       int err;
+
+       mutex_lock(&dma_list_mutex);
+       dmaengine_ref_count++;
+
+       /* try to grab channels */
+       list_for_each_entry_safe(device, _d, &dma_device_list, global_node) {
+               if (dma_has_cap(DMA_PRIVATE, device->cap_mask))
+                       continue;
+               list_for_each_entry(chan, &device->channels, device_node) {
+                       err = dma_chan_get(chan);
+                       if (err == -ENODEV) {
+                               /* module removed before we could use it */
+                               list_del_rcu(&device->global_node);
+                               break;
+                       } else if (err)
+                               pr_debug("%s: failed to get %s: (%d)\n",
+                                      __func__, dma_chan_name(chan), err);
+               }
+       }
+
+       /* if this is the first reference and there were channels
+        * waiting we need to rebalance to get those channels
+        * incorporated into the channel table
+        */
+       if (dmaengine_ref_count == 1)
+               dma_channel_rebalance();
+       mutex_unlock(&dma_list_mutex);
+}
+EXPORT_SYMBOL(dmaengine_get);
+
+/**
+ * dmaengine_put - let dma drivers be removed when ref_count == 0
+ */
+void dmaengine_put(void)
+{
+       struct dma_device *device;
+       struct dma_chan *chan;
+
+       mutex_lock(&dma_list_mutex);
+       dmaengine_ref_count--;
+       BUG_ON(dmaengine_ref_count < 0);
+       /* drop channel references */
+       list_for_each_entry(device, &dma_device_list, global_node) {
+               if (dma_has_cap(DMA_PRIVATE, device->cap_mask))
+                       continue;
+               list_for_each_entry(chan, &device->channels, device_node)
+                       dma_chan_put(chan);
+       }
+       mutex_unlock(&dma_list_mutex);
+}
+EXPORT_SYMBOL(dmaengine_put);
+
+static bool device_has_all_tx_types(struct dma_device *device)
+{
+       /* A device that satisfies this test has channels that will never cause
+        * an async_tx channel switch event as all possible operation types can
+        * be handled.
+        */
+       #ifdef CONFIG_ASYNC_TX_DMA
+       if (!dma_has_cap(DMA_INTERRUPT, device->cap_mask))
+               return false;
+       #endif
+
+       #if defined(CONFIG_ASYNC_MEMCPY) || defined(CONFIG_ASYNC_MEMCPY_MODULE)
+       if (!dma_has_cap(DMA_MEMCPY, device->cap_mask))
+               return false;
+       #endif
+
+       #if defined(CONFIG_ASYNC_XOR) || defined(CONFIG_ASYNC_XOR_MODULE)
+       if (!dma_has_cap(DMA_XOR, device->cap_mask))
+               return false;
+
+       #ifndef CONFIG_ASYNC_TX_DISABLE_XOR_VAL_DMA
+       if (!dma_has_cap(DMA_XOR_VAL, device->cap_mask))
+               return false;
+       #endif
+       #endif
+
+       #if defined(CONFIG_ASYNC_PQ) || defined(CONFIG_ASYNC_PQ_MODULE)
+       if (!dma_has_cap(DMA_PQ, device->cap_mask))
+               return false;
+
+       #ifndef CONFIG_ASYNC_TX_DISABLE_PQ_VAL_DMA
+       if (!dma_has_cap(DMA_PQ_VAL, device->cap_mask))
+               return false;
+       #endif
+       #endif
+
+       return true;
+}
+
+static int get_dma_id(struct dma_device *device)
+{
+       int rc;
+
+       mutex_lock(&dma_list_mutex);
+
+       rc = idr_alloc(&dma_idr, NULL, 0, 0, GFP_KERNEL);
+       if (rc >= 0)
+               device->dev_id = rc;
+
+       mutex_unlock(&dma_list_mutex);
+       return rc < 0 ? rc : 0;
+}
+
+/**
+ * dma_async_device_register - registers DMA devices found
+ * @device: &dma_device
+ */
+int dma_async_device_register(struct dma_device *device)
+{
+       int chancnt = 0, rc;
+       struct dma_chan* chan;
+       atomic_t *idr_ref;
+
+       if (!device)
+               return -ENODEV;
+
+       /* validate device routines */
+       BUG_ON(dma_has_cap(DMA_MEMCPY, device->cap_mask) &&
+               !device->device_prep_dma_memcpy);
+       BUG_ON(dma_has_cap(DMA_XOR, device->cap_mask) &&
+               !device->device_prep_dma_xor);
+       BUG_ON(dma_has_cap(DMA_XOR_VAL, device->cap_mask) &&
+               !device->device_prep_dma_xor_val);
+       BUG_ON(dma_has_cap(DMA_PQ, device->cap_mask) &&
+               !device->device_prep_dma_pq);
+       BUG_ON(dma_has_cap(DMA_PQ_VAL, device->cap_mask) &&
+               !device->device_prep_dma_pq_val);
+       BUG_ON(dma_has_cap(DMA_INTERRUPT, device->cap_mask) &&
+               !device->device_prep_dma_interrupt);
+       BUG_ON(dma_has_cap(DMA_SG, device->cap_mask) &&
