All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH v16 0/4] Add NVIDIA Tegra GPC-DMA driver
@ 2022-01-10 16:05 Akhil R
  2022-01-10 16:05 ` [PATCH v16 1/4] dt-bindings: dmaengine: Add doc for tegra gpcdma Akhil R
                   ` (3 more replies)
  0 siblings, 4 replies; 17+ messages in thread
From: Akhil R @ 2022-01-10 16:05 UTC (permalink / raw)
  To: dan.j.williams, devicetree, dmaengine, jonathanh, kyarlagadda,
	ldewangan, linux-kernel, linux-tegra, p.zabel, rgumasta, robh+dt,
	thierry.reding, vkoul
  Cc: akhilrajeev

Add support for NVIDIA Tegra general purpose DMA driver for
Tegra186 and Tegra194 platform.

v15 -> v16:
  * Added cyclic mode support
  * Added device pause and resume
  * Added sg_req array to handle multiple sgs
  * Request irq only when a dma chan is allocated
  * Removed locks and busy flag
  * Updates in terminate_all callback

v15 - https://lkml.org/lkml/2021/12/16/905

The additional changes were added to align with the features that got
added in the interim.
Keeping "Reviewed-by: Jon Hunter <jonathanh@nvidia.com>" as the changes
were reviewed internally.

Akhil R (4):
  dt-bindings: dmaengine: Add doc for tegra gpcdma
  dmaengine: tegra: Add tegra gpcdma driver
  arm64: defconfig: tegra: Enable GPCDMA
  arm64: tegra: Add GPCDMA node for tegra186 and tegra194

 .../bindings/dma/nvidia,tegra186-gpc-dma.yaml      |  110 ++
 arch/arm64/boot/dts/nvidia/tegra186.dtsi           |   42 +
 arch/arm64/boot/dts/nvidia/tegra194.dtsi           |   43 +
 arch/arm64/configs/defconfig                       |    1 +
 drivers/dma/Kconfig                                |   11 +
 drivers/dma/Makefile                               |    1 +
 drivers/dma/tegra186-gpc-dma.c                     | 1464 ++++++++++++++++++++
 7 files changed, 1672 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/dma/nvidia,tegra186-gpc-dma.yaml
 create mode 100644 drivers/dma/tegra186-gpc-dma.c

-- 
2.7.4


^ permalink raw reply	[flat|nested] 17+ messages in thread

* [PATCH v16 1/4] dt-bindings: dmaengine: Add doc for tegra gpcdma
  2022-01-10 16:05 [PATCH v16 0/4] Add NVIDIA Tegra GPC-DMA driver Akhil R
@ 2022-01-10 16:05 ` Akhil R
  2022-01-10 16:05 ` [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver Akhil R
                   ` (2 subsequent siblings)
  3 siblings, 0 replies; 17+ messages in thread
From: Akhil R @ 2022-01-10 16:05 UTC (permalink / raw)
  To: dan.j.williams, devicetree, dmaengine, jonathanh, kyarlagadda,
	ldewangan, linux-kernel, linux-tegra, p.zabel, rgumasta, robh+dt,
	thierry.reding, vkoul
  Cc: akhilrajeev

Add DT binding document for Nvidia Tegra GPCDMA controller.

Signed-off-by: Rajesh Gumasta <rgumasta@nvidia.com>
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-by: Rob Herring <robh@kernel.org>
---
 .../bindings/dma/nvidia,tegra186-gpc-dma.yaml      | 110 +++++++++++++++++++++
 1 file changed, 110 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/dma/nvidia,tegra186-gpc-dma.yaml

diff --git a/Documentation/devicetree/bindings/dma/nvidia,tegra186-gpc-dma.yaml b/Documentation/devicetree/bindings/dma/nvidia,tegra186-gpc-dma.yaml
new file mode 100644
index 0000000..9dd1476
--- /dev/null
+++ b/Documentation/devicetree/bindings/dma/nvidia,tegra186-gpc-dma.yaml
@@ -0,0 +1,110 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/dma/nvidia,tegra186-gpc-dma.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: NVIDIA Tegra GPC DMA Controller Device Tree Bindings
+
+description: |
+  The Tegra General Purpose Central (GPC) DMA controller is used for faster
+  data transfers between memory to memory, memory to device and device to
+  memory.
+
+maintainers:
+  - Jon Hunter <jonathanh@nvidia.com>
+  - Rajesh Gumasta <rgumasta@nvidia.com>
+
+allOf:
+  - $ref: "dma-controller.yaml#"
+
+properties:
+  compatible:
+    oneOf:
+      - const: nvidia,tegra186-gpcdma
+      - items:
+          - const: nvidia,tegra194-gpcdma
+          - const: nvidia,tegra186-gpcdma
+
+  "#dma-cells":
+    const: 1
+
+  reg:
+    maxItems: 1
+
+  interrupts:
+    description:
+      Should contain all of the per-channel DMA interrupts in
+      ascending order with respect to the DMA channel index.
+    minItems: 1
+    maxItems: 31
+
+  resets:
+    maxItems: 1
+
+  reset-names:
+    const: gpcdma
+
+  iommus:
+    maxItems: 1
+
+  dma-coherent: true
+
+required:
+  - compatible
+  - reg
+  - interrupts
+  - resets
+  - reset-names
+  - "#dma-cells"
+  - iommus
+
+additionalProperties: false
+
+examples:
+  - |
+    #include <dt-bindings/interrupt-controller/arm-gic.h>
+    #include <dt-bindings/memory/tegra186-mc.h>
+    #include <dt-bindings/reset/tegra186-reset.h>
+
+    dma-controller@2600000 {
+        compatible = "nvidia,tegra186-gpcdma";
+        reg = <0x2600000 0x210000>;
+        resets = <&bpmp TEGRA186_RESET_GPCDMA>;
+        reset-names = "gpcdma";
+        interrupts = <GIC_SPI 76 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 77 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 78 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 79 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 80 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 81 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 82 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 83 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 84 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 85 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 86 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 87 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 88 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 89 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 90 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 91 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 92 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 93 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 94 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 95 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 96 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 97 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 98 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 99 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 100 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 101 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 102 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 103 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 104 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 105 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 106 IRQ_TYPE_LEVEL_HIGH>;
+        #dma-cells = <1>;
+        iommus = <&smmu TEGRA186_SID_GPCDMA_0>;
+        dma-coherent;
+    };
+...
-- 
2.7.4


^ permalink raw reply related	[flat|nested] 17+ messages in thread

* [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-10 16:05 [PATCH v16 0/4] Add NVIDIA Tegra GPC-DMA driver Akhil R
  2022-01-10 16:05 ` [PATCH v16 1/4] dt-bindings: dmaengine: Add doc for tegra gpcdma Akhil R
@ 2022-01-10 16:05 ` Akhil R
  2022-01-10 20:31   ` kernel test robot
  2022-01-14 11:57   ` Dmitry Osipenko
  2022-01-10 16:05 ` [PATCH v16 3/4] arm64: defconfig: tegra: Enable GPCDMA Akhil R
  2022-01-10 16:05 ` [PATCH v16 4/4] arm64: tegra: Add GPCDMA node for tegra186 and tegra194 Akhil R
  3 siblings, 2 replies; 17+ messages in thread
From: Akhil R @ 2022-01-10 16:05 UTC (permalink / raw)
  To: dan.j.williams, devicetree, dmaengine, jonathanh, kyarlagadda,
	ldewangan, linux-kernel, linux-tegra, p.zabel, rgumasta, robh+dt,
	thierry.reding, vkoul
  Cc: akhilrajeev, Pavan Kunapuli

Adding GPC DMA controller driver for Tegra186 and Tegra194. The driver
supports dma transfers between memory to memory, IO peripheral to memory
and memory to IO peripheral.

Co-developed-by: Pavan Kunapuli <pkunapuli@nvidia.com>
Signed-off-by: Pavan Kunapuli <pkunapuli@nvidia.com>
Co-developed-by: Rajesh Gumasta <rgumasta@nvidia.com>
Signed-off-by: Rajesh Gumasta <rgumasta@nvidia.com>
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
---
 drivers/dma/Kconfig            |   11 +
 drivers/dma/Makefile           |    1 +
 drivers/dma/tegra186-gpc-dma.c | 1464 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 1476 insertions(+)
 create mode 100644 drivers/dma/tegra186-gpc-dma.c

diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig
index 6bcdb4e..dbe7664 100644
--- a/drivers/dma/Kconfig
+++ b/drivers/dma/Kconfig
@@ -629,6 +629,17 @@ config TXX9_DMAC
 	  Support the TXx9 SoC internal DMA controller.  This can be
 	  integrated in chips such as the Toshiba TX4927/38/39.
 
