diff options
author | Vinod Koul <vinod.koul@intel.com> | 2015-02-02 16:55:43 -0800 |
---|---|---|
committer | Vinod Koul <vinod.koul@intel.com> | 2015-02-02 16:55:43 -0800 |
commit | 46c2eb645991c00bcbdf991c2f74389f7efce918 (patch) | |
tree | 874cd6c0e0c72238b966cb08135e5af9e9b1c936 /drivers/dma/sh | |
parent | 2cd6f7928ca42c7744f493e0ae2a4acaf02331a0 (diff) | |
parent | ee4b876bbee2c5c53518110849f23c117eec099c (diff) |
Merge branch 'topic/rcar' into for-linus
Diffstat (limited to 'drivers/dma/sh')
-rw-r--r-- | drivers/dma/sh/Kconfig | 14 | ||||
-rw-r--r-- | drivers/dma/sh/Makefile | 1 | ||||
-rw-r--r-- | drivers/dma/sh/rcar-dmac.c | 1747 |
3 files changed, 1761 insertions, 1 deletions
diff --git a/drivers/dma/sh/Kconfig b/drivers/dma/sh/Kconfig index 0349125a2e20..8190ad225a1b 100644 --- a/drivers/dma/sh/Kconfig +++ b/drivers/dma/sh/Kconfig @@ -2,6 +2,10 @@ # DMA engine configuration for sh # +config RENESAS_DMA + bool + select DMA_ENGINE + # # DMA Engine Helpers # @@ -12,7 +16,7 @@ config SH_DMAE_BASE depends on !SUPERH || SH_DMA depends on !SH_DMA_API default y - select DMA_ENGINE + select RENESAS_DMA help Enable support for the Renesas SuperH DMA controllers. @@ -52,3 +56,11 @@ config RCAR_AUDMAC_PP depends on SH_DMAE_BASE help Enable support for the Renesas R-Car Audio DMAC Peripheral Peripheral controllers. + +config RCAR_DMAC + tristate "Renesas R-Car Gen2 DMA Controller" + depends on ARCH_SHMOBILE || COMPILE_TEST + select RENESAS_DMA + help + This driver supports the general purpose DMA controller found in the + Renesas R-Car second generation SoCs. diff --git a/drivers/dma/sh/Makefile b/drivers/dma/sh/Makefile index 0a5cfdb76e45..2852f9db61a4 100644 --- a/drivers/dma/sh/Makefile +++ b/drivers/dma/sh/Makefile @@ -16,3 +16,4 @@ obj-$(CONFIG_SH_DMAE) += shdma.o obj-$(CONFIG_SUDMAC) += sudmac.o obj-$(CONFIG_RCAR_HPB_DMAE) += rcar-hpbdma.o obj-$(CONFIG_RCAR_AUDMAC_PP) += rcar-audmapp.o +obj-$(CONFIG_RCAR_DMAC) += rcar-dmac.o diff --git a/drivers/dma/sh/rcar-dmac.c b/drivers/dma/sh/rcar-dmac.c new file mode 100644 index 000000000000..29dd09ad41ff --- /dev/null +++ b/drivers/dma/sh/rcar-dmac.c @@ -0,0 +1,1747 @@ +/* + * Renesas R-Car Gen2 DMA Controller Driver + * + * Copyright (C) 2014 Renesas Electronics Inc. + * + * Author: Laurent Pinchart <laurent.pinchart@ideasonboard.com> + * + * This is free software; you can redistribute it and/or modify + * it under the terms of version 2 of the GNU General Public License as + * published by the Free Software Foundation. + */ + +#include <linux/dma-mapping.h> +#include <linux/dmaengine.h> +#include <linux/interrupt.h> +#include <linux/list.h> +#include <linux/module.h> +#include <linux/mutex.h> +#include <linux/of.h> +#include <linux/of_dma.h> +#include <linux/of_platform.h> +#include <linux/platform_device.h> +#include <linux/pm_runtime.h> +#include <linux/slab.h> +#include <linux/spinlock.h> + +#include "../dmaengine.h" + +/* + * struct rcar_dmac_xfer_chunk - Descriptor for a hardware transfer + * @node: entry in the parent's chunks list + * @src_addr: device source address + * @dst_addr: device destination address + * @size: transfer size in bytes + */ +struct rcar_dmac_xfer_chunk { + struct list_head node; + + dma_addr_t src_addr; + dma_addr_t dst_addr; + u32 size; +}; + +/* + * struct rcar_dmac_hw_desc - Hardware descriptor for a transfer chunk + * @sar: value of the SAR register (source address) + * @dar: value of the DAR register (destination address) + * @tcr: value of the TCR register (transfer count) + */ +struct rcar_dmac_hw_desc { + u32 sar; + u32 dar; + u32 tcr; + u32 reserved; +} __attribute__((__packed__)); + +/* + * struct rcar_dmac_desc - R-Car Gen2 DMA Transfer Descriptor + * @async_tx: base DMA asynchronous transaction descriptor + * @direction: direction of the DMA transfer + * @xfer_shift: log2 of the transfer size + * @chcr: value of the channel configuration register for this transfer + * @node: entry in the channel's descriptors lists + * @chunks: