Merge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx

* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx: (28 commits)
  ioat: cleanup ->timer_fn() and ->cleanup_fn() prototypes
  ioat3: interrupt coalescing
  ioat: close potential BUG_ON race in the descriptor cleanup path
  ioat2: kill pending flag
  ioat3: use ioat2_quiesce()
  ioat3: cleanup, don't enable DCA completion writes
  DMAENGINE: COH 901 318 lli sg offset fix
  DMAENGINE: COH 901 318 configure channel direction
  DMAENGINE: COH 901 318 remove irq counting
  DMAENGINE: COH 901 318 descriptor pool refactoring
  DMAENGINE: COH 901 318 cleanups
  dma: Add MPC512x DMA driver
  Debugging options for the DMA engine subsystem
  iop-adma: redundant/wrong tests in iop_*_count()?
  dmatest: fix handling of an even number of xor_sources
  dmatest: correct raid6 PQ test
  fsldma: Fix cookie issues
  fsldma: Fix cookie issues
  dma: cases IPU_PIX_FMT_BGRA32, BGR32 and ABGR32 are the same in ipu_ch_param_set_size()
  dma: make Open Firmware device id constant
  ...

Documentation/powerpc/dts-bindings/fsl/dma.txt

@@ -44,21 +44,29 @@ Example:
 			compatible = "fsl,mpc8349-dma-channel", "fsl,elo-dma-channel";
 			cell-index = <0>;
 			reg = <0 0x80>;
+			interrupt-parent = <&ipic>;
+			interrupts = <71 8>;
 		};
 		dma-channel@80 {
 			compatible = "fsl,mpc8349-dma-channel", "fsl,elo-dma-channel";
 			cell-index = <1>;
 			reg = <0x80 0x80>;
+			interrupt-parent = <&ipic>;
+			interrupts = <71 8>;
 		};
 		dma-channel@100 {
 			compatible = "fsl,mpc8349-dma-channel", "fsl,elo-dma-channel";
 			cell-index = <2>;
 			reg = <0x100 0x80>;
+			interrupt-parent = <&ipic>;
+			interrupts = <71 8>;
 		};
 		dma-channel@180 {
 			compatible = "fsl,mpc8349-dma-channel", "fsl,elo-dma-channel";
 			cell-index = <3>;
 			reg = <0x180 0x80>;
+			interrupt-parent = <&ipic>;
+			interrupts = <71 8>;
 		};
 	};
 

arch/arm/include/asm/hardware/iop3xx-adma.h

@@ -366,8 +366,7 @@ static inline int iop_chan_xor_slot_count(size_t len, int src_cnt,
 		slot_cnt += *slots_per_op;
 	}
 
-	if (len)
-		slot_cnt += *slots_per_op;
+	slot_cnt += *slots_per_op;
 
 	return slot_cnt;
 }
@@ -389,8 +388,7 @@ static inline int iop_chan_zero_sum_slot_count(size_t len, int src_cnt,
 		slot_cnt += *slots_per_op;
 	}
 
-	if (len)
-		slot_cnt += *slots_per_op;
+	slot_cnt += *slots_per_op;
 
 	return slot_cnt;
 }
@@ -737,10 +735,8 @@ iop_desc_set_zero_sum_byte_count(struct iop_adma_desc_slot *desc, u32 len)
 			i += slots_per_op;
 		} while (len > IOP_ADMA_ZERO_SUM_MAX_BYTE_COUNT);
 
-		if (len) {
-			iter = iop_hw_desc_slot_idx(hw_desc, i);
-			iter->byte_count = len;
-		}
+		iter = iop_hw_desc_slot_idx(hw_desc, i);
+		iter->byte_count = len;
 	}
 }
 

arch/arm/mach-u300/include/mach/coh901318.h

@@ -53,7 +53,7 @@ struct coh901318_params {
  * struct coh_dma_channel - dma channel base
  * @name: ascii name of dma channel
  * @number: channel id number
- * @desc_nbr_max: number of preallocated descriptortors
+ * @desc_nbr_max: number of preallocated descriptors
  * @priority_high: prio of channel, 0 low otherwise high.
  * @param: configuration parameters
  * @dev_addr: physical address of periphal connected to channel

drivers/dma/Kconfig

@@ -13,6 +13,22 @@ menuconfig DMADEVICES
 	  DMA Device drivers supported by the configured arch, it may
 	  be empty in some cases.
 
+config DMADEVICES_DEBUG
+        bool "DMA Engine debugging"
+        depends on DMADEVICES != n
+        help
+          This is an option for use by developers; most people should
+          say N here.  This enables DMA engine core and driver debugging.
+
+config DMADEVICES_VDEBUG
+        bool "DMA Engine verbose debugging"
+        depends on DMADEVICES_DEBUG != n
+        help
+          This is an option for use by developers; most people should
+          say N here.  This enables deeper (more verbose) debugging of
+          the DMA engine core and drivers.
+
+
 if DMADEVICES
 
 comment "DMA Devices"
@@ -69,6 +85,13 @@ config FSL_DMA
 	  The Elo is the DMA controller on some 82xx and 83xx parts, and the
 	  Elo Plus is the DMA controller on 85xx and 86xx parts.
 
+config MPC512X_DMA
+	tristate "Freescale MPC512x built-in DMA engine support"
+	depends on PPC_MPC512x
+	select DMA_ENGINE
+	---help---
+	  Enable support for the Freescale MPC512x built-in DMA engine.
+
 config MV_XOR
 	bool "Marvell XOR engine support"
 	depends on PLAT_ORION

drivers/dma/Makefile

+ifeq ($(CONFIG_DMADEVICES_DEBUG),y)
+	EXTRA_CFLAGS	+= -DDEBUG
+endif
+ifeq ($(CONFIG_DMADEVICES_VDEBUG),y)
+	EXTRA_CFLAGS	+= -DVERBOSE_DEBUG
+endif
+
 obj-$(CONFIG_DMA_ENGINE) += dmaengine.o
 obj-$(CONFIG_NET_DMA) += iovlock.o
 obj-$(CONFIG_DMATEST) += dmatest.o
 obj-$(CONFIG_INTEL_IOATDMA) += ioat/
 obj-$(CONFIG_INTEL_IOP_ADMA) += iop-adma.o
 obj-$(CONFIG_FSL_DMA) += fsldma.o
+obj-$(CONFIG_MPC512X_DMA) += mpc512x_dma.o
 obj-$(CONFIG_MV_XOR) += mv_xor.o
 obj-$(CONFIG_DW_DMAC) += dw_dmac.o
 obj-$(CONFIG_AT_HDMAC) += at_hdmac.o

drivers/dma/coh901318.c

@@ -39,7 +39,6 @@ struct coh901318_desc {
 	unsigned int sg_len;
 	struct coh901318_lli *data;
 	enum dma_data_direction dir;
-	int pending_irqs;
 	unsigned long flags;
 };
 
@@ -72,7 +71,6 @@ struct coh901318_chan {
 
 	unsigned long nbr_active_done;
 	unsigned long busy;
-	int pending_irqs;
 
 	struct coh901318_base *base;
 };
@@ -80,18 +78,16 @@ struct coh901318_chan {
 static void coh901318_list_print(struct coh901318_chan *cohc,
 				 struct coh901318_lli *lli)
 {
-	struct coh901318_lli *l;
-	dma_addr_t addr =  virt_to_phys(lli);
+	struct coh901318_lli *l = lli;
 	int i = 0;
 
