ioat_dma.c

/*
 * Intel I/OAT DMA Linux driver
 * Copyright(c) 2004 - 2007 Intel Corporation.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * 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.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program; if not, write to the Free Software Foundation, Inc.,
 * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * The full GNU General Public License is included in this distribution in
 * the file called "COPYING".
 *
 */

/*
 * This driver supports an Intel I/OAT DMA engine, which does asynchronous
 * copy operations.
 */

#include <linux/init.h>
#include <linux/module.h>
#include <linux/pci.h>
#include <linux/interrupt.h>
#include <linux/dmaengine.h>
#include <linux/delay.h>
#include <linux/dma-mapping.h>
#include "ioatdma.h"
#include "ioatdma_registers.h"
#include "ioatdma_hw.h"

#define to_ioat_chan(chan) container_of(chan, struct ioat_dma_chan, common)
#define to_ioatdma_device(dev) container_of(dev, struct ioatdma_device, common)
#define to_ioat_desc(lh) container_of(lh, struct ioat_desc_sw, node)
#define tx_to_ioat_desc(tx) container_of(tx, struct ioat_desc_sw, async_tx)

static int ioat_pending_level = 4;
module_param(ioat_pending_level, int, 0644);
MODULE_PARM_DESC(ioat_pending_level,
		 "high-water mark for pushing ioat descriptors (default: 4)");

/* internal functions */
static void ioat_dma_start_null_desc(struct ioat_dma_chan *ioat_chan);
static void ioat_dma_memcpy_cleanup(struct ioat_dma_chan *ioat_chan);

static struct ioat_desc_sw *
ioat1_dma_get_next_descriptor(struct ioat_dma_chan *ioat_chan);
static struct ioat_desc_sw *
ioat2_dma_get_next_descriptor(struct ioat_dma_chan *ioat_chan);

static inline struct ioat_dma_chan *ioat_lookup_chan_by_index(
						struct ioatdma_device *device,
						int index)
{
	return device->idx[index];
}

/**
 * ioat_dma_do_interrupt - handler used for single vector interrupt mode
 * @irq: interrupt id
 * @data: interrupt data
 */
static irqreturn_t ioat_dma_do_interrupt(int irq, void *data)
{
	struct ioatdma_device *instance = data;
	struct ioat_dma_chan *ioat_chan;
	unsigned long attnstatus;
	int bit;
	u8 intrctrl;

	intrctrl = readb(instance->reg_base + IOAT_INTRCTRL_OFFSET);

	if (!(intrctrl & IOAT_INTRCTRL_MASTER_INT_EN))
		return IRQ_NONE;

	if (!(intrctrl & IOAT_INTRCTRL_INT_STATUS)) {
		writeb(intrctrl, instance->reg_base + IOAT_INTRCTRL_OFFSET);
		return IRQ_NONE;
	}

	attnstatus = readl(instance->reg_base + IOAT_ATTNSTATUS_OFFSET);
	for_each_bit(bit, &attnstatus, BITS_PER_LONG) {
		ioat_chan = ioat_lookup_chan_by_index(instance, bit);
		tasklet_schedule(&ioat_chan->cleanup_task);
	}

	writeb(intrctrl, instance->reg_base + IOAT_INTRCTRL_OFFSET);
	return IRQ_HANDLED;
}

/**
 * ioat_dma_do_interrupt_msix - handler used for vector-per-channel interrupt mode
 * @irq: interrupt id
 * @data: interrupt data
 */
static irqreturn_t ioat_dma_do_interrupt_msix(int irq, void *data)
{
	struct ioat_dma_chan *ioat_chan = data;

	tasklet_schedule(&ioat_chan->cleanup_task);

	return IRQ_HANDLED;
}

static void ioat_dma_cleanup_tasklet(unsigned long data);

/**
 * ioat_dma_enumerate_channels - find and initialize the device's channels
 * @device: the device to be enumerated
 */
static int ioat_dma_enumerate_channels(struct ioatdma_device *device)
{
	u8 xfercap_scale;
	u32 xfercap;
	int i;
	struct ioat_dma_chan *ioat_chan;

	device->common.chancnt = readb(device->reg_base + IOAT_CHANCNT_OFFSET);
	xfercap_scale = readb(device->reg_base + IOAT_XFERCAP_OFFSET);
	xfercap = (xfercap_scale == 0 ? -1 : (1UL << xfercap_scale));

	for (i = 0; i < device->common.chancnt; i++) {
		ioat_chan = kzalloc(sizeof(*ioat_chan), GFP_KERNEL);
		if (!ioat_chan) {
			device->common.chancnt = i;
			break;
		}

		ioat_chan->device = device;
		ioat_chan->reg_base = device->reg_base + (0x80 * (i + 1));
		ioat_chan->xfercap = xfercap;
		ioat_chan->desccount = 0;
		if (ioat_chan->device->version != IOAT_VER_1_2) {
			writel(IOAT_DCACTRL_CMPL_WRITE_ENABLE
					| IOAT_DMA_DCA_ANY_CPU,
				ioat_chan->reg_base + IOAT_DCACTRL_OFFSET);
		}
		spin_lock_init(&ioat_chan->cleanup_lock);
		spin_lock_init(&ioat_chan->desc_lock);
		INIT_LIST_HEAD(&ioat_chan->free_desc);
		INIT_LIST_HEAD(&ioat_chan->used_desc);
		/* This should be made common somewhere in dmaengine.c */
		ioat_chan->common.device = &device->common;
		list_add_tail(&ioat_chan->common.device_node,
			      &device->common.channels);
		device->idx[i] = ioat_chan;
		tasklet_init(&ioat_chan->cleanup_task,
			     ioat_dma_cleanup_tasklet,
			     (unsigned long) ioat_chan);
		tasklet_disable(&ioat_chan->cleanup_task);
	}
	return device->common.chancnt;
}

