diff options
Diffstat (limited to 'drivers/dma')
-rw-r--r-- | drivers/dma/Kconfig | 11 | ||||
-rw-r--r-- | drivers/dma/Makefile | 4 | ||||
-rw-r--r-- | drivers/dma/amba-pl08x.c | 1215 | ||||
-rw-r--r-- | drivers/dma/at_hdmac.c | 21 | ||||
-rw-r--r-- | drivers/dma/fsldma.c | 10 | ||||
-rw-r--r-- | drivers/dma/fsldma.h | 9 | ||||
-rw-r--r-- | drivers/dma/imx-dma.c | 28 | ||||
-rw-r--r-- | drivers/dma/imx-sdma.c | 264 | ||||
-rw-r--r-- | drivers/dma/intel_mid_dma.c | 47 | ||||
-rw-r--r-- | drivers/dma/ioat/Makefile | 2 | ||||
-rw-r--r-- | drivers/dma/iop-adma.c | 4 | ||||
-rw-r--r-- | drivers/dma/ipu/ipu_idmac.c | 50 | ||||
-rw-r--r-- | drivers/dma/mpc512x_dma.c | 187 | ||||
-rw-r--r-- | drivers/dma/mv_xor.c | 2 | ||||
-rw-r--r-- | drivers/dma/pch_dma.c | 34 | ||||
-rw-r--r-- | drivers/dma/ppc4xx/adma.c | 5 | ||||
-rw-r--r-- | drivers/dma/shdma.c | 131 | ||||
-rw-r--r-- | drivers/dma/shdma.h | 1 | ||||
-rw-r--r-- | drivers/dma/ste_dma40.c | 1437 | ||||
-rw-r--r-- | drivers/dma/ste_dma40_ll.c | 302 | ||||
-rw-r--r-- | drivers/dma/ste_dma40_ll.h | 66 |
21 files changed, 1965 insertions, 1865 deletions
diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig index 6ee23592700a..1c28816152fa 100644 --- a/drivers/dma/Kconfig +++ b/drivers/dma/Kconfig @@ -109,7 +109,7 @@ config FSL_DMA config MPC512X_DMA tristate "Freescale MPC512x built-in DMA engine support" - depends on PPC_MPC512x + depends on PPC_MPC512x || PPC_MPC831x select DMA_ENGINE ---help--- Enable support for the Freescale MPC512x built-in DMA engine. @@ -200,11 +200,16 @@ config PL330_DMA platform_data for a dma-pl330 device. config PCH_DMA - tristate "Topcliff (Intel EG20T) PCH DMA support" + tristate "Intel EG20T PCH / OKI SEMICONDUCTOR ML7213 IOH DMA support" depends on PCI && X86 select DMA_ENGINE help - Enable support for the Topcliff (Intel EG20T) PCH DMA engine. + Enable support for Intel EG20T PCH DMA engine. + + This driver also can be used for OKI SEMICONDUCTOR ML7213 IOH(Input/ + Output Hub) which is for IVI(In-Vehicle Infotainment) use. + ML7213 is companion chip for Intel Atom E6xx series. + ML7213 is completely compatible for Intel EG20T PCH. config IMX_SDMA tristate "i.MX SDMA support" diff --git a/drivers/dma/Makefile b/drivers/dma/Makefile index a8a84f4587f2..64b21f5cd740 100644 --- a/drivers/dma/Makefile +++ b/drivers/dma/Makefile @@ -1,8 +1,8 @@ ifeq ($(CONFIG_DMADEVICES_DEBUG),y) - EXTRA_CFLAGS += -DDEBUG + ccflags-y += -DDEBUG endif ifeq ($(CONFIG_DMADEVICES_VDEBUG),y) - EXTRA_CFLAGS += -DVERBOSE_DEBUG + ccflags-y += -DVERBOSE_DEBUG endif obj-$(CONFIG_DMA_ENGINE) += dmaengine.o diff --git a/drivers/dma/amba-pl08x.c b/drivers/dma/amba-pl08x.c index b605cc9ac3a2..07bca4970e50 100644 --- a/drivers/dma/amba-pl08x.c +++ b/drivers/dma/amba-pl08x.c @@ -19,14 +19,14 @@ * this program; if not, write to the Free Software Foundation, Inc., 59 * Temple Place - Suite 330, Boston, MA 02111-1307, USA. * - * The full GNU General Public License is iin this distribution in the - * file called COPYING. + * The full GNU General Public License is in this distribution in the file + * called COPYING. * * Documentation: ARM DDI 0196G == PL080 - * Documentation: ARM DDI 0218E == PL081 + * Documentation: ARM DDI 0218E == PL081 * - * PL080 & PL081 both have 16 sets of DMA signals that can be routed to - * any channel. + * PL080 & PL081 both have 16 sets of DMA signals that can be routed to any + * channel. * * The PL080 has 8 channels available for simultaneous use, and the PL081 * has only two channels. So on these DMA controllers the number of channels @@ -53,7 +53,23 @@ * * ASSUMES default (little) endianness for DMA transfers * - * Only DMAC flow control is implemented + * The PL08x has two flow control settings: + * - DMAC flow control: the transfer size defines the number of transfers + * which occur for the current LLI entry, and the DMAC raises TC at the + * end of every LLI entry. Observed behaviour shows the DMAC listening + * to both the BREQ and SREQ signals (contrary to documented), + * transferring data if either is active. The LBREQ and LSREQ signals + * are ignored. + * + * - Peripheral flow control: the transfer size is ignored (and should be + * zero). The data is transferred from the current LLI entry, until + * after the final transfer signalled by LBREQ or LSREQ. The DMAC + * will then move to the next LLI entry. + * + * Only the former works sanely with scatter lists, so we only implement + * the DMAC flow control method. However, peripherals which use the LBREQ + * and LSREQ signals (eg, MMCI) are unable to use this mode, which through + * these hardware restrictions prevents them from using scatter DMA. * * Global TODO: * - Break out common code from arch/arm/mach-s3c64xx and share @@ -61,50 +77,40 @@ #include <linux/device.h> #include <linux/init.h> #include <linux/module.h> -#include <linux/pci.h> #include <linux/interrupt.h> #include <linux/slab.h> +#include <linux/delay.h> #include <linux/dmapool.h> -#include <linux/amba/bus.h> #include <linux/dmaengine.h> +#include <linux/amba/bus.h> #include <linux/amba/pl08x.h> #include <linux/debugfs.h> #include <linux/seq_file.h> #include <asm/hardware/pl080.h> -#include <asm/dma.h> -#include <asm/mach/dma.h> -#include <asm/atomic.h> -#include <asm/processor.h> -#include <asm/cacheflush.h> #define DRIVER_NAME "pl08xdmac" /** - * struct vendor_data - vendor-specific config parameters - * for PL08x derivates - * @name: the name of this specific variant + * struct vendor_data - vendor-specific config parameters for PL08x derivatives * @channels: the number of channels available in this variant - * @dualmaster: whether this version supports dual AHB masters - * or not. + * @dualmaster: whether this version supports dual AHB masters or not. */ struct vendor_data { - char *name; u8 channels; bool dualmaster; }; /* * PL08X private data structures - * An LLI struct - see pl08x TRM - * Note that next uses bit[0] as a bus bit, - * start & end do not - their bus bit info - * is in cctl + * An LLI struct - see PL08x TRM. Note that next uses bit[0] as