list of transfer chunks for this transfer + * @running: the transfer chunk being currently processed + * @nchunks: number of transfer chunks for this transfer + * @hwdescs.use: whether the transfer descriptor uses hardware descriptors + * @hwdescs.mem: hardware descriptors memory for the transfer + * @hwdescs.dma: device address of the hardware descriptors memory + * @hwdescs.size: size of the hardware descriptors in bytes + * @size: transfer size in bytes + * @cyclic: when set indicates that the DMA transfer is cyclic + */ +struct rcar_dmac_desc { + struct dma_async_tx_descriptor async_tx; + enum dma_transfer_direction direction; + unsigned int xfer_shift; + u32 chcr; + + struct list_head node; + struct list_head chunks; + struct rcar_dmac_xfer_chunk *running; + unsigned int nchunks; + + struct { + bool use; + struct rcar_dmac_hw_desc *mem; + dma_addr_t dma; + size_t size; + } hwdescs; + + unsigned int size; + bool cyclic; +}; + +#define to_rcar_dmac_desc(d) container_of(d, struct rcar_dmac_desc, async_tx) + +/* + * struct rcar_dmac_desc_page - One page worth of descriptors + * @node: entry in the channel's pages list + * @descs: array of DMA descriptors + * @chunks: array of transfer chunk descriptors + */ +struct rcar_dmac_desc_page { + struct list_head node; + + union { + struct rcar_dmac_desc descs[0]; + struct rcar_dmac_xfer_chunk chunks[0]; + }; +}; + +#define RCAR_DMAC_DESCS_PER_PAGE \ + ((PAGE_SIZE - offsetof(struct rcar_dmac_desc_page, descs)) / \ + sizeof(struct rcar_dmac_desc)) +#define RCAR_DMAC_XFER_CHUNKS_PER_PAGE \ + ((PAGE_SIZE - offsetof(struct rcar_dmac_desc_page, chunks)) / \ + sizeof(struct rcar_dmac_xfer_chunk)) + +/* + * struct rcar_dmac_chan - R-Car Gen2 DMA Controller Channel + * @chan: base DMA channel object + * @iomem: channel I/O memory base + * @index: index of this channel in the controller + * @src_xfer_size: size (in bytes) of hardware transfers on the source side + * @dst_xfer_size: size (in bytes) of hardware transfers on the destination side + * @src_slave_addr: slave source memory address + * @dst_slave_addr: slave destination memory address + * @mid_rid: hardware MID/RID for the DMA client using this channel + * @lock: protects the channel CHCR register and the desc members + * @desc.free: list of free descriptors + * @desc.pending: list of pending descriptors (submitted with tx_submit) + * @desc.active: list of active descriptors (activated with issue_pending) + * @desc.done: list of completed descriptors + * @desc.wait: list of descriptors waiting for an ack + * @desc.running: the descriptor being processed (a member of the active list) + * @desc.chunks_free: list of free transfer chunk descriptors + * @desc.pages: list of pages used by allocated descriptors + */ +struct rcar_dmac_chan { + struct dma_chan chan; + void __iomem *iomem; + unsigned int index; + + unsigned int src_xfer_size; + unsigned int dst_xfer_size; + dma_addr_t src_slave_addr; + dma_addr_t dst_slave_addr; + int mid_rid; + + spinlock_t lock; + + struct { + struct list_head free; + struct list_head pending; + struct list_head active; + struct list_head done; + struct list_head wait; + struct rcar_dmac_desc *running; + + struct list_head chunks_free; + + struct list_head pages; + } desc; +}; + +#define to_rcar_dmac_chan(c) container_of(c, struct rcar_dmac_chan, chan) + +/* + * struct rcar_dmac - R-Car Gen2 DMA Controller + * @engine: base DMA engine object + * @dev: the hardware device + * @iomem: remapped I/O memory base + * @n_channels: number of available channels + * @channels: array of DMAC channels + * @modules: bitmask of client modules in use + */ +struct rcar_dmac { + struct dma_device engine; + struct device *dev; + void __iomem *iomem; + + unsigned int n_channels; + struct rcar_dmac_chan *channels; + + unsigned long modules[256 / BITS_PER_LONG]; +}; + +#define to_rcar_dmac(d) container_of(d, struct rcar_dmac, engine) + +/* ----------------------------------------------------------------------------- + * Registers + */ + +#define RCAR_DMAC_CHAN_OFFSET(i) (0x8000 + 0x80 * (i)) + +#define RCAR_DMAISTA 0x0020 +#define RCAR_DMASEC 0x0030 +#define RCAR_DMAOR 0x0060 +#define RCAR_DMAOR_PRI_FIXED (0 << 8) +#define RCAR_DMAOR_PRI_ROUND_ROBIN (3 << 8) +#define RCAR_DMAOR_AE (1 << 2) +#define RCAR_DMAOR_DME (1 << 0) +#define RCAR_DMACHCLR 0x0080 +#define RCAR_DMADPSEC 0x00a0 + +#define RCAR_DMASAR 0x0000 +#define RCAR_DMADAR 0x0004 +#define RCAR_DMATCR 0x0008 +#define RCAR_DMATCR_MASK 0x00ffffff +#define RCAR_DMATSR 0x0028 +#define RCAR_DMACHCR 0x000c +#define RCAR_DMACHCR_CAE (1 << 31) +#define RCAR_DMACHCR_CAIE (1 << 30) +#define RCAR_DMACHCR_DPM_DISABLED (0 << 28) +#define RCAR_DMACHCR_DPM_ENABLED (1 << 28) +#define RCAR_DMACHCR_DPM_REPEAT (2 << 28) +#define RCAR_DMACHCR_DPM_INFINITE (3 << 28) +#define RCAR_DMACHCR_RPT_SAR (1 << 27) +#define RCAR_DMACHCR_RPT_DAR (1 << 26) +#define RCAR_DMACHCR_RPT_TCR (1 << 25) +#define RCAR_DMACHCR_DPB (1 << 22) +#define RCAR_DMACHCR_DSE (1 << 19) +#define RCAR_DMACHCR_DSIE (1 << 18) +#define RCAR_DMACHCR_TS_1B ((0 << 20) | (0 << 3)) +#define RCAR_DMACHCR_TS_2B ((0 << 20) | (1 << 3)) +#define RCAR_DMACHCR_TS_4B ((0 << 20) | (2 << 3)) +#define RCAR_DMACHCR_TS_16B ((0 << 20) | (3 << 3)) +#define RCAR_DMACHCR_TS_32B ((1 << 20) | (0 << 3)) +#define RCAR_DMACHCR_TS_64B ((1 << 20) | (1 << 3)) +#define RCAR_DMACHCR_TS_8B ((1 << 20) | (3 << 3)) +#define RCAR_DMACHCR_DM_FIXED (0 << 14) +#define RCAR_DMACHCR_DM_INC (1 << 14) +#define RCAR_DMACHCR_DM_DEC (2 << 14) +#define RCAR_DMACHCR_SM_FIXED (0 << 12) +#define RCAR_DMACHCR_SM_INC (1 << 12) +#define RCAR_DMACHCR_SM_DEC (2 << 12) +#define RCAR_DMACHCR_RS_AUTO (4 << 8) +#define RCAR_DMACHCR_RS_DMARS (8 << 8) +#define RCAR_DMACHCR_IE (1 << 2) +#define RCAR_DMACHCR_TE (1 << 1) +#define RCAR_DMACHCR_DE (1 << 0) +#define RCAR_DMATCRB 0x0018 +#define RCAR_DMATSRB 0x0038 +#define RCAR_DMACHCRB 0x001c +#define RCAR_DMACHCRB_DCNT(n) ((n) << 24) +#define RCAR_DMACHCRB_DPTR_MASK (0xff << 16) +#define RCAR_DMACHCRB_DPTR_SHIFT 16 +#define RCAR_DMACHCRB_DRST (1 << 15) +#define RCAR_DMACHCRB_DTS (1 << 8) +#define RCAR_DMACHCRB_SLM_NORMAL (0 << 4) +#define RCAR_DMACHCRB_SLM_CLK(n) ((8 | (n)) << 4) +#define RCAR_DMACHCRB_PRI(n) ((n) << 0) +#define RCAR_DMARS 0x0040 +#define RCAR_DMABUFCR 0x0048 +#define RCAR_DMABUFCR_MBU(n) ((n) << 16) +#define RCAR_DMABUFCR_ULB(n) ((n) << 0) +#define RCAR_DMADPBASE 0x0050 +#define RCAR_DMADPBASE_MASK 0xfffffff0 +#define RCAR_DMADPBASE_SEL (1 << 0) +#define RCAR_DMADPCR 0x0054 +#define RCAR_DMADPCR_DIPT(n) ((n) << 24) +#define RCAR_DMAFIXSAR 0x0010 +#define RCAR_DMAFIXDAR 0x0014 +#define RCAR_DMAFIXDPBASE 0x0060 + +/* Hardcode the MEMCPY transfer size to 4 bytes. */ +#define RCAR_DMAC_MEMCPY_XFER_SIZE 4 + +/* ----------------------------------------------------------------------------- + * Device access + */ + +static void rcar_dmac_write(struct rcar_dmac *dmac, u32 reg, u32 data) +{ + if (reg == RCAR_DMAOR) + writew(data, dmac->iomem + reg); + else + writel(data, dmac->iomem + reg); +} + +static u32 rcar_dmac_read(struct rcar_dmac *dmac, u32 reg) +{ + if (reg == RCAR_DMAOR) + return readw(dmac->iomem + reg); + else + return readl(dmac->iomem + reg); +} + +static u32 rcar_dmac_chan_read(struct rcar_dmac_chan *chan, u32 reg) +{ + if (reg == RCAR_DMARS) + return readw(chan->iomem + reg); + else + return readl(chan->iomem + reg); +} + +static void rcar_dmac_chan_write(struct rcar_dmac_chan *chan, u32 reg, u32 data) +{ + if (reg == RCAR_DMARS) + writew(data, chan->iomem + reg); + else + writel(data, chan->iomem + reg); +} + +/* ----------------------------------------------------------------------------- + * Initialization and configuration + */ + +static bool rcar_dmac_chan_is_busy(struct rcar_dmac_chan *chan) +{ + u32 chcr = rcar_dmac_chan_read(chan, RCAR_DMACHCR); + + return (chcr & (RCAR_DMACHCR_DE | RCAR_DMACHCR_TE)) == RCAR_DMACHCR_DE; +} + +static void rcar_dmac_chan_start_xfer(struct rcar_dmac_chan *chan) +{ + struct rcar_dmac_desc *desc = chan->desc.running; + u32 chcr = desc->chcr; + + WARN_ON_ONCE(rcar_dmac_chan_is_busy(chan)); + + if (chan->mid_rid >= 0) + rcar_dmac_chan_write(chan, RCAR_DMARS, chan->mid_rid); + + if (desc->hwdescs.use) { + dev_dbg(chan->chan.device->dev, + "chan%u: queue desc %p: %u@%pad\n", + chan->index, desc, desc->nchunks, &desc->hwdescs.dma); + +#ifdef CONFIG_ARCH_DMA_ADDR_T_64BIT + rcar_dmac_chan_write(chan, RCAR_DMAFIXDPBASE, + desc->hwdescs.dma >> 32); +#endif + rcar_dmac_chan_write(chan, RCAR_DMADPBASE, + (desc->hwdescs.dma & 0xfffffff0) | + RCAR_DMADPBASE_SEL); + rcar_dmac_chan_write(chan, RCAR_DMACHCRB, + RCAR_DMACHCRB_DCNT(desc->nchunks - 1) | + RCAR_DMACHCRB_DRST); + + /* + * Program the descriptor stage interrupt to occur after the end + * of the first stage. + */ + rcar_dmac_chan_write(chan, RCAR_DMADPCR, RCAR_DMADPCR_DIPT(1)); + + chcr |= RCAR_DMACHCR_RPT_SAR | RCAR_DMACHCR_RPT_DAR + | RCAR_DMACHCR_RPT_TCR | RCAR_DMACHCR_DPB; + + /* + * If the descriptor isn't cyclic enable normal descriptor mode + * and the transfer completion interrupt. + */ + if (!desc->cyclic) + chcr |= RCAR_DMACHCR_DPM_ENABLED | RCAR_DMACHCR_IE; + /* + * If the descriptor is cyclic and has a callback enable the + * descriptor stage interrupt in infinite repeat mode. + */ + else if (desc->async_tx.callback) + chcr |= RCAR_DMACHCR_DPM_INFINITE | RCAR_DMACHCR_DSIE; + /* + * Otherwise just select infinite repeat mode without any + * interrupt. + */ + else + chcr |= RCAR_DMACHCR_DPM_INFINITE; + } else { + struct rcar_dmac_xfer_chunk *chunk = desc->running; + + dev_dbg(chan->chan.device->dev, + "chan%u: queue chunk %p: %u@%pad -> %pad\n", + chan->index, chunk, chunk->size, &chunk->src_addr, + &chunk->dst_addr); + +#ifdef CONFIG_ARCH_DMA_ADDR_T_64BIT + rcar_dmac_chan_write(chan, RCAR_DMAFIXSAR, + chunk->src_addr >> 32); + rcar_dmac_chan_write(chan, RCAR_DMAFIXDAR, + chunk->dst_addr >> 32); +#endif + rcar_dmac_chan_write(chan, RCAR_DMASAR, + chunk->src_addr & 0xffffffff); + rcar_dmac_chan_write(chan, RCAR_DMADAR, + chunk->dst_addr & 0xffffffff); + rcar_dmac_chan_write(chan, RCAR_DMATCR, + chunk->size >> desc->xfer_shift); + + chcr |= RCAR_DMACHCR_DPM_DISABLED | RCAR_DMACHCR_IE; + } + + rcar_dmac_chan_write(chan, RCAR_DMACHCR, chcr | RCAR_DMACHCR_DE); +} + +static int rcar_dmac_init(struct rcar_dmac *dmac) +{ + u16 dmaor; + + /* Clear all channels and enable the DMAC globally. */ + rcar_dmac_write(dmac, RCAR_DMACHCLR, 0x7fff); + rcar_dmac_write(dmac, RCAR_DMAOR, + RCAR_DMAOR_PRI_FIXED | RCAR_DMAOR_DME); + + dmaor = rcar_dmac_read(dmac, RCAR_DMAOR); + if ((dmaor & (RCAR_DMAOR_AE | RCAR_DMAOR_DME)) != RCAR_DMAOR_DME) { + dev_warn(dmac->dev, "DMAOR initialization failed.\n"); + return -EIO; + } + + return 0; +} + +/* ----------------------------------------------------------------------------- + * Descriptors submission + */ + +static dma_cookie_t rcar_dmac_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct rcar_dmac_chan *chan = to_rcar_dmac_chan(tx->chan); + struct rcar_dmac_desc *desc = to_rcar_dmac_desc(tx); + unsigned long flags; + dma_cookie_t cookie; + + spin_lock_irqsave(&chan->lock, flags); + + cookie = dma_cookie_assign(tx); + + dev_dbg(chan->chan.device->dev, "chan%u: submit #%d@%p\n", + chan->index, tx->cookie, desc); + + list_add_tail(&desc->node, &chan->desc.pending); + desc->running = list_first_entry(&desc->chunks, + struct rcar_dmac_xfer_chunk, node); + + spin_unlock_irqrestore(&chan->lock, flags); + + return cookie; +} + +/* ----------------------------------------------------------------------------- + * Descriptors allocation and free + */ + +/* + * rcar_dmac_desc_alloc - Allocate a page worth of DMA descriptors + * @chan: the DMA channel + * @gfp: allocation flags + */ +static int rcar_dmac_desc_alloc(struct rcar_dmac_chan *chan, gfp_t gfp) +{ + struct rcar_dmac_desc_page *page; + LIST_HEAD(list); + unsigned int i; + + page = (void *)get_zeroed_page(gfp); + if (!page) + return -ENOMEM; + + for (i = 0; i < RCAR_DMAC_DESCS_PER_PAGE; ++i) { + struct rcar_dmac_desc *desc = &page->descs[i]; + + dma_async_tx_descriptor_init(&desc->async_tx, &chan->chan); + desc->async_tx.tx_submit = rcar_dmac_tx_submit; + INIT_LIST_HEAD(&desc->chunks); + + list_add_tail(&desc->node, &list); + } + + spin_lock_irq(&chan->lock); + list_splice_tail(&list, &chan->desc.free); + list_add_tail(&page->node, &chan->desc.pages); + spin_unlock_irq(&chan->lock); + + return 0; +} + +/* + * rcar_dmac_desc_put - Release a DMA transfer descriptor + * @chan: the DMA channel + * @desc: the descriptor + * + * Put the descriptor and its transfer chunk descriptors back in the channel's + * free descriptors lists. The descriptor's chunks list will be reinitialized to + * an empty list as a result. + * + * The descriptor must have been removed from the channel's lists before calling + * this function. + * + * Locking: Must be called in non-atomic context. + */ +static void rcar_dmac_desc_put(struct rcar_dmac_chan *chan, + struct rcar_dmac_desc *desc) +{ + spin_lock_irq(&chan->lock); + list_splice_tail_init(&desc->chunks, &chan->desc.chunks_free); + list_add_tail(&desc->node, &chan->desc.free); + spin_unlock_irq(&chan->lock); +} + +static void rcar_dmac_desc_recycle_acked(struct rcar_dmac_chan *chan) +{ + struct rcar_dmac_desc *desc, *_desc; + LIST_HEAD(list); + + /* + * We have to temporarily move all descriptors from the wait list to a + * local list as iterating over the wait list, even with + * list_for_each_entry_safe, isn't safe if we release the channel lock + * around the rcar_dmac_desc_put() call. + */ + spin_lock_irq(&chan->lock); + list_splice_init(&chan->desc.wait, &list); + spin_unlock_irq(&chan->lock); + + list_for_each_entry_safe(desc, _desc, &list, node) { + if (async_tx_test_ack(&desc->async_tx)) { + list_del(&desc->node); + rcar_dmac_desc_put(chan, desc); + } + } + + if (list_empty(&list)) + return; + + /* Put the remaining descriptors back in the wait list. */ + spin_lock_irq(&chan->lock); + list_splice(&list, &chan->desc.wait); + spin_unlock_irq(&chan->lock); +} + +/* + * rcar_dmac_desc_get - Allocate a descriptor for a DMA transfer + * @chan: the DMA channel + * + * Locking: This function must be called in a non-atomic context. + * + * Return: A pointer to the allocated descriptor or NULL if no descriptor can + * be allocated. + */ +static struct rcar_dmac_desc *rcar_dmac_desc_get(struct rcar_dmac_chan *chan) +{ + struct rcar_dmac_desc *desc; + int ret; + + /* Recycle acked descriptors before attempting allocation. */ + rcar_dmac_desc_recycle_acked(chan); + + spin_lock_irq(&chan->lock); + + do { + if (list_empty(&chan->desc.free)) { + /* + * No free descriptors, allocate a page worth of them + * and try again, as someone else could race us to get + * the newly allocated descriptors. If the allocation + * fails return an error. + */ + spin_unlock_irq(&chan->lock); + ret = rcar_dmac_desc_alloc(chan, GFP_NOWAIT); + if (ret < 0) + return NULL; + spin_lock_irq(&chan->lock); + continue; + } + + desc = list_first_entry(&chan->desc.free, struct rcar_dmac_desc, + node); + list_del(&desc->node); + } while (!desc); + + spin_unlock_irq(&chan->lock); + + return desc; +} + +/* + * rcar_dmac_xfer_chunk_alloc - Allocate a page worth of transfer chunks + * @chan: the DMA channel + * @gfp: allocation flags + */ +static int rcar_dmac_xfer_chunk_alloc(struct rcar_dmac_chan *chan, gfp_t gfp) +{ + struct rcar_dmac_desc_page *page; + LIST_HEAD(list); + unsigned int i; + + page = (void *)get_zeroed_page(gfp); + if (!page) + return -ENOMEM; + + for (i = 0; i < RCAR_DMAC_XFER_CHUNKS_PER_PAGE; ++i) { + struct rcar_dmac_xfer_chunk *chunk = &page->chunks[i]; + + list_add_tail(&chunk->node, &list); + } + + spin_lock_irq(&chan->lock); + list_splice_tail(&list, &chan->desc.chunks_free); + list_add_tail(&page->node, &chan->desc.pages); + spin_unlock_irq(&chan->lock); + + return 0; +} + +/* + * rcar_dmac_xfer_chunk_get - Allocate a transfer chunk for a DMA transfer + * @chan: the DMA channel + * + * Locking: This function must be called in a non-atomic context. + * + * Return: A pointer to the allocated transfer chunk descriptor or NULL if no + * descriptor can be allocated. + */ +static struct rcar_dmac_xfer_chunk * +rcar_dmac_xfer_chunk_get(struct rcar_dmac_chan *chan) +{ + struct rcar_dmac_xfer_chunk *chunk; + int ret; + + spin_lock_irq(&chan->lock); + + do { + if (list_empty(&chan->desc.chunks_free)) { + /* + * No free descriptors, allocate a page worth of them + * and try again, as someone else could race us to get + * the newly allocated descriptors. If the allocation + * fails return an error. + */ + spin_unlock_irq(&chan->lock); + ret = rcar_dmac_xfer_chunk_alloc(chan, GFP_NOWAIT); + if (ret < 0) + return NULL; + spin_lock_irq(&chan->lock); + continue; + } + + chunk = list_first_entry(&chan->desc.chunks_free, + struct rcar_dmac_xfer_chunk, node); + list_del(&chunk->node); + } while (!chunk); + + spin_unlock_irq(&chan->lock); + + return chunk; +} + +static void rcar_dmac_realloc_hwdesc(struct rcar_dmac_chan *chan, + struct rcar_dmac_desc *desc, size_t size) +{ + /* + * dma_alloc_coherent() allocates memory in page size increments. To + * avoid reallocating the hardware descriptors when the allocated size + * wouldn't change align the requested size to a multiple of the page + * size. + */ + size = PAGE_ALIGN(size); + + if (desc->hwdescs.size == size) + return; + + if (desc->hwdescs.mem) { + dma_free_coherent(NULL, desc->hwdescs.size, desc->hwdescs.mem, + desc->hwdescs.dma); + desc->hwdescs.mem = NULL; + desc->hwdescs.size = 0; + } + + if (!size) + return; + + desc->hwdescs.mem = dma_alloc_coherent(NULL, size, &desc->hwdescs.dma, + GFP_NOWAIT); + if (!desc->hwdescs.mem) + return; + + desc->hwdescs.size = size; +} + +static int rcar_dmac_fill_hwdesc(struct rcar_dmac_chan *chan, + struct rcar_dmac_desc *desc) +{ + struct rcar_dmac_xfer_chunk *chunk; + struct rcar_dmac_hw_desc *hwdesc; + + rcar_dmac_realloc_hwdesc(chan, desc, desc->nchunks * sizeof(*hwdesc)); + + hwdesc = desc->hwdescs.mem; + if (!hwdesc) + return -ENOMEM; + + list_for_each_entry(chunk, &desc->chunks, node) { + hwdesc->sar = chunk->src_addr; + hwdesc->dar = chunk->dst_addr; + hwdesc->tcr = chunk->size >> desc->xfer_shift; + hwdesc++; + } + + return 0; +} + +/* ----------------------------------------------------------------------------- + * Stop and reset + */ + +static void rcar_dmac_chan_halt(struct rcar_dmac_chan *chan) +{ + u32 chcr = rcar_dmac_chan_read(chan, RCAR_DMACHCR); + + chcr &= ~(RCAR_DMACHCR_DSE | RCAR_DMACHCR_DSIE | RCAR_DMACHCR_IE | + RCAR_DMACHCR_TE | RCAR_DMACHCR_DE); + rcar_dmac_chan_write(chan, RCAR_DMACHCR, chcr); +} + +static void rcar_dmac_chan_reinit(struct rcar_dmac_chan *chan) +{ + struct rcar_dmac_desc *desc, *_desc; + unsigned long flags; + LIST_HEAD(descs); + + spin_lock_irqsave(&chan->lock, flags); + + /* Move all non-free descriptors to the local lists. */ + list_splice_init(&chan->desc.pending, &descs); + list_splice_init(&chan->desc.active, &descs); + list_splice_init(&chan->desc.done, &descs); + list_splice_init(&chan->desc.wait, &descs); + + chan->desc.running = NULL; + + spin_unlock_irqrestore(&chan->lock, flags); + + list_for_each_entry_safe(desc, _desc, &descs, node) { + list_del(&desc->node); + rcar_dmac_desc_put(chan, desc); + } +} + +static void rcar_dmac_stop(struct rcar_dmac *dmac) +{ + rcar_dmac_write(dmac, RCAR_DMAOR, 0); +} + +static void rcar_dmac_abort(struct rcar_dmac *dmac) +{ + unsigned int i; + + /* Stop all channels. */ + for (i = 0; i < dmac->n_channels; ++i) { + struct rcar_dmac_chan *chan = &dmac->channels[i]; + + /* Stop and reinitialize the channel. */ + spin_lock(&chan->lock); + rcar_dmac_chan_halt(chan); + spin_unlock(&chan->lock); + + rcar_dmac_chan_reinit(chan); + } +} + +/* ----------------------------------------------------------------------------- + * Descriptors preparation + */ + +static void rcar_dmac_chan_configure_desc(struct