Merge branch 'topic/cppi' into for-linus

Documentation/devicetree/bindings/usb/da8xx-usb.txt

@@ -18,10 +18,26 @@ Required properties:
 
  - phy-names: Should be "usb-phy"
 
+ - dmas: specifies the dma channels
+
+ - dma-names: specifies the names of the channels. Use "rxN" for receive
+   and "txN" for transmit endpoints. N specifies the endpoint number.
+
 Optional properties:
 ~~~~~~~~~~~~~~~~~~~~
  - vbus-supply: Phandle to a regulator providing the USB bus power.
 
+DMA
+~~~
+- compatible: ti,da830-cppi41
+- reg: offset and length of the following register spaces: CPPI DMA Controller,
+  CPPI DMA Scheduler, Queue Manager
+- reg-names: "controller", "scheduler", "queuemgr"
+- #dma-cells: should be set to 2. The first number represents the
+  channel number (0 … 3 for endpoints 1 … 4).
+  The second number is 0 for RX and 1 for TX transfers.
+- #dma-channels: should be set to 4 representing the 4 endpoints.
+
 Example:
 	usb_phy: usb-phy {
 		compatible = "ti,da830-usb-phy";
@@ -30,7 +46,10 @@ Example:
 	};
 	usb0: usb@200000 {
 		compatible = "ti,da830-musb";
-		reg =   <0x00200000 0x10000>;
+		reg = <0x00200000 0x1000>;
+		ranges;
+		#address-cells = <1>;
+		#size-cells = <1>;
 		interrupts = <58>;
 		interrupt-names = "mc";
 
@@ -39,5 +58,25 @@ Example:
 		phys = <&usb_phy 0>;
 		phy-names = "usb-phy";
 
+		dmas = <&cppi41dma 0 0 &cppi41dma 1 0
+			&cppi41dma 2 0 &cppi41dma 3 0
+			&cppi41dma 0 1 &cppi41dma 1 1
+			&cppi41dma 2 1 &cppi41dma 3 1>;
+		dma-names =
+			"rx1", "rx2", "rx3", "rx4",
+			"tx1", "tx2", "tx3", "tx4";
+
 		status = "okay";
+
+		cppi41dma: dma-controller@201000 {
+			compatible = "ti,da830-cppi41";
+			reg =  <0x201000 0x1000
+				0x202000 0x1000
+				0x204000 0x4000>;
+			reg-names = "controller", "scheduler", "queuemgr";
+			interrupts = <58>;
+			#dma-cells = <2>;
+			#dma-channels = <4>;
+		};
+
 	};

drivers/dma/Kconfig

@@ -514,12 +514,12 @@ config TIMB_DMA
 	  Enable support for the Timberdale FPGA DMA engine.
 
 config TI_CPPI41
-	tristate "AM33xx CPPI41 DMA support"
-	depends on ARCH_OMAP
+	tristate "CPPI 4.1 DMA support"
+	depends on (ARCH_OMAP || ARCH_DAVINCI_DA8XX)
 	select DMA_ENGINE
 	help
 	  The Communications Port Programming Interface (CPPI) 4.1 DMA engine
-	  is currently used by the USB driver on AM335x platforms.
+	  is currently used by the USB driver on AM335x and DA8xx platforms.