a bus bit, + * start & end do not - their bus bit info is in cctl. Also note that these + * are fixed 32-bit quantities. */ -struct lli { - dma_addr_t src; - dma_addr_t dst; - dma_addr_t next; +struct pl08x_lli { + u32 src; + u32 dst; + u32 lli; u32 cctl; }; @@ -119,6 +125,8 @@ struct lli { * @phy_chans: array of data for the physical channels * @pool: a pool for the LLI descriptors * @pool_ctr: counter of LLIs in the pool + * @lli_buses: bitmask to or in to LLI pointer selecting AHB port for LLI fetches + * @mem_buses: set to indicate memory transfers on AHB2. * @lock: a spinlock for this struct */ struct pl08x_driver_data { @@ -126,11 +134,13 @@ struct pl08x_driver_data { struct dma_device memcpy; void __iomem *base; struct amba_device *adev; - struct vendor_data *vd; + const struct vendor_data *vd; struct pl08x_platform_data *pd; struct pl08x_phy_chan *phy_chans; struct dma_pool *pool; int pool_ctr; + u8 lli_buses; + u8 mem_buses; spinlock_t lock; }; @@ -152,9 +162,9 @@ struct pl08x_driver_data { /* Size (bytes) of each LLI buffer allocated for one transfer */ # define PL08X_LLI_TSFR_SIZE 0x2000 -/* Maximimum times we call dma_pool_alloc on this pool without freeing */ +/* Maximum times we call dma_pool_alloc on this pool without freeing */ #define PL08X_MAX_ALLOCS 0x40 -#define MAX_NUM_TSFR_LLIS (PL08X_LLI_TSFR_SIZE/sizeof(struct lli)) +#define MAX_NUM_TSFR_LLIS (PL08X_LLI_TSFR_SIZE/sizeof(struct pl08x_lli)) #define PL08X_ALIGN 8 static inline struct pl08x_dma_chan *to_pl08x_chan(struct dma_chan *chan) @@ -162,6 +172,11 @@ static inline struct pl08x_dma_chan *to_pl08x_chan(struct dma_chan *chan) return container_of(chan, struct pl08x_dma_chan, chan); } +static inline struct pl08x_txd *to_pl08x_txd(struct dma_async_tx_descriptor *tx) +{ + return container_of(tx, struct pl08x_txd, tx); +} + /* * Physical channel handling */ @@ -177,103 +192,63 @@ static int pl08x_phy_channel_busy(struct pl08x_phy_chan *ch) /* * Set the initial DMA register values i.e. those for the first LLI - * The next lli pointer and the configuration interrupt bit have - * been set when the LLIs were constructed + * The next LLI pointer and the configuration interrupt bit have + * been set when the LLIs were constructed. Poke them into the hardware + * and start the transfer. */ -static void pl08x_set_cregs(struct pl08x_driver_data *pl08x, - struct pl08x_phy_chan *ch) +static void pl08x_start_txd(struct pl08x_dma_chan *plchan, + struct pl08x_txd *txd) { - /* Wait for channel inactive */ - while (pl08x_phy_channel_busy(ch)) - ; - - dev_vdbg(&pl08x->adev->dev, - "WRITE channel %d: csrc=%08x, cdst=%08x, " - "cctl=%08x, clli=%08x, ccfg=%08x\n", - ch->id, - ch->csrc, - ch->cdst, - ch->cctl, - ch->clli, - ch->ccfg); - - writel(ch->csrc, ch->base + PL080_CH_SRC_ADDR); - writel(ch->cdst, ch->base + PL080_CH_DST_ADDR); - writel(ch->clli, ch->base + PL080_CH_LLI); - writel(ch->cctl, ch->base + PL080_CH_CONTROL); - writel(ch->ccfg, ch->base + PL080_CH_CONFIG); -} - -static inline void pl08x_config_phychan_for_txd(struct pl08x_dma_chan *plchan) -{ - struct pl08x_channel_data *cd = plchan->cd; + struct pl08x_driver_data *pl08x = plchan->host; struct pl08x_phy_chan *phychan = plchan->phychan; - struct pl08x_txd *txd = plchan->at; - - /* Copy the basic control register calculated at transfer config */ - phychan->csrc = txd->csrc; - phychan->cdst = txd->cdst; - phychan->clli = txd->clli; - phychan->cctl = txd->cctl; - - /* Assign the signal to the proper control registers */ - phychan->ccfg = cd->ccfg; - phychan->ccfg &= ~PL080_CONFIG_SRC_SEL_MASK; - phychan->ccfg &= ~PL080_CONFIG_DST_SEL_MASK; - /* If it wasn't set from AMBA, ignore it */ - if (txd->direction == DMA_TO_DEVICE) - /* Select signal as destination */ - phychan->ccfg |= - (phychan->signal << PL080_CONFIG_DST_SEL_SHIFT); - else if (txd->direction == DMA_FROM_DEVICE) - /* Select signal as source */ - phychan->ccfg |= - (phychan->signal << PL080_CONFIG_SRC_SEL_SHIFT); - /* Always enable error interrupts */ - phychan->ccfg |= PL080_CONFIG_ERR_IRQ_MASK; - /* Always enable terminal interrupts */ - phychan->ccfg |= PL080_CONFIG_TC_IRQ_MASK; -} - -/* - * Enable the DMA channel - * Assumes all other configuration bits have been set - * as desired before this code is called - */ -static void pl08x_enable_phy_chan(struct pl08x_driver_data *pl08x, - struct pl08x_phy_chan *ch) -{ + struct pl08x_lli *lli = &txd->llis_va[0]; u32 val; - /* - * Do not access config register until channel shows as disabled - */ - while (readl(pl08x->base + PL080_EN_CHAN) & (1 << ch->id)) - ; + plchan->at = txd; - /* - * Do not access config register until channel shows as inactive - */ - val = readl(ch->base + PL080_CH_CONFIG); + /* Wait for channel inactive */ + while (pl08x_phy_channel_busy(phychan)) + cpu_relax(); + + dev_vdbg(&pl08x->adev->dev, + "WRITE channel %d: csrc=0x%08x, cdst=0x%08x, " + "clli=0x%08x, cctl=0x%08x, ccfg=0x%08x\n", + phychan->id, lli->src, lli->dst, lli->lli, lli->cctl, + txd->ccfg); + + writel(lli->src, phychan->base + PL080_CH_SRC_ADDR); + writel(lli->dst, phychan->base + PL080_CH_DST_ADDR); + writel(lli->lli, phychan->base + PL080_CH_LLI); + writel(lli->cctl, phychan->base + PL080_CH_CONTROL); + writel(txd->ccfg, phychan->base + PL080_CH_CONFIG); + + /* Enable the DMA channel */ + /* Do not access config register until channel shows as disabled */ + while (readl(pl08x->base + PL080_EN_CHAN) & (1 << phychan->id)) + cpu_relax(); + + /* Do not access config register until channel shows as inactive */ + val = readl(phychan->base + PL080_CH_CONFIG); while ((val & PL080_CONFIG_ACTIVE) || (val & PL080_CONFIG_ENABLE)) - val = readl(ch->base + PL080_CH_CONFIG); + val = readl(phychan->base + PL080_CH_CONFIG); - writel(val | PL080_CONFIG_ENABLE, ch->base + PL080_CH_CONFIG); + writel(val | PL080_CONFIG_ENABLE, phychan->base + PL080_CH_CONFIG); } /* - * Overall DMAC remains enabled always. - * - * Disabling individual channels could lose data. + * Pause the channel by setting the HALT bit. * - * Disable the peripheral DMA after disabling the DMAC - * in order to allow the DMAC FIFO to drain, and - * hence allow the channel to show inactive + * For M->P transfers, pause the DMAC first and then stop the peripheral - + * the FIFO can only drain if the peripheral is still requesting data. + * (note: this can still timeout if the DMAC FIFO never drains of data.) * + * For P->M transfers, disable the peripheral first to stop it filling + * the DMAC FIFO, and then pause the DMAC. */ static void pl08x_pause_phy_chan(struct pl08x_phy_chan *ch) { u32 val; + int timeout; /* Set the HALT bit and wait for the FIFO to drain */ val = readl(ch->base + PL080_CH_CONFIG); @@ -281,8 +256,13 @@ static void pl08x_pause_phy_chan(struct pl08x_phy_chan *ch) writel(val, ch->base + PL080_CH_CONFIG); /* Wait for channel inactive */ - while (pl08x_phy_channel_busy(ch)) - ; + for (timeout = 1000; timeout; timeout--) { + if (!pl08x_phy_channel_busy(ch)) + break; + udelay(1); + } + if (pl08x_phy_channel_busy(ch)) + pr_err("pl08x: channel%u timeout waiting for pause\n", ch->id); } static void pl08x_resume_phy_chan(struct pl08x_phy_chan *ch) @@ -296,19 +276,24 @@ static void pl08x_resume_phy_chan(struct pl08x_phy_chan *ch) } -/* Stops the channel */ -static void pl08x_stop_phy_chan(struct pl08x_phy_chan *ch) +/* + * pl08x_terminate_phy_chan() stops the channel, clears the FIFO and + * clears any pending interrupt status. This should not be used for + * an on-going transfer, but as a method of shutting down a channel + * (eg, when it's no longer used) or terminating a transfer. + */ +static void pl08x_terminate_phy_chan(struct pl08x_driver_data *pl08x, + struct pl08x_phy_chan *ch) { - u32 val; + u32 val = readl(ch->base + PL080_CH_CONFIG); - pl08x_pause_phy_chan(ch); + val &= ~(PL080_CONFIG_ENABLE | PL080_CONFIG_ERR_IRQ_MASK | + PL080_CONFIG_TC_IRQ_MASK); - /* Disable channel */ - val = readl(ch->base + PL080_CH_CONFIG); - val &= ~PL080_CONFIG_ENABLE; - val &= ~PL080_CONFIG_ERR_IRQ_MASK; - val &= ~PL080_CONFIG_TC_IRQ_MASK; writel(val, ch->base + PL080_CH_CONFIG); + + writel(1 << ch->id, pl08x->base + PL080_ERR_CLEAR); + writel(1 << ch->id, pl08x->base + PL080_TC_CLEAR); } static inline u32 get_bytes_in_cctl(u32 cctl) @@ -333,54 +318,56 @@ static inline u32 get_bytes_in_cctl(u32 cctl) static u32 pl08x_getbytes_chan(struct pl08x_dma_chan *plchan) { struct pl08x_phy_chan *ch; - struct pl08x_txd *txdi = NULL; struct pl08x_txd *txd; unsigned long flags; - u32 bytes = 0; + size_t bytes = 0; spin_lock_irqsave(&plchan->lock, flags); - ch = plchan->phychan; txd = plchan->at; /* - * Next follow the LLIs to get the number of pending bytes in the - * currently active transaction. + * Follow the LLIs to get the number of remaining + * bytes in the currently active transaction. */ if (ch && txd) { - struct lli *llis_va = txd->llis_va; - struct lli *llis_bus = (struct lli *) txd->llis_bus; - u32 clli = readl(ch->base + PL080_CH_LLI); + u32 clli = readl(ch->base + PL080_CH_LLI) & ~PL080_LLI_LM_AHB2; - /* First get the bytes in the current active LLI */ + /* First get the remaining bytes in the active transfer */ bytes = get_bytes_in_cctl(readl(ch->base + PL080_CH_CONTROL)); if (clli) { - int i = 0; + struct pl08x_lli *llis_va = txd->llis_va; + dma_addr_t llis_bus = txd->llis_bus; + int index; + + BUG_ON(clli < llis_bus || clli >= llis_bus + + sizeof(struct pl08x_lli) * MAX_NUM_TSFR_LLIS); + + /* + * Locate the next LLI - as this is an array, + * it's simple maths to find. + */ + index = (clli - llis_bus) / sizeof(struct pl08x_lli); - /* Forward to the LLI pointed to by clli */ - while ((clli != (u32) &(llis_bus[i])) && - (i < MAX_NUM_TSFR_LLIS)) - i++; + for (; index < MAX_NUM_TSFR_LLIS; index++) { + bytes += get_bytes_in_cctl(llis_va[index].cctl); - while (clli) { - bytes += get_bytes_in_cctl(llis_va[i].cctl); /* - * A clli of 0x00000000 will terminate the - * LLI list + * A LLI pointer of 0 terminates the LLI list */ - clli = llis_va[i].next; - i++; + if (!llis_va[index].lli) + break; } } } /* Sum up all queued transactions */ - if (!list_empty(&plchan->desc_list)) { - list_for_each_entry(txdi, &plchan->desc_list, node) { + if (!list_empty(&plchan->pend_list)) { + struct pl08x_txd *txdi; + list_for_each_entry(txdi, &plchan->pend_list, node) { bytes += txdi->len; } - } spin_unlock_irqrestore(&plchan->lock, flags); @@ -390,6 +377,10 @@ static u32 pl08x_getbytes_chan(struct pl08x_dma_chan *plchan) /* * Allocate a physical channel for a virtual channel + * + * Try to locate a physical channel to be used for this transfer. If all + * are taken return NULL and the requester will have to cope by using + * some fallback PIO mode or retrying later. */ static struct pl08x_phy_chan * pl08x_get_phy_channel(struct pl08x_driver_data *pl08x, @@ -399,12 +390,6 @@ pl08x_get_phy_channel(struct pl08x_driver_data *pl08x, unsigned long flags; int i; - /* - * Try to locate a physical channel to be used for - * this transfer. If all are taken return NULL and - * the requester will have to cope by using some fallback - * PIO mode or retrying later. - */ for (i = 0; i < pl08x->vd->channels; i++) { ch = &pl08x->phy_chans[i]; @@ -433,13 +418,12 @@ static inline void pl08x_put_phy_channel(struct pl08x_driver_data *pl08x, { unsigned long flags; + spin_lock_irqsave(&ch->lock, flags); + /* Stop the channel and clear its interrupts */ - pl08x_stop_phy_chan(ch); - writel((1 << ch->id), pl08x->base + PL080_ERR_CLEAR); - writel((1 << ch->id), pl08x->base + PL080_TC_CLEAR); + pl08x_terminate_phy_chan(pl08x, ch); /* Mark it as free */ - spin_lock_irqsave(&ch->lock, flags); ch->serving = NULL; spin_unlock_irqrestore(&ch->lock, flags); } @@ -465,11 +449,11 @@ static inline unsigned int pl08x_get_bytes_for_cctl(unsigned int coded) } static inline u32 pl08x_cctl_bits(u32 cctl, u8 srcwidth, u8 dstwidth, - u32 tsize) + size_t tsize) { u32 retbits = cctl; - /* Remove all src, dst and transfersize bits */ + /* Remove all src, dst and transfer size bits */ retbits &= ~PL080_CONTROL_DWIDTH_MASK; retbits &= ~PL080_CONTROL_SWIDTH_MASK; retbits &= ~PL080_CONTROL_TRANSFER_SIZE_MASK; @@ -509,95 +493,87 @@ static inline u32 pl08x_cctl_bits(u32 cctl, u8 srcwidth, u8 dstwidth, return retbits; } +struct pl08x_lli_build_data { + struct pl08x_txd *txd; + struct pl08x_driver_data *pl08x; + struct pl08x_bus_data srcbus; + struct pl08x_bus_data dstbus; + size_t remainder; +}; + /* - * Autoselect a master bus to use for the transfer - * this prefers the destination bus if both available - * if fixed address on one bus the other will be chosen + * Autoselect a master bus to use for the transfer this prefers the + * destination bus if both available if fixed address on one bus the + * other will be chosen */ -void pl08x_choose_master_bus(struct pl08x_bus_data *src_bus, - struct pl08x_bus_data *dst_bus, struct pl08x_bus_data **mbus, - struct pl08x_bus_data **sbus, u32 cctl) +static void pl08x_choose_master_bus(struct pl08x_lli_build_data *bd, + struct pl08x_bus_data **mbus, struct pl08x_bus_data **sbus, u32 cctl) { if (!(cctl & PL080_CONTROL_DST_INCR)) { - *mbus = src_bus; - *sbus = dst_bus; + *mbus = &bd->srcbus; + *sbus = &bd->dstbus; } else if (!(cctl & PL080_CONTROL_SRC_INCR)) { - *mbus = dst_bus; - *sbus = src_bus; + *mbus = &bd->dstbus; + *sbus = &bd->srcbus; } else { - if (dst_bus->buswidth == 4) { - *mbus = dst_bus; - *sbus = src_bus; - } else if (src_bus->buswidth == 4) { - *mbus = src_bus; - *sbus = dst_bus; - } else if (dst_bus->buswidth == 2) { - *mbus = dst_bus; - *sbus = src_bus; - } else if (src_bus->buswidth == 2) { - *mbus = src_bus; - *sbus = dst_bus; + if (bd->dstbus.buswidth == 4) { + *mbus = &bd->dstbus; + *sbus = &bd->srcbus; + } else if (bd->srcbus.buswidth == 4) { + *mbus = &bd->srcbus; + *sbus = &bd->dstbus; + } else if (bd->dstbus.buswidth == 2) { + *mbus = &bd->dstbus; + *sbus = &bd->srcbus; + } else if (bd->srcbus.buswidth == 2) { + *mbus = &bd->srcbus; + *sbus = &bd->dstbus; } else { - /* src_bus->buswidth == 1 */ - *mbus = dst_bus; - *sbus = src_bus; + /* bd->srcbus.buswidth == 1 */ + *mbus = &bd->dstbus; + *sbus = &bd->srcbus; } } } /* - * Fills in one LLI for a certain transfer descriptor - * and advance the counter + * Fills in one LLI for a certain transfer descriptor and advance the counter */ -int pl08x_fill_lli_for_desc(struct pl08x_driver_data *pl08x, - struct pl08x_txd *txd, int num_llis, int len, - u32 cctl, u32 *remainder) +static void pl08x_fill_lli_for_desc(struct pl08x_lli_build_data *bd, + int num_llis, int len, u32 cctl) { - struct lli *llis_va = txd->llis_va; - struct lli *llis_bus = (struct lli *) txd->llis_bus; + struct pl08x_lli *llis_va = bd->txd->llis_va; + dma_addr_t llis_bus = bd->txd->llis_bus; BUG_ON(num_llis >= MAX_NUM_TSFR_LLIS); - llis_va[num_llis].cctl = cctl; - llis_va[num_llis].src = txd->srcbus.addr; - llis_va[num_llis].dst = txd->dstbus.addr; - - /* - * On versions with dual masters, you can optionally AND on - * PL080_LLI_LM_AHB2 to the LLI to tell the hardware to read - * in new LLIs with that controller, but we always try to - * choose AHB1 to point into memory. The idea is to have AHB2 - * fixed on the peripheral and AHB1 messing around in the - * memory. So we don't manipulate this bit currently. - */ - - llis_va[num_llis].next = - (dma_addr_t)((u32) &(llis_bus[num_llis + 1])); + llis_va[num_llis].cctl = cctl; + llis_va[num_llis].src = bd->srcbus.addr; + llis_va[num_llis].dst = bd->dstbus.addr; + llis_va[num_llis].lli = llis_bus + (num_llis + 1) * sizeof(struct pl08x_lli); + if (bd->pl08x->lli_buses & PL08X_AHB2) + llis_va[num_llis].lli |= PL080_LLI_LM_AHB2; if (cctl & PL080_CONTROL_SRC_INCR) - txd->srcbus.addr += len; + bd->srcbus.addr += len; if (cctl & PL080_CONTROL_DST_INCR) - txd->dstbus.addr += len; + bd->dstbus.addr += len; - *remainder -= len; + BUG_ON(bd->remainder < len); - return num_llis + 1; + bd->remainder -= len; } /* - * Return number of bytes to fill to boundary, or len + * Return number of bytes to fill to boundary, or len. + * This calculation works for any value of addr. */ -static inline u32 pl08x_pre_boundary(u32 addr, u32 len) +static inline size_t pl08x_pre_boundary(u32 addr, size_t len) { - u32 boundary; + size_t boundary_len = PL08X_BOUNDARY_SIZE - + (addr & (PL08X_BOUNDARY_SIZE - 1)); - boundary = ((addr >> PL08X_BOUNDARY_SHIFT) + 1) - << PL08X_BOUNDARY_SHIFT; - - if (boundary < addr + len) - return boundary - addr; - else - return len; + return min(boundary_len, len); } /* @@ -608,20 +584,13 @@ static inline u32 pl08x_pre_boundary(u32 addr, u32 len) static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, struct pl08x_txd *txd) { - struct pl08x_channel_data *cd = txd->cd; struct pl08x_bus_data *mbus, *sbus; - u32 remainder; + struct pl08x_lli_build_data bd; int num_llis = 0; u32 cctl; - int max_bytes_per_lli; - int total_bytes = 0; - struct lli *llis_va; - struct lli *llis_bus; - - if (!txd) { - dev_err(&pl08x->adev->dev, "%s no descriptor\n", __func__); - return 0; - } + size_t max_bytes_per_lli; + size_t total_bytes = 0; + struct pl08x_lli *llis_va; txd->llis_va = dma_pool_alloc(pl08x->pool, GFP_NOWAIT, &txd->llis_bus); @@ -632,121 +601,79 @@ static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, pl08x->pool_ctr++; - /* - * Initialize bus values for this transfer - * from the passed optimal values - */ - if (!cd) { - dev_err(&pl08x->adev->dev, "%s no channel data\n", __func__); - return 0; - } - - /* Get the default CCTL from the platform data */ - cctl = cd->cctl; + /* Get the default CCTL */ + cctl = txd->cctl; - /* - * On the PL080 we have two bus masters and we - * should select one for source and one for - * destination. We try to use AHB2 for the - * bus which does not increment (typically the - * peripheral) else we just choose something. - */ - cctl &= ~(PL080_CONTROL_DST_AHB2 | PL080_CONTROL_SRC_AHB2); - if (pl08x->vd->dualmaster) { - if (cctl & PL080_CONTROL_SRC_INCR) - /* Source increments, use AHB2 for destination */ - cctl |= PL080_CONTROL_DST_AHB2; - else if (cctl & PL080_CONTROL_DST_INCR) - /* Destination increments, use AHB2 for source */ - cctl |= PL080_CONTROL_SRC_AHB2; - else - /* Just pick something, source AHB1 dest AHB2 */ - cctl |= PL080_CONTROL_DST_AHB2; - } + bd.txd = txd; + bd.pl08x = pl08x; + bd.srcbus.addr = txd->src_addr; + bd.dstbus.addr = txd->dst_addr; /* Find maximum width of the source bus */ - txd->srcbus.maxwidth = + bd.srcbus.maxwidth = pl08x_get_bytes_for_cctl((cctl & PL080_CONTROL_SWIDTH_MASK) >> PL080_CONTROL_SWIDTH_SHIFT); /* Find maximum width of the destination bus */ - txd->dstbus.maxwidth = + bd.dstbus.maxwidth = pl08x_get_bytes_for_cctl((cctl & PL080_CONTROL_DWIDTH_MASK) >> PL080_CONTROL_DWIDTH_SHIFT); /* Set up the bus widths to the maximum */ - txd->srcbus.buswidth = txd->srcbus.maxwidth; - txd->dstbus.buswidth = txd->dstbus.maxwidth; + bd.srcbus.buswidth = bd.srcbus.maxwidth; + bd.dstbus.buswidth = bd.dstbus.maxwidth; dev_vdbg(&pl08x->adev->dev, "%s source bus is %d bytes wide, dest bus is %d bytes wide\n", - __func__, txd->srcbus.buswidth, txd->dstbus.buswidth); + __func__, bd.srcbus.buswidth, bd.dstbus.buswidth); /* * Bytes transferred == tsize * MIN(buswidths), not max(buswidths) */ - max_bytes_per_lli = min(txd->srcbus.buswidth, txd->dstbus.buswidth) * + max_bytes_per_lli = min(bd.srcbus.buswidth, bd.dstbus.buswidth) * PL080_CONTROL_TRANSFER_SIZE_MASK; dev_vdbg(&pl08x->adev->dev, - "%s max bytes per lli = %d\n", + "%s max bytes per lli = %zu\n", __func__, max_bytes_per_lli); /* We need to count this down to zero */ - remainder = txd->len; + bd.remainder = txd->len; dev_vdbg(&pl08x->adev->dev, - "%s remainder = %d\n", - __func__, remainder); + "%s remainder = %zu\n", + __func__, bd.remainder); /* * Choose bus to align to * - prefers destination bus if both available * - if fixed address on one bus chooses other - * - modifies cctl to choose an apropriate master - */ - pl08x_choose_master_bus(&txd->srcbus, &txd->dstbus, - &mbus, &sbus, cctl); - - - /* - * The lowest bit of the LLI register - * is also used to indicate which master to - * use for reading the LLIs. */ + pl08x_choose_master_bus(&bd, &mbus, &sbus, cctl); if (txd->len < mbus->buswidth) { - /* - * Less than a bus width available - * - send as single bytes - */ - while (remainder) { + /* Less than a bus width available - send as single bytes */ + while (bd.remainder) { dev_vdbg(&pl08x->adev->dev, "%s single byte LLIs for a transfer of " - "less than a bus width (remain %08x)\n", - __func__, remainder); + "less than a bus width (remain 0x%08x)\n", + __func__, bd.remainder); cctl = pl08x_cctl_bits(cctl, 1, 1, 1); - num_llis = - pl08x_fill_lli_for_desc(pl08x, txd, num_llis, 1, - cctl, &remainder); + pl08x_fill_lli_for_desc(&bd, num_llis++, 1, cctl); total_bytes++; } } else { - /* - * Make one byte LLIs until master bus is aligned - * - slave will then be aligned also - */ + /* Make one byte LLIs until master bus is aligned */ while ((mbus->addr) % (mbus->buswidth)) { dev_vdbg(&pl08x->adev->dev, "%s adjustment lli for less than bus width " - "(remain %08x)\n", - __func__, remainder); + "(remain 0x%08x)\n", + __func__, bd.remainder); cctl = pl08x_cctl_bits(cctl, 1, 1, 1); - num_llis = pl08x_fill_lli_for_desc - (pl08x, txd, num_llis, 1, cctl, &remainder); + pl08x_fill_lli_for_desc(&bd, num_llis++, 1, cctl); total_bytes++; } /* - * Master now aligned + * Master now aligned * - if slave is not then we must set its width down */ if (sbus->addr % sbus->buswidth) { @@ -761,63 +688,51 @@ static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, * Make largest possible LLIs until less than one bus * width left */ - while (remainder > (mbus->buswidth - 1)) { - int lli_len, target_len; - int tsize; - int odd_bytes; + while (bd.remainder > (mbus->buswidth - 1)) { + size_t lli_len, target_len, tsize, odd_bytes; /* * If enough left try to send max possible, * otherwise try to send the remainder */ - target_len = remainder; - if (remainder > max_bytes_per_lli) - target_len = max_bytes_per_lli; + target_len = min(bd.remainder, max_bytes_per_lli); /* - * Set bus lengths for incrementing busses - * to number of bytes which fill to next memory - * boundary + * Set bus lengths for incrementing buses to the + * number of bytes which fill to next memory boundary, + * limiting on the target length calculated above. */ if (cctl & PL080_CONTROL_SRC_INCR) - txd->srcbus.fill_bytes = - pl08x_pre_boundary( - txd->srcbus.addr, - remainder); + bd.srcbus.fill_bytes = + pl08x_pre_boundary(bd.srcbus.addr, + target_len); else - txd->srcbus.fill_bytes = - max_bytes_per_lli; + bd.srcbus.fill_bytes = target_len; if (cctl & PL080_CONTROL_DST_INCR) - txd->dstbus.fill_bytes = - pl08x_pre_boundary( - txd->dstbus.addr, - remainder); + bd.dstbus.fill_bytes = + pl08x_pre_boundary(bd.dstbus.addr, + target_len); else - txd->dstbus.fill_bytes = - max_bytes_per_lli; + bd.dstbus.fill_bytes = target_len; - /* - * Find the nearest - */ - lli_len = min(txd->srcbus.fill_bytes, - txd->dstbus.fill_bytes); + /* Find the nearest */ + lli_len = min(bd.srcbus.fill_bytes, + bd.dstbus.fill_bytes); - BUG_ON(lli_len > remainder); + BUG_ON(lli_len > bd.remainder); if (lli_len <= 0) { dev_err(&pl08x->adev->dev, - "%s lli_len is %d, <= 0\n", + "%s lli_len is %zu, <= 0\n", __func__, lli_len); return 0; } if (lli_len == target_len) { /* - * Can send what we wanted - */ - /* - * Maintain alignment + * Can send what we wanted. + * Maintain alignment */ lli_len = (lli_len/mbus->buswidth) * mbus->buswidth; @@ -825,17 +740,14 @@ static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, } else { /* * So now we know how many bytes to transfer - * to get to the nearest boundary - * The next lli will past the boundary - * - however we may be working to a boundary - * on the slave bus - * We need to ensure the master stays aligned + * to get to the nearest boundary. The next + * LLI will past the boundary. However, we + * may be working to a boundary on the slave + * bus. We need to ensure the master stays + * aligned, and that we are working in + * multiples of the bus widths. */ odd_bytes = lli_len % mbus->buswidth; - /* - * - and that we are working in multiples - * of the bus widths - */ lli_len -= odd_bytes; } @@ -855,41 +767,38 @@ static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, if (target_len != lli_len) { dev_vdbg(&pl08x->adev->dev, - "%s can't send what we want. Desired %08x, lli of %08x bytes in txd of %08x\n", + "%s can't send what we want. Desired 0x%08zx, lli of 0x%08zx bytes in txd of 0x%08zx\n", __func__, target_len, lli_len, txd->len); } cctl = pl08x_cctl_bits(cctl, - txd->srcbus.buswidth, - txd->dstbus.buswidth, + bd.srcbus.buswidth, + bd.dstbus.buswidth, tsize); dev_vdbg(&pl08x->adev->dev, - "%s fill lli with single lli chunk of size %08x (remainder %08x)\n", - __func__, lli_len, remainder); - num_llis = pl08x_fill_lli_for_desc(pl08x, txd, - num_llis, lli_len, cctl, - &remainder); + "%s fill lli with single lli chunk of size 0x%08zx (remainder 0x%08zx)\n", + __func__, lli_len, bd.remainder); + pl08x_fill_lli_for_desc(&bd, num_llis++, + lli_len, cctl); total_bytes += lli_len; } if (odd_bytes) { /* - * Creep past the boundary, - * maintaining master alignment + * Creep past the boundary, maintaining + * master alignment */ int j; for (j = 0; (j < mbus->buswidth) - && (remainder); j++) { + && (bd.remainder); j++) { cctl = pl08x_cctl_bits(cctl, 1, 1, 1); dev_vdbg(&pl08x->adev->dev, - "%s align with boundardy, single byte (remain %08x)\n", - __func__, remainder); - num_llis = - pl08x_fill_lli_for_desc(pl08x, - txd, num_llis, 1, - cctl, &remainder); + "%s align with boundary, single byte (remain 0x%08zx)\n", + __func__, bd.remainder); + pl08x_fill_lli_for_desc(&bd, + num_llis++, 1, cctl); total_bytes++; } } @@ -898,25 +807,18 @@ static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, /* * Send any odd bytes */ - if (remainder < 0) { - dev_err(&pl08x->adev->dev, "%s remainder not fitted 0x%08x bytes\n", - __func__, remainder); - return 0; - } - - while (remainder) { + while (bd.remainder) { cctl = pl08x_cctl_bits(cctl, 1, 1, 1); dev_vdbg(&pl08x->adev->dev, - "%s align with boundardy, single odd byte (remain %d)\n", - __func__, remainder); - num_llis = pl08x_fill_lli_for_desc(pl08x, txd, num_llis, - 1, cctl, &remainder); + "%s align with boundary, single odd byte (remain %zu)\n", + __func__, bd.remainder); + pl08x_fill_lli_for_desc(&bd, num_llis++, 1, cctl); total_bytes++; } } if (total_bytes != txd->len) { dev_err(&pl08x->adev->dev, - "%s size of encoded lli:s don't match total txd, transferred 0x%08x from size 0x%08x\n", + "%s size of encoded lli:s don't match total txd, transferred 0x%08zx from size 0x%08zx\n", __func__, total_bytes, txd->len); return 0; } @@ -927,41 +829,12 @@ static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, __func__, (u32) MAX_NUM_TSFR_LLIS); return 0; } - /* - * Decide whether this is a loop or a terminated transfer - */ - llis_va = txd->llis_va; - llis_bus = (struct lli *) txd->llis_bus; - - if (cd->circular_buffer) { - /* - * Loop the circular buffer so that the next element - * points back to the beginning of the LLI. - */ - llis_va[num_llis - 1].next = - (dma_addr_t)((unsigned int)&(llis_bus[0])); - } else { - /* - * On non-circular buffers, the final LLI terminates - * the LLI. - */ - llis_va[num_llis - 1].next = 0; - /* - * The final LLI element shall also fire an interrupt - */ - llis_va[num_llis - 1].cctl |= PL080_CONTROL_TC_IRQ_EN; - } - - /* Now store the channel register values */ - txd->csrc = llis_va[0].src; - txd->cdst = llis_va[0].dst; - if (num_llis > 1) - txd->clli = llis_va[0].next; - else - txd->clli = 0; - txd->cctl = llis_va[0].cctl; - /* ccfg will be set at physical channel allocation time */ + llis_va = txd->llis_va; + /* The final LLI terminates the LLI. */ + llis_va[num_llis - 1].lli = 0; + /* The final LLI element shall also fire an interrupt. */ + llis_va[num_llis - 1].cctl |= PL080_CONTROL_TC_IRQ_EN; #ifdef VERBOSE_DEBUG { @@ -969,13 +842,13 @@ static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, for (i = 0; i < num_llis; i++) { dev_vdbg(&pl08x->adev->dev, - "lli %d @%p: csrc=%08x, cdst=%08x, cctl=%08x, clli=%08x\n", + "lli %d @%p: csrc=0x%08x, cdst=0x%08x, cctl=0x%08x, clli=0x%08x\n", i, &llis_va[i], llis_va[i].src, llis_va[i].dst, llis_va[i].cctl, - llis_va[i].next + llis_va[i].lli ); } } @@ -988,14 +861,8 @@ static int pl08x_fill_llis_for_desc(struct pl08x_driver_data *pl08x, static void pl08x_free_txd(struct pl08x_driver_data *pl08x, struct pl08x_txd *txd) { - if (!txd) - dev_err(&pl08x->adev->dev, - "%s no descriptor to free\n", - __func__); - /* Free the LLI */ - dma_pool_free(pl08x->pool, txd->llis_va, - txd->llis_bus); + dma_pool_free(pl08x->pool, txd->llis_va, txd->llis_bus); pl08x->pool_ctr--; @@ -1008,13 +875,12 @@ static void pl08x_free_txd_list(struct pl08x_driver_data *pl08x, struct pl08x_txd *txdi = NULL; struct pl08x_txd *next; - if (!list_empty(&plchan->desc_list)) { + if (!list_empty(&plchan->pend_list)) { list_for_each_entry_safe(txdi, - next, &plchan->desc_list, node) { + next, &plchan->pend_list, node) { list_del(&txdi->node); pl08x_free_txd(pl08x, txdi); } - } } @@ -1069,6 +935,12 @@ static int prep_phy_channel(struct pl08x_dma_chan *plchan, return -EBUSY; } ch->signal = ret; + + /* Assign the flow control signal to this channel */ + if (txd->direction == DMA_TO_DEVICE) + txd->ccfg |= ch->signal << PL080_CONFIG_DST_SEL_SHIFT; + else if (txd->direction == DMA_FROM_DEVICE) + txd->ccfg |= ch->signal << PL080_CONFIG_SRC_SEL_SHIFT; } dev_dbg(&pl08x->adev->dev, "allocated physical channel %d and signal %d for xfer on %s\n", @@ -1076,19 +948,54 @@ static int prep_phy_channel(struct pl08x_dma_chan *plchan, ch->signal, plchan->name); + plchan->phychan_hold++; plchan->phychan = ch; return 0; } +static void release_phy_channel(struct pl08x_dma_chan *plchan) +{ + struct pl08x_driver_data *pl08x = plchan->host; + + if ((plchan->phychan->signal >= 0) && pl08x->pd->put_signal) { + pl08x->pd->put_signal(plchan); + plchan->phychan->signal = -1; + } + pl08x_put_phy_channel(pl08x, plchan->phychan); + plchan->phychan = NULL; +} + static dma_cookie_t pl08x_tx_submit(struct dma_async_tx_descriptor *tx) { struct pl08x_dma_chan *plchan = to_pl08x_chan(tx->chan); + struct pl08x_txd *txd = to_pl08x_txd(tx); + unsigned