+               !device->device_prep_dma_sg);
+       BUG_ON(dma_has_cap(DMA_CYCLIC, device->cap_mask) &&
+               !device->device_prep_dma_cyclic);
+       BUG_ON(dma_has_cap(DMA_INTERLEAVE, device->cap_mask) &&
+               !device->device_prep_interleaved_dma);
+
+       BUG_ON(!device->device_tx_status);
+       BUG_ON(!device->device_issue_pending);
+       BUG_ON(!device->dev);
+
+       /* note: this only matters in the
+        * CONFIG_ASYNC_TX_ENABLE_CHANNEL_SWITCH=n case
+        */
+       if (device_has_all_tx_types(device))
+               dma_cap_set(DMA_ASYNC_TX, device->cap_mask);
+
+       idr_ref = kmalloc(sizeof(*idr_ref), GFP_KERNEL);
+       if (!idr_ref)
+               return -ENOMEM;
+       rc = get_dma_id(device);
+       if (rc != 0) {
+               kfree(idr_ref);
+               return rc;
+       }
+
+       atomic_set(idr_ref, 0);
+
+       /* represent channels in sysfs. Probably want devs too */
+       list_for_each_entry(chan, &device->channels, device_node) {
+               rc = -ENOMEM;
+               chan->local = alloc_percpu(typeof(*chan->local));
+               if (chan->local == NULL)
+                       goto err_out;
+               chan->dev = kzalloc(sizeof(*chan->dev), GFP_KERNEL);
+               if (chan->dev == NULL) {
+                       free_percpu(chan->local);
+                       chan->local = NULL;
+                       goto err_out;
+               }
+
+               chan->chan_id = chancnt++;
+               chan->dev->device.class = &dma_devclass;
+               chan->dev->device.parent = device->dev;
+               chan->dev->chan = chan;
+               chan->dev->idr_ref = idr_ref;
+               chan->dev->dev_id = device->dev_id;
+               atomic_inc(idr_ref);
+               dev_set_name(&chan->dev->device, "dma%dchan%d",
+                            device->dev_id, chan->chan_id);
+
+               rc = device_register(&chan->dev->device);
+               if (rc) {
+                       free_percpu(chan->local);
+                       chan->local = NULL;
+                       kfree(chan->dev);
+                       atomic_dec(idr_ref);
+                       goto err_out;
+               }
+               chan->client_count = 0;
+       }
+       device->chancnt = chancnt;
+
+       mutex_lock(&dma_list_mutex);
+       /* take references on public channels */
+       if (dmaengine_ref_count && !dma_has_cap(DMA_PRIVATE, device->cap_mask))
+               list_for_each_entry(chan, &device->channels, device_node) {
+                       /* if clients are already waiting for channels we need
+                        * to take references on their behalf
+                        */
+                       if (dma_chan_get(chan) == -ENODEV) {
+                               /* note we can only get here for the first
+                                * channel as the remaining channels are
+                                * guaranteed to get a reference
+                                */
+                               rc = -ENODEV;
+                               mutex_unlock(&dma_list_mutex);
+                               goto err_out;
+                       }
+               }
+       list_add_tail_rcu(&device->global_node, &dma_device_list);
+       if (dma_has_cap(DMA_PRIVATE, device->cap_mask))
+               device->privatecnt++;   /* Always private */
+       dma_channel_rebalance();
+       mutex_unlock(&dma_list_mutex);
+
+       return 0;
+
+err_out:
+       /* if we never registered a channel just release the idr */
+       if (atomic_read(idr_ref) == 0) {
+               mutex_lock(&dma_list_mutex);
+               idr_remove(&dma_idr, device->dev_id);
+               mutex_unlock(&dma_list_mutex);
+               kfree(idr_ref);
+               return rc;
+       }
+
+       list_for_each_entry(chan, &device->channels, device_node) {
+               if (chan->local == NULL)
+                       continue;
+               mutex_lock(&dma_list_mutex);
+               chan->dev->chan = NULL;
+               mutex_unlock(&dma_list_mutex);
+               device_unregister(&chan->dev->device);
+               free_percpu(chan->local);
+       }
+       return rc;
+}
+EXPORT_SYMBOL(dma_async_device_register);
+
+/**
+ * dma_async_device_unregister - unregister a DMA device
+ * @device: &dma_device
+ *
+ * This routine is called by dma driver exit routines, dmaengine holds module
+ * references to prevent it being called while channels are in use.