/**
 * ioat_dma_memcpy_issue_pending - push potentially unrecognized appended
 *                                 descriptors to hw
 * @chan: DMA channel handle
 */
static inline void __ioat1_dma_memcpy_issue_pending(
						struct ioat_dma_chan *ioat_chan)
{
	ioat_chan->pending = 0;
	writeb(IOAT_CHANCMD_APPEND, ioat_chan->reg_base + IOAT1_CHANCMD_OFFSET);
}

static void ioat1_dma_memcpy_issue_pending(struct dma_chan *chan)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);

	if (ioat_chan->pending != 0) {
		spin_lock_bh(&ioat_chan->desc_lock);
		__ioat1_dma_memcpy_issue_pending(ioat_chan);
		spin_unlock_bh(&ioat_chan->desc_lock);
	}
}

static inline void __ioat2_dma_memcpy_issue_pending(
						struct ioat_dma_chan *ioat_chan)
{
	ioat_chan->pending = 0;
	writew(ioat_chan->dmacount,
	       ioat_chan->reg_base + IOAT_CHAN_DMACOUNT_OFFSET);
}

static void ioat2_dma_memcpy_issue_pending(struct dma_chan *chan)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);

	if (ioat_chan->pending != 0) {
		spin_lock_bh(&ioat_chan->desc_lock);
		__ioat2_dma_memcpy_issue_pending(ioat_chan);
		spin_unlock_bh(&ioat_chan->desc_lock);
	}
}

static dma_cookie_t ioat1_tx_submit(struct dma_async_tx_descriptor *tx)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(tx->chan);
	struct ioat_desc_sw *first = tx_to_ioat_desc(tx);
	struct ioat_desc_sw *prev, *new;
	struct ioat_dma_descriptor *hw;
	dma_cookie_t cookie;
	LIST_HEAD(new_chain);
	u32 copy;
	size_t len;
	dma_addr_t src, dst;
	int orig_ack;
	unsigned int desc_count = 0;

	/* src and dest and len are stored in the initial descriptor */
	len = first->len;
	src = first->src;
	dst = first->dst;
	orig_ack = first->async_tx.ack;
	new = first;

	spin_lock_bh(&ioat_chan->desc_lock);
	prev = to_ioat_desc(ioat_chan->used_desc.prev);
	prefetch(prev->hw);
	do {
		copy = min_t(size_t, len, ioat_chan->xfercap);

		new->async_tx.ack = 1;

		hw = new->hw;
		hw->size = copy;
		hw->ctl = 0;
		hw->src_addr = src;
		hw->dst_addr = dst;
		hw->next = 0;

		/* chain together the physical address list for the HW */
		wmb();
		prev->hw->next = (u64) new->async_tx.phys;

		len -= copy;
		dst += copy;
		src += copy;

		list_add_tail(&new->node, &new_chain);
		desc_count++;
		prev = new;
	} while (len && (new = ioat1_dma_get_next_descriptor(ioat_chan)));

	hw->ctl = IOAT_DMA_DESCRIPTOR_CTL_CP_STS;
	if (new->async_tx.callback) {
		hw->ctl |= IOAT_DMA_DESCRIPTOR_CTL_INT_GN;
		if (first != new) {
			/* move callback into to last desc */
			new->async_tx.callback = first->async_tx.callback;
			new->async_tx.callback_param
					= first->async_tx.callback_param;
			first->async_tx.callback = NULL;
			first->async_tx.callback_param = NULL;
		}
	}

	new->tx_cnt = desc_count;
	new->async_tx.ack = orig_ack; /* client is in control of this ack */

	/* store the original values for use in later cleanup */
	if (new != first) {
		new->src = first->src;
		new->dst = first->dst;
		new->len = first->len;
	}

	/* cookie incr and addition to used_list must be atomic */
	cookie = ioat_chan->common.cookie;
	cookie++;
	if (cookie < 0)
		cookie = 1;
	ioat_chan->common.cookie = new->async_tx.cookie = cookie;

	/* write address into NextDescriptor field of last desc in chain */
	to_ioat_desc(ioat_chan->used_desc.prev)->hw->next =
							first->async_tx.phys;
	__list_splice(&new_chain, ioat_chan->used_desc.prev);

	ioat_chan->dmacount += desc_count;
	ioat_chan->pending += desc_count;
	if (ioat_chan->pending >= ioat_pending_level)
		__ioat1_dma_memcpy_issue_pending(ioat_chan);
	spin_unlock_bh(&ioat_chan->desc_lock);

	return cookie;
}

static dma_cookie_t ioat2_tx_submit(struct dma_async_tx_descriptor *tx)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(tx->chan);
	struct ioat_desc_sw *first = tx_to_ioat_desc(tx);
	struct ioat_desc_sw *new;
	struct ioat_dma_descriptor *hw;
	dma_cookie_t cookie;
	u32 copy;
	size_t len;
	dma_addr_t src, dst;
	int orig_ack;
	unsigned int desc_count = 0;

	/* src and dest and len are stored in the initial descriptor */
	len = first->len;
	src = first->src;
	dst = first->dst;
	orig_ack = first->async_tx.ack;
	new = first;

	/*
	 * ioat_chan->desc_lock is still in force in version 2 path
	 * it gets unlocked at end of this function
	 */
	do {
		copy = min_t(size_t, len, ioat_chan->xfercap);

		new->async_tx.ack = 1;

		hw = new->hw;
		hw->size = copy;
		hw->ctl = 0;
		hw->src_addr = src;
		hw->dst_addr = dst;

		len -= copy;
		dst += copy;
		src += copy;
		desc_count++;
	} while (len && (new = ioat2_dma_get_next_descriptor(ioat_chan)));

	hw->ctl = IOAT_DMA_DESCRIPTOR_CTL_CP_STS;
	if (new->async_tx.callback) {
		hw->ctl |= IOAT_DMA_DESCRIPTOR_CTL_INT_GN;
		if (first != new) {
			/* move callback into to last desc */
			new->async_tx.callback = first->async_tx.callback;
			new->async_tx.callback_param
					= first->async_tx.callback_param;
			first->async_tx.callback = NULL;
			first->async_tx.callback_param = NULL;
		}
	}

	new->tx_cnt = desc_count;
	new->async_tx.ack = orig_ack; /* client is in control of this ack */