-	while (addr) {
-		l = phys_to_virt(addr);
+	while (l) {
 		dev_vdbg(COHC_2_DEV(cohc), "i %d, lli %p, ctrl 0x%x, src 0x%x"
-			 ", dst 0x%x, link 0x%x link_virt 0x%p\n",
+			 ", dst 0x%x, link 0x%x virt_link_addr 0x%p\n",
 			 i, l, l->control, l->src_addr, l->dst_addr,
-			 l->link_addr, phys_to_virt(l->link_addr));
+			 l->link_addr, l->virt_link_addr);
 		i++;
-		addr = l->link_addr;
+		l = l->virt_link_addr;
 	}
 }
 
@@ -125,7 +121,7 @@ static int coh901318_debugfs_read(struct file *file, char __user *buf,
 		goto err_kmalloc;
 	tmp = dev_buf;
 
-	tmp += sprintf(tmp, "DMA -- enable dma channels\n");
+	tmp += sprintf(tmp, "DMA -- enabled dma channels\n");
 
 	for (i = 0; i < debugfs_dma_base->platform->max_channels; i++)
 		if (started_channels & (1 << i))
@@ -337,16 +333,22 @@ coh901318_desc_get(struct coh901318_chan *cohc)
 		 * TODO: alloc a pile of descs instead of just one,
 		 * avoid many small allocations.
 		 */
-		desc = kmalloc(sizeof(struct coh901318_desc), GFP_NOWAIT);
+		desc = kzalloc(sizeof(struct coh901318_desc), GFP_NOWAIT);
 		if (desc == NULL)
 			goto out;
 		INIT_LIST_HEAD(&desc->node);
+		dma_async_tx_descriptor_init(&desc->desc, &cohc->chan);
 	} else {
 		/* Reuse an old desc. */
 		desc = list_first_entry(&cohc->free,
 					struct coh901318_desc,
 					node);
 		list_del(&desc->node);
+		/* Initialize it a bit so it's not insane */
+		desc->sg = NULL;
+		desc->sg_len = 0;
+		desc->desc.callback = NULL;
+		desc->desc.callback_param = NULL;
 	}
 
  out:
@@ -364,10 +366,6 @@ static void
 coh901318_desc_submit(struct coh901318_chan *cohc, struct coh901318_desc *desc)
 {
 	list_add_tail(&desc->node, &cohc->active);
-
-	BUG_ON(cohc->pending_irqs != 0);
-
-	cohc->pending_irqs = desc->pending_irqs;
 }
 
 static struct coh901318_desc *
@@ -592,6 +590,10 @@ static struct coh901318_desc *coh901318_queue_start(struct coh901318_chan *cohc)
 	return cohd_que;
 }
 
+/*
+ * This tasklet is called from the interrupt handler to
+ * handle each descriptor (DMA job) that is sent to a channel.
+ */
 static void dma_tasklet(unsigned long data)
 {
 	struct coh901318_chan *cohc = (struct coh901318_chan *) data;
@@ -600,55 +602,58 @@ static void dma_tasklet(unsigned long data)
 	dma_async_tx_callback callback;
 	void *callback_param;
 
+	dev_vdbg(COHC_2_DEV(cohc), "[%s] chan_id %d"
+		 " nbr_active_done %ld\n", __func__,
+		 cohc->id, cohc->nbr_active_done);
+
 	spin_lock_irqsave(&cohc->lock, flags);
 
-	/* get first active entry from list */
+	/* get first active descriptor entry from list */
 	cohd_fin = coh901318_first_active_get(cohc);
 
-	BUG_ON(cohd_fin->pending_irqs == 0);
-
 	if (cohd_fin == NULL)
 		goto err;
 
-	cohd_fin->pending_irqs--;
-	cohc->completed = cohd_fin->desc.cookie;
+	/* locate callback to client */
+	callback = cohd_fin->desc.callback;
+	callback_param = cohd_fin->desc.callback_param;
 
-	if (cohc->nbr_active_done == 0)
-		return;
+	/* sign this job as completed on the channel */
+	cohc->completed = cohd_fin->desc.cookie;
 
-	if (!cohd_fin->pending_irqs) {
-		/* release the lli allocation*/
-		coh901318_lli_free(&cohc->base->pool, &cohd_fin->data);
-	}
+	/* release the lli allocation and remove the descriptor */
+	coh901318_lli_free(&cohc->base->pool, &cohd_fin->data);
 
-	dev_vdbg(COHC_2_DEV(cohc), "[%s] chan_id %d pending_irqs %d"
-		 " nbr_active_done %ld\n", __func__,
-		 cohc->id, cohc->pending_irqs, cohc->nbr_active_done);
+	/* return desc to free-list */
+	coh901318_desc_remove(cohd_fin);
+	coh901318_desc_free(cohc, cohd_fin);
 
-	/* callback to client */
-	callback = cohd_fin->desc.callback;
-	callback_param = cohd_fin->desc.callback_param;
-
-	if (!cohd_fin->pending_irqs) {
-		coh901318_desc_remove(cohd_fin);
+	spin_unlock_irqrestore(&cohc->lock, flags);
 
-		/* return desc to free-list */
-		coh901318_desc_free(cohc, cohd_fin);
-	}
+	/* Call the callback when we're done */
+	if (callback)
+		callback(callback_param);
 
-	if (cohc->nbr_active_done)
-		cohc->nbr_active_done--;
+	spin_lock_irqsave(&cohc->lock, flags);
 
+	/*
+	 * If another interrupt fired while the tasklet was scheduling,
+	 * we don't get called twice, so we have this number of active
+	 * counter that keep track of the number of IRQs expected to
+	 * be handled for this channel. If there happen to be more than
+	 * one IRQ to be ack:ed, we simply schedule this tasklet again.
+	 */
+	cohc->nbr_active_done--;
 	if (cohc->nbr_active_done) {
+		dev_dbg(COHC_2_DEV(cohc), "scheduling tasklet again, new IRQs "
+			"came in while we were scheduling this tasklet\n");
 		if (cohc_chan_conf(cohc)->priority_high)
 			tasklet_hi_schedule(&cohc->tasklet);
 		else
 			tasklet_schedule(&cohc->tasklet);
 	}
-	spin_unlock_irqrestore(&cohc->lock, flags);
 
-	if (callback)
-		callback(callback_param);
+	spin_unlock_irqrestore(&cohc->lock, flags);
 
 	return;
 
@@ -667,16 +672,17 @@ static void dma_tc_handle(struct coh901318_chan *cohc)
 	if (!cohc->allocated)
 		return;
 
-	BUG_ON(cohc->pending_irqs == 0);
+	spin_lock(&cohc->lock);
 
-	cohc->pending_irqs--;
 	cohc->nbr_active_done++;
 
-	if (cohc->pending_irqs == 0 && coh901318_queue_start(cohc) == NULL)
+	if (coh901318_queue_start(cohc) == NULL)
 		cohc->busy = 0;
 