+config TEGRA186_GPC_DMA
+	tristate "NVIDIA Tegra GPC DMA support"
+	depends on ARCH_TEGRA || COMPILE_TEST
+	select DMA_ENGINE
+	help
+	  Support for the NVIDIA Tegra General Purpose Central DMA controller.
+	  The DMA controller has multiple DMA channels which can be configured
+	  for different peripherals like UART, SPI, etc which are on APB bus.
+	  This DMA controller transfers data from memory to peripheral FIFO
+	  or vice versa. It also supports memory to memory data transfer.
+
 config TEGRA20_APB_DMA
 	tristate "NVIDIA Tegra20 APB DMA support"
 	depends on ARCH_TEGRA || COMPILE_TEST
diff --git a/drivers/dma/Makefile b/drivers/dma/Makefile
index 616d926..2f1b87f 100644
--- a/drivers/dma/Makefile
+++ b/drivers/dma/Makefile
@@ -72,6 +72,7 @@ obj-$(CONFIG_STM32_MDMA) += stm32-mdma.o
 obj-$(CONFIG_SPRD_DMA) += sprd-dma.o
 obj-$(CONFIG_S3C24XX_DMAC) += s3c24xx-dma.o
 obj-$(CONFIG_TXX9_DMAC) += txx9dmac.o
+obj-$(CONFIG_TEGRA186_GPC_DMA) += tegra186-gpc-dma.o
 obj-$(CONFIG_TEGRA20_APB_DMA) += tegra20-apb-dma.o
 obj-$(CONFIG_TEGRA210_ADMA) += tegra210-adma.o
 obj-$(CONFIG_TIMB_DMA) += timb_dma.o
diff --git a/drivers/dma/tegra186-gpc-dma.c b/drivers/dma/tegra186-gpc-dma.c
new file mode 100644
index 0000000..ef2541a
--- /dev/null
+++ b/drivers/dma/tegra186-gpc-dma.c
@@ -0,0 +1,1464 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * DMA driver for NVIDIA Tegra186 and Tegra194 GPC DMA controller.
+ *
+ * Copyright (c) 2014-2021, NVIDIA CORPORATION.  All rights reserved.
+ */
+
+#include <linux/bitfield.h>
+#include <linux/dmaengine.h>
+#include <linux/dma-mapping.h>
+#include <linux/interrupt.h>
+#include <linux/iommu.h>
+#include <linux/iopoll.h>
+#include <linux/minmax.h>
+#include <linux/module.h>
+#include <linux/of_device.h>
+#include <linux/of_dma.h>
+#include <linux/platform_device.h>
+#include <linux/reset.h>
+#include <linux/slab.h>
+#include <linux/version.h>
+#include <dt-bindings/memory/tegra186-mc.h>
+#include "virt-dma.h"
+
+/* CSR register */
+#define TEGRA_GPCDMA_CHAN_CSR			0x00
+#define TEGRA_GPCDMA_CSR_ENB			BIT(31)
+#define TEGRA_GPCDMA_CSR_IE_EOC			BIT(30)
+#define TEGRA_GPCDMA_CSR_ONCE			BIT(27)
+
+#define TEGRA_GPCDMA_CSR_FC_MODE		GENMASK(25, 24)
+#define TEGRA_GPCDMA_CSR_FC_MODE_NO_MMIO	\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_FC_MODE, 0)
+#define TEGRA_GPCDMA_CSR_FC_MODE_ONE_MMIO	\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_FC_MODE, 1)
+#define TEGRA_GPCDMA_CSR_FC_MODE_TWO_MMIO	\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_FC_MODE, 2)
+#define TEGRA_GPCDMA_CSR_FC_MODE_FOUR_MMIO	\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_FC_MODE, 3)
+
+#define TEGRA_GPCDMA_CSR_DMA			GENMASK(23, 21)
+#define TEGRA_GPCDMA_CSR_DMA_IO2MEM_NO_FC	\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_DMA, 0)
+#define TEGRA_GPCDMA_CSR_DMA_IO2MEM_FC		\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_DMA, 1)
+#define TEGRA_GPCDMA_CSR_DMA_MEM2IO_NO_FC	\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_DMA, 2)
+#define TEGRA_GPCDMA_CSR_DMA_MEM2IO_FC		\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_DMA, 3)
+#define TEGRA_GPCDMA_CSR_DMA_MEM2MEM		\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_DMA, 4)
+#define TEGRA_GPCDMA_CSR_DMA_FIXED_PAT		\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_DMA, 6)
+
+#define TEGRA_GPCDMA_CSR_REQ_SEL_MASK		GENMASK(20, 16)
+#define TEGRA_GPCDMA_CSR_REQ_SEL_UNUSED		\
+					FIELD_PREP(TEGRA_GPCDMA_CSR_REQ_SEL_MASK, 4)
+#define TEGRA_GPCDMA_CSR_IRQ_MASK		BIT(15)
+#define TEGRA_GPCDMA_CSR_WEIGHT			GENMASK(13, 10)
+
+/* STATUS register */
+#define TEGRA_GPCDMA_CHAN_STATUS		0x004
+#define TEGRA_GPCDMA_STATUS_BUSY		BIT(31)
+#define TEGRA_GPCDMA_STATUS_ISE_EOC		BIT(30)
+#define TEGRA_GPCDMA_STATUS_PING_PONG		BIT(28)
+#define TEGRA_GPCDMA_STATUS_DMA_ACTIVITY	BIT(27)
+#define TEGRA_GPCDMA_STATUS_CHANNEL_PAUSE	BIT(26)
+#define TEGRA_GPCDMA_STATUS_CHANNEL_RX		BIT(25)
+#define TEGRA_GPCDMA_STATUS_CHANNEL_TX		BIT(24)
+#define TEGRA_GPCDMA_STATUS_IRQ_INTR_STA	BIT(23)
+#define TEGRA_GPCDMA_STATUS_IRQ_STA		BIT(21)
+#define TEGRA_GPCDMA_STATUS_IRQ_TRIG_STA	BIT(20)
+
+#define TEGRA_GPCDMA_CHAN_CSRE			0x008
+#define TEGRA_GPCDMA_CHAN_CSRE_PAUSE		BIT(31)
+
+/* Source address */
+#define TEGRA_GPCDMA_CHAN_SRC_PTR		0x00C
+
+/* Destination address */
+#define TEGRA_GPCDMA_CHAN_DST_PTR		0x010
+
+/* High address pointer */
+#define TEGRA_GPCDMA_CHAN_HIGH_ADDR_PTR		0x014
+#define TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR		GENMASK(7, 0)
+#define TEGRA_GPCDMA_HIGH_ADDR_DST_PTR		GENMASK(23, 16)
+
+/* MC sequence register */
+#define TEGRA_GPCDMA_CHAN_MCSEQ			0x18
+#define TEGRA_GPCDMA_MCSEQ_DATA_SWAP		BIT(31)
+#define TEGRA_GPCDMA_MCSEQ_REQ_COUNT		GENMASK(30, 25)
+#define TEGRA_GPCDMA_MCSEQ_BURST		GENMASK(24, 23)
+#define TEGRA_GPCDMA_MCSEQ_BURST_2		\
+					FIELD_PREP(TEGRA_GPCDMA_MCSEQ_BURST, 0)
+#define TEGRA_GPCDMA_MCSEQ_BURST_16		\
+					FIELD_PREP(TEGRA_GPCDMA_MCSEQ_BURST, 3)
+#define TEGRA_GPCDMA_MCSEQ_WRAP1		GENMASK(22, 20)
+#define TEGRA_GPCDMA_MCSEQ_WRAP0		GENMASK(19, 17)
+#define TEGRA_GPCDMA_MCSEQ_WRAP_NONE		0
+
+#define TEGRA_GPCDMA_MCSEQ_STREAM_ID1_MASK	GENMASK(13, 7)
+#define TEGRA_GPCDMA_MCSEQ_STREAM_ID0_MASK	GENMASK(6, 0)
+
+/* MMIO sequence register */
+#define TEGRA_GPCDMA_CHAN_MMIOSEQ			0x01c
+#define TEGRA_GPCDMA_MMIOSEQ_DBL_BUF		BIT(31)
+#define TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH		GENMASK(30, 28)
+#define TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH_8	\
+					FIELD_PREP(TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH, 0)
+#define TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH_16	\
+					FIELD_PREP(TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH, 1)
+#define TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH_32	\
+					FIELD_PREP(TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH, 2)
+#define TEGRA_GPCDMA_MMIOSEQ_DATA_SWAP		BIT(27)
+#define TEGRA_GPCDMA_MMIOSEQ_BURST		GENMASK(26, 23)
+#define TEGRA_GPCDMA_MMIOSEQ_BURST_SHIFT	23
+#define TEGRA_GPCDMA_MMIOSEQ_BURST_MIN		1U
+#define TEGRA_GPCDMA_MMIOSEQ_BURST_MAX		16U
+#define TEGRA_GPCDMA_MMIOSEQ_MASTER_ID		GENMASK(22, 19)
+#define TEGRA_GPCDMA_MMIOSEQ_WRAP_WORD		GENMASK(18, 16)
+#define TEGRA_GPCDMA_MMIOSEQ_MMIO_PROT		GENMASK(8, 7)
+
+/* Channel WCOUNT */
+#define TEGRA_GPCDMA_CHAN_WCOUNT		0x20
+
+/* Transfer count */
+#define TEGRA_GPCDMA_CHAN_XFER_COUNT		0x24
+
+/* DMA byte count status */
+#define TEGRA_GPCDMA_CHAN_DMA_BYTE_STATUS	0x28
+
+/* Error Status Register */
+#define TEGRA_GPCDMA_CHAN_ERR_STATUS		0x30
+#define TEGRA_GPCDMA_CHAN_ERR_TYPE_SHIFT	8
+#define TEGRA_GPCDMA_CHAN_ERR_TYPE_MASK	0xF
+#define TEGRA_GPCDMA_CHAN_ERR_TYPE(err)	(			\
+		((err) >> TEGRA_GPCDMA_CHAN_ERR_TYPE_SHIFT) &	\
+		TEGRA_GPCDMA_CHAN_ERR_TYPE_MASK)
+#define TEGRA_DMA_BM_FIFO_FULL_ERR		0xF
+#define TEGRA_DMA_PERIPH_FIFO_FULL_ERR		0xE
+#define TEGRA_DMA_PERIPH_ID_ERR			0xD
+#define TEGRA_DMA_STREAM_ID_ERR			0xC
+#define TEGRA_DMA_MC_SLAVE_ERR			0xB
+#define TEGRA_DMA_MMIO_SLAVE_ERR		0xA
+
+/* Fixed Pattern */
+#define TEGRA_GPCDMA_CHAN_FIXED_PATTERN		0x34
+
+#define TEGRA_GPCDMA_CHAN_TZ			0x38
+#define TEGRA_GPCDMA_CHAN_TZ_MMIO_PROT_1	BIT(0)
+#define TEGRA_GPCDMA_CHAN_TZ_MC_PROT_1		BIT(1)
+
+#define TEGRA_GPCDMA_CHAN_SPARE			0x3c
+#define TEGRA_GPCDMA_CHAN_SPARE_EN_LEGACY_FC	BIT(16)
+
+/*
+ * If any burst is in flight and DMA paused then this is the time to complete
+ * on-flight burst and update DMA status register.
+ */
+#define TEGRA_GPCDMA_BURST_COMPLETE_TIME	20
+#define TEGRA_GPCDMA_BURST_COMPLETION_TIMEOUT	100
+
+/* Channel base address offset from GPCDMA base address */
+#define TEGRA_GPCDMA_CHANNEL_BASE_ADD_OFFSET	0x20000
+
+struct tegra_dma;
+
+/*
+ * tegra_dma_chip_data Tegra chip specific DMA data
+ * @nr_channels: Number of channels available in the controller.
+ * @channel_reg_size: Channel register size.
+ * @max_dma_count: Maximum DMA transfer count supported by DMA controller.
+ * @hw_support_pause: DMA HW engine support pause of the channel.
+ */
+struct tegra_dma_chip_data {
+	unsigned int nr_channels;
+	unsigned int channel_reg_size;
+	unsigned int max_dma_count;
+	bool hw_support_pause;
+};
+
+/* DMA channel registers */
+struct tegra_dma_channel_regs {
+	unsigned long csr;
+	unsigned long src_ptr;
+	unsigned long dst_ptr;
+	unsigned long high_addr_ptr;
+	unsigned long mc_seq;
+	unsigned long mmio_seq;
+	unsigned long wcount;
+	unsigned long fixed_pattern;
+};
+
+/*
+ * tegra_dma_sg_req: Dma request details to configure hardware. This
+ * contains the details for one transfer to configure DMA hw.
+ * The client's request for data transfer can be broken into multiple
+ * sub-transfer as per requester details and hw support. This sub transfer
+ * get added as an array in Tegra DMA desc which manages the transfer details.
+ */
+struct tegra_dma_sg_req {
+	unsigned int len;
+	struct tegra_dma_channel_regs ch_regs;
+};
+
+/*
+ * tegra_dma_desc: Tegra DMA descriptors which uses virt_dma_desc to
+ * manage client request and keep track of transfer status, callbacks
+ * and request counts etc.
+ */
+struct tegra_dma_desc {
+	struct virt_dma_desc vd;
+	unsigned int bytes_requested;
+	unsigned int bytes_transferred;
+	unsigned int sg_idx;
+	unsigned int sg_count;
+	bool cyclic;
+	struct tegra_dma_channel *tdc;
+	struct tegra_dma_sg_req sg_req[];
+};
+
+/*
+ * tegra_dma_channel: Channel specific information
+ */
+struct tegra_dma_channel {
+	struct virt_dma_chan vc;
+	struct tegra_dma_desc *dma_desc;
+	char name[30];
+	bool config_init;
+	int id;
+	int irq;
+	unsigned int stream_id;
+	unsigned long chan_base_offset;
+	struct tegra_dma *tdma;
+	int slave_id;
+	enum dma_transfer_direction sid_dir;
+	struct dma_slave_config dma_sconfig;
+};
+
+/*
+ * tegra_dma: Tegra DMA specific information
+ */
+struct tegra_dma {
+	struct dma_device dma_dev;
+	struct device *dev;
+	void __iomem *base_addr;
+	const struct tegra_dma_chip_data *chip_data;
+	struct reset_control *rst;
+	unsigned long sid_m2d_reserved;
+	unsigned long sid_d2m_reserved;
+	struct tegra_dma_channel channels[];
+};
+
+static inline void tdc_write(struct tegra_dma_channel *tdc,
+			     u32 reg, u32 val)
+{
+	writel_relaxed(val, tdc->tdma->base_addr + tdc->chan_base_offset + reg);
+}
+
+static inline u32 tdc_read(struct tegra_dma_channel *tdc, u32 reg)
+{
+	return readl_relaxed(tdc->tdma->base_addr + tdc->chan_base_offset + reg);
+}
+
+static inline struct tegra_dma_channel *to_tegra_dma_chan(struct dma_chan *dc)
+{
+	return container_of(dc, struct tegra_dma_channel, vc.chan);
+}
+
+static inline struct tegra_dma_desc *vd_to_tegra_dma_desc(struct virt_dma_desc *vd)
+{
+	return container_of(vd, struct tegra_dma_desc, vd);
+}
+
+static inline struct device *tdc2dev(struct tegra_dma_channel *tdc)
+{
+	return tdc->vc.chan.device->dev;
+}
+
+static void tegra_dma_dump_chan_regs(struct tegra_dma_channel *tdc)
+{
+	dev_dbg(tdc2dev(tdc), "DMA Channel %d name %s register dump:\n",
+		tdc->id, tdc->name);
+	dev_dbg(tdc2dev(tdc), "CSR %x STA %x CSRE %x SRC %x DST %x\n",
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSR),
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS),
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSRE),
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_SRC_PTR),
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_DST_PTR)
+	);
+	dev_dbg(tdc2dev(tdc), "MCSEQ %x IOSEQ %x WCNT %x XFER %x BSTA %x\n",
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_MCSEQ),
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_MMIOSEQ),
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_WCOUNT),
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT),
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_DMA_BYTE_STATUS)
+	);
+	dev_dbg(tdc2dev(tdc), "DMA ERR_STA %x\n",
+		tdc_read(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS));
+}
+
+static int tegra_dma_sid_reserve(struct tegra_dma_channel *tdc,
+				 enum dma_transfer_direction direction)
+{
+	struct tegra_dma *tdma = tdc->tdma;
+	unsigned int sid = tdc->slave_id;
+
+	if (!is_slave_direction(direction))
+		return 0;
+
+	switch (direction) {
+	case DMA_MEM_TO_DEV:
+		if (test_and_set_bit(sid, &tdma->sid_m2d_reserved)) {
+			dev_err(tdma->dev, "slave id already in use\n");
+			return -EINVAL;
+		}
+		break;
+	case DMA_DEV_TO_MEM:
+		if (test_and_set_bit(sid, &tdma->sid_d2m_reserved)) {
+			dev_err(tdma->dev, "slave id already in use\n");
+			return -EINVAL;
+		}
+		break;
+	default:
+		break;
+	}
+
+	tdc->sid_dir = direction;
+
+	return 0;
+}
+
+static void tegra_dma_sid_free(struct tegra_dma_channel *tdc)
+{
+	struct tegra_dma *tdma = tdc->tdma;
+	unsigned int sid = tdc->slave_id;
+
+	switch (tdc->sid_dir) {
+	case DMA_MEM_TO_DEV:
+		clear_bit(sid,  &tdma->sid_m2d_reserved);
+		break;
+	case DMA_DEV_TO_MEM:
+		clear_bit(sid,  &tdma->sid_d2m_reserved);
+		break;
+	default:
+		break;
+	}
+
+	tdc->sid_dir = DMA_TRANS_NONE;
+}
+
+static void tegra_dma_desc_free(struct virt_dma_desc *vd)
+{
+	kfree(container_of(vd, struct tegra_dma_desc, vd));
+}
+
+static int tegra_dma_slave_config(struct dma_chan *dc,
+				  struct dma_slave_config *sconfig)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+
+	memcpy(&tdc->dma_sconfig, sconfig, sizeof(*sconfig));
+	tdc->config_init = true;
+
+	return 0;
+}
+
+static int tegra_dma_pause(struct dma_chan *dc)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	unsigned long wcount;
+	u32 val;
+	int ret;
+
+	if (!tdc->tdma->chip_data->hw_support_pause)
+		return 0;
+
+	val = tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSRE);
+	val |= TEGRA_GPCDMA_CHAN_CSRE_PAUSE;
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSRE, val);
+
+	/* Wait until busy bit is de-asserted */
+	ret = readl_relaxed_poll_timeout_atomic(tdc->tdma->base_addr +
+			tdc->chan_base_offset + TEGRA_GPCDMA_CHAN_STATUS,
+			val,
+			!(val & TEGRA_GPCDMA_STATUS_BUSY),
+			TEGRA_GPCDMA_BURST_COMPLETE_TIME,
+			TEGRA_GPCDMA_BURST_COMPLETION_TIMEOUT);
+
+	if (ret) {
+		dev_err(tdc2dev(tdc), "DMA pause timed out\n");
+		return ret;
+	}
+
+	synchronize_irq(tdc->irq);
+
+	wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
+	tdc->dma_desc->bytes_transferred +=
+			tdc->dma_desc->bytes_requested - (wcount * 4);
+
+	return ret;
+}
+
+static int tegra_dma_resume(struct dma_chan *dc)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	u32 val;
+
+	if (!tdc->tdma->chip_data->hw_support_pause)
+		return 0;
+
+	val = tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSRE);
+	val &= ~TEGRA_GPCDMA_CHAN_CSRE_PAUSE;
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSRE, val);
+
+	return 0;
+}
+
+static void tegra_dma_stop(struct tegra_dma_channel *tdc)
+{
+	u32 csr, status;
+
+	csr = tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSR);
+
+	/* Disable interrupts */
+	csr &= ~TEGRA_GPCDMA_CSR_IE_EOC;
+
+	/* Disable DMA */
+	csr &= ~TEGRA_GPCDMA_CSR_ENB;
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSR, csr);
+
+	/* Clear interrupt status if it is there */
+	status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS);
+	if (status & TEGRA_GPCDMA_STATUS_ISE_EOC) {
+		dev_dbg(tdc2dev(tdc), "%s():clearing interrupt\n", __func__);
+		tdc_write(tdc, TEGRA_GPCDMA_CHAN_STATUS, status);
+	}
+}
+
+static void tegra_dma_configure_next_sg(struct tegra_dma_channel *tdc)
+{
+	struct tegra_dma_desc *dma_desc = tdc->dma_desc;
+	struct tegra_dma_channel_regs *ch_regs;
+