 
 config TI_DMA_CROSSBAR
 	bool

drivers/dma/cppi41.c

@@ -68,7 +68,6 @@
 #define QMGR_MEMCTRL_IDX_SH	16
 #define QMGR_MEMCTRL_DESC_SH	8
 
-#define QMGR_NUM_PEND	5
 #define QMGR_PEND(x)	(0x90 + (x) * 4)
 
 #define QMGR_PENDING_SLOT_Q(x)	(x / 32)
@@ -131,7 +130,6 @@ struct cppi41_dd {
 	u32 first_td_desc;
 	struct cppi41_channel *chan_busy[ALLOC_DECS_NUM];
 
-	void __iomem *usbss_mem;
 	void __iomem *ctrl_mem;
 	void __iomem *sched_mem;
 	void __iomem *qmgr_mem;
@@ -139,6 +137,10 @@ struct cppi41_dd {
 	const struct chan_queues *queues_rx;
 	const struct chan_queues *queues_tx;
 	struct chan_queues td_queue;
+	u16 first_completion_queue;
+	u16 qmgr_num_pend;
+	u32 n_chans;
+	u8 platform;
 
 	struct list_head pending;	/* Pending queued transfers */
 	spinlock_t lock;		/* Lock for pending list */
@@ -149,8 +151,7 @@ struct cppi41_dd {
 	bool is_suspended;
 };
 
-#define FIST_COMPLETION_QUEUE	93
-static struct chan_queues usb_queues_tx[] = {
+static struct chan_queues am335x_usb_queues_tx[] = {
 	/* USB0 ENDP 1 */
 	[ 0] = { .submit = 32, .complete =  93},
 	[ 1] = { .submit = 34, .complete =  94},
@@ -186,7 +187,7 @@ static struct chan_queues usb_queues_tx[] = {
 	[29] = { .submit = 90, .complete = 139},
 };
 
-static const struct chan_queues usb_queues_rx[] = {
+static const struct chan_queues am335x_usb_queues_rx[] = {
 	/* USB0 ENDP 1 */
 	[ 0] = { .submit =  1, .complete = 109},
 	[ 1] = { .submit =  2, .complete = 110},
@@ -222,11 +223,26 @@ static const struct chan_queues usb_queues_rx[] = {
 	[29] = { .submit = 30, .complete = 155},
 };
 
+static const struct chan_queues da8xx_usb_queues_tx[] = {
+	[0] = { .submit =  16, .complete = 24},
+	[1] = { .submit =  18, .complete = 24},
+	[2] = { .submit =  20, .complete = 24},
+	[3] = { .submit =  22, .complete = 24},
+};
+
+static const struct chan_queues da8xx_usb_queues_rx[] = {
+	[0] = { .submit =  1, .complete = 26},
+	[1] = { .submit =  3, .complete = 26},
+	[2] = { .submit =  5, .complete = 26},
+	[3] = { .submit =  7, .complete = 26},
+};
+
 struct cppi_glue_infos {
-	irqreturn_t (*isr)(int irq, void *data);
 	const struct chan_queues *queues_rx;
 	const struct chan_queues *queues_tx;
 	struct chan_queues td_queue;
+	u16 first_completion_queue;
+	u16 qmgr_num_pend;
 };
 
 static struct cppi41_channel *to_cpp41_chan(struct dma_chan *c)
@@ -285,19 +301,21 @@ static u32 cppi41_pop_desc(struct cppi41_dd *cdd, unsigned queue_num)
 static irqreturn_t cppi41_irq(int irq, void *data)
 {
 	struct cppi41_dd *cdd = data;
+	u16 first_completion_queue = cdd->first_completion_queue;
+	u16 qmgr_num_pend = cdd->qmgr_num_pend;
 	struct cppi41_channel *c;
 	int i;
 
-	for (i = QMGR_PENDING_SLOT_Q(FIST_COMPLETION_QUEUE); i < QMGR_NUM_PEND;
+	for (i = QMGR_PENDING_SLOT_Q(first_completion_queue); i < qmgr_num_pend;
 			i++) {
 		u32 val;
 		u32 q_num;
 
 		val = cppi_readl(cdd->qmgr_mem + QMGR_PEND(i));
-		if (i == QMGR_PENDING_SLOT_Q(FIST_COMPLETION_QUEUE) && val) {
+		if (i == QMGR_PENDING_SLOT_Q(first_completion_queue) && val) {