 	BUG_ON(list_empty(&cohc->active));
 
+	spin_unlock(&cohc->lock);
+
 	if (cohc_chan_conf(cohc)->priority_high)
 		tasklet_hi_schedule(&cohc->tasklet);
 	else
@@ -870,6 +876,7 @@ coh901318_prep_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 	struct coh901318_chan *cohc = to_coh901318_chan(chan);
 	int lli_len;
 	u32 ctrl_last = cohc_chan_param(cohc)->ctrl_lli_last;
+	int ret;
 
 	spin_lock_irqsave(&cohc->lock, flg);
 
@@ -890,22 +897,19 @@ coh901318_prep_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
 	if (data == NULL)
 		goto err;
 
-	cohd = coh901318_desc_get(cohc);
-	cohd->sg = NULL;
-	cohd->sg_len = 0;
-	cohd->data = data;
-
-	cohd->pending_irqs =
-		coh901318_lli_fill_memcpy(
-				&cohc->base->pool, data, src, size, dest,
-				cohc_chan_param(cohc)->ctrl_lli_chained,
-				ctrl_last);
-	cohd->flags = flags;
+	ret = coh901318_lli_fill_memcpy(
+		&cohc->base->pool, data, src, size, dest,
+		cohc_chan_param(cohc)->ctrl_lli_chained,
+		ctrl_last);
+	if (ret)
+		goto err;
 
 	COH_DBG(coh901318_list_print(cohc, data));
 
-	dma_async_tx_descriptor_init(&cohd->desc, chan);
-
+	/* Pick a descriptor to handle this transfer */
+	cohd = coh901318_desc_get(cohc);
+	cohd->data = data;
+	cohd->flags = flags;
 	cohd->desc.tx_submit = coh901318_tx_submit;
 
 	spin_unlock_irqrestore(&cohc->lock, flg);
@@ -924,6 +928,7 @@ coh901318_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl,
 	struct coh901318_chan *cohc = to_coh901318_chan(chan);
 	struct coh901318_lli *data;
 	struct coh901318_desc *cohd;
+	const struct coh901318_params *params;
 	struct scatterlist *sg;
 	int len = 0;
 	int size;
@@ -931,7 +936,9 @@ coh901318_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl,
 	u32 ctrl_chained = cohc_chan_param(cohc)->ctrl_lli_chained;
 	u32 ctrl = cohc_chan_param(cohc)->ctrl_lli;
 	u32 ctrl_last = cohc_chan_param(cohc)->ctrl_lli_last;
+	u32 config;
 	unsigned long flg;
+	int ret;
 
 	if (!sgl)
 		goto out;
@@ -947,15 +954,14 @@ coh901318_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl,
 		/* Trigger interrupt after last lli */
 		ctrl_last |= COH901318_CX_CTRL_TC_IRQ_ENABLE;
 
-	cohd = coh901318_desc_get(cohc);
-	cohd->sg = NULL;
-	cohd->sg_len = 0;
-	cohd->dir = direction;
+	params = cohc_chan_param(cohc);
+	config = params->config;
 
 	if (direction == DMA_TO_DEVICE) {
 		u32 tx_flags = COH901318_CX_CTRL_PRDD_SOURCE |
 			COH901318_CX_CTRL_SRC_ADDR_INC_ENABLE;
 
+		config |= COH901318_CX_CFG_RM_MEMORY_TO_PRIMARY;
 		ctrl_chained |= tx_flags;
 		ctrl_last |= tx_flags;
 		ctrl |= tx_flags;
@@ -963,16 +969,14 @@ coh901318_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl,
 		u32 rx_flags = COH901318_CX_CTRL_PRDD_DEST |
 			COH901318_CX_CTRL_DST_ADDR_INC_ENABLE;
 
+		config |= COH901318_CX_CFG_RM_PRIMARY_TO_MEMORY;
 		ctrl_chained |= rx_flags;
 		ctrl_last |= rx_flags;
 		ctrl |= rx_flags;
 	} else
 		goto err_direction;
 
-	dma_async_tx_descriptor_init(&cohd->desc, chan);
-
-	cohd->desc.tx_submit = coh901318_tx_submit;
-
+	coh901318_set_conf(cohc, config);
 
 	/* The dma only supports transmitting packages up to
 	 * MAX_DMA_PACKET_SIZE. Calculate to total number of
@@ -994,32 +998,37 @@ coh901318_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl,
 		len += factor;
 	}
 
+	pr_debug("Allocate %d lli:s for this transfer\n", len);
 	data = coh901318_lli_alloc(&cohc->base->pool, len);
 
 	if (data == NULL)
 		goto err_dma_alloc;
 
 	/* initiate allocated data list */
-	cohd->pending_irqs =
-		coh901318_lli_fill_sg(&cohc->base->pool, data, sgl, sg_len,
-				      cohc_dev_addr(cohc),
-				      ctrl_chained,
-				      ctrl,
-				      ctrl_last,
-				      direction, COH901318_CX_CTRL_TC_IRQ_ENABLE);
-	cohd->data = data;
-
-	cohd->flags = flags;
+	ret = coh901318_lli_fill_sg(&cohc->base->pool, data, sgl, sg_len,
+				    cohc_dev_addr(cohc),
+				    ctrl_chained,
+				    ctrl,
+				    ctrl_last,
+				    direction, COH901318_CX_CTRL_TC_IRQ_ENABLE);
+	if (ret)
+		goto err_lli_fill;
 
 	COH_DBG(coh901318_list_print(cohc, data));
 
+	/* Pick a descriptor to handle this transfer */
+	cohd = coh901318_desc_get(cohc);
+	cohd->dir = direction;
+	cohd->flags = flags;
+	cohd->desc.tx_submit = coh901318_tx_submit;
+	cohd->data = data;
+
 	spin_unlock_irqrestore(&cohc->lock, flg);
 
 	return &cohd->desc;
+ err_lli_fill:
  err_dma_alloc:
  err_direction:
-	coh901318_desc_remove(cohd);
-	coh901318_desc_free(cohc, cohd);
 	spin_unlock_irqrestore(&cohc->lock, flg);
  out:
 	return NULL;
@@ -1092,9 +1101,8 @@ coh901318_terminate_all(struct dma_chan *chan)
 		/* release the lli allocation*/
 		coh901318_lli_free(&cohc->base->pool, &cohd->data);
 
-		coh901318_desc_remove(cohd);
-
 		/* return desc to free-list */
+		coh901318_desc_remove(cohd);
 		coh901318_desc_free(cohc, cohd);
 	}
 
@@ -1102,16 +1110,14 @@ coh901318_terminate_all(struct dma_chan *chan)
 		/* release the lli allocation*/
 		coh901318_lli_free(&cohc->base->pool, &cohd->data);
 
-		coh901318_desc_remove(cohd);
-
 		/* return desc to free-list */
+		coh901318_desc_remove(cohd);
 		coh901318_desc_free(cohc, cohd);
 	}
 
 
 	cohc->nbr_active_done = 0;
 	cohc->busy = 0;
-	cohc->pending_irqs = 0;
 
 	spin_unlock_irqrestore(&cohc->lock, flags);
 }
@@ -1138,7 +1144,6 @@ void coh901318_base_init(struct dma_device *dma, const int *pick_chans,
 
 			spin_lock_init(&cohc->lock);
 