	/* store the original values for use in later cleanup */
	if (new != first) {
		new->src = first->src;
		new->dst = first->dst;
		new->len = first->len;
	}

	/* cookie incr and addition to used_list must be atomic */
	cookie = ioat_chan->common.cookie;
	cookie++;
	if (cookie < 0)
		cookie = 1;
	ioat_chan->common.cookie = new->async_tx.cookie = cookie;

	ioat_chan->dmacount += desc_count;
	ioat_chan->pending += desc_count;
	if (ioat_chan->pending >= ioat_pending_level)
		__ioat2_dma_memcpy_issue_pending(ioat_chan);
	spin_unlock_bh(&ioat_chan->desc_lock);

	return cookie;
}

/**
 * ioat_dma_alloc_descriptor - allocate and return a sw and hw descriptor pair
 * @ioat_chan: the channel supplying the memory pool for the descriptors
 * @flags: allocation flags
 */
static struct ioat_desc_sw *ioat_dma_alloc_descriptor(
					struct ioat_dma_chan *ioat_chan,
					gfp_t flags)
{
	struct ioat_dma_descriptor *desc;
	struct ioat_desc_sw *desc_sw;
	struct ioatdma_device *ioatdma_device;
	dma_addr_t phys;

	ioatdma_device = to_ioatdma_device(ioat_chan->common.device);
	desc = pci_pool_alloc(ioatdma_device->dma_pool, flags, &phys);
	if (unlikely(!desc))
		return NULL;

	desc_sw = kzalloc(sizeof(*desc_sw), flags);
	if (unlikely(!desc_sw)) {
		pci_pool_free(ioatdma_device->dma_pool, desc, phys);
		return NULL;
	}

	memset(desc, 0, sizeof(*desc));
	dma_async_tx_descriptor_init(&desc_sw->async_tx, &ioat_chan->common);
	switch (ioat_chan->device->version) {
	case IOAT_VER_1_2:
		desc_sw->async_tx.tx_submit = ioat1_tx_submit;
		break;
	case IOAT_VER_2_0:
		desc_sw->async_tx.tx_submit = ioat2_tx_submit;
		break;
	}
	INIT_LIST_HEAD(&desc_sw->async_tx.tx_list);

	desc_sw->hw = desc;
	desc_sw->async_tx.phys = phys;

	return desc_sw;
}

static int ioat_initial_desc_count = 256;
module_param(ioat_initial_desc_count, int, 0644);
MODULE_PARM_DESC(ioat_initial_desc_count,
		 "initial descriptors per channel (default: 256)");

/**
 * ioat2_dma_massage_chan_desc - link the descriptors into a circle
 * @ioat_chan: the channel to be massaged
 */
static void ioat2_dma_massage_chan_desc(struct ioat_dma_chan *ioat_chan)
{
	struct ioat_desc_sw *desc, *_desc;

	/* setup used_desc */
	ioat_chan->used_desc.next = ioat_chan->free_desc.next;
	ioat_chan->used_desc.prev = NULL;

	/* pull free_desc out of the circle so that every node is a hw
	 * descriptor, but leave it pointing to the list
	 */
	ioat_chan->free_desc.prev->next = ioat_chan->free_desc.next;
	ioat_chan->free_desc.next->prev = ioat_chan->free_desc.prev;

	/* circle link the hw descriptors */
	desc = to_ioat_desc(ioat_chan->free_desc.next);
	desc->hw->next = to_ioat_desc(desc->node.next)->async_tx.phys;
	list_for_each_entry_safe(desc, _desc, ioat_chan->free_desc.next, node) {
		desc->hw->next = to_ioat_desc(desc->node.next)->async_tx.phys;
	}
}

/**
 * ioat_dma_alloc_chan_resources - returns the number of allocated descriptors
 * @chan: the channel to be filled out
 */
static int ioat_dma_alloc_chan_resources(struct dma_chan *chan)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
	struct ioat_desc_sw *desc;
	u16 chanctrl;
	u32 chanerr;
	int i;
	LIST_HEAD(tmp_list);

	/* have we already been set up? */
	if (!list_empty(&ioat_chan->free_desc))
		return ioat_chan->desccount;

	/* Setup register to interrupt and write completion status on error */
	chanctrl = IOAT_CHANCTRL_ERR_INT_EN |
		IOAT_CHANCTRL_ANY_ERR_ABORT_EN |
		IOAT_CHANCTRL_ERR_COMPLETION_EN;
	writew(chanctrl, ioat_chan->reg_base + IOAT_CHANCTRL_OFFSET);

	chanerr = readl(ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
	if (chanerr) {
		dev_err(&ioat_chan->device->pdev->dev,
			"CHANERR = %x, clearing\n", chanerr);
		writel(chanerr, ioat_chan->reg_base + IOAT_CHANERR_OFFSET);
	}

	/* Allocate descriptors */
	for (i = 0; i < ioat_initial_desc_count; i++) {
		desc = ioat_dma_alloc_descriptor(ioat_chan, GFP_KERNEL);
		if (!desc) {
			dev_err(&ioat_chan->device->pdev->dev,
				"Only %d initial descriptors\n", i);
			break;
		}
		list_add_tail(&desc->node, &tmp_list);
	}
	spin_lock_bh(&ioat_chan->desc_lock);
	ioat_chan->desccount = i;
	list_splice(&tmp_list, &ioat_chan->free_desc);
	if (ioat_chan->device->version != IOAT_VER_1_2)
		ioat2_dma_massage_chan_desc(ioat_chan);
	spin_unlock_bh(&ioat_chan->desc_lock);

