dmaengine: amd: qdma: Add AMD QDMA driver

Adds driver to enable PCIe board which uses AMD QDMA (the Queue-based
Direct Memory Access) subsystem. For example, Xilinx Alveo V70 AI
Accelerator devices.
    https://www.xilinx.com/applications/data-center/v70.html

The QDMA subsystem is used in conjunction with the PCI Express IP block
to provide high performance data transfer between host memory and the
card's DMA subsystem.

            +-------+       +-------+       +-----------+
   PCIe     |       |       |       |       |           |
   Tx/Rx    |       |       |       |  AXI  |           |
 <=======>  | PCIE  | <===> | QDMA  | <====>| User Logic|
            |       |       |       |       |           |
            +-------+       +-------+       +-----------+

The primary mechanism to transfer data using the QDMA is for the QDMA
engine to operate on instructions (descriptors) provided by the host
operating system. Using the descriptors, the QDMA can move data in both
the Host to Card (H2C) direction, or the Card to Host (C2H) direction.
The QDMA provides a per-queue basis option whether DMA traffic goes
to an AXI4 memory map (MM) interface or to an AXI4-Stream interface.

The hardware detail is provided by
    https://docs.xilinx.com/r/en-US/pg302-qdma

Implements dmaengine APIs to support MM DMA transfers.
- probe the available DMA channels
- use dma_slave_map for channel lookup
- use virtual channel to manage dmaengine tx descriptors
- implement device_prep_slave_sg callback to handle host scatter gather
  list

Signed-off-by: Nishad Saraf <nishads@amd.com>
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Link: https://lore.kernel.org/r/20240819211948.688786-2-lizhi.hou@amd.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

MAINTAINERS

@@ -1128,6 +1128,14 @@ L:	dmaengine@vger.kernel.org
 S:	Maintained
 F:	drivers/dma/ptdma/
 
+AMD QDMA DRIVER
+M:	Nishad Saraf <nishads@amd.com>
+M:	Lizhi Hou <lizhi.hou@amd.com>
+L:	dmaengine@vger.kernel.org
+S:	Supported
+F:	drivers/dma/amd/qdma/
+F:	include/linux/platform_data/amd_qdma.h
+
 AMD SEATTLE DEVICE TREE SUPPORT
 M:	Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
 M:	Tom Lendacky <thomas.lendacky@amd.com>

drivers/dma/Kconfig

@@ -716,6 +716,8 @@ config XILINX_ZYNQMP_DPDMA
 	  display driver.
 
 # driver files
+source "drivers/dma/amd/Kconfig"
+
 source "drivers/dma/bestcomm/Kconfig"
 
 source "drivers/dma/mediatek/Kconfig"

drivers/dma/Makefile

@@ -83,6 +83,7 @@ obj-$(CONFIG_ST_FDMA) += st_fdma.o
 obj-$(CONFIG_FSL_DPAA2_QDMA) += fsl-dpaa2-qdma/
 obj-$(CONFIG_INTEL_LDMA) += lgm/
 
+obj-y += amd/
 obj-y += mediatek/
 obj-y += qcom/
 obj-y += stm32/

drivers/dma/amd/Kconfig

@@ -0,0 +1,14 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+config AMD_QDMA
+	tristate "AMD Queue-based DMA"
+	depends on HAS_IOMEM
+	select DMA_ENGINE
+	select DMA_VIRTUAL_CHANNELS
+	select REGMAP_MMIO
+	help
+	  Enable support for the AMD Queue-based DMA subsystem. The primary
+	  mechanism to transfer data using the QDMA is for the QDMA engine to
+	  operate on instructions (descriptors) provided by the host operating
+	  system. Using the descriptors, the QDMA can move data in either the
+	  Host to Card (H2C) direction or the Card to Host (C2H) direction.

drivers/dma/amd/Makefile

@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0
+
+obj-$(CONFIG_AMD_QDMA) += qdma/

drivers/dma/amd/qdma/Makefile

@@ -0,0 +1,5 @@
+# SPDX-License-Identifier: GPL-2.0
+
+obj-$(CONFIG_AMD_QDMA)			+= amd-qdma.o
+
+amd-qdma-$(CONFIG_AMD_QDMA)		:= qdma.o qdma-comm-regs.o