-			cohc->pending_irqs = 0;
 			cohc->nbr_active_done = 0;
 			cohc->busy = 0;
 			INIT_LIST_HEAD(&cohc->free);
@@ -1254,12 +1259,17 @@ static int __init coh901318_probe(struct platform_device *pdev)
 	base->dma_memcpy.device_issue_pending = coh901318_issue_pending;
 	base->dma_memcpy.device_terminate_all = coh901318_terminate_all;
 	base->dma_memcpy.dev = &pdev->dev;
+	/*
+	 * This controller can only access address at even 32bit boundaries,
+	 * i.e. 2^2
+	 */
+	base->dma_memcpy.copy_align = 2;
 	err = dma_async_device_register(&base->dma_memcpy);
 
 	if (err)
 		goto err_register_memcpy;
 
-	dev_dbg(&pdev->dev, "Initialized COH901318 DMA on virtual base 0x%08x\n",
+	dev_info(&pdev->dev, "Initialized COH901318 DMA on virtual base 0x%08x\n",
 		(u32) base->virtbase);
 
 	return err;

drivers/dma/coh901318_lli.c

@@ -74,6 +74,8 @@ coh901318_lli_alloc(struct coh901318_pool *pool, unsigned int len)
 
 	lli = head;
 	lli->phy_this = phy;
+	lli->link_addr = 0x00000000;
+	lli->virt_link_addr = 0x00000000U;
 
 	for (i = 1; i < len; i++) {
 		lli_prev = lli;
@@ -85,13 +87,13 @@ coh901318_lli_alloc(struct coh901318_pool *pool, unsigned int len)
 
 		DEBUGFS_POOL_COUNTER_ADD(pool, 1);
 		lli->phy_this = phy;
+		lli->link_addr = 0x00000000;
+		lli->virt_link_addr = 0x00000000U;
 
 		lli_prev->link_addr = phy;
 		lli_prev->virt_link_addr = lli;
 	}
 
-	lli->link_addr = 0x00000000U;
-
 	spin_unlock(&pool->lock);
 
 	return head;
@@ -166,8 +168,7 @@ coh901318_lli_fill_memcpy(struct coh901318_pool *pool,
 	lli->src_addr = src;
 	lli->dst_addr = dst;
 
-	/* One irq per single transfer */
-	return 1;
+	return 0;
 }
 
 int
@@ -223,8 +224,7 @@ coh901318_lli_fill_single(struct coh901318_pool *pool,
 	lli->src_addr = src;
 	lli->dst_addr = dst;
 
-	/* One irq per single transfer */
-	return 1;
+	return 0;
 }
 
 int
@@ -240,7 +240,6 @@ coh901318_lli_fill_sg(struct coh901318_pool *pool,
 	u32 ctrl_sg;
 	dma_addr_t src = 0;
 	dma_addr_t dst = 0;
-	int nbr_of_irq = 0;
 	u32 bytes_to_transfer;
 	u32 elem_size;
 
@@ -269,15 +268,12 @@ coh901318_lli_fill_sg(struct coh901318_pool *pool,
 			ctrl_sg = ctrl ? ctrl : ctrl_last;
 
 
-		if ((ctrl_sg & ctrl_irq_mask))
-			nbr_of_irq++;
-
 		if (dir == DMA_TO_DEVICE)
 			/* increment source address */
-			src = sg_dma_address(sg);
+			src = sg_phys(sg);
 		else
 			/* increment destination address */
-			dst =  sg_dma_address(sg);
+			dst =  sg_phys(sg);
 
 		bytes_to_transfer = sg_dma_len(sg);
 
@@ -310,8 +306,7 @@ coh901318_lli_fill_sg(struct coh901318_pool *pool,
 	}
 	spin_unlock(&pool->lock);
 
-	/* There can be many IRQs per sg transfer */
-	return nbr_of_irq;
+	return 0;
  err:
 	spin_unlock(&pool->lock);
 	return -EINVAL;

drivers/dma/dmatest.c

@@ -237,7 +237,7 @@ static int dmatest_func(void *data)
 	dma_cookie_t		cookie;
 	enum dma_status		status;
 	enum dma_ctrl_flags 	flags;
-	u8			pq_coefs[pq_sources];
+	u8			pq_coefs[pq_sources + 1];
 	int			ret;
 	int			src_cnt;
 	int			dst_cnt;
@@ -257,7 +257,7 @@ static int dmatest_func(void *data)
 	} else if (thread->type == DMA_PQ) {
 		src_cnt = pq_sources | 1; /* force odd to ensure dst = src */
 		dst_cnt = 2;
-		for (i = 0; i < pq_sources; i++)
+		for (i = 0; i < src_cnt; i++)
 			pq_coefs[i] = 1;
 	} else
 		goto err_srcs;
@@ -347,7 +347,7 @@ static int dmatest_func(void *data)
 		else if (thread->type == DMA_XOR)
 			tx = dev->device_prep_dma_xor(chan,
 						      dma_dsts[0] + dst_off,
-						      dma_srcs, xor_sources,
+						      dma_srcs, src_cnt,
 						      len, flags);
 		else if (thread->type == DMA_PQ) {
 			dma_addr_t dma_pq[dst_cnt];
@@ -355,7 +355,7 @@ static int dmatest_func(void *data)
 			for (i = 0; i < dst_cnt; i++)
 				dma_pq[i] = dma_dsts[i] + dst_off;
 			tx = dev->device_prep_dma_pq(chan, dma_pq, dma_srcs,
-						     pq_sources, pq_coefs,
+						     src_cnt, pq_coefs,
 						     len, flags);
 		}
 

drivers/dma/fsldma.h

@@ -92,11 +92,9 @@ struct fsl_desc_sw {
 	struct list_head node;
 	struct list_head tx_list;
 	struct dma_async_tx_descriptor async_tx;
-	struct list_head *ld;
-	void *priv;
 } __attribute__((aligned(32)));
 
-struct fsl_dma_chan_regs {
+struct fsldma_chan_regs {
 	u32 mr;	/* 0x00 - Mode Register */
 	u32 sr;	/* 0x04 - Status Register */
 	u64 cdar;	/* 0x08 - Current descriptor address register */
@@ -106,20 +104,19 @@ struct fsl_dma_chan_regs {
 	u64 ndar;	/* 0x24 - Next Descriptor Address Register */
 };
 
-struct fsl_dma_chan;
+struct fsldma_chan;
 #define FSL_DMA_MAX_CHANS_PER_DEVICE 4
 
-struct fsl_dma_device {
-	void __iomem *reg_base;	/* DGSR register base */
-	struct resource reg;	/* Resource for register */
+struct fsldma_device {
+	void __iomem *regs;	/* DGSR register base */
 	struct device *dev;
 	struct dma_device common;
-	struct fsl_dma_chan *chan[FSL_DMA_MAX_CHANS_PER_DEVICE];
+	struct fsldma_chan *chan[FSL_DMA_MAX_CHANS_PER_DEVICE];
 	u32 feature;		/* The same as DMA channels */
 	int irq;		/* Channel IRQ */
 };
 