	/* allocate a completion writeback area */
	/* doing 2 32bit writes to mmio since 1 64b write doesn't work */
	ioat_chan->completion_virt =
		pci_pool_alloc(ioat_chan->device->completion_pool,
			       GFP_KERNEL,
			       &ioat_chan->completion_addr);
	memset(ioat_chan->completion_virt, 0,
	       sizeof(*ioat_chan->completion_virt));
	writel(((u64) ioat_chan->completion_addr) & 0x00000000FFFFFFFF,
	       ioat_chan->reg_base + IOAT_CHANCMP_OFFSET_LOW);
	writel(((u64) ioat_chan->completion_addr) >> 32,
	       ioat_chan->reg_base + IOAT_CHANCMP_OFFSET_HIGH);

	tasklet_enable(&ioat_chan->cleanup_task);
	ioat_dma_start_null_desc(ioat_chan);  /* give chain to dma device */
	return ioat_chan->desccount;
}

/**
 * ioat_dma_free_chan_resources - release all the descriptors
 * @chan: the channel to be cleaned
 */
static void ioat_dma_free_chan_resources(struct dma_chan *chan)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
	struct ioatdma_device *ioatdma_device = to_ioatdma_device(chan->device);
	struct ioat_desc_sw *desc, *_desc;
	int in_use_descs = 0;

	tasklet_disable(&ioat_chan->cleanup_task);
	ioat_dma_memcpy_cleanup(ioat_chan);

	/* Delay 100ms after reset to allow internal DMA logic to quiesce
	 * before removing DMA descriptor resources.
	 */
	writeb(IOAT_CHANCMD_RESET,
	       ioat_chan->reg_base
			+ IOAT_CHANCMD_OFFSET(ioat_chan->device->version));
	mdelay(100);

	spin_lock_bh(&ioat_chan->desc_lock);
	switch (ioat_chan->device->version) {
	case IOAT_VER_1_2:
		list_for_each_entry_safe(desc, _desc,
					 &ioat_chan->used_desc, node) {
			in_use_descs++;
			list_del(&desc->node);
			pci_pool_free(ioatdma_device->dma_pool, desc->hw,
				      desc->async_tx.phys);
			kfree(desc);
		}
		list_for_each_entry_safe(desc, _desc,
					 &ioat_chan->free_desc, node) {
			list_del(&desc->node);
			pci_pool_free(ioatdma_device->dma_pool, desc->hw,
				      desc->async_tx.phys);
			kfree(desc);
		}
		break;
	case IOAT_VER_2_0:
		list_for_each_entry_safe(desc, _desc,
					 ioat_chan->free_desc.next, node) {
			list_del(&desc->node);
			pci_pool_free(ioatdma_device->dma_pool, desc->hw,
				      desc->async_tx.phys);
			kfree(desc);
		}
		desc = to_ioat_desc(ioat_chan->free_desc.next);
		pci_pool_free(ioatdma_device->dma_pool, desc->hw,
			      desc->async_tx.phys);
		kfree(desc);
		INIT_LIST_HEAD(&ioat_chan->free_desc);
		INIT_LIST_HEAD(&ioat_chan->used_desc);
		break;
	}
	spin_unlock_bh(&ioat_chan->desc_lock);

	pci_pool_free(ioatdma_device->completion_pool,
		      ioat_chan->completion_virt,
		      ioat_chan->completion_addr);

	/* one is ok since we left it on there on purpose */
	if (in_use_descs > 1)
		dev_err(&ioat_chan->device->pdev->dev,
			"Freeing %d in use descriptors!\n",
			in_use_descs - 1);

	ioat_chan->last_completion = ioat_chan->completion_addr = 0;
	ioat_chan->pending = 0;
	ioat_chan->dmacount = 0;
}

/**
 * ioat_dma_get_next_descriptor - return the next available descriptor
 * @ioat_chan: IOAT DMA channel handle
 *
 * Gets the next descriptor from the chain, and must be called with the
 * channel's desc_lock held.  Allocates more descriptors if the channel
 * has run out.
 */
static struct ioat_desc_sw *
ioat1_dma_get_next_descriptor(struct ioat_dma_chan *ioat_chan)
{
	struct ioat_desc_sw *new;

	if (!list_empty(&ioat_chan->free_desc)) {
		new = to_ioat_desc(ioat_chan->free_desc.next);
		list_del(&new->node);
	} else {
		/* try to get another desc */
		new = ioat_dma_alloc_descriptor(ioat_chan, GFP_ATOMIC);
		if (!new) {
			dev_err(&ioat_chan->device->pdev->dev,
				"alloc failed\n");
			return NULL;
		}
	}

	prefetch(new->hw);
	return new;
}

static struct ioat_desc_sw *
ioat2_dma_get_next_descriptor(struct ioat_dma_chan *ioat_chan)
{
	struct ioat_desc_sw *new;

	/*
	 * used.prev points to where to start processing
	 * used.next points to next free descriptor
	 * if used.prev == NULL, there are none waiting to be processed
	 * if used.next == used.prev.prev, there is only one free descriptor,
	 *      and we need to use it to as a noop descriptor before
	 *      linking in a new set of descriptors, since the device
	 *      has probably already read the pointer to it
	 */
	if (ioat_chan->used_desc.prev &&
	    ioat_chan->used_desc.next == ioat_chan->used_desc.prev->prev) {

		struct ioat_desc_sw *desc;
		struct ioat_desc_sw *noop_desc;
		int i;

		/* set up the noop descriptor */
		noop_desc = to_ioat_desc(ioat_chan->used_desc.next);
		noop_desc->hw->size = 0;
		noop_desc->hw->ctl = IOAT_DMA_DESCRIPTOR_NUL;
		noop_desc->hw->src_addr = 0;
		noop_desc->hw->dst_addr = 0;

		ioat_chan->used_desc.next = ioat_chan->used_desc.next->next;
		ioat_chan->pending++;
		ioat_chan->dmacount++;