+	ch_regs = &dma_desc->sg_req[dma_desc->sg_idx].ch_regs;
+
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_WCOUNT, ch_regs->wcount);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_SRC_PTR, ch_regs->src_ptr);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_DST_PTR, ch_regs->dst_ptr);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_HIGH_ADDR_PTR, ch_regs->high_addr_ptr);
+
+	/* Start DMA */
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSR,
+		  ch_regs->csr | TEGRA_GPCDMA_CSR_ENB);
+
+	dma_desc->sg_idx++;
+
+	/* Reset the sg index for cyclic transfers */
+	if (dma_desc->sg_idx == dma_desc->sg_count)
+		dma_desc->sg_idx = 0;
+}
+
+static void tegra_dma_start(struct tegra_dma_channel *tdc)
+{
+	struct tegra_dma_desc *dma_desc = tdc->dma_desc;
+	struct tegra_dma_channel_regs *ch_regs;
+	struct virt_dma_desc *vdesc;
+	u32 val;
+
+	if (!dma_desc) {
+		vdesc = vchan_next_desc(&tdc->vc);
+		if (!vdesc)
+			return;
+		dma_desc = vd_to_tegra_dma_desc(vdesc);
+		list_del(&vdesc->node);
+
+		dma_desc->tdc = tdc;
+		tdc->dma_desc = dma_desc;
+	}
+
+	ch_regs = &dma_desc->sg_req[dma_desc->sg_idx].ch_regs;
+
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_WCOUNT, ch_regs->wcount);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSR, 0);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_SRC_PTR, ch_regs->src_ptr);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_DST_PTR, ch_regs->dst_ptr);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_HIGH_ADDR_PTR, ch_regs->high_addr_ptr);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_FIXED_PATTERN, ch_regs->fixed_pattern);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_MMIOSEQ, ch_regs->mmio_seq);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_MCSEQ, ch_regs->mc_seq);
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSR, ch_regs->csr);
+
+	val = tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSRE);
+	val &= ~TEGRA_GPCDMA_CHAN_CSRE_PAUSE;
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSRE, val);
+
+	/* Start DMA */
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSR,
+		  ch_regs->csr | TEGRA_GPCDMA_CSR_ENB);
+
+	dma_desc->sg_idx++;
+}
+
+static void tegra_dma_xfer_complete(struct tegra_dma_channel *tdc)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&tdc->vc.lock, flags);
+	vchan_cookie_complete(&tdc->dma_desc->vd);
+	spin_unlock_irqrestore(&tdc->vc.lock, flags);
+
+	tegra_dma_sid_free(tdc);
+	tdc->dma_desc = NULL;
+}
+
+static void tegra_dma_chan_decode_error(struct tegra_dma_channel *tdc,
+					unsigned int err_status)
+{
+	switch (TEGRA_GPCDMA_CHAN_ERR_TYPE(err_status)) {
+	case TEGRA_DMA_BM_FIFO_FULL_ERR:
+		dev_err(tdc->tdma->dev,
+			"GPCDMA CH%d bm fifo full\n", tdc->id);
+		break;
+
+	case TEGRA_DMA_PERIPH_FIFO_FULL_ERR:
+		dev_err(tdc->tdma->dev,
+			"GPCDMA CH%d peripheral fifo full\n", tdc->id);
+		break;
+
+	case TEGRA_DMA_PERIPH_ID_ERR:
+		dev_err(tdc->tdma->dev,
+			"GPCDMA CH%d illegal peripheral id\n", tdc->id);
+		break;
+
+	case TEGRA_DMA_STREAM_ID_ERR:
+		dev_err(tdc->tdma->dev,
+			"GPCDMA CH%d illegal stream id\n", tdc->id);
+		break;
+
+	case TEGRA_DMA_MC_SLAVE_ERR:
+		dev_err(tdc->tdma->dev,
+			"GPCDMA CH%d mc slave error\n", tdc->id);
+		break;
+
+	case TEGRA_DMA_MMIO_SLAVE_ERR:
+		dev_err(tdc->tdma->dev,
+			"GPCDMA CH%d mmio slave error\n", tdc->id);
+		break;
+
+	default:
+		dev_err(tdc->tdma->dev,
+			"GPCDMA CH%d security violation %x\n", tdc->id,
+			err_status);
+	}
+}
+
+static irqreturn_t tegra_dma_isr(int irq, void *dev_id)
+{
+	struct tegra_dma_channel *tdc = dev_id;
+	struct tegra_dma_desc *dma_desc = tdc->dma_desc;
+	struct tegra_dma_sg_req *sg_req;
+	unsigned long flags;
+	u32 status;
+
+	/* Check channel error status register */
+	status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS);
+	if (status) {
+		tegra_dma_chan_decode_error(tdc, status);
+		tegra_dma_dump_chan_regs(tdc);
+		tdc_write(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS, 0xFFFFFFFF);
+	}
+
+	status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS);
+	if (!(status & TEGRA_GPCDMA_STATUS_ISE_EOC) || !dma_desc)
+		return IRQ_HANDLED;
+
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_STATUS,
+		  TEGRA_GPCDMA_STATUS_ISE_EOC);
+
+	sg_req = dma_desc->sg_req;
+	dma_desc->bytes_transferred += sg_req[dma_desc->sg_idx - 1].len;
+
+	if (dma_desc->sg_idx == dma_desc->sg_count) {
+		/* Mark complete */
+		tegra_dma_xfer_complete(tdc);
+	} else if (dma_desc->cyclic) {
+		spin_lock_irqsave(&tdc->vc.lock, flags);
+		vchan_cyclic_callback(&dma_desc->vd);
+		spin_unlock_irqrestore(&tdc->vc.lock, flags);
+		tegra_dma_configure_next_sg(tdc);
+	} else {
+		tegra_dma_start(tdc);
+	}
+
+	return IRQ_HANDLED;
+}
+
+static void tegra_dma_issue_pending(struct dma_chan *dc)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	unsigned long flags;
+	int err;
+	u32 val;
+
+	if (tdc->dma_desc)
+		return;
+
+	spin_lock_irqsave(&tdc->vc.lock, flags);
+	if (vchan_issue_pending(&tdc->vc))
+		tegra_dma_start(tdc);
+
+	spin_unlock_irqrestore(&tdc->vc.lock, flags);
+
+	if (tdc->dma_desc && tdc->dma_desc->cyclic) {
+		/*
+		 * For cyclic dma transfers, program the second
+		 * transfer parameters as soon as the first dma
+		 * transfer is started inorder for the dma
+		 * controller to trigger the second transfer
+		 * with the correct parameters. Poll for the
+		 * channel busy bit and start the transfer.
+		 */
+		err = readl_relaxed_poll_timeout_atomic(tdc->tdma->base_addr +
+				tdc->chan_base_offset + TEGRA_GPCDMA_CHAN_STATUS,
+				val,
+				(val & TEGRA_GPCDMA_STATUS_BUSY),
+				TEGRA_GPCDMA_BURST_COMPLETE_TIME,
+				TEGRA_GPCDMA_BURST_COMPLETION_TIMEOUT);
+		if (!err)
+			tegra_dma_configure_next_sg(tdc);
+	}
+}
+
+static int tegra_dma_stop_client(struct tegra_dma_channel *tdc)
+{
+	int ret;
+	unsigned long status, wcount;
+	u32 csr = tdc_read(tdc, TEGRA_GPCDMA_CHAN_CSR);
+
+	/* Before Reading DMA status to figure out number
+	 * of bytes transferred by DMA channel:
+	 * Change the client associated with the DMA channel
+	 * to stop DMA engine from starting any more bursts for
+	 * the given client and wait for in flight bursts to complete
+	 */
+	csr &= ~(TEGRA_GPCDMA_CSR_REQ_SEL_MASK);
+	csr |= TEGRA_GPCDMA_CSR_REQ_SEL_UNUSED;
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_CSR, csr);
+
+	/* Wait for in flight data transfer to finish */
+	udelay(TEGRA_GPCDMA_BURST_COMPLETE_TIME);
+
+	/* If TX/RX path is still active wait till it becomes
+	 * inactive
+	 */
+
+	ret = readl_relaxed_poll_timeout_atomic(tdc->tdma->base_addr +
+				tdc->chan_base_offset +
+				TEGRA_GPCDMA_CHAN_STATUS,
+				status,
+				!(status & (TEGRA_GPCDMA_STATUS_CHANNEL_TX |
+				TEGRA_GPCDMA_STATUS_CHANNEL_RX)),
+				5,
+				TEGRA_GPCDMA_BURST_COMPLETION_TIMEOUT);
+	if (ret) {
+		dev_err(tdc2dev(tdc), "Timeout waiting for DMA burst completion!\n");
+		tegra_dma_dump_chan_regs(tdc);
+	}
+
+	wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
+	tdc->dma_desc->bytes_transferred +=
+			tdc->dma_desc->bytes_requested - (wcount * 4);
+
+	return ret;
+}
+
+static int tegra_dma_terminate_all(struct dma_chan *dc)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	unsigned long flags;
+	LIST_HEAD(head);
+	int err;
+
+	if (tdc->dma_desc) {
+		err = tegra_dma_stop_client(tdc);
+		if (err)
+			return err;
+
+		tegra_dma_stop(tdc);
+		tdc->dma_desc = NULL;
+	}
+
+	tegra_dma_sid_free(tdc);
+
+	spin_lock_irqsave(&tdc->vc.lock, flags);
+	vchan_get_all_descriptors(&tdc->vc, &head);
+	spin_unlock_irqrestore(&tdc->vc.lock, flags);
+
+	vchan_dma_desc_free_list(&tdc->vc, &head);
+
+	return 0;
+}
+
+static enum dma_status tegra_dma_tx_status(struct dma_chan *dc,
+					   dma_cookie_t cookie,
+					   struct dma_tx_state *txstate)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	struct tegra_dma_desc *dma_desc;
+	struct virt_dma_desc *vd;
+	unsigned int residual;
+	enum dma_status ret;
+
+	ret = dma_cookie_status(dc, cookie, txstate);
+	if (ret == DMA_COMPLETE)
+		return ret;
+
+	vd = vchan_find_desc(&tdc->vc, cookie);
+	if (vd) {
+		dma_desc = vd_to_tegra_dma_desc(vd);
+		residual = dma_desc->bytes_requested;
+		dma_set_residue(txstate, residual);
+	} else if (tdc->dma_desc && tdc->dma_desc->vd.tx.cookie == cookie) {
+		dma_desc = tdc->dma_desc;
+		residual =  dma_desc->bytes_requested -
+					(dma_desc->bytes_transferred %
+					dma_desc->bytes_requested);
+		dma_set_residue(txstate, residual);
+	} else {
+		dev_err(tdc2dev(tdc), "cookie %d is not found\n", cookie);
+	}
+
+	return ret;
+}
+
+static inline int get_bus_width(struct tegra_dma_channel *tdc,
+				enum dma_slave_buswidth slave_bw)
+{
+	switch (slave_bw) {
+	case DMA_SLAVE_BUSWIDTH_1_BYTE:
+		return TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH_8;
+	case DMA_SLAVE_BUSWIDTH_2_BYTES:
+		return TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH_16;
+	case DMA_SLAVE_BUSWIDTH_4_BYTES:
+		return TEGRA_GPCDMA_MMIOSEQ_BUS_WIDTH_32;
+	default:
+		dev_err(tdc2dev(tdc), "given slave bus width is not supported\n");
+		return -EINVAL;
+	}
+}
+
+static unsigned int get_burst_size(struct tegra_dma_channel *tdc,
+				   u32 burst_size, enum dma_slave_buswidth slave_bw,
+				   int len)
+{
+	unsigned int burst_mmio_width, burst_byte;
+
+	/*
+	 * burst_size from client is in terms of the bus_width.
+	 * convert that into words.
+	 * If burst_size is not specified from client, then use
+	 * len to calculate the optimum burst size
+	 */
+	burst_byte = burst_size ? burst_size * slave_bw : len;
+	burst_mmio_width = burst_byte / 4;
+
+	clamp(burst_mmio_width, TEGRA_GPCDMA_MMIOSEQ_BURST_MIN,
+	      TEGRA_GPCDMA_MMIOSEQ_BURST_MAX);
+
+	return (fls(burst_mmio_width) - 1) << TEGRA_GPCDMA_MMIOSEQ_BURST_SHIFT;
+}
+
+static int get_transfer_param(struct tegra_dma_channel *tdc,
+			      enum dma_transfer_direction direction,
+			      unsigned long *apb_addr,
+			      unsigned long *mmio_seq,
+			      unsigned long *csr,
+			      unsigned int *burst_size,
+			      enum dma_slave_buswidth *slave_bw)
+{
+	switch (direction) {
+	case DMA_MEM_TO_DEV:
+		*apb_addr = tdc->dma_sconfig.dst_addr;
+		*mmio_seq = get_bus_width(tdc, tdc->dma_sconfig.dst_addr_width);
+		*burst_size = tdc->dma_sconfig.dst_maxburst;
+		*slave_bw = tdc->dma_sconfig.dst_addr_width;
+		*csr = TEGRA_GPCDMA_CSR_DMA_MEM2IO_FC;
+		return 0;
+	case DMA_DEV_TO_MEM:
+		*apb_addr = tdc->dma_sconfig.src_addr;
+		*mmio_seq = get_bus_width(tdc, tdc->dma_sconfig.src_addr_width);
+		*burst_size = tdc->dma_sconfig.src_maxburst;
+		*slave_bw = tdc->dma_sconfig.src_addr_width;
+		*csr = TEGRA_GPCDMA_CSR_DMA_IO2MEM_FC;
+		return 0;
+	case DMA_MEM_TO_MEM:
+		*burst_size = tdc->dma_sconfig.src_addr_width;
+		*csr = TEGRA_GPCDMA_CSR_DMA_MEM2MEM;
+		return 0;
+	default:
+		dev_err(tdc2dev(tdc), "DMA direction is not supported\n");
+	}
+
+	return -EINVAL;
+}
+
+static struct dma_async_tx_descriptor *
+tegra_dma_prep_dma_memset(struct dma_chan *dc, dma_addr_t dest, int value,
+			  size_t len, unsigned long flags)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	unsigned int max_dma_count = tdc->tdma->chip_data->max_dma_count;
+	struct tegra_dma_sg_req *sg_req;
+	struct tegra_dma_desc *dma_desc;
+	unsigned long csr, mc_seq;
+
+	if ((len & 3) || (dest & 3) || len > max_dma_count) {
+		dev_err(tdc2dev(tdc),
+			"DMA length/memory address is not supported\n");
+		return NULL;
+	}
+
+	/* Set dma mode to fixed pattern */
+	csr = TEGRA_GPCDMA_CSR_DMA_FIXED_PAT;
+	/* Enable once or continuous mode */
+	csr |= TEGRA_GPCDMA_CSR_ONCE;
+	/* Enable IRQ mask */
+	csr |= TEGRA_GPCDMA_CSR_IRQ_MASK;
+	/* Enable the dma interrupt */
+	if (flags & DMA_PREP_INTERRUPT)
+		csr |= TEGRA_GPCDMA_CSR_IE_EOC;
+	/* Configure default priority weight for the channel */
+	csr |= FIELD_PREP(TEGRA_GPCDMA_CSR_WEIGHT, 1);
+
+	mc_seq =  tdc_read(tdc, TEGRA_GPCDMA_CHAN_MCSEQ);
+	/* retain stream-id and clean rest */
+	mc_seq &= TEGRA_GPCDMA_MCSEQ_STREAM_ID0_MASK;
+
+	/* Set the address wrapping */
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP0,
+						TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP1,
+						TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
+
+	/* Program outstanding MC requests */
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_REQ_COUNT, 1);
+	/* Set burst size */
+	mc_seq |= TEGRA_GPCDMA_MCSEQ_BURST_16;
+
+	dma_desc = kzalloc(struct_size(dma_desc, sg_req, 1), GFP_NOWAIT);
+	if (!dma_desc)
+		return NULL;
+
+	dma_desc->bytes_requested = len;
+	dma_desc->sg_count = 1;
+	sg_req = dma_desc->sg_req;
+
+	sg_req[0].ch_regs.src_ptr = 0;
+	sg_req[0].ch_regs.dst_ptr = dest;
+	sg_req[0].ch_regs.high_addr_ptr =
+			FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
+	sg_req[0].ch_regs.fixed_pattern = value;
+	/* Word count reg takes value as (N +1) words */
+	sg_req[0].ch_regs.wcount = ((len - 4) >> 2);
+	sg_req[0].ch_regs.csr = csr;
+	sg_req[0].ch_regs.mmio_seq = 0;
+	sg_req[0].ch_regs.mc_seq = mc_seq;
+	sg_req[0].len = len;
+
+	dma_desc->cyclic = false;
+	return vchan_tx_prep(&tdc->vc, &dma_desc->vd, flags);
+}
+
+static struct dma_async_tx_descriptor *
+tegra_dma_prep_dma_memcpy(struct dma_chan *dc, dma_addr_t dest,
+			  dma_addr_t src, size_t len, unsigned long flags)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	unsigned int max_dma_count = tdc->tdma->chip_data->max_dma_count;
+	struct tegra_dma_sg_req *sg_req;
+	struct tegra_dma_desc *dma_desc;
+	unsigned long csr, mc_seq;
+
+	if ((len & 3) || (src & 3) || (dest & 3) || len > max_dma_count) {
+		dev_err(tdc2dev(tdc),
+			"DMA length/memory address is not supported\n");
+		return NULL;
+	}
+
+	/* Set dma mode to memory to memory transfer */
+	csr = TEGRA_GPCDMA_CSR_DMA_MEM2MEM;
+	/* Enable once or continuous mode */
+	csr |= TEGRA_GPCDMA_CSR_ONCE;
+	/* Enable IRQ mask */
+	csr |= TEGRA_GPCDMA_CSR_IRQ_MASK;
+	/* Enable the dma interrupt */
+	if (flags & DMA_PREP_INTERRUPT)
+		csr |= TEGRA_GPCDMA_CSR_IE_EOC;
+	/* Configure default priority weight for the channel */
+	csr |= FIELD_PREP(TEGRA_GPCDMA_CSR_WEIGHT, 1);
+
+	mc_seq =  tdc_read(tdc, TEGRA_GPCDMA_CHAN_MCSEQ);
+	/* retain stream-id and clean rest */
+	mc_seq &= (TEGRA_GPCDMA_MCSEQ_STREAM_ID0_MASK) |
+		  (TEGRA_GPCDMA_MCSEQ_STREAM_ID1_MASK);
+
+	/* Set the address wrapping */
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP0,
+			     TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP1,
+			     TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
+
+	/* Program outstanding MC requests */
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_REQ_COUNT, 1);
+	/* Set burst size */
+	mc_seq |= TEGRA_GPCDMA_MCSEQ_BURST_16;
+
+	dma_desc = kzalloc(struct_size(dma_desc, sg_req, 1), GFP_NOWAIT);
+	if (!dma_desc)
+		return NULL;
+
+	dma_desc->bytes_requested = len;
+	dma_desc->sg_count = 1;
+	sg_req = dma_desc->sg_req;
+
+	sg_req[0].ch_regs.src_ptr = src;
+	sg_req[0].ch_regs.dst_ptr = dest;
+	sg_req[0].ch_regs.high_addr_ptr =
+		FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (src >> 32));
+	sg_req[0].ch_regs.high_addr_ptr |=
+		FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
+	/* Word count reg takes value as (N +1) words */
+	sg_req[0].ch_regs.wcount = ((len - 4) >> 2);
+	sg_req[0].ch_regs.csr = csr;
+	sg_req[0].ch_regs.mmio_seq = 0;
+	sg_req[0].ch_regs.mc_seq = mc_seq;
+	sg_req[0].len = len;
+
+	dma_desc->cyclic = false;
+	return vchan_tx_prep(&tdc->vc, &dma_desc->vd, flags);
+}
+
+static struct dma_async_tx_descriptor *