+ */
+void dma_async_device_unregister(struct dma_device *device)
+{
+       struct dma_chan *chan;
+
+       mutex_lock(&dma_list_mutex);
+       list_del_rcu(&device->global_node);
+       dma_channel_rebalance();
+       mutex_unlock(&dma_list_mutex);
+
+       list_for_each_entry(chan, &device->channels, device_node) {
+               WARN_ONCE(chan->client_count,
+                         "%s called while %d clients hold a reference\n",
+                         __func__, chan->client_count);
+               mutex_lock(&dma_list_mutex);
+               chan->dev->chan = NULL;
+               mutex_unlock(&dma_list_mutex);
+               device_unregister(&chan->dev->device);
+               free_percpu(chan->local);
+       }
+}
+EXPORT_SYMBOL(dma_async_device_unregister);
+
+struct dmaengine_unmap_pool {
+       struct kmem_cache *cache;
+       const char *name;
+       mempool_t *pool;
+       size_t size;
+};
+
+#define __UNMAP_POOL(x) { .size = x, .name = "dmaengine-unmap-" __stringify(x) }
+static struct dmaengine_unmap_pool unmap_pool[] = {
+       __UNMAP_POOL(2),
+       #if IS_ENABLED(CONFIG_DMA_ENGINE_RAID)
+       __UNMAP_POOL(16),
+       __UNMAP_POOL(128),
+       __UNMAP_POOL(256),
+       #endif
+};
+
+static struct dmaengine_unmap_pool *__get_unmap_pool(int nr)
+{
+       int order = get_count_order(nr);
+
+       switch (order) {
+       case 0 ... 1:
+               return &unmap_pool[0];
+       case 2 ... 4:
+               return &unmap_pool[1];
+       case 5 ... 7:
+               return &unmap_pool[2];
+       case 8:
+               return &unmap_pool[3];
+       default:
+               BUG();
+               return NULL;
+       }
+}
+
+static void dmaengine_unmap(struct kref *kref)
+{
+       struct dmaengine_unmap_data *unmap = container_of(kref, typeof(*unmap), kref);
+       struct device *dev = unmap->dev;
+       int cnt, i;
+
+       cnt = unmap->to_cnt;
+       for (i = 0; i < cnt; i++)
+               dma_unmap_page(dev, unmap->addr[i], unmap->len,
+                              DMA_TO_DEVICE);
+       cnt += unmap->from_cnt;
+       for (; i < cnt; i++)
+               dma_unmap_page(dev, unmap->addr[i], unmap->len,
+                              DMA_FROM_DEVICE);
+       cnt += unmap->bidi_cnt;
+       for (; i < cnt; i++) {
+               if (unmap->addr[i] == 0)
+                       continue;
+               dma_unmap_page(dev, unmap->addr[i], unmap->len,
+                              DMA_BIDIRECTIONAL);
+       }
+       cnt = unmap->map_cnt;
+       mempool_free(unmap, __get_unmap_pool(cnt)->pool);
+}
+
+void dmaengine_unmap_put(struct dmaengine_unmap_data *unmap)
+{
+       if (unmap)
+               kref_put(&unmap->kref, dmaengine_unmap);
+}
+EXPORT_SYMBOL_GPL(dmaengine_unmap_put);
+
+static void dmaengine_destroy_unmap_pool(void)
+{
+       int i;
+
+       for (i = 0; i < ARRAY_SIZE(unmap_pool); i++) {
+               struct dmaengine_unmap_pool *p = &unmap_pool[i];
+
+               if (p->pool)
+                       mempool_destroy(p->pool);
+               p->pool = NULL;
+               if (p->cache)
+                       kmem_cache_destroy(p->cache);
+               p->cache = NULL;
+       }
+}
+
+static int __init dmaengine_init_unmap_pool(void)
+{
+       int i;
+
+       for (i = 0; i < ARRAY_SIZE(unmap_pool); i++) {
+               struct dmaengine_unmap_pool *p = &unmap_pool[i];
+               size_t size;
+
+               size = sizeof(struct dmaengine_unmap_data) +
+                      sizeof(dma_addr_t) * p->size;