 			u32 mask;
 			/* set corresponding bit for completetion Q 93 */
-			mask = 1 << QMGR_PENDING_BIT_Q(FIST_COMPLETION_QUEUE);
+			mask = 1 << QMGR_PENDING_BIT_Q(first_completion_queue);
 			/* not set all bits for queues less than Q 93 */
 			mask--;
 			/* now invert and keep only Q 93+ set */
@@ -402,11 +420,9 @@ static enum dma_status cppi41_dma_tx_status(struct dma_chan *chan,
 	struct cppi41_channel *c = to_cpp41_chan(chan);
 	enum dma_status ret;
 
-	/* lock */
 	ret = dma_cookie_status(chan, cookie, txstate);
-	if (txstate && ret == DMA_COMPLETE)
-		txstate->residue = c->residue;
-	/* unlock */
+
+	dma_set_residue(txstate, c->residue);
 
 	return ret;
 }
@@ -630,7 +646,7 @@ static int cppi41_tear_down_chan(struct cppi41_channel *c)
 		if (!c->is_tx) {
 			reg |= GCR_STARV_RETRY;
 			reg |= GCR_DESC_TYPE_HOST;
-			reg |= c->q_comp_num;
+			reg |= cdd->td_queue.complete;
 		}
 		reg |= GCR_TEARDOWN;
 		cppi_writel(reg, c->gcr_reg);
@@ -641,7 +657,7 @@ static int cppi41_tear_down_chan(struct cppi41_channel *c)
 	if (!c->td_seen || !c->td_desc_seen) {
 
 		desc_phys = cppi41_pop_desc(cdd, cdd->td_queue.complete);
-		if (!desc_phys)
+		if (!desc_phys && c->is_tx)
 			desc_phys = cppi41_pop_desc(cdd, c->q_comp_num);
 
 		if (desc_phys == c->desc_phys) {
@@ -723,39 +739,24 @@ static int cppi41_stop_chan(struct dma_chan *chan)
 	return 0;
 }
 
-static void cleanup_chans(struct cppi41_dd *cdd)
-{
-	while (!list_empty(&cdd->ddev.channels)) {
-		struct cppi41_channel *cchan;
-
-		cchan = list_first_entry(&cdd->ddev.channels,
-				struct cppi41_channel, chan.device_node);
-		list_del(&cchan->chan.device_node);
-		kfree(cchan);
-	}
-}
-
 static int cppi41_add_chans(struct device *dev, struct cppi41_dd *cdd)
 {
-	struct cppi41_channel *cchan;
+	struct cppi41_channel *cchan, *chans;
 	int i;
-	int ret;
-	u32 n_chans;
+	u32 n_chans = cdd->n_chans;
 
-	ret = of_property_read_u32(dev->of_node, "#dma-channels",
-			&n_chans);
-	if (ret)
-		return ret;
 	/*
 	 * The channels can only be used as TX or as RX. So we add twice
 	 * that much dma channels because USB can only do RX or TX.
 	 */
 	n_chans *= 2;
 
+	chans = devm_kcalloc(dev, n_chans, sizeof(*chans), GFP_KERNEL);
+	if (!chans)
+		return -ENOMEM;
+
 	for (i = 0; i < n_chans; i++) {
-		cchan = kzalloc(sizeof(*cchan), GFP_KERNEL);
-		if (!cchan)
-			goto err;
+		cchan = &chans[i];
 
 		cchan->cdd = cdd;
 		if (i & 1) {
@@ -775,9 +776,6 @@ static int cppi41_add_chans(struct device *dev, struct cppi41_dd *cdd)
 	cdd->first_td_desc = n_chans;
 
 	return 0;
-err:
-	cleanup_chans(cdd);
-	return -ENOMEM;
 }
 
 static void purge_descs(struct device *dev, struct cppi41_dd *cdd)
@@ -859,7 +857,7 @@ static void init_sched(struct cppi41_dd *cdd)
 
 	word = 0;
 	cppi_writel(0, cdd->sched_mem + DMA_SCHED_CTRL);
-	for (ch = 0; ch < 15 * 2; ch += 2) {
+	for (ch = 0; ch < cdd->n_chans; ch += 2) {
 
 		reg = SCHED_ENTRY0_CHAN(ch);
 		reg |= SCHED_ENTRY1_CHAN(ch) | SCHED_ENTRY1_IS_RX;