+tegra_dma_prep_slave_sg(struct dma_chan *dc, struct scatterlist *sgl,
+			unsigned int sg_len, enum dma_transfer_direction direction,
+			unsigned long flags, void *context)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	unsigned int max_dma_count = tdc->tdma->chip_data->max_dma_count;
+	unsigned long csr, mc_seq, apb_ptr = 0, mmio_seq = 0;
+	enum dma_slave_buswidth slave_bw;
+	struct tegra_dma_sg_req *sg_req;
+	struct tegra_dma_desc *dma_desc;
+	struct scatterlist *sg;
+	u32 burst_size;
+	unsigned int i;
+	int ret;
+
+	if (!tdc->config_init) {
+		dev_err(tdc2dev(tdc), "DMA channel is not configured\n");
+		return NULL;
+	}
+	if (sg_len < 1) {
+		dev_err(tdc2dev(tdc), "Invalid segment length %d\n", sg_len);
+		return NULL;
+	}
+
+	ret = tegra_dma_sid_reserve(tdc, direction);
+	if (ret)
+		return NULL;
+
+	ret = get_transfer_param(tdc, direction, &apb_ptr, &mmio_seq, &csr,
+				 &burst_size, &slave_bw);
+	if (ret < 0)
+		return NULL;
+
+	/* Enable once or continuous mode */
+	csr |= TEGRA_GPCDMA_CSR_ONCE;
+	/* Program the slave id in requestor select */
+	csr |= FIELD_PREP(TEGRA_GPCDMA_CSR_REQ_SEL_MASK, tdc->slave_id);
+	/* Enable IRQ mask */
+	csr |= TEGRA_GPCDMA_CSR_IRQ_MASK;
+	/* Configure default priority weight for the channel*/
+	csr |= FIELD_PREP(TEGRA_GPCDMA_CSR_WEIGHT, 1);
+
+	/* Enable the dma interrupt */
+	if (flags & DMA_PREP_INTERRUPT)
+		csr |= TEGRA_GPCDMA_CSR_IE_EOC;
+
+	mc_seq =  tdc_read(tdc, TEGRA_GPCDMA_CHAN_MCSEQ);
+	/* retain stream-id and clean rest */
+	mc_seq &= TEGRA_GPCDMA_MCSEQ_STREAM_ID0_MASK;
+
+	/* Set the address wrapping on both MC and MMIO side */
+
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP0,
+			     TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP1,
+			     TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
+	mmio_seq |= FIELD_PREP(TEGRA_GPCDMA_MMIOSEQ_WRAP_WORD, 1);
+
+	/* Program 2 MC outstanding requests by default. */
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_REQ_COUNT, 1);
+
+	/* Setting MC burst size depending on MMIO burst size */
+	if (burst_size == 64)
+		mc_seq |= TEGRA_GPCDMA_MCSEQ_BURST_16;
+	else
+		mc_seq |= TEGRA_GPCDMA_MCSEQ_BURST_2;
+
+	dma_desc = kzalloc(struct_size(dma_desc, sg_req, 1), GFP_NOWAIT);
+	if (!dma_desc)
+		return NULL;
+
+	dma_desc->sg_count = sg_len;
+	sg_req = dma_desc->sg_req;
+
+	/* Make transfer requests */
+	for_each_sg(sgl, sg, sg_len, i) {
+		u32 len;
+		dma_addr_t mem;
+
+		mem = sg_dma_address(sg);
+		len = sg_dma_len(sg);
+
+		if ((len & 3) || (mem & 3) || len > max_dma_count) {
+			dev_err(tdc2dev(tdc),
+				"DMA length/memory address is not supported\n");
+			kfree(dma_desc);
+			return NULL;
+		}
+
+		mmio_seq |= get_burst_size(tdc, burst_size, slave_bw, len);
+		dma_desc->bytes_requested += len;
+
+		if (direction == DMA_MEM_TO_DEV) {
+			sg_req[i].ch_regs.src_ptr = mem;
+			sg_req[i].ch_regs.dst_ptr = apb_ptr;
+			sg_req[i].ch_regs.high_addr_ptr =
+				FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (mem >> 32));
+		} else if (direction == DMA_DEV_TO_MEM) {
+			sg_req[i].ch_regs.src_ptr = apb_ptr;
+			sg_req[i].ch_regs.dst_ptr = mem;
+			sg_req[i].ch_regs.high_addr_ptr =
+				FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (mem >> 32));
+		}
+
+		/*
+		 * Word count register takes input in words. Writing a value
+		 * of N into word count register means a req of (N+1) words.
+		 */
+		sg_req[i].ch_regs.wcount = ((len - 4) >> 2);
+		sg_req[i].ch_regs.csr = csr;
+		sg_req[i].ch_regs.mmio_seq = mmio_seq;
+		sg_req[i].ch_regs.mc_seq = mc_seq;
+		sg_req[i].len = len;
+	}
+
+	dma_desc->cyclic = false;
+	return vchan_tx_prep(&tdc->vc, &dma_desc->vd, flags);
+}
+
+static struct dma_async_tx_descriptor *
+tegra_dma_prep_dma_cyclic(struct dma_chan *dc, dma_addr_t buf_addr, size_t buf_len,
+			  size_t period_len, enum dma_transfer_direction direction,
+			  unsigned long flags)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	unsigned int max_dma_count = tdc->tdma->chip_data->max_dma_count;
+	struct tegra_dma_desc *dma_desc;
+	struct tegra_dma_sg_req *sg_req;
+	unsigned long csr, mc_seq, apb_ptr = 0, mmio_seq = 0;
+	unsigned int len, period_count, i;
+	dma_addr_t mem = buf_addr;
+	u32 burst_size;
+	enum dma_slave_buswidth slave_bw;
+	int ret;
+
+	if (!buf_len || !period_len) {
+		dev_err(tdc2dev(tdc), "Invalid buffer/period len\n");
+		return NULL;
+	}
+
+	if (!tdc->config_init) {
+		dev_err(tdc2dev(tdc), "DMA slave is not configured\n");
+		return NULL;
+	}
+
+	ret = tegra_dma_sid_reserve(tdc, direction);
+	if (ret)
+		return NULL;
+
+	/*
+	 * We only support cycle transfer when buf_len is multiple of
+	 * period_len.
+	 */
+	if (buf_len % period_len) {
+		dev_err(tdc2dev(tdc), "buf_len is not multiple of period_len\n");
+		return NULL;
+	}
+
+	len = period_len;
+	if ((len & 3) || (buf_addr & 3) || len > max_dma_count) {
+		dev_err(tdc2dev(tdc), "Req len/mem address is not correct\n");
+		return NULL;
+	}
+
+	ret = get_transfer_param(tdc, direction, &apb_ptr, &mmio_seq, &csr,
+				 &burst_size, &slave_bw);
+	if (ret < 0)
+		return NULL;
+
+	/* Enable once or continuous mode */
+	csr &= ~TEGRA_GPCDMA_CSR_ONCE;
+	/* Program the slave id in requestor select */
+	csr |= FIELD_PREP(TEGRA_GPCDMA_CSR_REQ_SEL_MASK, tdc->slave_id);
+	/* Enable IRQ mask */
+	csr |= TEGRA_GPCDMA_CSR_IRQ_MASK;
+	/* Configure default priority weight for the channel*/
+	csr |= FIELD_PREP(TEGRA_GPCDMA_CSR_WEIGHT, 1);
+
+	/* Enable the dma interrupt */
+	if (flags & DMA_PREP_INTERRUPT)
+		csr |= TEGRA_GPCDMA_CSR_IE_EOC;
+
+	mmio_seq |= FIELD_PREP(TEGRA_GPCDMA_MMIOSEQ_WRAP_WORD, 1);
+
+	mc_seq =  tdc_read(tdc, TEGRA_GPCDMA_CHAN_MCSEQ);
+	/* retain stream-id and clean rest */
+	mc_seq &= TEGRA_GPCDMA_MCSEQ_STREAM_ID0_MASK;
+
+	/* Set the address wrapping on both MC and MMIO side */
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP0,
+			     TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP1,
+			     TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
+
+	/* Program 2 MC outstanding requests by default. */
+	mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_REQ_COUNT, 1);
+	/* Setting MC burst size depending on MMIO burst size */
+	if (burst_size == 64)
+		mc_seq |= TEGRA_GPCDMA_MCSEQ_BURST_16;
+	else
+		mc_seq |= TEGRA_GPCDMA_MCSEQ_BURST_2;
+
+	period_count = buf_len / period_len;
+	dma_desc = kzalloc(struct_size(dma_desc, sg_req, period_count),
+			   GFP_NOWAIT);
+	if (!dma_desc)
+		return NULL;
+
+	dma_desc->bytes_requested = buf_len;
+	dma_desc->sg_count = period_count;
+
+	/* Split transfer equal to period size */
+	for (i = 0; i < period_count; i++) {
+		mmio_seq |= get_burst_size(tdc, burst_size, slave_bw, len);
+		if (direction == DMA_MEM_TO_DEV) {
+			sg_req[i].ch_regs.src_ptr = mem;
+			sg_req[i].ch_regs.dst_ptr = apb_ptr;
+			sg_req[i].ch_regs.high_addr_ptr =
+				FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (mem >> 32));
+		} else if (direction == DMA_DEV_TO_MEM) {
+			sg_req[i].ch_regs.src_ptr = apb_ptr;
+			sg_req[i].ch_regs.dst_ptr = mem;
+			sg_req[i].ch_regs.high_addr_ptr =
+				FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (mem >> 32));
+		}
+		/*
+		 * Word count register takes input in words. Writing a value
+		 * of N into word count register means a req of (N+1) words.
+		 */
+		sg_req[i].ch_regs.wcount = ((len - 4) >> 2);
+		sg_req[i].ch_regs.csr = csr;
+		sg_req[i].ch_regs.mmio_seq = mmio_seq;
+		sg_req[i].ch_regs.mc_seq = mc_seq;
+		sg_req[i].len = len;
+
+		mem += len;
+	}
+
+	dma_desc->cyclic = true;
+
+	return vchan_tx_prep(&tdc->vc, &dma_desc->vd, flags);
+}
+
+static int tegra_dma_alloc_chan_resources(struct dma_chan *dc)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+	int ret;
+
+	ret = request_irq(tdc->irq, tegra_dma_isr, 0, tdc->name, tdc);
+	if (ret) {
+		dev_err(tdc2dev(tdc), "request_irq failed for %s\n", tdc->name);
+		return ret;
+	}
+
+	dma_cookie_init(&tdc->vc.chan);
+	tdc->config_init = false;
+	return 0;
+}
+
+static void tegra_dma_chan_synchronize(struct dma_chan *dc)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+
+	vchan_synchronize(&tdc->vc);
+}
+
+static void tegra_dma_free_chan_resources(struct dma_chan *dc)
+{
+	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+
+	dev_dbg(tdc2dev(tdc), "Freeing channel %d\n", tdc->id);
+
+	tegra_dma_terminate_all(dc);
+
+	tasklet_kill(&tdc->vc.task);
+	tdc->config_init = false;
+	tdc->slave_id = -1;
+	tdc->sid_dir = DMA_TRANS_NONE;
+	free_irq(tdc->irq, tdc);
+	vchan_free_chan_resources(&tdc->vc);
+}
+
+static struct dma_chan *tegra_dma_of_xlate(struct of_phandle_args *dma_spec,
+					   struct of_dma *ofdma)
+{
+	struct tegra_dma *tdma = ofdma->of_dma_data;
+	struct tegra_dma_channel *tdc;
+	struct dma_chan *chan;
+
+	chan = dma_get_any_slave_channel(&tdma->dma_dev);
+	if (!chan)
+		return NULL;
+
+	tdc = to_tegra_dma_chan(chan);
+	tdc->slave_id = dma_spec->args[0];
+
+	return chan;
+}
+
+static const struct tegra_dma_chip_data tegra186_dma_chip_data = {
+	.nr_channels = 31,
+	.channel_reg_size = SZ_64K,
+	.max_dma_count = SZ_1G,
+	.hw_support_pause = false,
+};
+
+static const struct tegra_dma_chip_data tegra194_dma_chip_data = {
+	.nr_channels = 31,
+	.channel_reg_size = SZ_64K,
+	.max_dma_count = SZ_1G,
+	.hw_support_pause = true,
+};
+
+static const struct of_device_id tegra_dma_of_match[] = {
+	{
+		.compatible = "nvidia,tegra186-gpcdma",
+		.data = &tegra186_dma_chip_data,
+	}, {
+		.compatible = "nvidia,tegra194-gpcdma",
+		.data = &tegra194_dma_chip_data,
+	}, {
+	},
+};
+MODULE_DEVICE_TABLE(of, tegra_dma_of_match);
+
+static int tegra_dma_program_sid(struct tegra_dma_channel *tdc,
+				 int chan, int stream_id)
+{
+	unsigned int reg_val =  tdc_read(tdc, TEGRA_GPCDMA_CHAN_MCSEQ);
+
+	reg_val &= ~(TEGRA_GPCDMA_MCSEQ_STREAM_ID0_MASK);
+	reg_val &= ~(TEGRA_GPCDMA_MCSEQ_STREAM_ID1_MASK);
+
+	reg_val |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_STREAM_ID0_MASK, stream_id);
+	reg_val |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_STREAM_ID1_MASK, stream_id);
+
+	tdc_write(tdc, TEGRA_GPCDMA_CHAN_MCSEQ, reg_val);
+	return 0;
+}
+
+static int tegra_dma_probe(struct platform_device *pdev)
+{
+	const struct tegra_dma_chip_data *cdata = NULL;
+	struct iommu_fwspec *iommu_spec;
+	unsigned int stream_id, i;
+	struct tegra_dma *tdma;
+	struct resource	*res;
+	int ret;
+
+	cdata = of_device_get_match_data(&pdev->dev);
+
+	tdma = devm_kzalloc(&pdev->dev, sizeof(*tdma) + cdata->nr_channels *
+			sizeof(struct tegra_dma_channel), GFP_KERNEL);
+	if (!tdma)
+		return -ENOMEM;
+
+	tdma->dev = &pdev->dev;
+	tdma->chip_data = cdata;
+	platform_set_drvdata(pdev, tdma);
+
+	tdma->base_addr = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(tdma->base_addr))
+		return PTR_ERR(tdma->base_addr);
+
+	tdma->rst = devm_reset_control_get_exclusive(&pdev->dev, "gpcdma");
+	if (IS_ERR(tdma->rst)) {
+		return dev_err_probe(&pdev->dev, PTR_ERR(tdma->rst),
+			      "Missing controller reset\n");
+	}
+	reset_control_reset(tdma->rst);
+
+	tdma->dma_dev.dev = &pdev->dev;
+
+	iommu_spec = dev_iommu_fwspec_get(&pdev->dev);
+	if (!iommu_spec) {
+		dev_err(&pdev->dev, "Missing iommu stream-id\n");
+		return -EINVAL;
+	}
+	stream_id = iommu_spec->ids[0] & 0xffff;
+
+	INIT_LIST_HEAD(&tdma->dma_dev.channels);
+	for (i = 0; i < cdata->nr_channels; i++) {
+		struct tegra_dma_channel *tdc = &tdma->channels[i];
+
+		tdc->chan_base_offset = TEGRA_GPCDMA_CHANNEL_BASE_ADD_OFFSET +
+					i * cdata->channel_reg_size;
+		res = platform_get_resource(pdev, IORESOURCE_IRQ, i);
+		if (!res) {
+			dev_err(&pdev->dev, "No irq resource for chan %d\n", i);
+			return -EINVAL;
+		}
+		tdc->irq = res->start;
+		snprintf(tdc->name, sizeof(tdc->name), "gpcdma.%d", i);
+
+		tdc->tdma = tdma;
+		tdc->id = i;
+		tdc->slave_id = -1;
+
+		vchan_init(&tdc->vc, &tdma->dma_dev);
+		tdc->vc.desc_free = tegra_dma_desc_free;
+
+		/* program stream-id for this channel */
+		tegra_dma_program_sid(tdc, i, stream_id);
+		tdc->stream_id = stream_id;
+	}
+
+	dma_cap_set(DMA_SLAVE, tdma->dma_dev.cap_mask);
+	dma_cap_set(DMA_PRIVATE, tdma->dma_dev.cap_mask);
+	dma_cap_set(DMA_MEMCPY, tdma->dma_dev.cap_mask);
+	dma_cap_set(DMA_MEMSET, tdma->dma_dev.cap_mask);
+	dma_cap_set(DMA_CYCLIC, tdma->dma_dev.cap_mask);
+
+	/*
+	 * Only word aligned transfers are supported. Set the copy
+	 * alignment shift.
+	 */
+	tdma->dma_dev.copy_align = 2;
+	tdma->dma_dev.fill_align = 2;
+	tdma->dma_dev.device_alloc_chan_resources =
+					tegra_dma_alloc_chan_resources;
+	tdma->dma_dev.device_free_chan_resources =
+					tegra_dma_free_chan_resources;
+	tdma->dma_dev.device_prep_slave_sg = tegra_dma_prep_slave_sg;
+	tdma->dma_dev.device_prep_dma_memcpy = tegra_dma_prep_dma_memcpy;
+	tdma->dma_dev.device_prep_dma_memset = tegra_dma_prep_dma_memset;
+	tdma->dma_dev.device_prep_dma_cyclic = tegra_dma_prep_dma_cyclic;
+	tdma->dma_dev.device_config = tegra_dma_slave_config;
+	tdma->dma_dev.device_terminate_all = tegra_dma_terminate_all;
+	tdma->dma_dev.device_tx_status = tegra_dma_tx_status;
+	tdma->dma_dev.device_issue_pending = tegra_dma_issue_pending;
+	tdma->dma_dev.device_pause = tegra_dma_pause;
+	tdma->dma_dev.device_resume = tegra_dma_resume;
+	tdma->dma_dev.device_synchronize = tegra_dma_chan_synchronize;
+	tdma->dma_dev.residue_granularity = DMA_RESIDUE_GRANULARITY_BURST;
+
+	ret = dma_async_device_register(&tdma->dma_dev);
+	if (ret < 0) {
+		dev_err_probe(&pdev->dev, ret,
+			      "GPC DMA driver registration failed\n");
+		return ret;
+	}
+
+	ret = of_dma_controller_register(pdev->dev.of_node,
+					 tegra_dma_of_xlate, tdma);
+	if (ret < 0) {
+		dev_err_probe(&pdev->dev, ret,
+			      "GPC DMA OF registration failed\n");
+
+		dma_async_device_unregister(&tdma->dma_dev);
+		return ret;
+	}
+
+	dev_info(&pdev->dev, "GPC DMA driver register %d channels\n",
+		 cdata->nr_channels);
+
+	return 0;
+}
+
+static int tegra_dma_remove(struct platform_device *pdev)
+{
+	struct tegra_dma *tdma = platform_get_drvdata(pdev);
+
+	of_dma_controller_free(pdev->dev.of_node);
+	dma_async_device_unregister(&tdma->dma_dev);
+
+	return 0;
+}
+
+static int __maybe_unused tegra_dma_pm_suspend(struct device *dev)
+{
+	struct tegra_dma *tdma = dev_get_drvdata(dev);
+	unsigned int i;
+
+	for (i = 0; i < tdma->chip_data->nr_channels; i++) {
+		struct tegra_dma_channel *tdc = &tdma->channels[i];
+
+		if (tdc->dma_desc) {
+			dev_err(tdma->dev, "channel %u busy\n", i);
+			return -EBUSY;
+		}
+	}
+
+	return 0;
+}
+
+static int __maybe_unused tegra_dma_pm_resume(struct device *dev)
+{
+	struct tegra_dma *tdma = dev_get_drvdata(dev);
+	unsigned int i;
+
+	reset_control_reset(tdma->rst);
+
+	for (i = 0; i < tdma->chip_data->nr_channels; i++) {
+		struct tegra_dma_channel *tdc = &tdma->channels[i];
+
+		tegra_dma_program_sid(tdc, i, tdc->stream_id);
+	}
+
+	return 0;
+}
+
+static const struct __maybe_unused dev_pm_ops tegra_dma_dev_pm_ops = {
+	SET_SYSTEM_SLEEP_PM_OPS(tegra_dma_pm_suspend, tegra_dma_pm_resume)
+};
+
+static struct platform_driver tegra_dmac_driver = {
+	.driver = {
+		.name	= "tegra-gpcdma",
+		.pm	= &tegra_dma_dev_pm_ops,
+		.of_match_table = tegra_dma_of_match,
+	},
+	.probe		= tegra_dma_probe,
+	.remove		= tegra_dma_remove,
+};
+
+module_platform_driver(tegra_dmac_driver);
+
+MODULE_DESCRIPTION("NVIDIA Tegra GPC DMA Controller driver");
+MODULE_AUTHOR("Pavan Kunapuli <pkunapuli@nvidia.com>");
+MODULE_AUTHOR("Rajesh Gumasta <rgumasta@nvidia.com>");
+MODULE_LICENSE("GPL v2");
-- 
2.7.4