rcar_dmac_chan *chan, + struct rcar_dmac_desc *desc) +{ + static const u32 chcr_ts[] = { + RCAR_DMACHCR_TS_1B, RCAR_DMACHCR_TS_2B, + RCAR_DMACHCR_TS_4B, RCAR_DMACHCR_TS_8B, + RCAR_DMACHCR_TS_16B, RCAR_DMACHCR_TS_32B, + RCAR_DMACHCR_TS_64B, + }; + + unsigned int xfer_size; + u32 chcr; + + switch (desc->direction) { + case DMA_DEV_TO_MEM: + chcr = RCAR_DMACHCR_DM_INC | RCAR_DMACHCR_SM_FIXED + | RCAR_DMACHCR_RS_DMARS; + xfer_size = chan->src_xfer_size; + break; + + case DMA_MEM_TO_DEV: + chcr = RCAR_DMACHCR_DM_FIXED | RCAR_DMACHCR_SM_INC + | RCAR_DMACHCR_RS_DMARS; + xfer_size = chan->dst_xfer_size; + break; + + case DMA_MEM_TO_MEM: + default: + chcr = RCAR_DMACHCR_DM_INC | RCAR_DMACHCR_SM_INC + | RCAR_DMACHCR_RS_AUTO; + xfer_size = RCAR_DMAC_MEMCPY_XFER_SIZE; + break; + } + + desc->xfer_shift = ilog2(xfer_size); + desc->chcr = chcr | chcr_ts[desc->xfer_shift]; +} + +/* + * rcar_dmac_chan_prep_sg - prepare transfer descriptors from an SG list + * + * Common routine for public (MEMCPY) and slave DMA. The MEMCPY case is also + * converted to scatter-gather to guarantee consistent locking and a correct + * list manipulation. For slave DMA direction carries the usual meaning, and, + * logically, the SG list is RAM and the addr variable contains slave address, + * e.g., the FIFO I/O register. For MEMCPY direction equals DMA_MEM_TO_MEM + * and the SG list contains only one element and points at the source buffer. + */ +static struct dma_async_tx_descriptor * +rcar_dmac_chan_prep_sg(struct rcar_dmac_chan *chan, struct scatterlist *sgl, + unsigned int sg_len, dma_addr_t dev_addr, + enum dma_transfer_direction dir, unsigned long dma_flags, + bool cyclic) +{ + struct rcar_dmac_xfer_chunk *chunk; + struct rcar_dmac_desc *desc; + struct scatterlist *sg; + unsigned int nchunks = 0; + unsigned int max_chunk_size; + unsigned int full_size = 0; + bool highmem = false; + unsigned int i; + + desc = rcar_dmac_desc_get(chan); + if (!desc) + return NULL; + + desc->async_tx.flags = dma_flags; + desc->async_tx.cookie = -EBUSY; + + desc->cyclic = cyclic; + desc->direction = dir; + + rcar_dmac_chan_configure_desc(chan, desc); + + max_chunk_size = (RCAR_DMATCR_MASK + 1) << desc->xfer_shift; + + /* + * Allocate and fill the transfer chunk descriptors. We own the only + * reference to the DMA descriptor, there's no need for locking. + */ + for_each_sg(sgl, sg, sg_len, i) { + dma_addr_t mem_addr = sg_dma_address(sg); + unsigned int len = sg_dma_len(sg); + + full_size += len; + + while (len) { + unsigned int size = min(len, max_chunk_size); + +#ifdef CONFIG_ARCH_DMA_ADDR_T_64BIT + /* + * Prevent individual transfers from crossing 4GB + * boundaries. + */ + if (dev_addr >> 32 != (dev_addr + size - 1) >> 32) + size = ALIGN(dev_addr, 1ULL << 32) - dev_addr; + if (mem_addr >> 32 != (mem_addr + size - 1) >> 32) + size = ALIGN(mem_addr, 1ULL << 32) - mem_addr; + + /* + * Check if either of the source or destination address + * can't be expressed in 32 bits. If so we can't use + * hardware descriptor lists. + */ + if (dev_addr >> 32 || mem_addr >> 32) + highmem = true; +#endif + + chunk = rcar_dmac_xfer_chunk_get(chan); + if (!chunk) { + rcar_dmac_desc_put(chan, desc); + return NULL; + } + + if (dir == DMA_DEV_TO_MEM) { + chunk->src_addr = dev_addr; + chunk->dst_addr = mem_addr; + } else { + chunk->src_addr = mem_addr; + chunk->dst_addr = dev_addr; + } + + chunk->size = size; + + dev_dbg(chan->chan.device->dev, + "chan%u: chunk %p/%p sgl %u@%p, %u/%u %pad -> %pad\n", + chan->index, chunk, desc, i, sg, size, len, + &chunk->src_addr, &chunk->dst_addr); + + mem_addr += size; + if (dir == DMA_MEM_TO_MEM) + dev_addr += size; + + len -= size; + + list_add_tail(&chunk->node, &desc->chunks); + nchunks++; + } + } + + desc->nchunks = nchunks; + desc->size = full_size; + + /* + * Use hardware descriptor lists if possible when more than one chunk + * needs to be transferred (otherwise they don't make much sense). + * + * The highmem check currently covers the whole