-/* Define macros for fsl_dma_chan->feature property */
+/* Define macros for fsldma_chan->feature property */
 #define FSL_DMA_LITTLE_ENDIAN	0x00000000
 #define FSL_DMA_BIG_ENDIAN	0x00000001
 
@@ -130,28 +127,28 @@ struct fsl_dma_device {
 #define FSL_DMA_CHAN_PAUSE_EXT	0x00001000
 #define FSL_DMA_CHAN_START_EXT	0x00002000
 
-struct fsl_dma_chan {
-	struct fsl_dma_chan_regs __iomem *reg_base;
+struct fsldma_chan {
+	struct fsldma_chan_regs __iomem *regs;
 	dma_cookie_t completed_cookie;	/* The maximum cookie completed */
 	spinlock_t desc_lock;		/* Descriptor operation lock */
-	struct list_head ld_queue;	/* Link descriptors queue */
+	struct list_head ld_pending;	/* Link descriptors queue */
+	struct list_head ld_running;	/* Link descriptors queue */
 	struct dma_chan common;		/* DMA common channel */
 	struct dma_pool *desc_pool;	/* Descriptors pool */
 	struct device *dev;		/* Channel device */
-	struct resource reg;		/* Resource for register */
 	int irq;			/* Channel IRQ */
 	int id;				/* Raw id of this channel */
 	struct tasklet_struct tasklet;
 	u32 feature;
 
-	void (*toggle_ext_pause)(struct fsl_dma_chan *fsl_chan, int enable);
-	void (*toggle_ext_start)(struct fsl_dma_chan *fsl_chan, int enable);
-	void (*set_src_loop_size)(struct fsl_dma_chan *fsl_chan, int size);
-	void (*set_dest_loop_size)(struct fsl_dma_chan *fsl_chan, int size);
-	void (*set_request_count)(struct fsl_dma_chan *fsl_chan, int size);
+	void (*toggle_ext_pause)(struct fsldma_chan *fsl_chan, int enable);
+	void (*toggle_ext_start)(struct fsldma_chan *fsl_chan, int enable);
+	void (*set_src_loop_size)(struct fsldma_chan *fsl_chan, int size);
+	void (*set_dst_loop_size)(struct fsldma_chan *fsl_chan, int size);
+	void (*set_request_count)(struct fsldma_chan *fsl_chan, int size);
 };
 
-#define to_fsl_chan(chan) container_of(chan, struct fsl_dma_chan, common)
+#define to_fsl_chan(chan) container_of(chan, struct fsldma_chan, common)
 #define to_fsl_desc(lh) container_of(lh, struct fsl_desc_sw, node)
 #define tx_to_fsl_desc(tx) container_of(tx, struct fsl_desc_sw, async_tx)
 

drivers/dma/ioat/dma.c

@@ -94,16 +94,12 @@ static irqreturn_t ioat_dma_do_interrupt_msix(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
-static void ioat1_cleanup_tasklet(unsigned long data);
-
 /* common channel initialization */
-void ioat_init_channel(struct ioatdma_device *device,
-		       struct ioat_chan_common *chan, int idx,
-		       void (*timer_fn)(unsigned long),
-		       void (*tasklet)(unsigned long),
-		       unsigned long ioat)
+void ioat_init_channel(struct ioatdma_device *device, struct ioat_chan_common *chan, int idx)
 {
 	struct dma_device *dma = &device->common;
+	struct dma_chan *c = &chan->common;
+	unsigned long data = (unsigned long) c;
 
 	chan->device = device;
 	chan->reg_base = device->reg_base + (0x80 * (idx + 1));
@@ -112,14 +108,12 @@ void ioat_init_channel(struct ioatdma_device *device,
 	list_add_tail(&chan->common.device_node, &dma->channels);
 	device->idx[idx] = chan;
 	init_timer(&chan->timer);
-	chan->timer.function = timer_fn;
-	chan->timer.data = ioat;
-	tasklet_init(&chan->cleanup_task, tasklet, ioat);
+	chan->timer.function = device->timer_fn;
+	chan->timer.data = data;
+	tasklet_init(&chan->cleanup_task, device->cleanup_fn, data);
 	tasklet_disable(&chan->cleanup_task);
 }
 
-static void ioat1_timer_event(unsigned long data);
-
 /**
  * ioat1_dma_enumerate_channels - find and initialize the device's channels
  * @device: the device to be enumerated
@@ -155,10 +149,7 @@ static int ioat1_enumerate_channels(struct ioatdma_device *device)
 		if (!ioat)
 			break;
 
-		ioat_init_channel(device, &ioat->base, i,
-				  ioat1_timer_event,
-				  ioat1_cleanup_tasklet,
-				  (unsigned long) ioat);
+		ioat_init_channel(device, &ioat->base, i);
 		ioat->xfercap = xfercap;
 		spin_lock_init(&ioat->desc_lock);
 		INIT_LIST_HEAD(&ioat->free_desc);
@@ -532,12 +523,12 @@ ioat1_dma_prep_memcpy(struct dma_chan *c, dma_addr_t dma_dest,
 	return &desc->txd;
 }
 
-static void ioat1_cleanup_tasklet(unsigned long data)
+static void ioat1_cleanup_event(unsigned long data)
 {
-	struct ioat_dma_chan *chan = (void *)data;
+	struct ioat_dma_chan *ioat = to_ioat_chan((void *) data);
 
-	ioat1_cleanup(chan);
-	writew(IOAT_CHANCTRL_RUN, chan->base.reg_base + IOAT_CHANCTRL_OFFSET);
+	ioat1_cleanup(ioat);
+	writew(IOAT_CHANCTRL_RUN, ioat->base.reg_base + IOAT_CHANCTRL_OFFSET);
 }
 
 void ioat_dma_unmap(struct ioat_chan_common *chan, enum dma_ctrl_flags flags,
@@ -687,7 +678,7 @@ static void ioat1_cleanup(struct ioat_dma_chan *ioat)
 
 static void ioat1_timer_event(unsigned long data)
 {
-	struct ioat_dma_chan *ioat = (void *) data;
+	struct ioat_dma_chan *ioat = to_ioat_chan((void *) data);
 	struct ioat_chan_common *chan = &ioat->base;
 
 	dev_dbg(to_dev(chan), "%s: state: %lx\n", __func__, chan->state);
@@ -734,16 +725,17 @@ static void ioat1_timer_event(unsigned long data)
 	spin_unlock_bh(&chan->cleanup_lock);
 }
 
-static enum dma_status
-ioat1_dma_is_complete(struct dma_chan *c, dma_cookie_t cookie,
+enum dma_status
+ioat_is_dma_complete(struct dma_chan *c, dma_cookie_t cookie,
 		      dma_cookie_t *done, dma_cookie_t *used)
 {
-	struct ioat_dma_chan *ioat = to_ioat_chan(c);
+	struct ioat_chan_common *chan = to_chan_common(c);
+	struct ioatdma_device *device = chan->device;
 
 	if (ioat_is_complete(c, cookie, done, used) == DMA_SUCCESS)
 		return DMA_SUCCESS;
 