		/* try to get a few more descriptors */
		for (i = 16; i; i--) {
			desc = ioat_dma_alloc_descriptor(ioat_chan, GFP_ATOMIC);
			if (!desc) {
				dev_err(&ioat_chan->device->pdev->dev,
					"alloc failed\n");
				break;
			}
			list_add_tail(&desc->node, ioat_chan->used_desc.next);

			desc->hw->next
				= to_ioat_desc(desc->node.next)->async_tx.phys;
			to_ioat_desc(desc->node.prev)->hw->next
				= desc->async_tx.phys;
			ioat_chan->desccount++;
		}

		ioat_chan->used_desc.next = noop_desc->node.next;
	}
	new = to_ioat_desc(ioat_chan->used_desc.next);
	prefetch(new);
	ioat_chan->used_desc.next = new->node.next;

	if (ioat_chan->used_desc.prev == NULL)
		ioat_chan->used_desc.prev = &new->node;

	prefetch(new->hw);
	return new;
}

static struct ioat_desc_sw *ioat_dma_get_next_descriptor(
						struct ioat_dma_chan *ioat_chan)
{
	if (!ioat_chan)
		return NULL;

	switch (ioat_chan->device->version) {
	case IOAT_VER_1_2:
		return ioat1_dma_get_next_descriptor(ioat_chan);
		break;
	case IOAT_VER_2_0:
		return ioat2_dma_get_next_descriptor(ioat_chan);
		break;
	}
	return NULL;
}

static struct dma_async_tx_descriptor *ioat1_dma_prep_memcpy(
						struct dma_chan *chan,
						dma_addr_t dma_dest,
						dma_addr_t dma_src,
						size_t len,
						unsigned long flags)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
	struct ioat_desc_sw *new;

	spin_lock_bh(&ioat_chan->desc_lock);
	new = ioat_dma_get_next_descriptor(ioat_chan);
	spin_unlock_bh(&ioat_chan->desc_lock);

	if (new) {
		new->len = len;
		new->dst = dma_dest;
		new->src = dma_src;
		return &new->async_tx;
	} else
		return NULL;
}

static struct dma_async_tx_descriptor *ioat2_dma_prep_memcpy(
						struct dma_chan *chan,
						dma_addr_t dma_dest,
						dma_addr_t dma_src,
						size_t len,
						unsigned long flags)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
	struct ioat_desc_sw *new;

	spin_lock_bh(&ioat_chan->desc_lock);
	new = ioat2_dma_get_next_descriptor(ioat_chan);

	/*
	 * leave ioat_chan->desc_lock set in ioat 2 path
	 * it will get unlocked at end of tx_submit
	 */

	if (new) {
		new->len = len;
		new->dst = dma_dest;
		new->src = dma_src;
		return &new->async_tx;
	} else
		return NULL;
}

static void ioat_dma_cleanup_tasklet(unsigned long data)
{
	struct ioat_dma_chan *chan = (void *)data;
	ioat_dma_memcpy_cleanup(chan);
	writew(IOAT_CHANCTRL_INT_DISABLE,
	       chan->reg_base + IOAT_CHANCTRL_OFFSET);
}

/**
 * ioat_dma_memcpy_cleanup - cleanup up finished descriptors
 * @chan: ioat channel to be cleaned up
 */
static void ioat_dma_memcpy_cleanup(struct ioat_dma_chan *ioat_chan)
{
	unsigned long phys_complete;
	struct ioat_desc_sw *desc, *_desc;
	dma_cookie_t cookie = 0;
	unsigned long desc_phys;
	struct ioat_desc_sw *latest_desc;

	prefetch(ioat_chan->completion_virt);

	if (!spin_trylock_bh(&ioat_chan->cleanup_lock))
		return;

	/* The completion writeback can happen at any time,
	   so reads by the driver need to be atomic operations
	   The descriptor physical addresses are limited to 32-bits
	   when the CPU can only do a 32-bit mov */

#if (BITS_PER_LONG == 64)
	phys_complete =
		ioat_chan->completion_virt->full
		& IOAT_CHANSTS_COMPLETED_DESCRIPTOR_ADDR;
#else
	phys_complete =
		ioat_chan->completion_virt->low & IOAT_LOW_COMPLETION_MASK;
#endif

	if ((ioat_chan->completion_virt->full
		& IOAT_CHANSTS_DMA_TRANSFER_STATUS) ==
				IOAT_CHANSTS_DMA_TRANSFER_STATUS_HALTED) {
		dev_err(&ioat_chan->device->pdev->dev,
			"Channel halted, chanerr = %x\n",
			readl(ioat_chan->reg_base + IOAT_CHANERR_OFFSET));

		/* TODO do something to salvage the situation */
	}

	if (phys_complete == ioat_chan->last_completion) {
		spin_unlock_bh(&ioat_chan->cleanup_lock);
		return;
	}

	cookie = 0;
	spin_lock_bh(&ioat_chan->desc_lock);
	switch (ioat_chan->device->version) {
	case IOAT_VER_1_2:
		list_for_each_entry_safe(desc, _desc,
					 &ioat_chan->used_desc, node) {

			/*
			 * Incoming DMA requests may use multiple descriptors,
			 * due to exceeding xfercap, perhaps. If so, only the
			 * last one will have a cookie, and require unmapping.
			 */
			if (desc->async_tx.cookie) {
				cookie = desc->async_tx.cookie;

				/*
				 * yes we are unmapping both _page and _single
				 * alloc'd regions with unmap_page. Is this
				 * *really* that bad?
				 */
				pci_unmap_page(ioat_chan->device->pdev,
						pci_unmap_addr(desc, dst),
						pci_unmap_len(desc, len),
						PCI_DMA_FROMDEVICE);
				pci_unmap_page(ioat_chan->device->pdev,
						pci_unmap_addr(desc, src),
						pci_unmap_len(desc, len),
						PCI_DMA_TODEVICE);

				if (desc->async_tx.callback) {
					desc->async_tx.callback(desc->async_tx.callback_param);
					desc->async_tx.callback = NULL;
				}
			}

			if (desc->async_tx.phys != phys_complete) {
				/*
				 * a completed entry, but not the last, so clean
				 * up if the client is done with the descriptor
				 */
				if (desc->async_tx.ack) {
					list_del(&desc->node);
					list_add_tail(&desc->node,
						      &ioat_chan->free_desc);
				} else
					desc->async_tx.cookie = 0;
			} else {
				/*
				 * last used desc. Do not remove, so we can
				 * append from it, but don't look at it next
				 * time, either
				 */
				desc->async_tx.cookie = 0;