+
+               p->cache = kmem_cache_create(p->name, size, 0,
+                                            SLAB_HWCACHE_ALIGN, NULL);
+               if (!p->cache)
+                       break;
+               p->pool = mempool_create_slab_pool(1, p->cache);
+               if (!p->pool)
+                       break;
+       }
+
+       if (i == ARRAY_SIZE(unmap_pool))
+               return 0;
+
+       dmaengine_destroy_unmap_pool();
+       return -ENOMEM;
+}
+
+struct dmaengine_unmap_data *
+dmaengine_get_unmap_data(struct device *dev, int nr, gfp_t flags)
+{
+       struct dmaengine_unmap_data *unmap;
+
+       unmap = mempool_alloc(__get_unmap_pool(nr)->pool, flags);
+       if (!unmap)
+               return NULL;
+
+       memset(unmap, 0, sizeof(*unmap));
+       kref_init(&unmap->kref);
+       unmap->dev = dev;
+       unmap->map_cnt = nr;
+
+       return unmap;
+}
+EXPORT_SYMBOL(dmaengine_get_unmap_data);
+
+void dma_async_tx_descriptor_init(struct dma_async_tx_descriptor *tx,
+       struct dma_chan *chan)
+{
+       tx->chan = chan;
+       #ifdef CONFIG_ASYNC_TX_ENABLE_CHANNEL_SWITCH
+       spin_lock_init(&tx->lock);
+       #endif
+}
+EXPORT_SYMBOL(dma_async_tx_descriptor_init);
+
+/* dma_wait_for_async_tx - spin wait for a transaction to complete
+ * @tx: in-flight transaction to wait on
+ */
+enum dma_status
+dma_wait_for_async_tx(struct dma_async_tx_descriptor *tx)
+{
+       unsigned long dma_sync_wait_timeout = jiffies + msecs_to_jiffies(5000);
+
+       if (!tx)
+               return DMA_COMPLETE;
+
+       while (tx->cookie == -EBUSY) {
+               if (time_after_eq(jiffies, dma_sync_wait_timeout)) {
+                       pr_err("%s timeout waiting for descriptor submission\n",
+                              __func__);
+                       return DMA_ERROR;
+               }
+               cpu_relax();
+       }
+       return dma_sync_wait(tx->chan, tx->cookie);
+}
+EXPORT_SYMBOL_GPL(dma_wait_for_async_tx);
+
+/* dma_run_dependencies - helper routine for dma drivers to process
+ *     (start) dependent operations on their target channel
+ * @tx: transaction with dependencies
+ */
+void dma_run_dependencies(struct dma_async_tx_descriptor *tx)
+{
+       struct dma_async_tx_descriptor *dep = txd_next(tx);
+       struct dma_async_tx_descriptor *dep_next;
+       struct dma_chan *chan;
+
+       if (!dep)
+               return;
+
+       /* we'll submit tx->next now, so clear the link */
+       txd_clear_next(tx);
+       chan = dep->chan;
+
+       /* keep submitting up until a channel switch is detected
+        * in that case we will be called again as a result of
+        * processing the interrupt from async_tx_channel_switch
+        */
+       for (; dep; dep = dep_next) {
+               txd_lock(dep);
+               txd_clear_parent(dep);
+               dep_next = txd_next(dep);
+               if (dep_next && dep_next->chan == chan)
+                       txd_clear_next(dep); /* ->next will be submitted */
+               else
+                       dep_next = NULL; /* submit current dep and terminate */
+               txd_unlock(dep);
+
+               dep->tx_submit(dep);
+       }
+
+       chan->device->device_issue_pending(chan);
+}
+EXPORT_SYMBOL_GPL(dma_run_dependencies);
+
+static int __init dma_bus_init(void)
+{
+       int err = dmaengine_init_unmap_pool();
+
+       if (err)
+               return err;
+       return class_register(&dma_devclass);
+}
+arch_initcall(dma_bus_init);
+
+