-	ioat1_cleanup(ioat);
+	device->cleanup_fn((unsigned long) c);
 
 	return ioat_is_complete(c, cookie, done, used);
 }
@@ -1199,12 +1191,14 @@ int __devinit ioat1_dma_probe(struct ioatdma_device *device, int dca)
 	device->intr_quirk = ioat1_intr_quirk;
 	device->enumerate_channels = ioat1_enumerate_channels;
 	device->self_test = ioat_dma_self_test;
+	device->timer_fn = ioat1_timer_event;
+	device->cleanup_fn = ioat1_cleanup_event;
 	dma = &device->common;
 	dma->device_prep_dma_memcpy = ioat1_dma_prep_memcpy;
 	dma->device_issue_pending = ioat1_dma_memcpy_issue_pending;
 	dma->device_alloc_chan_resources = ioat1_dma_alloc_chan_resources;
 	dma->device_free_chan_resources = ioat1_dma_free_chan_resources;
-	dma->device_is_tx_complete = ioat1_dma_is_complete;
+	dma->device_is_tx_complete = ioat_is_dma_complete;
 
 	err = ioat_probe(device);
 	if (err)

drivers/dma/ioat/dma.h

@@ -61,7 +61,7 @@
  * @intr_quirk: interrupt setup quirk (for ioat_v1 devices)
  * @enumerate_channels: hw version specific channel enumeration
  * @reset_hw: hw version specific channel (re)initialization
- * @cleanup_tasklet: select between the v2 and v3 cleanup routines
+ * @cleanup_fn: select between the v2 and v3 cleanup routines
  * @timer_fn: select between the v2 and v3 timer watchdog routines
  * @self_test: hardware version specific self test for each supported op type
  *
@@ -80,7 +80,7 @@ struct ioatdma_device {
 	void (*intr_quirk)(struct ioatdma_device *device);
 	int (*enumerate_channels)(struct ioatdma_device *device);
 	int (*reset_hw)(struct ioat_chan_common *chan);
-	void (*cleanup_tasklet)(unsigned long data);
+	void (*cleanup_fn)(unsigned long data);
 	void (*timer_fn)(unsigned long data);
 	int (*self_test)(struct ioatdma_device *device);
 };
@@ -337,10 +337,9 @@ struct dca_provider * __devinit ioat_dca_init(struct pci_dev *pdev,
 					      void __iomem *iobase);
 unsigned long ioat_get_current_completion(struct ioat_chan_common *chan);
 void ioat_init_channel(struct ioatdma_device *device,
-		       struct ioat_chan_common *chan, int idx,
-		       void (*timer_fn)(unsigned long),
-		       void (*tasklet)(unsigned long),
-		       unsigned long ioat);
+		       struct ioat_chan_common *chan, int idx);
+enum dma_status ioat_is_dma_complete(struct dma_chan *c, dma_cookie_t cookie,
+				     dma_cookie_t *done, dma_cookie_t *used);
 void ioat_dma_unmap(struct ioat_chan_common *chan, enum dma_ctrl_flags flags,
 		    size_t len, struct ioat_dma_descriptor *hw);
 bool ioat_cleanup_preamble(struct ioat_chan_common *chan,

drivers/dma/ioat/dma_v2.c

@@ -51,48 +51,40 @@ MODULE_PARM_DESC(ioat_ring_max_alloc_order,
 
 void __ioat2_issue_pending(struct ioat2_dma_chan *ioat)
 {
-	void * __iomem reg_base = ioat->base.reg_base;
+	struct ioat_chan_common *chan = &ioat->base;
 
-	ioat->pending = 0;
 	ioat->dmacount += ioat2_ring_pending(ioat);
 	ioat->issued = ioat->head;
 	/* make descriptor updates globally visible before notifying channel */
 	wmb();
-	writew(ioat->dmacount, reg_base + IOAT_CHAN_DMACOUNT_OFFSET);
-	dev_dbg(to_dev(&ioat->base),
+	writew(ioat->dmacount, chan->reg_base + IOAT_CHAN_DMACOUNT_OFFSET);
+	dev_dbg(to_dev(chan),
 		"%s: head: %#x tail: %#x issued: %#x count: %#x\n",
 		__func__, ioat->head, ioat->tail, ioat->issued, ioat->dmacount);
 }
 
-void ioat2_issue_pending(struct dma_chan *chan)
+void ioat2_issue_pending(struct dma_chan *c)
 {
-	struct ioat2_dma_chan *ioat = to_ioat2_chan(chan);
+	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);
 
-	spin_lock_bh(&ioat->ring_lock);
-	if (ioat->pending == 1)
+	if (ioat2_ring_pending(ioat)) {
+		spin_lock_bh(&ioat->ring_lock);
 		__ioat2_issue_pending(ioat);
-	spin_unlock_bh(&ioat->ring_lock);
+		spin_unlock_bh(&ioat->ring_lock);
+	}
 }
 
 /**
  * ioat2_update_pending - log pending descriptors
  * @ioat: ioat2+ channel
  *
- * set pending to '1' unless pending is already set to '2', pending == 2
- * indicates that submission is temporarily blocked due to an in-flight
- * reset.  If we are already above the ioat_pending_level threshold then
- * just issue pending.
- *
- * called with ring_lock held
+ * Check if the number of unsubmitted descriptors has exceeded the
+ * watermark.  Called with ring_lock held
  */
 static void ioat2_update_pending(struct ioat2_dma_chan *ioat)
 {
-	if (unlikely(ioat->pending == 2))
-		return;
-	else if (ioat2_ring_pending(ioat) > ioat_pending_level)
+	if (ioat2_ring_pending(ioat) > ioat_pending_level)
 		__ioat2_issue_pending(ioat);
-	else
-		ioat->pending = 1;
 }
 
 static void __ioat2_start_null_desc(struct ioat2_dma_chan *ioat)
@@ -166,7 +158,7 @@ static void __cleanup(struct ioat2_dma_chan *ioat, unsigned long phys_complete)
 			seen_current = true;
 	}
 	ioat->tail += i;
-	BUG_ON(!seen_current); /* no active descs have written a completion? */
+	BUG_ON(active && !seen_current); /* no active descs have written a completion? */
 
 	chan->last_completion = phys_complete;
 	if (ioat->head == ioat->tail) {
@@ -207,9 +199,9 @@ static void ioat2_cleanup(struct ioat2_dma_chan *ioat)
 	spin_unlock_bh(&chan->cleanup_lock);
 }
 
-void ioat2_cleanup_tasklet(unsigned long data)
+void ioat2_cleanup_event(unsigned long data)
 {
-	struct ioat2_dma_chan *ioat = (void *) data;
+	struct ioat2_dma_chan *ioat = to_ioat2_chan((void *) data);
 
 	ioat2_cleanup(ioat);
 	writew(IOAT_CHANCTRL_RUN, ioat->base.reg_base + IOAT_CHANCTRL_OFFSET);
@@ -291,7 +283,7 @@ static void ioat2_restart_channel(struct ioat2_dma_chan *ioat)
 
 void ioat2_timer_event(unsigned long data)
 {
-	struct ioat2_dma_chan *ioat = (void *) data;
+	struct ioat2_dma_chan *ioat = to_ioat2_chan((void *) data);
 	struct ioat_chan_common *chan = &ioat->base;
 
 	spin_lock_bh(&chan->cleanup_lock);
@@ -397,10 +389,7 @@ int ioat2_enumerate_channels(struct ioatdma_device *device)
 		if (!ioat)
 			break;
 