transfer. As an + * optimization we could use descriptor lists for consecutive lowmem + * chunks and direct manual mode for highmem chunks. Whether the + * performance improvement would be significant enough compared to the + * additional complexity remains to be investigated. + */ + desc->hwdescs.use = !highmem && nchunks > 1; + if (desc->hwdescs.use) { + if (rcar_dmac_fill_hwdesc(chan, desc) < 0) + desc->hwdescs.use = false; + } + + return &desc->async_tx; +} + +/* ----------------------------------------------------------------------------- + * DMA engine operations + */ + +static int rcar_dmac_alloc_chan_resources(struct dma_chan *chan) +{ + struct rcar_dmac_chan *rchan = to_rcar_dmac_chan(chan); + int ret; + + INIT_LIST_HEAD(&rchan->desc.free); + INIT_LIST_HEAD(&rchan->desc.pending); + INIT_LIST_HEAD(&rchan->desc.active); + INIT_LIST_HEAD(&rchan->desc.done); + INIT_LIST_HEAD(&rchan->desc.wait); + INIT_LIST_HEAD(&rchan->desc.chunks_free); + INIT_LIST_HEAD(&rchan->desc.pages); + + /* Preallocate descriptors. */ + ret = rcar_dmac_xfer_chunk_alloc(rchan, GFP_KERNEL); + if (ret < 0) + return -ENOMEM; + + ret = rcar_dmac_desc_alloc(rchan, GFP_KERNEL); + if (ret < 0) + return -ENOMEM; + + return pm_runtime_get_sync(chan->device->dev); +} + +static void rcar_dmac_free_chan_resources(struct dma_chan *chan) +{ + struct rcar_dmac_chan *rchan = to_rcar_dmac_chan(chan); + struct rcar_dmac *dmac = to_rcar_dmac(chan->device); + struct rcar_dmac_desc_page *page, *_page; + struct rcar_dmac_desc *desc; + LIST_HEAD(list); + + /* Protect against ISR */ + spin_lock_irq(&rchan->lock); + rcar_dmac_chan_halt(rchan); + spin_unlock_irq(&rchan->lock); + + /* Now no new interrupts will occur */ + + if (rchan->mid_rid >= 0) { + /* The caller is holding dma_list_mutex */ + clear_bit(rchan->mid_rid, dmac->modules); + rchan->mid_rid = -EINVAL; + } + + list_splice(&rchan->desc.free, &list); + list_splice(&rchan->desc.pending, &list); + list_splice(&rchan->desc.active, &list); + list_splice(&rchan->desc.done, &list); + list_splice(&rchan->desc.wait, &list); + + list_for_each_entry(desc, &list, node) + rcar_dmac_realloc_hwdesc(rchan, desc, 0); + + list_for_each_entry_safe(page, _page, &rchan->desc.pages, node) { + list_del(&page->node); + free_page((unsigned long)page); + } + + pm_runtime_put(chan->device->dev); +} + +static struct dma_async_tx_descriptor * +rcar_dmac_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dma_dest, + dma_addr_t dma_src, size_t len, unsigned long flags) +{ + struct rcar_dmac_chan *rchan = to_rcar_dmac_chan(chan); + struct scatterlist sgl; + + if (!len) + return NULL; + + sg_init_table(&sgl, 1); + sg_set_page(&sgl, pfn_to_page(PFN_DOWN(dma_src)), len, + offset_in_page(dma_src)); + sg_dma_address(&sgl) = dma_src; + sg_dma_len(&sgl) = len; + + return rcar_dmac_chan_prep_sg(rchan, &sgl, 1, dma_dest, + DMA_MEM_TO_MEM, flags, false); +} + +static struct dma_async_tx_descriptor * +rcar_dmac_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, + unsigned int sg_len, enum dma_transfer_direction dir, + unsigned long flags, void *context) +{ + struct rcar_dmac_chan *rchan = to_rcar_dmac_chan(chan); + dma_addr_t dev_addr; + + /* Someone calling slave DMA on a generic channel? */ + if (rchan->mid_rid < 0 || !sg_len) { + dev_warn(chan->device->dev, + "%s: bad parameter: len=%d, id=%d\n", + __func__, sg_len, rchan->mid_rid); + return NULL; + } + + dev_addr = dir == DMA_DEV_TO_MEM + ? rchan->src_slave_addr : rchan->dst_slave_addr; + return rcar_dmac_chan_prep_sg(rchan, sgl, sg_len, dev_addr, + dir, flags, false); +} + +#define RCAR_DMAC_MAX_SG_LEN 32 + +static struct dma_async_tx_descriptor * +rcar_dmac_prep_dma_cyclic(struct dma_chan *chan, dma_addr_t buf_addr, + size_t buf_len, size_t period_len, + enum dma_transfer_direction dir, unsigned long flags) +{ + struct rcar_dmac_chan *rchan = to_rcar_dmac_chan(chan); + struct dma_async_tx_descriptor *desc; + struct scatterlist *sgl; + dma_addr_t dev_addr; + unsigned int sg_len; + |