				/* TODO check status bits? */
				break;
			}
		}
		break;
	case IOAT_VER_2_0:
		/* has some other thread has already cleaned up? */
		if (ioat_chan->used_desc.prev == NULL)
			break;

		/* work backwards to find latest finished desc */
		desc = to_ioat_desc(ioat_chan->used_desc.next);
		latest_desc = NULL;
		do {
			desc = to_ioat_desc(desc->node.prev);
			desc_phys = (unsigned long)desc->async_tx.phys
				       & IOAT_CHANSTS_COMPLETED_DESCRIPTOR_ADDR;
			if (desc_phys == phys_complete) {
				latest_desc = desc;
				break;
			}
		} while (&desc->node != ioat_chan->used_desc.prev);

		if (latest_desc != NULL) {

			/* work forwards to clear finished descriptors */
			for (desc = to_ioat_desc(ioat_chan->used_desc.prev);
			     &desc->node != latest_desc->node.next &&
			     &desc->node != ioat_chan->used_desc.next;
			     desc = to_ioat_desc(desc->node.next)) {
				if (desc->async_tx.cookie) {
					cookie = desc->async_tx.cookie;
					desc->async_tx.cookie = 0;

					pci_unmap_page(ioat_chan->device->pdev,
						      pci_unmap_addr(desc, dst),
						      pci_unmap_len(desc, len),
						      PCI_DMA_FROMDEVICE);
					pci_unmap_page(ioat_chan->device->pdev,
						      pci_unmap_addr(desc, src),
						      pci_unmap_len(desc, len),
						      PCI_DMA_TODEVICE);

					if (desc->async_tx.callback) {
						desc->async_tx.callback(desc->async_tx.callback_param);
						desc->async_tx.callback = NULL;
					}
				}
			}

			/* move used.prev up beyond those that are finished */
			if (&desc->node == ioat_chan->used_desc.next)
				ioat_chan->used_desc.prev = NULL;
			else
				ioat_chan->used_desc.prev = &desc->node;
		}
		break;
	}

	spin_unlock_bh(&ioat_chan->desc_lock);

	ioat_chan->last_completion = phys_complete;
	if (cookie != 0)
		ioat_chan->completed_cookie = cookie;

	spin_unlock_bh(&ioat_chan->cleanup_lock);
}

static void ioat_dma_dependency_added(struct dma_chan *chan)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
	spin_lock_bh(&ioat_chan->desc_lock);
	if (ioat_chan->pending == 0) {
		spin_unlock_bh(&ioat_chan->desc_lock);
		ioat_dma_memcpy_cleanup(ioat_chan);
	} else
		spin_unlock_bh(&ioat_chan->desc_lock);
}

/**
 * ioat_dma_is_complete - poll the status of a IOAT DMA transaction
 * @chan: IOAT DMA channel handle
 * @cookie: DMA transaction identifier
 * @done: if not %NULL, updated with last completed transaction
 * @used: if not %NULL, updated with last used transaction
 */
static enum dma_status ioat_dma_is_complete(struct dma_chan *chan,
					    dma_cookie_t cookie,
					    dma_cookie_t *done,
					    dma_cookie_t *used)
{
	struct ioat_dma_chan *ioat_chan = to_ioat_chan(chan);
	dma_cookie_t last_used;
	dma_cookie_t last_complete;
	enum dma_status ret;

	last_used = chan->cookie;
	last_complete = ioat_chan->completed_cookie;

	if (done)
		*done = last_complete;
	if (used)
		*used = last_used;

	ret = dma_async_is_complete(cookie, last_complete, last_used);
	if (ret == DMA_SUCCESS)
		return ret;

	ioat_dma_memcpy_cleanup(ioat_chan);

	last_used = chan->cookie;
	last_complete = ioat_chan->completed_cookie;

	if (done)
		*done = last_complete;
	if (used)
		*used = last_used;

	return dma_async_is_complete(cookie, last_complete, last_used);
}

static void ioat_dma_start_null_desc(struct ioat_dma_chan *ioat_chan)
{
	struct ioat_desc_sw *desc;

	spin_lock_bh(&ioat_chan->desc_lock);

	desc = ioat_dma_get_next_descriptor(ioat_chan);
	desc->hw->ctl = IOAT_DMA_DESCRIPTOR_NUL
				| IOAT_DMA_DESCRIPTOR_CTL_INT_GN
				| IOAT_DMA_DESCRIPTOR_CTL_CP_STS;
	desc->hw->size = 0;
	desc->hw->src_addr = 0;
	desc->hw->dst_addr = 0;
	desc->async_tx.ack = 1;
	switch (ioat_chan->device->version) {
	case IOAT_VER_1_2:
		desc->hw->next = 0;
		list_add_tail(&desc->node, &ioat_chan->used_desc);

		writel(((u64) desc->async_tx.phys) & 0x00000000FFFFFFFF,
		       ioat_chan->reg_base + IOAT1_CHAINADDR_OFFSET_LOW);
		writel(((u64) desc->async_tx.phys) >> 32,
		       ioat_chan->reg_base + IOAT1_CHAINADDR_OFFSET_HIGH);

		writeb(IOAT_CHANCMD_START, ioat_chan->reg_base
			+ IOAT_CHANCMD_OFFSET(ioat_chan->device->version));
		break;
	case IOAT_VER_2_0:
		writel(((u64) desc->async_tx.phys) & 0x00000000FFFFFFFF,
		       ioat_chan->reg_base + IOAT2_CHAINADDR_OFFSET_LOW);
		writel(((u64) desc->async_tx.phys) >> 32,
		       ioat_chan->reg_base + IOAT2_CHAINADDR_OFFSET_HIGH);

		ioat_chan->dmacount++;
		__ioat2_dma_memcpy_issue_pending(ioat_chan);
		break;
	}
	spin_unlock_bh(&ioat_chan->desc_lock);
}

/*
 * Perform a IOAT transaction to verify the HW works.
 */
#define IOAT_TEST_SIZE 2000

static void ioat_dma_test_callback(void *dma_async_param)
{
	printk(KERN_ERR "ioatdma: ioat_dma_test_callback(%p)\n",
		dma_async_param);
}