^ permalink raw reply related	[flat|nested] 17+ messages in thread

* [PATCH v16 3/4] arm64: defconfig: tegra: Enable GPCDMA
  2022-01-10 16:05 [PATCH v16 0/4] Add NVIDIA Tegra GPC-DMA driver Akhil R
  2022-01-10 16:05 ` [PATCH v16 1/4] dt-bindings: dmaengine: Add doc for tegra gpcdma Akhil R
  2022-01-10 16:05 ` [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver Akhil R
@ 2022-01-10 16:05 ` Akhil R
  2022-01-10 16:05 ` [PATCH v16 4/4] arm64: tegra: Add GPCDMA node for tegra186 and tegra194 Akhil R
  3 siblings, 0 replies; 17+ messages in thread
From: Akhil R @ 2022-01-10 16:05 UTC (permalink / raw)
  To: dan.j.williams, devicetree, dmaengine, jonathanh, kyarlagadda,
	ldewangan, linux-kernel, linux-tegra, p.zabel, rgumasta, robh+dt,
	thierry.reding, vkoul
  Cc: akhilrajeev

Enable TEGRA_GPC_DMA in defconfig for Tegra186 and Tegra196 gpc
dma controller driver

Signed-off-by: Rajesh Gumasta <rgumasta@nvidia.com>
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
---
 arch/arm64/configs/defconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
index ee4bd77..96a796d 100644
--- a/arch/arm64/configs/defconfig
+++ b/arch/arm64/configs/defconfig
@@ -956,6 +956,7 @@ CONFIG_MV_XOR=y
 CONFIG_MV_XOR_V2=y
 CONFIG_OWL_DMA=y
 CONFIG_PL330_DMA=y
+CONFIG_TEGRA186_GPC_DMA=m
 CONFIG_TEGRA20_APB_DMA=y
 CONFIG_TEGRA210_ADMA=m
 CONFIG_QCOM_BAM_DMA=y
-- 
2.7.4


^ permalink raw reply related	[flat|nested] 17+ messages in thread

* [PATCH v16 4/4] arm64: tegra: Add GPCDMA node for tegra186 and tegra194
  2022-01-10 16:05 [PATCH v16 0/4] Add NVIDIA Tegra GPC-DMA driver Akhil R
                   ` (2 preceding siblings ...)
  2022-01-10 16:05 ` [PATCH v16 3/4] arm64: defconfig: tegra: Enable GPCDMA Akhil R
@ 2022-01-10 16:05 ` Akhil R
  3 siblings, 0 replies; 17+ messages in thread
From: Akhil R @ 2022-01-10 16:05 UTC (permalink / raw)
  To: dan.j.williams, devicetree, dmaengine, jonathanh, kyarlagadda,
	ldewangan, linux-kernel, linux-tegra, p.zabel, rgumasta, robh+dt,
	thierry.reding, vkoul
  Cc: akhilrajeev

Add device tree node for GPCDMA controller on Tegra186 target
and Tegra194 target.

Signed-off-by: Rajesh Gumasta <rgumasta@nvidia.com>
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
---
 arch/arm64/boot/dts/nvidia/tegra186.dtsi | 42 +++++++++++++++++++++++++++++++
 arch/arm64/boot/dts/nvidia/tegra194.dtsi | 43 ++++++++++++++++++++++++++++++++
 2 files changed, 85 insertions(+)

diff --git a/arch/arm64/boot/dts/nvidia/tegra186.dtsi b/arch/arm64/boot/dts/nvidia/tegra186.dtsi
index c91afff..6bec977 100644
--- a/arch/arm64/boot/dts/nvidia/tegra186.dtsi
+++ b/arch/arm64/boot/dts/nvidia/tegra186.dtsi
@@ -73,6 +73,48 @@
 		snps,rxpbl = <8>;
 	};
 
+	dma-controller@2600000 {
+		compatible = "nvidia,tegra186-gpcdma";
+		reg = <0x0 0x2600000 0x0 0x210000>;
+		resets = <&bpmp TEGRA186_RESET_GPCDMA>;
+		reset-names = "gpcdma";
+		interrupts = <GIC_SPI 76 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 77 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 78 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 79 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 80 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 81 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 82 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 83 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 84 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 85 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 86 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 87 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 88 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 89 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 90 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 91 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 92 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 93 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 94 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 95 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 96 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 97 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 98 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 99 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 100 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 101 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 102 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 103 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 104 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 105 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 106 IRQ_TYPE_LEVEL_HIGH>;
+		#dma-cells = <1>;
+		iommus = <&smmu TEGRA186_SID_GPCDMA_0>;
+		dma-coherent;
+		status = "okay";
+	};
+
 	aconnect@2900000 {
 		compatible = "nvidia,tegra186-aconnect",
 			     "nvidia,tegra210-aconnect";
diff --git a/arch/arm64/boot/dts/nvidia/tegra194.dtsi b/arch/arm64/boot/dts/nvidia/tegra194.dtsi
index 3c4acfc..491a02f 100644
--- a/arch/arm64/boot/dts/nvidia/tegra194.dtsi
+++ b/arch/arm64/boot/dts/nvidia/tegra194.dtsi
@@ -115,6 +115,49 @@
 			snps,rxpbl = <8>;
 		};
 
+		dma-controller@2600000 {
+			compatible = "nvidia,tegra194-gpcdma",
+				     "nvidia,tegra186-gpcdma";
+			reg = <0x2600000 0x210000>;
+			resets = <&bpmp TEGRA194_RESET_GPCDMA>;
+			reset-names = "gpcdma";
+			interrupts = <GIC_SPI 76 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 77 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 78 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 79 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 80 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 81 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 82 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 83 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 84 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 85 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 86 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 87 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 88 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 89 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 90 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 91 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 92 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 93 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 94 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 95 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 96 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 97 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 98 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 99 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 100 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 101 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 102 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 103 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 104 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 105 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 106 IRQ_TYPE_LEVEL_HIGH>;
+			#dma-cells = <1>;
+			iommus = <&smmu TEGRA194_SID_GPCDMA_0>;
+			dma-coherent;
+			status = "okay";
+		};
+
 		aconnect@2900000 {
 			compatible = "nvidia,tegra194-aconnect",
 				     "nvidia,tegra210-aconnect";
-- 
2.7.4


^ permalink raw reply related	[flat|nested] 17+ messages in thread

* Re: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-10 16:05 ` [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver Akhil R
@ 2022-01-10 20:31   ` kernel test robot
  2022-01-14 11:57   ` Dmitry Osipenko
  1 sibling, 0 replies; 17+ messages in thread
From: kernel test robot @ 2022-01-10 20:31 UTC (permalink / raw)
  To: kbuild-all

[-- Attachment #1: Type: text/plain, Size: 14112 bytes --]

Hi Akhil,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on vkoul-dmaengine/next]
[also build test WARNING on robh/for-next arm64/for-next/core v5.16 next-20220110]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Akhil-R/Add-NVIDIA-Tegra-GPC-DMA-driver/20220111-000734
base:   https://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine.git next
config: parisc-allyesconfig (https://download.01.org/0day-ci/archive/20220111/202201110405.jnZaAyE4-lkp(a)intel.com/config)
compiler: hppa-linux-gcc (GCC) 11.2.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/c8498ed60baaec5b689dbe5d3882342c88981eb3
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Akhil-R/Add-NVIDIA-Tegra-GPC-DMA-driver/20220111-000734
        git checkout c8498ed60baaec5b689dbe5d3882342c88981eb3
        # save the config file to linux build tree
        mkdir build_dir
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross O=build_dir ARCH=parisc SHELL=/bin/bash drivers/dma/

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All warnings (new ones prefixed by >>):