-		ioat_init_channel(device, &ioat->base, i,
-				  device->timer_fn,
-				  device->cleanup_tasklet,
-				  (unsigned long) ioat);
+		ioat_init_channel(device, &ioat->base, i);
 		ioat->xfercap_log = xfercap_log;
 		spin_lock_init(&ioat->ring_lock);
 		if (device->reset_hw(&ioat->base)) {
@@ -546,7 +535,6 @@ int ioat2_alloc_chan_resources(struct dma_chan *c)
 	ioat->head = 0;
 	ioat->issued = 0;
 	ioat->tail = 0;
-	ioat->pending = 0;
 	ioat->alloc_order = order;
 	spin_unlock_bh(&ioat->ring_lock);
 
@@ -701,7 +689,7 @@ int ioat2_alloc_and_lock(u16 *idx, struct ioat2_dma_chan *ioat, int num_descs)
 
 			mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT);
 			spin_unlock_bh(&chan->cleanup_lock);
-			device->timer_fn((unsigned long) ioat);
+			device->timer_fn((unsigned long) &chan->common);
 		} else
 			spin_unlock_bh(&chan->cleanup_lock);
 		return -ENOMEM;
@@ -785,7 +773,7 @@ void ioat2_free_chan_resources(struct dma_chan *c)
 
 	tasklet_disable(&chan->cleanup_task);
 	del_timer_sync(&chan->timer);
-	device->cleanup_tasklet((unsigned long) ioat);
+	device->cleanup_fn((unsigned long) c);
 	device->reset_hw(chan);
 
 	spin_lock_bh(&ioat->ring_lock);
@@ -815,25 +803,9 @@ void ioat2_free_chan_resources(struct dma_chan *c)
 
 	chan->last_completion = 0;
 	chan->completion_dma = 0;
-	ioat->pending = 0;
 	ioat->dmacount = 0;
 }
 
-enum dma_status
-ioat2_is_complete(struct dma_chan *c, dma_cookie_t cookie,
-		     dma_cookie_t *done, dma_cookie_t *used)
-{
-	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);
-	struct ioatdma_device *device = ioat->base.device;
-
-	if (ioat_is_complete(c, cookie, done, used) == DMA_SUCCESS)
-		return DMA_SUCCESS;
-
-	device->cleanup_tasklet((unsigned long) ioat);
-
-	return ioat_is_complete(c, cookie, done, used);
-}
-
 static ssize_t ring_size_show(struct dma_chan *c, char *page)
 {
 	struct ioat2_dma_chan *ioat = to_ioat2_chan(c);
@@ -874,7 +846,7 @@ int __devinit ioat2_dma_probe(struct ioatdma_device *device, int dca)
 
 	device->enumerate_channels = ioat2_enumerate_channels;
 	device->reset_hw = ioat2_reset_hw;
-	device->cleanup_tasklet = ioat2_cleanup_tasklet;
+	device->cleanup_fn = ioat2_cleanup_event;
 	device->timer_fn = ioat2_timer_event;
 	device->self_test = ioat_dma_self_test;
 	dma = &device->common;
@@ -882,7 +854,7 @@ int __devinit ioat2_dma_probe(struct ioatdma_device *device, int dca)
 	dma->device_issue_pending = ioat2_issue_pending;
 	dma->device_alloc_chan_resources = ioat2_alloc_chan_resources;
 	dma->device_free_chan_resources = ioat2_free_chan_resources;
-	dma->device_is_tx_complete = ioat2_is_complete;
+	dma->device_is_tx_complete = ioat_is_dma_complete;
 
 	err = ioat_probe(device);
 	if (err)

drivers/dma/ioat/dma_v2.h

@@ -47,7 +47,6 @@ extern int ioat_ring_alloc_order;
  * @head: allocated index
  * @issued: hardware notification point
  * @tail: cleanup index
- * @pending: lock free indicator for issued != head
  * @dmacount: identical to 'head' except for occasionally resetting to zero
  * @alloc_order: log2 of the number of allocated descriptors
  * @ring: software ring buffer implementation of hardware ring
@@ -61,7 +60,6 @@ struct ioat2_dma_chan {
 	u16 tail;
 	u16 dmacount;
 	u16 alloc_order;
-	int pending;
 	struct ioat_ring_ent **ring;
 	spinlock_t ring_lock;
 };
@@ -178,12 +176,10 @@ ioat2_dma_prep_memcpy_lock(struct dma_chan *c, dma_addr_t dma_dest,
 void ioat2_issue_pending(struct dma_chan *chan);
 int ioat2_alloc_chan_resources(struct dma_chan *c);
 void ioat2_free_chan_resources(struct dma_chan *c);
-enum dma_status ioat2_is_complete(struct dma_chan *c, dma_cookie_t cookie,
-				  dma_cookie_t *done, dma_cookie_t *used);
 void __ioat2_restart_chan(struct ioat2_dma_chan *ioat);
 bool reshape_ring(struct ioat2_dma_chan *ioat, int order);
 void __ioat2_issue_pending(struct ioat2_dma_chan *ioat);
-void ioat2_cleanup_tasklet(unsigned long data);
+void ioat2_cleanup_event(unsigned long data);
 void ioat2_timer_event(unsigned long data);
 int ioat2_quiesce(struct ioat_chan_common *chan, unsigned long tmo);
 int ioat2_reset_sync(struct ioat_chan_common *chan, unsigned long tmo);

drivers/dma/ioat/dma_v3.c

@@ -293,17 +293,25 @@ static void __cleanup(struct ioat2_dma_chan *ioat, unsigned long phys_complete)
 		}
 	}
 	ioat->tail += i;
-	BUG_ON(!seen_current); /* no active descs have written a completion? */
+	BUG_ON(active && !seen_current); /* no active descs have written a completion? */
 	chan->last_completion = phys_complete;
-	if (ioat->head == ioat->tail) {
+
+	active = ioat2_ring_active(ioat);
+	if (active == 0) {
 		dev_dbg(to_dev(chan), "%s: cancel completion timeout\n",
 			__func__);
 		clear_bit(IOAT_COMPLETION_PENDING, &chan->state);
 		mod_timer(&chan->timer, jiffies + IDLE_TIMEOUT);
 	}
+	/* 5 microsecond delay per pending descriptor */
+	writew(min((5 * active), IOAT_INTRDELAY_MASK),
+	       chan->device->reg_base + IOAT_INTRDELAY_OFFSET);
 }
 
-static void ioat3_cleanup(struct ioat2_dma_chan *ioat)
+/* try to cleanup, but yield (via spin_trylock) to incoming submissions
+ * with the expectation that we will immediately poll again shortly
+ */
+static void ioat3_cleanup_poll(struct ioat2_dma_chan *ioat)
 {
 	struct ioat_chan_common *chan = &ioat->base;
 	unsigned long phys_complete;
@@ -329,29 +337,41 @@ static void ioat3_cleanup(struct ioat2_dma_chan *ioat)
 	spin_unlock_bh(&chan->cleanup_lock);
 }
 