@@ -869,7 +867,7 @@ static void init_sched(struct cppi41_dd *cdd)
 		cppi_writel(reg, cdd->sched_mem + DMA_SCHED_WORD(word));
 		word++;
 	}
-	reg = 15 * 2 * 2 - 1;
+	reg = cdd->n_chans * 2 - 1;
 	reg |= DMA_SCHED_CTRL_EN;
 	cppi_writel(reg, cdd->sched_mem + DMA_SCHED_CTRL);
 }
@@ -885,7 +883,7 @@ static int init_cppi41(struct device *dev, struct cppi41_dd *cdd)
 		return -ENOMEM;
 
 	cppi_writel(cdd->scratch_phys, cdd->qmgr_mem + QMGR_LRAM0_BASE);
-	cppi_writel(QMGR_SCRATCH_SIZE, cdd->qmgr_mem + QMGR_LRAM_SIZE);
+	cppi_writel(TOTAL_DESCS_NUM, cdd->qmgr_mem + QMGR_LRAM_SIZE);
 	cppi_writel(0, cdd->qmgr_mem + QMGR_LRAM1_BASE);
 
 	ret = init_descs(dev, cdd);
@@ -894,6 +892,7 @@ static int init_cppi41(struct device *dev, struct cppi41_dd *cdd)
 
 	cppi_writel(cdd->td_queue.submit, cdd->ctrl_mem + DMA_TDFDQ);
 	init_sched(cdd);
+
 	return 0;
 err_td:
 	deinit_cppi41(dev, cdd);
@@ -933,8 +932,9 @@ static bool cpp41_dma_filter_fn(struct dma_chan *chan, void *param)
 	else
 		queues = cdd->queues_rx;
 
-	BUILD_BUG_ON(ARRAY_SIZE(usb_queues_rx) != ARRAY_SIZE(usb_queues_tx));
-	if (WARN_ON(cchan->port_num > ARRAY_SIZE(usb_queues_rx)))
+	BUILD_BUG_ON(ARRAY_SIZE(am335x_usb_queues_rx) !=
+		     ARRAY_SIZE(am335x_usb_queues_tx));
+	if (WARN_ON(cchan->port_num > ARRAY_SIZE(am335x_usb_queues_rx)))
 		return false;
 
 	cchan->q_num = queues[cchan->port_num].submit;
@@ -962,15 +962,25 @@ static struct dma_chan *cppi41_dma_xlate(struct of_phandle_args *dma_spec,
 			&dma_spec->args[0]);
 }
 
-static const struct cppi_glue_infos usb_infos = {
-	.isr = cppi41_irq,
-	.queues_rx = usb_queues_rx,
-	.queues_tx = usb_queues_tx,
+static const struct cppi_glue_infos am335x_usb_infos = {
+	.queues_rx = am335x_usb_queues_rx,
+	.queues_tx = am335x_usb_queues_tx,
 	.td_queue = { .submit = 31, .complete = 0 },
+	.first_completion_queue = 93,
+	.qmgr_num_pend = 5,
+};
+
+static const struct cppi_glue_infos da8xx_usb_infos = {
+	.queues_rx = da8xx_usb_queues_rx,
+	.queues_tx = da8xx_usb_queues_tx,
+	.td_queue = { .submit = 31, .complete = 0 },
+	.first_completion_queue = 24,
+	.qmgr_num_pend = 2,
 };
 
 static const struct of_device_id cppi41_dma_ids[] = {
-	{ .compatible = "ti,am3359-cppi41", .data = &usb_infos},
+	{ .compatible = "ti,am3359-cppi41", .data = &am335x_usb_infos},
+	{ .compatible = "ti,da830-cppi41", .data = &da8xx_usb_infos},
 	{},
 };
 MODULE_DEVICE_TABLE(of, cppi41_dma_ids);
@@ -995,6 +1005,8 @@ static int cppi41_dma_probe(struct platform_device *pdev)
 	struct cppi41_dd *cdd;
 	struct device *dev = &pdev->dev;
 	const struct cppi_glue_infos *glue_info;
+	struct resource *mem;
+	int index;
 	int irq;
 	int ret;
 
@@ -1021,19 +1033,31 @@ static int cppi41_dma_probe(struct platform_device *pdev)
 	INIT_LIST_HEAD(&cdd->ddev.channels);
 	cpp41_dma_info.dma_cap = cdd->ddev.cap_mask;
 
-	cdd->usbss_mem = of_iomap(dev->of_node, 0);
-	cdd->ctrl_mem = of_iomap(dev->of_node, 1);
-	cdd->sched_mem = of_iomap(dev->of_node, 2);
-	cdd->qmgr_mem = of_iomap(dev->of_node, 3);