   In file included from <command-line>:
   drivers/dma/tegra186-gpc-dma.c: In function 'tegra_dma_prep_dma_memset':
>> drivers/dma/tegra186-gpc-dma.c:861:74: warning: right shift count >= width of type [-Wshift-count-overflow]
     861 |                         FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
         |                                                                          ^~
   include/linux/compiler_types.h:315:23: note: in definition of macro '__compiletime_assert'
     315 |                 if (!(condition))                                       \
         |                       ^~~~~~~~~
   include/linux/compiler_types.h:335:9: note: in expansion of macro '_compiletime_assert'
     335 |         _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
         |         ^~~~~~~~~~~~~~~~~~~
   include/linux/build_bug.h:39:37: note: in expansion of macro 'compiletime_assert'
      39 | #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
         |                                     ^~~~~~~~~~~~~~~~~~
   include/linux/bitfield.h:49:17: note: in expansion of macro 'BUILD_BUG_ON_MSG'
      49 |                 BUILD_BUG_ON_MSG(__builtin_constant_p(_val) ?           \
         |                 ^~~~~~~~~~~~~~~~
   include/linux/bitfield.h:94:17: note: in expansion of macro '__BF_FIELD_CHECK'
      94 |                 __BF_FIELD_CHECK(_mask, 0ULL, _val, "FIELD_PREP: ");    \
         |                 ^~~~~~~~~~~~~~~~
   drivers/dma/tegra186-gpc-dma.c:861:25: note: in expansion of macro 'FIELD_PREP'
     861 |                         FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
         |                         ^~~~~~~~~~
   In file included from drivers/dma/tegra186-gpc-dma.c:8:
>> drivers/dma/tegra186-gpc-dma.c:861:74: warning: right shift count >= width of type [-Wshift-count-overflow]
     861 |                         FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
         |                                                                          ^~
   include/linux/bitfield.h:95:34: note: in definition of macro 'FIELD_PREP'
      95 |                 ((typeof(_mask))(_val) << __bf_shf(_mask)) & (_mask);   \
         |                                  ^~~~
   In file included from <command-line>:
   drivers/dma/tegra186-gpc-dma.c: In function 'tegra_dma_prep_dma_memcpy':
   drivers/dma/tegra186-gpc-dma.c:929:65: warning: right shift count >= width of type [-Wshift-count-overflow]
     929 |                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (src >> 32));
         |                                                                 ^~
   include/linux/compiler_types.h:315:23: note: in definition of macro '__compiletime_assert'
     315 |                 if (!(condition))                                       \
         |                       ^~~~~~~~~
   include/linux/compiler_types.h:335:9: note: in expansion of macro '_compiletime_assert'
     335 |         _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
         |         ^~~~~~~~~~~~~~~~~~~
   include/linux/build_bug.h:39:37: note: in expansion of macro 'compiletime_assert'
      39 | #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
         |                                     ^~~~~~~~~~~~~~~~~~
   include/linux/bitfield.h:49:17: note: in expansion of macro 'BUILD_BUG_ON_MSG'
      49 |                 BUILD_BUG_ON_MSG(__builtin_constant_p(_val) ?           \
         |                 ^~~~~~~~~~~~~~~~
   include/linux/bitfield.h:94:17: note: in expansion of macro '__BF_FIELD_CHECK'
      94 |                 __BF_FIELD_CHECK(_mask, 0ULL, _val, "FIELD_PREP: ");    \
         |                 ^~~~~~~~~~~~~~~~
   drivers/dma/tegra186-gpc-dma.c:929:17: note: in expansion of macro 'FIELD_PREP'
     929 |                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (src >> 32));
         |                 ^~~~~~~~~~
   In file included from drivers/dma/tegra186-gpc-dma.c:8:
   drivers/dma/tegra186-gpc-dma.c:929:65: warning: right shift count >= width of type [-Wshift-count-overflow]
     929 |                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (src >> 32));
         |                                                                 ^~
   include/linux/bitfield.h:95:34: note: in definition of macro 'FIELD_PREP'
      95 |                 ((typeof(_mask))(_val) << __bf_shf(_mask)) & (_mask);   \
         |                                  ^~~~
   In file included from <command-line>:
   drivers/dma/tegra186-gpc-dma.c:931:66: warning: right shift count >= width of type [-Wshift-count-overflow]
     931 |                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
         |                                                                  ^~
   include/linux/compiler_types.h:315:23: note: in definition of macro '__compiletime_assert'
     315 |                 if (!(condition))                                       \
         |                       ^~~~~~~~~
   include/linux/compiler_types.h:335:9: note: in expansion of macro '_compiletime_assert'
     335 |         _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
         |         ^~~~~~~~~~~~~~~~~~~
   include/linux/build_bug.h:39:37: note: in expansion of macro 'compiletime_assert'
      39 | #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
         |                                     ^~~~~~~~~~~~~~~~~~
   include/linux/bitfield.h:49:17: note: in expansion of macro 'BUILD_BUG_ON_MSG'
      49 |                 BUILD_BUG_ON_MSG(__builtin_constant_p(_val) ?           \
         |                 ^~~~~~~~~~~~~~~~
   include/linux/bitfield.h:94:17: note: in expansion of macro '__BF_FIELD_CHECK'
      94 |                 __BF_FIELD_CHECK(_mask, 0ULL, _val, "FIELD_PREP: ");    \
         |                 ^~~~~~~~~~~~~~~~
   drivers/dma/tegra186-gpc-dma.c:931:17: note: in expansion of macro 'FIELD_PREP'
     931 |                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
         |                 ^~~~~~~~~~
   In file included from drivers/dma/tegra186-gpc-dma.c:8:
   drivers/dma/tegra186-gpc-dma.c:931:66: warning: right shift count >= width of type [-Wshift-count-overflow]
     931 |                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
         |                                                                  ^~
   include/linux/bitfield.h:95:34: note: in definition of macro 'FIELD_PREP'
      95 |                 ((typeof(_mask))(_val) << __bf_shf(_mask)) & (_mask);   \
         |                                  ^~~~
   In file included from <command-line>:
   drivers/dma/tegra186-gpc-dma.c: In function 'tegra_dma_prep_slave_sg':
   drivers/dma/tegra186-gpc-dma.c:1040:81: warning: right shift count >= width of type [-Wshift-count-overflow]
    1040 |                                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (mem >> 32));
         |                                                                                 ^~
   include/linux/compiler_types.h:315:23: note: in definition of macro '__compiletime_assert'
     315 |                 if (!(condition))                                       \
         |                       ^~~~~~~~~
   include/linux/compiler_types.h:335:9: note: in expansion of macro '_compiletime_assert'
     335 |         _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
         |         ^~~~~~~~~~~~~~~~~~~
   include/linux/build_bug.h:39:37: note: in expansion of macro 'compiletime_assert'
      39 | #define BUILD_BUG_ON_MSG(cond, msg) compiletime_assert(!(cond), msg)
         |                                     ^~~~~~~~~~~~~~~~~~
   include/linux/bitfield.h:49:17: note: in expansion of macro 'BUILD_BUG_ON_MSG'
      49 |                 BUILD_BUG_ON_MSG(__builtin_constant_p(_val) ?           \
         |                 ^~~~~~~~~~~~~~~~
   include/linux/bitfield.h:94:17: note: in expansion of macro '__BF_FIELD_CHECK'
      94 |                 __BF_FIELD_CHECK(_mask, 0ULL, _val, "FIELD_PREP: ");    \
         |                 ^~~~~~~~~~~~~~~~
   drivers/dma/tegra186-gpc-dma.c:1040:33: note: in expansion of macro 'FIELD_PREP'
    1040 |                                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (mem >> 32));
         |                                 ^~~~~~~~~~
   In file included from drivers/dma/tegra186-gpc-dma.c:8:
   drivers/dma/tegra186-gpc-dma.c:1040:81: warning: right shift count >= width of type [-Wshift-count-overflow]
    1040 |                                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_SRC_PTR, (mem >> 32));
         |                                                                                 ^~
   include/linux/bitfield.h:95:34: note: in definition of macro 'FIELD_PREP'
      95 |                 ((typeof(_mask))(_val) << __bf_shf(_mask)) & (_mask);   \
         |                                  ^~~~
   In file included from <command-line>:
   drivers/dma/tegra186-gpc-dma.c:1045:81: warning: right shift count >= width of type [-Wshift-count-overflow]
    1045 |                                 FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (mem >> 32));
         |                                                                                 ^~
   include/linux/compiler_types.h:315:23: note: in definition of macro '__compiletime_assert'
     315 |                 if (!(condition))                                       \


vim +861 drivers/dma/tegra186-gpc-dma.c

   806	
   807	static struct dma_async_tx_descriptor *
   808	tegra_dma_prep_dma_memset(struct dma_chan *dc, dma_addr_t dest, int value,
   809				  size_t len, unsigned long flags)
   810	{
   811		struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
   812		unsigned int max_dma_count = tdc->tdma->chip_data->max_dma_count;
   813		struct tegra_dma_sg_req *sg_req;
   814		struct tegra_dma_desc *dma_desc;
   815		unsigned long csr, mc_seq;
   816	
   817		if ((len & 3) || (dest & 3) || len > max_dma_count) {
   818			dev_err(tdc2dev(tdc),
   819				"DMA length/memory address is not supported\n");
   820			return NULL;
   821		}
   822	
   823		/* Set dma mode to fixed pattern */
   824		csr = TEGRA_GPCDMA_CSR_DMA_FIXED_PAT;
   825		/* Enable once or continuous mode */
   826		csr |= TEGRA_GPCDMA_CSR_ONCE;
   827		/* Enable IRQ mask */
   828		csr |= TEGRA_GPCDMA_CSR_IRQ_MASK;
   829		/* Enable the dma interrupt */
   830		if (flags & DMA_PREP_INTERRUPT)
   831			csr |= TEGRA_GPCDMA_CSR_IE_EOC;
   832		/* Configure default priority weight for the channel */
   833		csr |= FIELD_PREP(TEGRA_GPCDMA_CSR_WEIGHT, 1);
   834	
   835		mc_seq =  tdc_read(tdc, TEGRA_GPCDMA_CHAN_MCSEQ);
   836		/* retain stream-id and clean rest */
   837		mc_seq &= TEGRA_GPCDMA_MCSEQ_STREAM_ID0_MASK;
   838	
   839		/* Set the address wrapping */
   840		mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP0,
   841							TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
   842		mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_WRAP1,
   843							TEGRA_GPCDMA_MCSEQ_WRAP_NONE);
   844	
   845		/* Program outstanding MC requests */
   846		mc_seq |= FIELD_PREP(TEGRA_GPCDMA_MCSEQ_REQ_COUNT, 1);
   847		/* Set burst size */
   848		mc_seq |= TEGRA_GPCDMA_MCSEQ_BURST_16;
   849	
   850		dma_desc = kzalloc(struct_size(dma_desc, sg_req, 1), GFP_NOWAIT);
   851		if (!dma_desc)
   852			return NULL;
   853	
   854		dma_desc->bytes_requested = len;
   855		dma_desc->sg_count = 1;
   856		sg_req = dma_desc->sg_req;
   857	
   858		sg_req[0].ch_regs.src_ptr = 0;
   859		sg_req[0].ch_regs.dst_ptr = dest;
   860		sg_req[0].ch_regs.high_addr_ptr =
 > 861				FIELD_PREP(TEGRA_GPCDMA_HIGH_ADDR_DST_PTR, (dest >> 32));
   862		sg_req[0].ch_regs.fixed_pattern = value;
   863		/* Word count reg takes value as (N +1) words */
   864		sg_req[0].ch_regs.wcount = ((len - 4) >> 2);
   865		sg_req[0].ch_regs.csr = csr;
   866		sg_req[0].ch_regs.mmio_seq = 0;
   867		sg_req[0].ch_regs.mc_seq = mc_seq;
   868		sg_req[0].len = len;
   869	
   870		dma_desc->cyclic = false;
   871		return vchan_tx_prep(&tdc->vc, &dma_desc->vd, flags);
   872	}
   873	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-10 16:05 ` [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver Akhil R
  2022-01-10 20:31   ` kernel test robot
@ 2022-01-14 11:57   ` Dmitry Osipenko
  2022-01-17  7:02     ` Akhil R
  1 sibling, 1 reply; 17+ messages in thread
From: Dmitry Osipenko @ 2022-01-14 11:57 UTC (permalink / raw)
  To: Akhil R, dan.j.williams, devicetree, dmaengine, jonathanh,
	kyarlagadda, ldewangan, linux-kernel, linux-tegra, p.zabel,
	rgumasta, robh+dt, thierry.reding, vkoul
  Cc: Pavan Kunapuli

10.01.2022 19:05, Akhil R пишет:
> +static int tegra_dma_terminate_all(struct dma_chan *dc)
> +{
> +	struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> +	unsigned long flags;
> +	LIST_HEAD(head);
> +	int err;
> +
> +	if (tdc->dma_desc) {

Needs locking protection against racing with the interrupt handler.

^ permalink raw reply	[flat|nested] 17+ messages in thread

* RE: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-14 11:57   ` Dmitry Osipenko
@ 2022-01-17  7:02     ` Akhil R
  2022-01-17 15:41       ` Dmitry Osipenko
  0 siblings, 1 reply; 17+ messages in thread
From: Akhil R @ 2022-01-17  7:02 UTC (permalink / raw)
  To: Dmitry Osipenko, dan.j.williams, devicetree, dmaengine,
	Jonathan Hunter, Krishna Yarlagadda, Laxman Dewangan,
	linux-kernel, linux-tegra, p.zabel, Rajesh Gumasta, robh+dt,
	thierry.reding, vkoul
  Cc: Pavan Kunapuli

> 10.01.2022 19:05, Akhil R пишет:
> > +static int tegra_dma_terminate_all(struct dma_chan *dc)
> > +{
> > +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> > +     unsigned long flags;
> > +     LIST_HEAD(head);
> > +     int err;
> > +
> > +     if (tdc->dma_desc) {
> 
> Needs locking protection against racing with the interrupt handler.
tegra_dma_stop_client() waits for the in-flight transfer 
to complete and prevents any additional transfer to start. 
Wouldn't it manage the race? Do you see any potential issue there?

Thanks,
Akhil

--
nvpublic

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-17  7:02     ` Akhil R
@ 2022-01-17 15:41       ` Dmitry Osipenko
  2022-01-18  5:36         ` Akhil R
  0 siblings, 1 reply; 17+ messages in thread
From: Dmitry Osipenko @ 2022-01-17 15:41 UTC (permalink / raw)
  To: Akhil R, dan.j.williams, devicetree, dmaengine, Jonathan Hunter,
	Krishna Yarlagadda, Laxman Dewangan, linux-kernel, linux-tegra,
	p.zabel, Rajesh Gumasta, robh+dt, thierry.reding, vkoul
  Cc: Pavan Kunapuli

17.01.2022 10:02, Akhil R пишет:
>> 10.01.2022 19:05, Akhil R пишет:
>>> +static int tegra_dma_terminate_all(struct dma_chan *dc)
>>> +{
>>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>>> +     unsigned long flags;
>>> +     LIST_HEAD(head);
>>> +     int err;
>>> +
>>> +     if (tdc->dma_desc) {
>>
>> Needs locking protection against racing with the interrupt handler.
> tegra_dma_stop_client() waits for the in-flight transfer 
> to complete and prevents any additional transfer to start. 
> Wouldn't it manage the race? Do you see any potential issue there?

You should consider interrupt handler like a process running in a
parallel thread. The interrupt handler sets tdc->dma_desc to NULL, hence
you'll get NULL dereference in tegra_dma_stop_client().

^ permalink raw reply	[flat|nested] 17+ messages in thread

* RE: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-17 15:41       ` Dmitry Osipenko
@ 2022-01-18  5:36         ` Akhil R
  2022-01-18 13:48           ` Dmitry Osipenko
  0 siblings, 1 reply; 17+ messages in thread
From: Akhil R @ 2022-01-18  5:36 UTC (permalink / raw)
  To: Dmitry Osipenko, dan.j.williams, devicetree, dmaengine,
	Jonathan Hunter, Krishna Yarlagadda, Laxman Dewangan,
	linux-kernel, linux-tegra, p.zabel, Rajesh Gumasta, robh+dt,
	thierry.reding, vkoul
  Cc: Pavan Kunapuli

> 17.01.2022 10:02, Akhil R пишет:
> >> 10.01.2022 19:05, Akhil R пишет:
> >>> +static int tegra_dma_terminate_all(struct dma_chan *dc)
> >>> +{
> >>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> >>> +     unsigned long flags;
> >>> +     LIST_HEAD(head);
> >>> +     int err;
> >>> +
> >>> +     if (tdc->dma_desc) {
> >>
> >> Needs locking protection against racing with the interrupt handler.
> > tegra_dma_stop_client() waits for the in-flight transfer
> > to complete and prevents any additional transfer to start.
> > Wouldn't it manage the race? Do you see any potential issue there?
> 
> You should consider interrupt handler like a process running in a
> parallel thread. The interrupt handler sets tdc->dma_desc to NULL, hence
> you'll get NULL dereference in tegra_dma_stop_client().

Is it better if I remove the below part from tegra_dma_stop_client() so
that dma_desc is not accessed at all?

+	wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
+	tdc->dma_desc->bytes_transferred +=
+			tdc->dma_desc->bytes_requested - (wcount * 4);

Because I don't see a point in updating the value there. dma_desc is set
to NULL in the next step in terminate_all() anyway.