-static void ioat3_cleanup_tasklet(unsigned long data)
+/* run cleanup now because we already delayed the interrupt via INTRDELAY */
+static void ioat3_cleanup_sync(struct ioat2_dma_chan *ioat)
+{
+	struct ioat_chan_common *chan = &ioat->base;
+	unsigned long phys_complete;
+
+	prefetch(chan->completion);
+
+	spin_lock_bh(&chan->cleanup_lock);
+	if (!ioat_cleanup_preamble(chan, &phys_complete)) {
+		spin_unlock_bh(&chan->cleanup_lock);
+		return;
+	}
+	spin_lock_bh(&ioat->ring_lock);
+
+	__cleanup(ioat, phys_complete);
+
+	spin_unlock_bh(&ioat->ring_lock);
+	spin_unlock_bh(&chan->cleanup_lock);
+}
+
+static void ioat3_cleanup_event(unsigned long data)
 {
-	struct ioat2_dma_chan *ioat = (void *) data;
+	struct ioat2_dma_chan *ioat = to_ioat2_chan((void *) data);
 
-	ioat3_cleanup(ioat);
-	writew(IOAT_CHANCTRL_RUN | IOAT3_CHANCTRL_COMPL_DCA_EN,
-	       ioat->base.reg_base + IOAT_CHANCTRL_OFFSET);
+	ioat3_cleanup_sync(ioat);
+	writew(IOAT_CHANCTRL_RUN, ioat->base.reg_base + IOAT_CHANCTRL_OFFSET);
 }
 
 static void ioat3_restart_channel(struct ioat2_dma_chan *ioat)
 {
 	struct ioat_chan_common *chan = &ioat->base;
 	unsigned long phys_complete;
-	u32 status;
-
-	status = ioat_chansts(chan);
-	if (is_ioat_active(status) || is_ioat_idle(status))
-		ioat_suspend(chan);
-	while (is_ioat_active(status) || is_ioat_idle(status)) {
-		status = ioat_chansts(chan);
-		cpu_relax();
-	}
 
+	ioat2_quiesce(chan, 0);
 	if (ioat_cleanup_preamble(chan, &phys_complete))
 		__cleanup(ioat, phys_complete);
 
@@ -360,7 +380,7 @@ static void ioat3_restart_channel(struct ioat2_dma_chan *ioat)
 
 static void ioat3_timer_event(unsigned long data)
 {
-	struct ioat2_dma_chan *ioat = (void *) data;
+	struct ioat2_dma_chan *ioat = to_ioat2_chan((void *) data);
 	struct ioat_chan_common *chan = &ioat->base;
 
 	spin_lock_bh(&chan->cleanup_lock);
@@ -426,7 +446,7 @@ ioat3_is_complete(struct dma_chan *c, dma_cookie_t cookie,
 	if (ioat_is_complete(c, cookie, done, used) == DMA_SUCCESS)
 		return DMA_SUCCESS;
 
-	ioat3_cleanup(ioat);
+	ioat3_cleanup_poll(ioat);
 
 	return ioat_is_complete(c, cookie, done, used);
 }
@@ -1239,11 +1259,11 @@ int __devinit ioat3_dma_probe(struct ioatdma_device *device, int dca)
 
 	if (is_raid_device) {
 		dma->device_is_tx_complete = ioat3_is_complete;
-		device->cleanup_tasklet = ioat3_cleanup_tasklet;
+		device->cleanup_fn = ioat3_cleanup_event;
 		device->timer_fn = ioat3_timer_event;
 	} else {
-		dma->device_is_tx_complete = ioat2_is_complete;
-		device->cleanup_tasklet = ioat2_cleanup_tasklet;
+		dma->device_is_tx_complete = ioat_is_dma_complete;
+		device->cleanup_fn = ioat2_cleanup_event;
 		device->timer_fn = ioat2_timer_event;
 	}
 

drivers/dma/ioat/registers.h

@@ -60,7 +60,7 @@
 #define IOAT_PERPORTOFFSET_OFFSET		0x0A	/* 16-bit */
 
 #define IOAT_INTRDELAY_OFFSET			0x0C	/* 16-bit */
-#define IOAT_INTRDELAY_INT_DELAY_MASK		0x3FFF	/* Interrupt Delay Time */
+#define IOAT_INTRDELAY_MASK			0x3FFF	/* Interrupt Delay Time */
 #define IOAT_INTRDELAY_COALESE_SUPPORT		0x8000	/* Interrupt Coalescing Supported */
 
 #define IOAT_DEVICE_STATUS_OFFSET		0x0E	/* 16-bit */

drivers/dma/ipu/ipu_idmac.c

@@ -348,6 +348,7 @@ static void ipu_ch_param_set_size(union chan_param_mem *params,
 		break;
 	case IPU_PIX_FMT_BGRA32:
 	case IPU_PIX_FMT_BGR32:
+	case IPU_PIX_FMT_ABGR32:
 		params->ip.bpp	= 0;
 		params->ip.pfs	= 4;
 		params->ip.npb	= 7;
@@ -376,20 +377,6 @@ static void ipu_ch_param_set_size(union chan_param_mem *params,
 		params->ip.wid2	= 7;		/* Blue bit width - 1 */
 		params->ip.wid3	= 7;		/* Alpha bit width - 1 */
 		break;
-	case IPU_PIX_FMT_ABGR32:
-		params->ip.bpp	= 0;
-		params->ip.pfs	= 4;
-		params->ip.npb	= 7;
-		params->ip.sat	= 2;		/* SAT = 32-bit access */
-		params->ip.ofs0	= 8;		/* Red bit offset */
-		params->ip.ofs1	= 16;		/* Green bit offset */
-		params->ip.ofs2	= 24;		/* Blue bit offset */
-		params->ip.ofs3	= 0;		/* Alpha bit offset */
-		params->ip.wid0	= 7;		/* Red bit width - 1 */
-		params->ip.wid1	= 7;		/* Green bit width - 1 */
-		params->ip.wid2	= 7;		/* Blue bit width - 1 */
-		params->ip.wid3	= 7;		/* Alpha bit width - 1 */
-		break;
 	case IPU_PIX_FMT_UYVY:
 		params->ip.bpp	= 2;
 		params->ip.pfs	= 6;

drivers/dma/ppc4xx/adma.c

@@ -4940,7 +4940,7 @@ out_free:
 	return ret;
 }
 
-static struct of_device_id __devinitdata ppc440spe_adma_of_match[] = {
+static const struct of_device_id ppc440spe_adma_of_match[] __devinitconst = {
 	{ .compatible	= "ibm,dma-440spe", },
 	{ .compatible	= "amcc,xor-accelerator", },
 	{},

include/linux/dmaengine.h

@@ -31,6 +31,8 @@
  * if dma_cookie_t is >0 it's a DMA request cookie, <0 it's an error code
  */
 typedef s32 dma_cookie_t;
+#define DMA_MIN_COOKIE	1
+#define DMA_MAX_COOKIE	INT_MAX
 
 #define dma_submit_error(cookie) ((cookie) < 0 ? 1 : 0)