+	index = of_property_match_string(dev->of_node,
+					 "reg-names", "controller");
+	if (index < 0)
+		return index;
+
+	mem = platform_get_resource(pdev, IORESOURCE_MEM, index);
+	cdd->ctrl_mem = devm_ioremap_resource(dev, mem);
+	if (IS_ERR(cdd->ctrl_mem))
+		return PTR_ERR(cdd->ctrl_mem);
+
+	mem = platform_get_resource(pdev, IORESOURCE_MEM, index + 1);
+	cdd->sched_mem = devm_ioremap_resource(dev, mem);
+	if (IS_ERR(cdd->sched_mem))
+		return PTR_ERR(cdd->sched_mem);
+
+	mem = platform_get_resource(pdev, IORESOURCE_MEM, index + 2);
+	cdd->qmgr_mem = devm_ioremap_resource(dev, mem);
+	if (IS_ERR(cdd->qmgr_mem))
+		return PTR_ERR(cdd->qmgr_mem);
+
 	spin_lock_init(&cdd->lock);
 	INIT_LIST_HEAD(&cdd->pending);
 
 	platform_set_drvdata(pdev, cdd);
 
-	if (!cdd->usbss_mem || !cdd->ctrl_mem || !cdd->sched_mem ||
-			!cdd->qmgr_mem)
-		return -ENXIO;
-
 	pm_runtime_enable(dev);
 	pm_runtime_set_autosuspend_delay(dev, 100);
 	pm_runtime_use_autosuspend(dev);
@@ -1044,6 +1068,13 @@ static int cppi41_dma_probe(struct platform_device *pdev)
 	cdd->queues_rx = glue_info->queues_rx;
 	cdd->queues_tx = glue_info->queues_tx;
 	cdd->td_queue = glue_info->td_queue;
+	cdd->qmgr_num_pend = glue_info->qmgr_num_pend;
+	cdd->first_completion_queue = glue_info->first_completion_queue;
+
+	ret = of_property_read_u32(dev->of_node,
+				   "#dma-channels", &cdd->n_chans);
+	if (ret)
+		goto err_get_n_chans;
 
 	ret = init_cppi41(dev, cdd);
 	if (ret)
@@ -1056,18 +1087,18 @@ static int cppi41_dma_probe(struct platform_device *pdev)
 	irq = irq_of_parse_and_map(dev->of_node, 0);
 	if (!irq) {
 		ret = -EINVAL;
-		goto err_irq;
+		goto err_chans;
 	}
 
-	ret = devm_request_irq(&pdev->dev, irq, glue_info->isr, IRQF_SHARED,
+	ret = devm_request_irq(&pdev->dev, irq, cppi41_irq, IRQF_SHARED,
 			dev_name(dev), cdd);
 	if (ret)
-		goto err_irq;
+		goto err_chans;
 	cdd->irq = irq;
 
 	ret = dma_async_device_register(&cdd->ddev);
 	if (ret)
-		goto err_dma_reg;
+		goto err_chans;
 
 	ret = of_dma_controller_register(dev->of_node,
 			cppi41_dma_xlate, &cpp41_dma_info);
@@ -1080,20 +1111,14 @@ static int cppi41_dma_probe(struct platform_device *pdev)
 	return 0;
 err_of:
 	dma_async_device_unregister(&cdd->ddev);
-err_dma_reg:
-err_irq:
-	cleanup_chans(cdd);
 err_chans:
 	deinit_cppi41(dev, cdd);
 err_init_cppi:
 	pm_runtime_dont_use_autosuspend(dev);
+err_get_n_chans:
 err_get_sync:
 	pm_runtime_put_sync(dev);
 	pm_runtime_disable(dev);
-	iounmap(cdd->usbss_mem);
-	iounmap(cdd->ctrl_mem);
-	iounmap(cdd->sched_mem);
-	iounmap(cdd->qmgr_mem);
 	return ret;
 }
 
@@ -1110,12 +1135,7 @@ static int cppi41_dma_remove(struct platform_device *pdev)
 	dma_async_device_unregister(&cdd->ddev);
 
 	devm_free_irq(&pdev->dev, cdd->irq, cdd);
-	cleanup_chans(cdd);
 	deinit_cppi41(&pdev->dev, cdd);
-	iounmap(cdd->usbss_mem);
-	iounmap(cdd->ctrl_mem);
-	iounmap(cdd->sched_mem);
-	iounmap(cdd->qmgr_mem);
 	pm_runtime_dont_use_autosuspend(&pdev->dev);
 	pm_runtime_put_sync(&pdev->dev);
 	pm_runtime_disable(&pdev->dev);