Thanks,
Akhil

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-18  5:36         ` Akhil R
@ 2022-01-18 13:48           ` Dmitry Osipenko
  2022-01-21 16:24             ` Akhil R
  0 siblings, 1 reply; 17+ messages in thread
From: Dmitry Osipenko @ 2022-01-18 13:48 UTC (permalink / raw)
  To: Akhil R, dan.j.williams, devicetree, dmaengine, Jonathan Hunter,
	Krishna Yarlagadda, Laxman Dewangan, linux-kernel, linux-tegra,
	p.zabel, Rajesh Gumasta, robh+dt, thierry.reding, vkoul
  Cc: Pavan Kunapuli

18.01.2022 08:36, Akhil R пишет:
>> 17.01.2022 10:02, Akhil R пишет:
>>>> 10.01.2022 19:05, Akhil R пишет:
>>>>> +static int tegra_dma_terminate_all(struct dma_chan *dc)
>>>>> +{
>>>>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>>>>> +     unsigned long flags;
>>>>> +     LIST_HEAD(head);
>>>>> +     int err;
>>>>> +
>>>>> +     if (tdc->dma_desc) {
>>>>
>>>> Needs locking protection against racing with the interrupt handler.
>>> tegra_dma_stop_client() waits for the in-flight transfer
>>> to complete and prevents any additional transfer to start.
>>> Wouldn't it manage the race? Do you see any potential issue there?
>>
>> You should consider interrupt handler like a process running in a
>> parallel thread. The interrupt handler sets tdc->dma_desc to NULL, hence
>> you'll get NULL dereference in tegra_dma_stop_client().
> 
> Is it better if I remove the below part from tegra_dma_stop_client() so
> that dma_desc is not accessed at all?
> 
> +	wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
> +	tdc->dma_desc->bytes_transferred +=
> +			tdc->dma_desc->bytes_requested - (wcount * 4);
> 
> Because I don't see a point in updating the value there. dma_desc is set
> to NULL in the next step in terminate_all() anyway.

That isn't going help you much because you also can't release DMA
descriptor while interrupt handler still may be running and using that
descriptor.

^ permalink raw reply	[flat|nested] 17+ messages in thread

* RE: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-18 13:48           ` Dmitry Osipenko
@ 2022-01-21 16:24             ` Akhil R
  2022-01-22 18:31               ` Dmitry Osipenko
  0 siblings, 1 reply; 17+ messages in thread
From: Akhil R @ 2022-01-21 16:24 UTC (permalink / raw)
  To: Dmitry Osipenko, dan.j.williams, devicetree, dmaengine,
	Jonathan Hunter, Krishna Yarlagadda, Laxman Dewangan,
	linux-kernel, linux-tegra, p.zabel, Rajesh Gumasta, robh+dt,
	thierry.reding, vkoul
  Cc: Pavan Kunapuli

> >>>>> +static int tegra_dma_terminate_all(struct dma_chan *dc)
> >>>>> +{
> >>>>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> >>>>> +     unsigned long flags;
> >>>>> +     LIST_HEAD(head);
> >>>>> +     int err;
> >>>>> +
> >>>>> +     if (tdc->dma_desc) {
> >>>>
> >>>> Needs locking protection against racing with the interrupt handler.
> >>> tegra_dma_stop_client() waits for the in-flight transfer
> >>> to complete and prevents any additional transfer to start.
> >>> Wouldn't it manage the race? Do you see any potential issue there?
> >>
> >> You should consider interrupt handler like a process running in a
> >> parallel thread. The interrupt handler sets tdc->dma_desc to NULL, hence
> >> you'll get NULL dereference in tegra_dma_stop_client().
> >
> > Is it better if I remove the below part from tegra_dma_stop_client() so
> > that dma_desc is not accessed at all?
> >
> > +     wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
> > +     tdc->dma_desc->bytes_transferred +=
> > +                     tdc->dma_desc->bytes_requested - (wcount * 4);
> >
> > Because I don't see a point in updating the value there. dma_desc is set
> > to NULL in the next step in terminate_all() anyway.
> 
> That isn't going help you much because you also can't release DMA
> descriptor while interrupt handler still may be running and using that
> descriptor.

Does the below functions look good to resolve the issue, provided
tegra_dma_stop_client() doesn't access dma_desc?

+static int tegra_dma_terminate_all(struct dma_chan *dc)
+{
+       struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
+       unsigned long flags;
+       LIST_HEAD(head);
+       int err;
+
+       err = tegra_dma_stop_client(tdc);
+       if (err)
+               return err;
+
+       tegra_dma_stop(tdc);
+
+       spin_lock_irqsave(&tdc->vc.lock, flags);
+       tegra_dma_sid_free(tdc);
+       tdc->dma_desc = NULL;
+
+       vchan_get_all_descriptors(&tdc->vc, &head);
+       spin_unlock_irqrestore(&tdc->vc.lock, flags);
+
+       vchan_dma_desc_free_list(&tdc->vc, &head);
+
+       return 0;
+}

+static irqreturn_t tegra_dma_isr(int irq, void *dev_id)
+{
+       struct tegra_dma_channel *tdc = dev_id;
+       struct tegra_dma_desc *dma_desc = tdc->dma_desc;
+       struct tegra_dma_sg_req *sg_req;
+       u32 status;
+
+       /* Check channel error status register */
+       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS);
+       if (status) {
+               tegra_dma_chan_decode_error(tdc, status);
+               tegra_dma_dump_chan_regs(tdc);
+               tdc_write(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS, 0xFFFFFFFF);
+       }
+
+       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS);
+       if (!(status & TEGRA_GPCDMA_STATUS_ISE_EOC))
+               return IRQ_HANDLED;
+
+       tdc_write(tdc, TEGRA_GPCDMA_CHAN_STATUS,
+                 TEGRA_GPCDMA_STATUS_ISE_EOC);
+
+       spin_lock(&tdc->vc.lock);
+       if (!dma_desc)
+               goto irq_done;
+
+       sg_req = dma_desc->sg_req;
+       dma_desc->bytes_transferred += sg_req[dma_desc->sg_idx - 1].len;
+
+       if (dma_desc->sg_idx == dma_desc->sg_count) {
+               tegra_dma_xfer_complete(tdc);
+       } else if (dma_desc->cyclic) {
+               vchan_cyclic_callback(&dma_desc->vd);
+               tegra_dma_configure_next_sg(tdc);
+       } else {
+               tegra_dma_start(tdc);
+       }
+
+irq_done:
+       spin_unlock(&tdc->vc.lock);
+       return IRQ_HANDLED;
+}


Thanks,
Akhil

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-21 16:24             ` Akhil R
@ 2022-01-22 18:31               ` Dmitry Osipenko
  2022-01-23 16:49                 ` Akhil R
  0 siblings, 1 reply; 17+ messages in thread
From: Dmitry Osipenko @ 2022-01-22 18:31 UTC (permalink / raw)
  To: Akhil R, dan.j.williams, devicetree, dmaengine, Jonathan Hunter,
	Krishna Yarlagadda, Laxman Dewangan, linux-kernel, linux-tegra,
	p.zabel, Rajesh Gumasta, robh+dt, thierry.reding, vkoul
  Cc: Pavan Kunapuli

21.01.2022 19:24, Akhil R пишет:
>>>>>>> +static int tegra_dma_terminate_all(struct dma_chan *dc)
>>>>>>> +{
>>>>>>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>>>>>>> +     unsigned long flags;
>>>>>>> +     LIST_HEAD(head);
>>>>>>> +     int err;
>>>>>>> +
>>>>>>> +     if (tdc->dma_desc) {
>>>>>>
>>>>>> Needs locking protection against racing with the interrupt handler.
>>>>> tegra_dma_stop_client() waits for the in-flight transfer
>>>>> to complete and prevents any additional transfer to start.
>>>>> Wouldn't it manage the race? Do you see any potential issue there?
>>>>
>>>> You should consider interrupt handler like a process running in a
>>>> parallel thread. The interrupt handler sets tdc->dma_desc to NULL, hence
>>>> you'll get NULL dereference in tegra_dma_stop_client().
>>>
>>> Is it better if I remove the below part from tegra_dma_stop_client() so
>>> that dma_desc is not accessed at all?
>>>
>>> +     wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
>>> +     tdc->dma_desc->bytes_transferred +=
>>> +                     tdc->dma_desc->bytes_requested - (wcount * 4);
>>>
>>> Because I don't see a point in updating the value there. dma_desc is set
>>> to NULL in the next step in terminate_all() anyway.
>>
>> That isn't going help you much because you also can't release DMA
>> descriptor while interrupt handler still may be running and using that
>> descriptor.
> 
> Does the below functions look good to resolve the issue, provided
> tegra_dma_stop_client() doesn't access dma_desc?

Stop shall not race with the start.

> +static int tegra_dma_terminate_all(struct dma_chan *dc)
> +{
> +       struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> +       unsigned long flags;
> +       LIST_HEAD(head);
> +       int err;
> +
> +       err = tegra_dma_stop_client(tdc);
> +       if (err)
> +               return err;
> +
> +       tegra_dma_stop(tdc);
> +
> +       spin_lock_irqsave(&tdc->vc.lock, flags);
> +       tegra_dma_sid_free(tdc);
> +       tdc->dma_desc = NULL;
> +
> +       vchan_get_all_descriptors(&tdc->vc, &head);
> +       spin_unlock_irqrestore(&tdc->vc.lock, flags);
> +
> +       vchan_dma_desc_free_list(&tdc->vc, &head);
> +
> +       return 0;
> +}
> 
> +static irqreturn_t tegra_dma_isr(int irq, void *dev_id)
> +{
> +       struct tegra_dma_channel *tdc = dev_id;
> +       struct tegra_dma_desc *dma_desc = tdc->dma_desc;
> +       struct tegra_dma_sg_req *sg_req;
> +       u32 status;
> +
> +       /* Check channel error status register */
> +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS);
> +       if (status) {
> +               tegra_dma_chan_decode_error(tdc, status);
> +               tegra_dma_dump_chan_regs(tdc);
> +               tdc_write(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS, 0xFFFFFFFF);
> +       }
> +
> +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS);
> +       if (!(status & TEGRA_GPCDMA_STATUS_ISE_EOC))
> +               return IRQ_HANDLED;
> +
> +       tdc_write(tdc, TEGRA_GPCDMA_CHAN_STATUS,
> +                 TEGRA_GPCDMA_STATUS_ISE_EOC);
> +
> +       spin_lock(&tdc->vc.lock);
> +       if (!dma_desc)
All checks and assignments must be done inside of critical section.


> +               goto irq_done;

^ permalink raw reply	[flat|nested] 17+ messages in thread

* RE: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-22 18:31               ` Dmitry Osipenko
@ 2022-01-23 16:49                 ` Akhil R
  2022-01-23 21:07                   ` Dmitry Osipenko
  0 siblings, 1 reply; 17+ messages in thread
From: Akhil R @ 2022-01-23 16:49 UTC (permalink / raw)
  To: Dmitry Osipenko, dan.j.williams, devicetree, dmaengine,
	Jonathan Hunter, Krishna Yarlagadda, Laxman Dewangan,
	linux-kernel, linux-tegra, p.zabel, Rajesh Gumasta, robh+dt,
	thierry.reding, vkoul
  Cc: Pavan Kunapuli

> 21.01.2022 19:24, Akhil R пишет:
> >>>>>>> +static int tegra_dma_terminate_all(struct dma_chan *dc) {
> >>>>>>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> >>>>>>> +     unsigned long flags;
> >>>>>>> +     LIST_HEAD(head);
> >>>>>>> +     int err;
> >>>>>>> +
> >>>>>>> +     if (tdc->dma_desc) {
> >>>>>>
> >>>>>> Needs locking protection against racing with the interrupt handler.
> >>>>> tegra_dma_stop_client() waits for the in-flight transfer to
> >>>>> complete and prevents any additional transfer to start.
> >>>>> Wouldn't it manage the race? Do you see any potential issue there?
> >>>>
> >>>> You should consider interrupt handler like a process running in a
> >>>> parallel thread. The interrupt handler sets tdc->dma_desc to NULL,
> >>>> hence you'll get NULL dereference in tegra_dma_stop_client().
> >>>
> >>> Is it better if I remove the below part from tegra_dma_stop_client()
> >>> so that dma_desc is not accessed at all?
> >>>
> >>> +     wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
> >>> +     tdc->dma_desc->bytes_transferred +=
> >>> +                     tdc->dma_desc->bytes_requested - (wcount * 4);
> >>>
> >>> Because I don't see a point in updating the value there. dma_desc is
> >>> set to NULL in the next step in terminate_all() anyway.
> >>
> >> That isn't going help you much because you also can't release DMA
> >> descriptor while interrupt handler still may be running and using
> >> that descriptor.
> >
> > Does the below functions look good to resolve the issue, provided
> > tegra_dma_stop_client() doesn't access dma_desc?
> 
> Stop shall not race with the start.
> 
> > +static int tegra_dma_terminate_all(struct dma_chan *dc) {
> > +       struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> > +       unsigned long flags;
> > +       LIST_HEAD(head);
> > +       int err;
> > +
> > +       err = tegra_dma_stop_client(tdc);
> > +       if (err)
> > +               return err;
> > +
> > +       tegra_dma_stop(tdc);
> > +
> > +       spin_lock_irqsave(&tdc->vc.lock, flags);
> > +       tegra_dma_sid_free(tdc);
> > +       tdc->dma_desc = NULL;
> > +
> > +       vchan_get_all_descriptors(&tdc->vc, &head);
> > +       spin_unlock_irqrestore(&tdc->vc.lock, flags);
> > +
> > +       vchan_dma_desc_free_list(&tdc->vc, &head);
> > +
> > +       return 0;
> > +}
> >
> > +static irqreturn_t tegra_dma_isr(int irq, void *dev_id) {
> > +       struct tegra_dma_channel *tdc = dev_id;
> > +       struct tegra_dma_desc *dma_desc = tdc->dma_desc;
> > +       struct tegra_dma_sg_req *sg_req;
> > +       u32 status;
> > +
> > +       /* Check channel error status register */
> > +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS);
> > +       if (status) {
> > +               tegra_dma_chan_decode_error(tdc, status);
> > +               tegra_dma_dump_chan_regs(tdc);
> > +               tdc_write(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS, 0xFFFFFFFF);
> > +       }
> > +
> > +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS);
> > +       if (!(status & TEGRA_GPCDMA_STATUS_ISE_EOC))
> > +               return IRQ_HANDLED;
> > +
> > +       tdc_write(tdc, TEGRA_GPCDMA_CHAN_STATUS,
> > +                 TEGRA_GPCDMA_STATUS_ISE_EOC);
> > +
> > +       spin_lock(&tdc->vc.lock);
> > +       if (!dma_desc)
> All checks and assignments must be done inside of critical section.

Okay. So, the lock should be held throughout the function.
Do you think tegra_dma_pause should also hold a lock
and remove irq_synchronize? That function also writes
to CSR register.