/**
 * ioat_dma_self_test - Perform a IOAT transaction to verify the HW works.
 * @device: device to be tested
 */
static int ioat_dma_self_test(struct ioatdma_device *device)
{
	int i;
	u8 *src;
	u8 *dest;
	struct dma_chan *dma_chan;
	struct dma_async_tx_descriptor *tx;
	dma_addr_t dma_dest, dma_src;
	dma_cookie_t cookie;
	int err = 0;

	src = kzalloc(sizeof(u8) * IOAT_TEST_SIZE, GFP_KERNEL);
	if (!src)
		return -ENOMEM;
	dest = kzalloc(sizeof(u8) * IOAT_TEST_SIZE, GFP_KERNEL);
	if (!dest) {
		kfree(src);
		return -ENOMEM;
	}

	/* Fill in src buffer */
	for (i = 0; i < IOAT_TEST_SIZE; i++)
		src[i] = (u8)i;

	/* Start copy, using first DMA channel */
	dma_chan = container_of(device->common.channels.next,
				struct dma_chan,
				device_node);
	if (device->common.device_alloc_chan_resources(dma_chan) < 1) {
		dev_err(&device->pdev->dev,
			"selftest cannot allocate chan resource\n");
		err = -ENODEV;
		goto out;
	}

	dma_src = dma_map_single(dma_chan->device->dev, src, IOAT_TEST_SIZE,
				 DMA_TO_DEVICE);
	dma_dest = dma_map_single(dma_chan->device->dev, dest, IOAT_TEST_SIZE,
				  DMA_FROM_DEVICE);
	tx = device->common.device_prep_dma_memcpy(dma_chan, dma_dest, dma_src,
						   IOAT_TEST_SIZE, 0);
	if (!tx) {
		dev_err(&device->pdev->dev,
			"Self-test prep failed, disabling\n");
		err = -ENODEV;
		goto free_resources;
	}

	async_tx_ack(tx);
	tx->callback = ioat_dma_test_callback;
	tx->callback_param = (void *)0x8086;
	cookie = tx->tx_submit(tx);
	if (cookie < 0) {
		dev_err(&device->pdev->dev,
			"Self-test setup failed, disabling\n");
		err = -ENODEV;
		goto free_resources;
	}
	device->common.device_issue_pending(dma_chan);
	msleep(1);

	if (device->common.device_is_tx_complete(dma_chan, cookie, NULL, NULL)
					!= DMA_SUCCESS) {
		dev_err(&device->pdev->dev,
			"Self-test copy timed out, disabling\n");
		err = -ENODEV;
		goto free_resources;
	}
	if (memcmp(src, dest, IOAT_TEST_SIZE)) {
		dev_err(&device->pdev->dev,
			"Self-test copy failed compare, disabling\n");
		err = -ENODEV;
		goto free_resources;
	}

free_resources:
	device->common.device_free_chan_resources(dma_chan);
out:
	kfree(src);
	kfree(dest);
	return err;
}

static char ioat_interrupt_style[32] = "msix";
module_param_string(ioat_interrupt_style, ioat_interrupt_style,
		    sizeof(ioat_interrupt_style), 0644);
MODULE_PARM_DESC(ioat_interrupt_style,
		 "set ioat interrupt style: msix (default), "
		 "msix-single-vector, msi, intx)");

/**
 * ioat_dma_setup_interrupts - setup interrupt handler
 * @device: ioat device
 */
static int ioat_dma_setup_interrupts(struct ioatdma_device *device)
{
	struct ioat_dma_chan *ioat_chan;
	int err, i, j, msixcnt;
	u8 intrctrl = 0;

	if (!strcmp(ioat_interrupt_style, "msix"))
		goto msix;
	if (!strcmp(ioat_interrupt_style, "msix-single-vector"))
		goto msix_single_vector;
	if (!strcmp(ioat_interrupt_style, "msi"))
		goto msi;
	if (!strcmp(ioat_interrupt_style, "intx"))
		goto intx;
	dev_err(&device->pdev->dev, "invalid ioat_interrupt_style %s\n",
		ioat_interrupt_style);
	goto err_no_irq;

msix:
	/* The number of MSI-X vectors should equal the number of channels */
	msixcnt = device->common.chancnt;
	for (i = 0; i < msixcnt; i++)
		device->msix_entries[i].entry = i;

	err = pci_enable_msix(device->pdev, device->msix_entries, msixcnt);
	if (err < 0)
		goto msi;
	if (err > 0)
		goto msix_single_vector;

	for (i = 0; i < msixcnt; i++) {
		ioat_chan = ioat_lookup_chan_by_index(device, i);
		err = request_irq(device->msix_entries[i].vector,
				  ioat_dma_do_interrupt_msix,
				  0, "ioat-msix", ioat_chan);
		if (err) {
			for (j = 0; j < i; j++) {
				ioat_chan =
					ioat_lookup_chan_by_index(device, j);
				free_irq(device->msix_entries[j].vector,
					 ioat_chan);
			}
			goto msix_single_vector;
		}
	}
	intrctrl |= IOAT_INTRCTRL_MSIX_VECTOR_CONTROL;
	device->irq_mode = msix_multi_vector;
	goto done;

msix_single_vector:
	device->msix_entries[0].entry = 0;
	err = pci_enable_msix(device->pdev, device->msix_entries, 1);
	if (err)
		goto msi;

	err = request_irq(device->msix_entries[0].vector, ioat_dma_do_interrupt,
			  0, "ioat-msix", device);
	if (err) {
		pci_disable_msix(device->pdev);
		goto msi;
	}
	device->irq_mode = msix_single_vector;
	goto done;

msi:
	err = pci_enable_msi(device->pdev);
	if (err)
		goto intx;