drivers/dma/amd/qdma/qdma-comm-regs.c

@@ -0,0 +1,64 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2023-2024, Advanced Micro Devices, Inc.
+ */
+
+#ifndef __QDMA_REGS_DEF_H
+#define __QDMA_REGS_DEF_H
+
+#include "qdma.h"
+
+const struct qdma_reg qdma_regos_default[QDMA_REGO_MAX] = {
+	[QDMA_REGO_CTXT_DATA] = QDMA_REGO(0x804, 8),
+	[QDMA_REGO_CTXT_CMD] = QDMA_REGO(0x844, 1),
+	[QDMA_REGO_CTXT_MASK] = QDMA_REGO(0x824, 8),
+	[QDMA_REGO_MM_H2C_CTRL] = QDMA_REGO(0x1004, 1),
+	[QDMA_REGO_MM_C2H_CTRL] = QDMA_REGO(0x1204, 1),
+	[QDMA_REGO_QUEUE_COUNT] = QDMA_REGO(0x120, 1),
+	[QDMA_REGO_RING_SIZE] = QDMA_REGO(0x204, 1),
+	[QDMA_REGO_H2C_PIDX] = QDMA_REGO(0x18004, 1),
+	[QDMA_REGO_C2H_PIDX] = QDMA_REGO(0x18008, 1),
+	[QDMA_REGO_INTR_CIDX] = QDMA_REGO(0x18000, 1),
+	[QDMA_REGO_FUNC_ID] = QDMA_REGO(0x12c, 1),
+	[QDMA_REGO_ERR_INT] = QDMA_REGO(0xb04, 1),
+	[QDMA_REGO_ERR_STAT] = QDMA_REGO(0x248, 1),
+};
+
+const struct qdma_reg_field qdma_regfs_default[QDMA_REGF_MAX] = {
+	/* QDMA_REGO_CTXT_DATA fields */
+	[QDMA_REGF_IRQ_ENABLE] = QDMA_REGF(53, 53),
+	[QDMA_REGF_WBK_ENABLE] = QDMA_REGF(52, 52),
+	[QDMA_REGF_WBI_CHECK] = QDMA_REGF(34, 34),
+	[QDMA_REGF_IRQ_ARM] = QDMA_REGF(16, 16),
+	[QDMA_REGF_IRQ_VEC] = QDMA_REGF(138, 128),
+	[QDMA_REGF_IRQ_AGG] = QDMA_REGF(139, 139),
+	[QDMA_REGF_WBI_INTVL_ENABLE] = QDMA_REGF(35, 35),
+	[QDMA_REGF_MRKR_DISABLE] = QDMA_REGF(62, 62),
+	[QDMA_REGF_QUEUE_ENABLE] = QDMA_REGF(32, 32),
+	[QDMA_REGF_QUEUE_MODE] = QDMA_REGF(63, 63),
+	[QDMA_REGF_DESC_BASE] = QDMA_REGF(127, 64),
+	[QDMA_REGF_DESC_SIZE] = QDMA_REGF(49, 48),
+	[QDMA_REGF_RING_ID] = QDMA_REGF(47, 44),
+	[QDMA_REGF_QUEUE_BASE] = QDMA_REGF(11, 0),
+	[QDMA_REGF_QUEUE_MAX] = QDMA_REGF(44, 32),
+	[QDMA_REGF_FUNCTION_ID] = QDMA_REGF(24, 17),
+	[QDMA_REGF_INTR_AGG_BASE] = QDMA_REGF(66, 15),
+	[QDMA_REGF_INTR_VECTOR] = QDMA_REGF(11, 1),
+	[QDMA_REGF_INTR_SIZE] = QDMA_REGF(69, 67),
+	[QDMA_REGF_INTR_VALID] = QDMA_REGF(0, 0),
+	[QDMA_REGF_INTR_COLOR] = QDMA_REGF(14, 14),
+	[QDMA_REGF_INTR_FUNCTION_ID] = QDMA_REGF(125, 114),
+	/* QDMA_REGO_CTXT_CMD fields */
+	[QDMA_REGF_CMD_INDX] = QDMA_REGF(19, 7),
+	[QDMA_REGF_CMD_CMD] = QDMA_REGF(6, 5),
+	[QDMA_REGF_CMD_TYPE] = QDMA_REGF(4, 1),
+	[QDMA_REGF_CMD_BUSY] = QDMA_REGF(0, 0),
+	/* QDMA_REGO_QUEUE_COUNT fields */
+	[QDMA_REGF_QUEUE_COUNT] = QDMA_REGF(11, 0),
+	/* QDMA_REGO_ERR_INT fields */
+	[QDMA_REGF_ERR_INT_FUNC] = QDMA_REGF(11, 0),
+	[QDMA_REGF_ERR_INT_VEC] = QDMA_REGF(22, 12),
+	[QDMA_REGF_ERR_INT_ARM] = QDMA_REGF(24, 24),
+};
+
+#endif	/* __QDMA_REGS_DEF_H */