Regards,
Akhil

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-23 16:49                 ` Akhil R
@ 2022-01-23 21:07                   ` Dmitry Osipenko
  2022-01-27 16:29                     ` Akhil R
  0 siblings, 1 reply; 17+ messages in thread
From: Dmitry Osipenko @ 2022-01-23 21:07 UTC (permalink / raw)
  To: Akhil R, dan.j.williams, devicetree, dmaengine, Jonathan Hunter,
	Krishna Yarlagadda, Laxman Dewangan, linux-kernel, linux-tegra,
	p.zabel, Rajesh Gumasta, robh+dt, thierry.reding, vkoul
  Cc: Pavan Kunapuli

23.01.2022 19:49, Akhil R пишет:
>> 21.01.2022 19:24, Akhil R пишет:
>>>>>>>>> +static int tegra_dma_terminate_all(struct dma_chan *dc) {
>>>>>>>>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>>>>>>>>> +     unsigned long flags;
>>>>>>>>> +     LIST_HEAD(head);
>>>>>>>>> +     int err;
>>>>>>>>> +
>>>>>>>>> +     if (tdc->dma_desc) {
>>>>>>>>
>>>>>>>> Needs locking protection against racing with the interrupt handler.
>>>>>>> tegra_dma_stop_client() waits for the in-flight transfer to
>>>>>>> complete and prevents any additional transfer to start.
>>>>>>> Wouldn't it manage the race? Do you see any potential issue there?
>>>>>>
>>>>>> You should consider interrupt handler like a process running in a
>>>>>> parallel thread. The interrupt handler sets tdc->dma_desc to NULL,
>>>>>> hence you'll get NULL dereference in tegra_dma_stop_client().
>>>>>
>>>>> Is it better if I remove the below part from tegra_dma_stop_client()
>>>>> so that dma_desc is not accessed at all?
>>>>>
>>>>> +     wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
>>>>> +     tdc->dma_desc->bytes_transferred +=
>>>>> +                     tdc->dma_desc->bytes_requested - (wcount * 4);
>>>>>
>>>>> Because I don't see a point in updating the value there. dma_desc is
>>>>> set to NULL in the next step in terminate_all() anyway.
>>>>
>>>> That isn't going help you much because you also can't release DMA
>>>> descriptor while interrupt handler still may be running and using
>>>> that descriptor.
>>>
>>> Does the below functions look good to resolve the issue, provided
>>> tegra_dma_stop_client() doesn't access dma_desc?
>>
>> Stop shall not race with the start.
>>
>>> +static int tegra_dma_terminate_all(struct dma_chan *dc) {
>>> +       struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>>> +       unsigned long flags;
>>> +       LIST_HEAD(head);
>>> +       int err;
>>> +
>>> +       err = tegra_dma_stop_client(tdc);
>>> +       if (err)
>>> +               return err;
>>> +
>>> +       tegra_dma_stop(tdc);
>>> +
>>> +       spin_lock_irqsave(&tdc->vc.lock, flags);
>>> +       tegra_dma_sid_free(tdc);
>>> +       tdc->dma_desc = NULL;
>>> +
>>> +       vchan_get_all_descriptors(&tdc->vc, &head);
>>> +       spin_unlock_irqrestore(&tdc->vc.lock, flags);
>>> +
>>> +       vchan_dma_desc_free_list(&tdc->vc, &head);
>>> +
>>> +       return 0;
>>> +}
>>>
>>> +static irqreturn_t tegra_dma_isr(int irq, void *dev_id) {
>>> +       struct tegra_dma_channel *tdc = dev_id;
>>> +       struct tegra_dma_desc *dma_desc = tdc->dma_desc;
>>> +       struct tegra_dma_sg_req *sg_req;
>>> +       u32 status;
>>> +
>>> +       /* Check channel error status register */
>>> +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS);
>>> +       if (status) {
>>> +               tegra_dma_chan_decode_error(tdc, status);
>>> +               tegra_dma_dump_chan_regs(tdc);
>>> +               tdc_write(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS, 0xFFFFFFFF);
>>> +       }
>>> +
>>> +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS);
>>> +       if (!(status & TEGRA_GPCDMA_STATUS_ISE_EOC))
>>> +               return IRQ_HANDLED;
>>> +
>>> +       tdc_write(tdc, TEGRA_GPCDMA_CHAN_STATUS,
>>> +                 TEGRA_GPCDMA_STATUS_ISE_EOC);
>>> +
>>> +       spin_lock(&tdc->vc.lock);
>>> +       if (!dma_desc)
>> All checks and assignments must be done inside of critical section.
> 
> Okay. So, the lock should be held throughout the function.
> Do you think tegra_dma_pause should also hold a lock
> and remove irq_synchronize? That function also writes
> to CSR register.

Interrupt handler shall not unpause channel in a case of race condition,
it should handle completed transfer and check whether channel is paused
before issuing next transfer.
So yes, pause also needs a lock.

^ permalink raw reply	[flat|nested] 17+ messages in thread

* RE: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-23 21:07                   ` Dmitry Osipenko
@ 2022-01-27 16:29                     ` Akhil R
  2022-01-27 23:15                       ` Dmitry Osipenko
  0 siblings, 1 reply; 17+ messages in thread
From: Akhil R @ 2022-01-27 16:29 UTC (permalink / raw)
  To: Dmitry Osipenko, dan.j.williams, devicetree, dmaengine,
	Jonathan Hunter, Krishna Yarlagadda, Laxman Dewangan,
	linux-kernel, linux-tegra, p.zabel, Rajesh Gumasta, robh+dt,
	thierry.reding, vkoul
  Cc: Pavan Kunapuli

> 23.01.2022 19:49, Akhil R пишет:
> >> 21.01.2022 19:24, Akhil R пишет:
> >>>>>>>>> +static int tegra_dma_terminate_all(struct dma_chan *dc) {
> >>>>>>>>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> >>>>>>>>> +     unsigned long flags;
> >>>>>>>>> +     LIST_HEAD(head);
> >>>>>>>>> +     int err;
> >>>>>>>>> +
> >>>>>>>>> +     if (tdc->dma_desc) {
> >>>>>>>>
> >>>>>>>> Needs locking protection against racing with the interrupt handler.
> >>>>>>> tegra_dma_stop_client() waits for the in-flight transfer to
> >>>>>>> complete and prevents any additional transfer to start.
> >>>>>>> Wouldn't it manage the race? Do you see any potential issue there?
> >>>>>>
> >>>>>> You should consider interrupt handler like a process running in a
> >>>>>> parallel thread. The interrupt handler sets tdc->dma_desc to NULL,
> >>>>>> hence you'll get NULL dereference in tegra_dma_stop_client().
> >>>>>
> >>>>> Is it better if I remove the below part from tegra_dma_stop_client()
> >>>>> so that dma_desc is not accessed at all?
> >>>>>
> >>>>> +     wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
> >>>>> +     tdc->dma_desc->bytes_transferred +=
> >>>>> +                     tdc->dma_desc->bytes_requested - (wcount * 4);
> >>>>>
> >>>>> Because I don't see a point in updating the value there. dma_desc is
> >>>>> set to NULL in the next step in terminate_all() anyway.
> >>>>
> >>>> That isn't going help you much because you also can't release DMA
> >>>> descriptor while interrupt handler still may be running and using
> >>>> that descriptor.
> >>>
> >>> Does the below functions look good to resolve the issue, provided
> >>> tegra_dma_stop_client() doesn't access dma_desc?
> >>
> >> Stop shall not race with the start.
> >>
> >>> +static int tegra_dma_terminate_all(struct dma_chan *dc) {
> >>> +       struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
> >>> +       unsigned long flags;
> >>> +       LIST_HEAD(head);
> >>> +       int err;
> >>> +
> >>> +       err = tegra_dma_stop_client(tdc);
> >>> +       if (err)
> >>> +               return err;
> >>> +
> >>> +       tegra_dma_stop(tdc);
> >>> +
> >>> +       spin_lock_irqsave(&tdc->vc.lock, flags);
> >>> +       tegra_dma_sid_free(tdc);
> >>> +       tdc->dma_desc = NULL;
> >>> +
> >>> +       vchan_get_all_descriptors(&tdc->vc, &head);
> >>> +       spin_unlock_irqrestore(&tdc->vc.lock, flags);
> >>> +
> >>> +       vchan_dma_desc_free_list(&tdc->vc, &head);
> >>> +
> >>> +       return 0;
> >>> +}
> >>>
> >>> +static irqreturn_t tegra_dma_isr(int irq, void *dev_id) {
> >>> +       struct tegra_dma_channel *tdc = dev_id;
> >>> +       struct tegra_dma_desc *dma_desc = tdc->dma_desc;
> >>> +       struct tegra_dma_sg_req *sg_req;
> >>> +       u32 status;
> >>> +
> >>> +       /* Check channel error status register */
> >>> +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS);
> >>> +       if (status) {
> >>> +               tegra_dma_chan_decode_error(tdc, status);
> >>> +               tegra_dma_dump_chan_regs(tdc);
> >>> +               tdc_write(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS, 0xFFFFFFFF);
> >>> +       }
> >>> +
> >>> +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS);
> >>> +       if (!(status & TEGRA_GPCDMA_STATUS_ISE_EOC))
> >>> +               return IRQ_HANDLED;
> >>> +
> >>> +       tdc_write(tdc, TEGRA_GPCDMA_CHAN_STATUS,
> >>> +                 TEGRA_GPCDMA_STATUS_ISE_EOC);
> >>> +
> >>> +       spin_lock(&tdc->vc.lock);
> >>> +       if (!dma_desc)
> >> All checks and assignments must be done inside of critical section.
> >
> > Okay. So, the lock should be held throughout the function.
> > Do you think tegra_dma_pause should also hold a lock
> > and remove irq_synchronize? That function also writes
> > to CSR register.
> 
> Interrupt handler shall not unpause channel in a case of race condition,
> it should handle completed transfer and check whether channel is paused
> before issuing next transfer.
> So yes, pause also needs a lock.
Thanks for the points. I collected more details on this scenario and found that
DMA can only be paused if there is an in-flight transfer (i.e no interrupt will be
pending). And there cannot be an interrupt after DMA is paused. Hence, there
is no case of ISR unpausing a paused transfer. So, I guess, the check or a lock
might not be required there.

One thing I am thinking of is to wait for DMA to be busy so that it can be paused.
I would add these changes and send out a new patch.

Thanks,
Akhil


^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver
  2022-01-27 16:29                     ` Akhil R
@ 2022-01-27 23:15                       ` Dmitry Osipenko
  0 siblings, 0 replies; 17+ messages in thread
From: Dmitry Osipenko @ 2022-01-27 23:15 UTC (permalink / raw)
  To: Akhil R, dan.j.williams, devicetree, dmaengine, Jonathan Hunter,
	Krishna Yarlagadda, Laxman Dewangan, linux-kernel, linux-tegra,
	p.zabel, Rajesh Gumasta, robh+dt, thierry.reding, vkoul
  Cc: Pavan Kunapuli

27.01.2022 19:29, Akhil R пишет:
>> 23.01.2022 19:49, Akhil R пишет:
>>>> 21.01.2022 19:24, Akhil R пишет:
>>>>>>>>>>> +static int tegra_dma_terminate_all(struct dma_chan *dc) {
>>>>>>>>>>> +     struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>>>>>>>>>>> +     unsigned long flags;
>>>>>>>>>>> +     LIST_HEAD(head);
>>>>>>>>>>> +     int err;
>>>>>>>>>>> +
>>>>>>>>>>> +     if (tdc->dma_desc) {
>>>>>>>>>>
>>>>>>>>>> Needs locking protection against racing with the interrupt handler.
>>>>>>>>> tegra_dma_stop_client() waits for the in-flight transfer to
>>>>>>>>> complete and prevents any additional transfer to start.
>>>>>>>>> Wouldn't it manage the race? Do you see any potential issue there?
>>>>>>>>
>>>>>>>> You should consider interrupt handler like a process running in a
>>>>>>>> parallel thread. The interrupt handler sets tdc->dma_desc to NULL,
>>>>>>>> hence you'll get NULL dereference in tegra_dma_stop_client().
>>>>>>>
>>>>>>> Is it better if I remove the below part from tegra_dma_stop_client()
>>>>>>> so that dma_desc is not accessed at all?
>>>>>>>
>>>>>>> +     wcount = tdc_read(tdc, TEGRA_GPCDMA_CHAN_XFER_COUNT);
>>>>>>> +     tdc->dma_desc->bytes_transferred +=
>>>>>>> +                     tdc->dma_desc->bytes_requested - (wcount * 4);
>>>>>>>
>>>>>>> Because I don't see a point in updating the value there. dma_desc is
>>>>>>> set to NULL in the next step in terminate_all() anyway.
>>>>>>
>>>>>> That isn't going help you much because you also can't release DMA
>>>>>> descriptor while interrupt handler still may be running and using
>>>>>> that descriptor.
>>>>>
>>>>> Does the below functions look good to resolve the issue, provided
>>>>> tegra_dma_stop_client() doesn't access dma_desc?
>>>>
>>>> Stop shall not race with the start.
>>>>
>>>>> +static int tegra_dma_terminate_all(struct dma_chan *dc) {
>>>>> +       struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>>>>> +       unsigned long flags;
>>>>> +       LIST_HEAD(head);
>>>>> +       int err;
>>>>> +
>>>>> +       err = tegra_dma_stop_client(tdc);
>>>>> +       if (err)
>>>>> +               return err;
>>>>> +
>>>>> +       tegra_dma_stop(tdc);
>>>>> +
>>>>> +       spin_lock_irqsave(&tdc->vc.lock, flags);
>>>>> +       tegra_dma_sid_free(tdc);
>>>>> +       tdc->dma_desc = NULL;
>>>>> +
>>>>> +       vchan_get_all_descriptors(&tdc->vc, &head);
>>>>> +       spin_unlock_irqrestore(&tdc->vc.lock, flags);
>>>>> +
>>>>> +       vchan_dma_desc_free_list(&tdc->vc, &head);
>>>>> +
>>>>> +       return 0;
>>>>> +}
>>>>>
>>>>> +static irqreturn_t tegra_dma_isr(int irq, void *dev_id) {
>>>>> +       struct tegra_dma_channel *tdc = dev_id;
>>>>> +       struct tegra_dma_desc *dma_desc = tdc->dma_desc;
>>>>> +       struct tegra_dma_sg_req *sg_req;
>>>>> +       u32 status;
>>>>> +
>>>>> +       /* Check channel error status register */
>>>>> +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS);
>>>>> +       if (status) {
>>>>> +               tegra_dma_chan_decode_error(tdc, status);
>>>>> +               tegra_dma_dump_chan_regs(tdc);
>>>>> +               tdc_write(tdc, TEGRA_GPCDMA_CHAN_ERR_STATUS, 0xFFFFFFFF);
>>>>> +       }
>>>>> +
>>>>> +       status = tdc_read(tdc, TEGRA_GPCDMA_CHAN_STATUS);
>>>>> +       if (!(status & TEGRA_GPCDMA_STATUS_ISE_EOC))
>>>>> +               return IRQ_HANDLED;
>>>>> +
>>>>> +       tdc_write(tdc, TEGRA_GPCDMA_CHAN_STATUS,
>>>>> +                 TEGRA_GPCDMA_STATUS_ISE_EOC);
>>>>> +
>>>>> +       spin_lock(&tdc->vc.lock);
>>>>> +       if (!dma_desc)
>>>> All checks and assignments must be done inside of critical section.
>>>
>>> Okay. So, the lock should be held throughout the function.
>>> Do you think tegra_dma_pause should also hold a lock
>>> and remove irq_synchronize? That function also writes
>>> to CSR register.
>>
>> Interrupt handler shall not unpause channel in a case of race condition,
>> it should handle completed transfer and check whether channel is paused
>> before issuing next transfer.
>> So yes, pause also needs a lock.
> Thanks for the points. I collected more details on this scenario and found that
> DMA can only be paused if there is an in-flight transfer (i.e no interrupt will be
> pending). And there cannot be an interrupt after DMA is paused. Hence, there
> is no case of ISR unpausing a paused transfer. So, I guess, the check or a lock
> might not be required there.

If there is in-flight transfer, then interrupt may fire at any time. All
CSRE modifications must be protected.

> One thing I am thinking of is to wait for DMA to be busy so that it can be paused.
> I would add these changes and send out a new patch.

You should assume that interrupt handler may run with a delay. Then
pause() will see that channel isn't busy, but delayed interrupt will
start the next transfer.

Also, the tegra_dma_pause() will crash if tdc->dma_desc=NULL.

Everything must be done under lock.

^ permalink raw reply	[flat|nested] 17+ messages in thread

end of thread, other threads:[~2022-01-27 23:15 UTC | newest]

Thread overview: 17+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-10 16:05 [PATCH v16 0/4] Add NVIDIA Tegra GPC-DMA driver Akhil R
2022-01-10 16:05 ` [PATCH v16 1/4] dt-bindings: dmaengine: Add doc for tegra gpcdma Akhil R
2022-01-10 16:05 ` [PATCH v16 2/4] dmaengine: tegra: Add tegra gpcdma driver Akhil R
2022-01-10 20:31   ` kernel test robot
2022-01-14 11:57   ` Dmitry Osipenko
2022-01-17  7:02     ` Akhil R
2022-01-17 15:41       ` Dmitry Osipenko
2022-01-18  5:36         ` Akhil R
2022-01-18 13:48           ` Dmitry Osipenko
2022-01-21 16:24             ` Akhil R
2022-01-22 18:31               ` Dmitry Osipenko
2022-01-23 16:49                 ` Akhil R
2022-01-23 21:07                   ` Dmitry Osipenko
2022-01-27 16:29                     ` Akhil R
2022-01-27 23:15                       ` Dmitry Osipenko
2022-01-10 16:05 ` [PATCH v16 3/4] arm64: defconfig: tegra: Enable GPCDMA Akhil R
2022-01-10 16:05 ` [PATCH v16 4/4] arm64: tegra: Add GPCDMA node for tegra186 and tegra194 Akhil R

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.