	err = request_irq(device->pdev->irq, ioat_dma_do_interrupt,
			  0, "ioat-msi", device);
	if (err) {
		pci_disable_msi(device->pdev);
		goto intx;
	}
	/*
	 * CB 1.2 devices need a bit set in configuration space to enable MSI
	 */
	if (device->version == IOAT_VER_1_2) {
		u32 dmactrl;
		pci_read_config_dword(device->pdev,
				      IOAT_PCI_DMACTRL_OFFSET, &dmactrl);
		dmactrl |= IOAT_PCI_DMACTRL_MSI_EN;
		pci_write_config_dword(device->pdev,
				       IOAT_PCI_DMACTRL_OFFSET, dmactrl);
	}
	device->irq_mode = msi;
	goto done;

intx:
	err = request_irq(device->pdev->irq, ioat_dma_do_interrupt,
			  IRQF_SHARED, "ioat-intx", device);
	if (err)
		goto err_no_irq;
	device->irq_mode = intx;

done:
	intrctrl |= IOAT_INTRCTRL_MASTER_INT_EN;
	writeb(intrctrl, device->reg_base + IOAT_INTRCTRL_OFFSET);
	return 0;

err_no_irq:
	/* Disable all interrupt generation */
	writeb(0, device->reg_base + IOAT_INTRCTRL_OFFSET);
	dev_err(&device->pdev->dev, "no usable interrupts\n");
	device->irq_mode = none;
	return -1;
}

/**
 * ioat_dma_remove_interrupts - remove whatever interrupts were set
 * @device: ioat device
 */
static void ioat_dma_remove_interrupts(struct ioatdma_device *device)
{
	struct ioat_dma_chan *ioat_chan;
	int i;

	/* Disable all interrupt generation */
	writeb(0, device->reg_base + IOAT_INTRCTRL_OFFSET);

	switch (device->irq_mode) {
	case msix_multi_vector:
		for (i = 0; i < device->common.chancnt; i++) {
			ioat_chan = ioat_lookup_chan_by_index(device, i);
			free_irq(device->msix_entries[i].vector, ioat_chan);
		}
		pci_disable_msix(device->pdev);
		break;
	case msix_single_vector:
		free_irq(device->msix_entries[0].vector, device);
		pci_disable_msix(device->pdev);
		break;
	case msi:
		free_irq(device->pdev->irq, device);
		pci_disable_msi(device->pdev);
		break;
	case intx:
		free_irq(device->pdev->irq, device);
		break;
	case none:
		dev_warn(&device->pdev->dev,
			 "call to %s without interrupts setup\n", __func__);
	}
	device->irq_mode = none;
}

struct ioatdma_device *ioat_dma_probe(struct pci_dev *pdev,
				      void __iomem *iobase)
{
	int err;
	struct ioatdma_device *device;

	device = kzalloc(sizeof(*device), GFP_KERNEL);
	if (!device) {
		err = -ENOMEM;
		goto err_kzalloc;
	}
	device->pdev = pdev;
	device->reg_base = iobase;
	device->version = readb(device->reg_base + IOAT_VER_OFFSET);

	/* DMA coherent memory pool for DMA descriptor allocations */
	device->dma_pool = pci_pool_create("dma_desc_pool", pdev,
					   sizeof(struct ioat_dma_descriptor),
					   64, 0);
	if (!device->dma_pool) {
		err = -ENOMEM;
		goto err_dma_pool;
	}

	device->completion_pool = pci_pool_create("completion_pool", pdev,
						  sizeof(u64), SMP_CACHE_BYTES,
						  SMP_CACHE_BYTES);
	if (!device->completion_pool) {
		err = -ENOMEM;
		goto err_completion_pool;
	}

	INIT_LIST_HEAD(&device->common.channels);
	ioat_dma_enumerate_channels(device);

	device->common.device_alloc_chan_resources =
						ioat_dma_alloc_chan_resources;
	device->common.device_free_chan_resources =
						ioat_dma_free_chan_resources;
	device->common.dev = &pdev->dev;

	dma_cap_set(DMA_MEMCPY, device->common.cap_mask);
	device->common.device_is_tx_complete = ioat_dma_is_complete;
	device->common.device_dependency_added = ioat_dma_dependency_added;
	switch (device->version) {
	case IOAT_VER_1_2:
		device->common.device_prep_dma_memcpy = ioat1_dma_prep_memcpy;
		device->common.device_issue_pending =
						ioat1_dma_memcpy_issue_pending;
		break;
	case IOAT_VER_2_0:
		device->common.device_prep_dma_memcpy = ioat2_dma_prep_memcpy;
		device->common.device_issue_pending =
						ioat2_dma_memcpy_issue_pending;
		break;
	}

	dev_err(&device->pdev->dev,
		"Intel(R) I/OAT DMA Engine found,"
		" %d channels, device version 0x%02x, driver version %s\n",
		device->common.chancnt, device->version, IOAT_DMA_VERSION);

	err = ioat_dma_setup_interrupts(device);
	if (err)
		goto err_setup_interrupts;

	err = ioat_dma_self_test(device);
	if (err)
		goto err_self_test;

	dma_async_device_register(&device->common);

	return device;

err_self_test:
	ioat_dma_remove_interrupts(device);
err_setup_interrupts:
	pci_pool_destroy(device->completion_pool);
err_completion_pool:
	pci_pool_destroy(device->dma_pool);
err_dma_pool:
	kfree(device);
err_kzalloc:
	dev_err(&pdev->dev,
		"Intel(R) I/OAT DMA Engine initialization failed\n");
	return NULL;
}

void ioat_dma_remove(struct ioatdma_device *device)
{
	struct dma_chan *chan, *_chan;
	struct ioat_dma_chan *ioat_chan;

	ioat_dma_remove_interrupts(device);

	dma_async_device_unregister(&device->common);

	pci_pool_destroy(device->dma_pool);
	pci_pool_destroy(device->completion_pool);

	iounmap(device->reg_base);
	pci_release_regions(device->pdev);
	pci_disable_device(device->pdev);

	list_for_each_entry_safe(chan, _chan,
				 &device->common.channels, device_node) {
		ioat_chan = to_ioat_chan(chan);
		list_del(&chan->device_node);
		kfree(ioat_chan);
	}
	kfree(device);
}