long flags; + + spin_lock_irqsave(&plchan->lock, flags); + + plchan->chan.cookie += 1; + if (plchan->chan.cookie < 0) + plchan->chan.cookie = 1; + tx->cookie = plchan->chan.cookie; + + /* Put this onto the pending list */ + list_add_tail(&txd->node, &plchan->pend_list); + + /* + * If there was no physical channel available for this memcpy, + * stack the request up and indicate that the channel is waiting + * for a free physical channel. + */ + if (!plchan->slave && !plchan->phychan) { + /* Do this memcpy whenever there is a channel ready */ + plchan->state = PL08X_CHAN_WAITING; + plchan->waiting = txd; + } else { + plchan->phychan_hold--; + } - atomic_inc(&plchan->last_issued); - tx->cookie = atomic_read(&plchan->last_issued); - /* This unlock follows the lock in the prep() function */ - spin_unlock_irqrestore(&plchan->lock, plchan->lockflags); + spin_unlock_irqrestore(&plchan->lock, flags); return tx->cookie; } @@ -1102,10 +1009,9 @@ static struct dma_async_tx_descriptor *pl08x_prep_dma_interrupt( } /* - * Code accessing dma_async_is_complete() in a tight loop - * may give problems - could schedule where indicated. - * If slaves are relying on interrupts to signal completion this - * function must not be called with interrupts disabled + * Code accessing dma_async_is_complete() in a tight loop may give problems. + * If slaves are relying on interrupts to signal completion this function + * must not be called with interrupts disabled. */ static enum dma_status pl08x_dma_tx_status(struct dma_chan *chan, @@ -1118,7 +1024,7 @@ pl08x_dma_tx_status(struct dma_chan *chan, enum dma_status ret; u32 bytesleft = 0; - last_used = atomic_read(&plchan->last_issued); + last_used = plchan->chan.cookie; last_complete = plchan->lc; ret = dma_async_is_complete(cookie, last_complete, last_used); @@ -1128,13 +1034,9 @@ pl08x_dma_tx_status(struct dma_chan *chan, } /* - * schedule(); could be inserted here - */ - - /* * This cookie not complete yet */ - last_used = atomic_read(&plchan->last_issued); + last_used = plchan->chan.cookie; last_complete = plchan->lc; /* Get number of bytes left in the active transactions and queue */ @@ -1199,37 +1101,35 @@ static const struct burst_table burst_sizes[] = { }, }; -static void dma_set_runtime_config(struct dma_chan *chan, - struct dma_slave_config *config) +static int dma_set_runtime_config(struct dma_chan *chan, + struct dma_slave_config *config) { struct pl08x_dma_chan *plchan = to_pl08x_chan(chan); struct pl08x_driver_data *pl08x = plchan->host; struct pl08x_channel_data *cd = plchan->cd; enum dma_slave_buswidth addr_width; + dma_addr_t addr; u32 maxburst; u32 cctl = 0; - /* Mask out all except src and dst channel */ - u32 ccfg = cd->ccfg & 0x000003DEU; - int i = 0; + int i; + + if (!plchan->slave) + return -EINVAL; /* Transfer direction */ plchan->runtime_direction = config->direction; if (config->direction == DMA_TO_DEVICE) { - plchan->runtime_addr = config->dst_addr; - cctl |= PL080_CONTROL_SRC_INCR; - ccfg |= PL080_FLOW_MEM2PER << PL080_CONFIG_FLOW_CONTROL_SHIFT; + addr = config->dst_addr; addr_width = config->dst_addr_width; maxburst = config->dst_maxburst; } else if (config->direction == DMA_FROM_DEVICE) { - plchan->runtime_addr = config->src_addr; - cctl |= PL080_CONTROL_DST_INCR; - ccfg |= PL080_FLOW_PER2MEM << PL080_CONFIG_FLOW_CONTROL_SHIFT; + addr = config->src_addr; addr_width = config->src_addr_width; maxburst = config->src_maxburst; } else { dev_err(&pl08x->adev->dev, "bad runtime_config: alien transfer direction\n"); - return; + return -EINVAL; } switch (addr_width) { @@ -1248,42 +1148,40 @@ static void dma_set_runtime_config(struct dma_chan *chan, default: dev_err(&pl08x->adev->dev, "bad runtime_config: alien address width\n"); - return; + return -EINVAL; } /* * Now decide on a maxburst: - * If this channel will only request single transfers, set - * this down to ONE element. + * If this channel will only request single transfers, set this + * down to ONE element. Also select one element if no maxburst + * is specified. */ - if (plchan->cd->single) { + if (plchan->cd->single || maxburst == 0) { cctl |= (PL080_BSIZE_1 << PL080_CONTROL_SB_SIZE_SHIFT) | (PL080_BSIZE_1 << PL080_CONTROL_DB_SIZE_SHIFT); } else { - while (i < ARRAY_SIZE(burst_sizes)) { + for (i = 0; i < ARRAY_SIZE(burst_sizes); i++) if (burst_sizes[i].burstwords <= maxburst) break; - i++; - } cctl |= burst_sizes[i].reg; } - /* Access the cell in privileged mode, non-bufferable, non-cacheable */ - cctl &= ~PL080_CONTROL_PROT_MASK; - cctl |= PL080_CONTROL_PROT_SYS; + plchan->runtime_addr = addr; /* Modify the default channel data to fit PrimeCell request */ cd->cctl = cctl; - cd->ccfg = ccfg; dev_dbg(&pl08x->adev->dev, "configured channel %s (%s) for %s, data width %d, " - "maxburst %d words, LE, CCTL=%08x, CCFG=%08x\n", + "maxburst %d words, LE, CCTL=0x%08x\n", dma_chan_name(chan), plchan->name, (config->direction == DMA_FROM_DEVICE) ? "RX" : "TX", addr_width, maxburst, - cctl, ccfg); + cctl); + + return 0; } /* @@ -1293,35 +1191,26 @@ static void dma_set_runtime_config(struct dma_chan *chan, static void pl08x_issue_pending(struct dma_chan *chan) { struct pl08x_dma_chan *plchan = to_pl08x_chan(chan); - struct pl08x_driver_data *pl08x = plchan->host; unsigned long flags; spin_lock_irqsave(&plchan->lock, flags); - /* Something is already active */ - if (plchan->at) { - spin_unlock_irqrestore(&plchan->lock, flags); - return; - } - - /* Didn't get a physical channel so waiting for it ... */ - if (plchan->state == PL08X_CHAN_WAITING) + /* Something is already active, or we're waiting for a channel... */ + if (plchan->at || plchan->state == PL08X_CHAN_WAITING) { + spin_unlock_irqrestore(&plchan->lock, flags); return; + } /* Take the first element in the queue and execute it */ - if (!list_empty(&plchan->desc_list)) { + if (!list_empty(&plchan->pend_list)) { struct pl08x_txd *next; - next = list_first_entry(&plchan->desc_list, + next = list_first_entry(&plchan->pend_list, struct pl08x_txd, node); list_del(&next->node); - plchan->at = next; plchan->state = PL08X_CHAN_RUNNING; - /* Configure the physical channel for the active txd */ - pl08x_config_phychan_for_txd(plchan); - pl08x_set_cregs(pl08x, plchan->phychan); - pl08x_enable_phy_chan(pl08x, plchan->phychan); + pl08x_start_txd(plchan, next); } spin_unlock_irqrestore(&plchan->lock, flags); @@ -1330,30 +1219,17 @@ static void pl08x_issue_pending(struct dma_chan *chan) static int pl08x_prep_channel_resources(struct pl08x_dma_chan *plchan, struct pl08x_txd *txd) { - int num_llis; struct pl08x_driver_data *pl08x = plchan->host; - int ret; + unsigned long flags; + int num_llis, ret; num_llis = pl08x_fill_llis_for_desc(pl08x, txd); - - if (!num_llis) + if (!num_llis) { + kfree(txd); return -EINVAL; + } - spin_lock_irqsave(&plchan->lock, plchan->lockflags); - - /* - * If this device is not using a circular buffer then - * queue this new descriptor for transfer. - * The descriptor for a circular buffer continues - * to be used until the channel is freed. - */ - if (txd->cd->circular_buffer) - dev_err(&pl08x->adev->dev, - "%s attempting to queue a circular buffer\n", - __func__); - else - list_add_tail(&txd->node, - &plchan->desc_list); + spin_lock_irqsave(&plchan->lock, flags); /* * See if we already have a physical channel allocated, @@ -1362,45 +1238,74 @@ static int pl08x_prep_channel_resources(struct pl08x_dma_chan *plchan, ret = prep_phy_channel(plchan, txd); if (ret) { /* - * No physical channel available, we will - * stack up the memcpy channels until there is a channel - * available to handle it whereas slave transfers may - * have been denied due to platform channel muxing restrictions - * and since there is no guarantee that this will ever be - * resolved, and since the signal must be aquired AFTER - * aquiring the physical channel, we will let them be NACK:ed - * with -EBUSY here. The drivers can alway retry the prep() - * call if they are eager on doing this using DMA. + * No physical channel was available. + * + * memcpy transfers can be sorted out at submission time. + * + * Slave transfers may have been denied due to platform + * channel muxing restrictions. Since there is no guarantee + * that this will ever be resolved, and the signal must be + * acquired AFTER acquiring the physical channel, we will let + * them be NACK:ed with -EBUSY here. The drivers can retry + * the prep() call if they are eager on doing this using DMA. */ if (plchan->slave) { pl08x_free_txd_list(pl08x, plchan); - spin_unlock_irqrestore(&plchan->lock, plchan->lockflags); + pl08x_free_txd(pl08x, txd); + spin_unlock_irqrestore(&plchan->lock, flags); return -EBUSY; } - /* Do this memcpy whenever there is a channel ready */ - plchan->state = PL08X_CHAN_WAITING; - plchan->waiting = txd; } else /* - * Else we're all set, paused and ready to roll, - * status will switch to PL08X_CHAN_RUNNING when - * we call issue_pending(). If there is something - * running on the channel already we don't change - * its state. + * Else we're all set, paused and ready to roll, status + * will switch to PL08X_CHAN_RUNNING when we call + * issue_pending(). If there is something running on the + * channel already we don't change its state. */ if (plchan->state == PL08X_CHAN_IDLE) plchan->state = PL08X_CHAN_PAUSED; - /* - * Notice that we leave plchan->lock locked on purpose: - * it will be unlocked in the subsequent tx_submit() - * call. This is a consequence of the current API. - */ + spin_unlock_irqrestore(&plchan->lock, flags); return 0; } /* + * Given the source and destination available bus masks, select which + * will be routed to each port. We try to have source and destination + * on separate ports, but always respect the allowable settings. + */ +static u32 pl08x_select_bus(struct pl08x_driver_data *pl08x, u8 src, u8 dst) +{ + u32 cctl = 0; + + if (!(dst & PL08X_AHB1) || ((dst & PL08X_AHB2) && (src & PL08X_AHB1))) + cctl |= PL080_CONTROL_DST_AHB2; + if (!(src & PL08X_AHB1) || ((src & PL08X_AHB2) && !(dst & PL08X_AHB2))) + cctl |= PL080_CONTROL_SRC_AHB2; + + return cctl; +} + +static struct pl08x_txd *pl08x_get_txd(struct pl08x_dma_chan *plchan, + unsigned long flags) +{ + struct pl08x_txd *txd = kzalloc(sizeof(struct pl08x_txd), GFP_NOWAIT); + + if (txd) { + dma_async_tx_descriptor_init(&txd->tx, &plchan->chan); + txd->tx.flags = flags; + txd->tx.tx_submit = pl08x_tx_submit; + INIT_LIST_HEAD(&txd->node); + + /* Always enable error and terminal interrupts */ + txd->ccfg = PL080_CONFIG_ERR_IRQ_MASK | + PL080_CONFIG_TC_IRQ_MASK; + } + return txd; +} + +/* * Initialize a descriptor to be used by memcpy submit */ static struct dma_async_tx_descriptor *pl08x_prep_dma_memcpy( @@ -1412,40 +1317,38 @@ static struct dma_async_tx_descriptor *pl08x_prep_dma_memcpy( struct pl08x_txd *txd; int ret; - txd = kzalloc(sizeof(struct pl08x_txd), GFP_NOWAIT); + txd = pl08x_get_txd(plchan, flags); if (!txd) { dev_err(&pl08x->adev->dev, "%s no memory for descriptor\n", __func__); return NULL; } - dma_async_tx_descriptor_init(&txd->tx, chan); txd->direction = DMA_NONE; - txd->srcbus.addr = src; - txd->dstbus.addr = dest; + txd->src_addr = src; + txd->dst_addr = dest; + txd->len = len; /* Set platform data for m2m */ - txd->cd = &pl08x->pd->memcpy_channel; + txd->ccfg |= PL080_FLOW_MEM2MEM << PL080_CONFIG_FLOW_CONTROL_SHIFT; + txd->cctl = pl08x->pd->memcpy_channel.cctl & + ~(PL080_CONTROL_DST_AHB2 | PL080_CONTROL_SRC_AHB2); + /* Both to be incremented or the code will break */ - txd->cd->cctl |= PL080_CONTROL_SRC_INCR | PL080_CONTROL_DST_INCR; - txd->tx.tx_submit = pl08x_tx_submit; - txd->tx.callback = NULL; - txd->tx.callback_param = NULL; - txd->len = len; + txd->cctl |= PL080_CONTROL_SRC_INCR | PL080_CONTROL_DST_INCR; + + if (pl08x->vd->dualmaster) + txd->cctl |= pl08x_select_bus(pl08x, + pl08x->mem_buses, pl08x->mem_buses); - INIT_LIST_HEAD(&txd->node); ret = pl08x_prep_channel_resources(plchan, txd); if (ret) return NULL; - /* - * NB: the channel lock is held at this point so tx_submit() - * must be called in direct succession. - */ return &txd->tx; } -struct dma_async_tx_descriptor *pl08x_prep_slave_sg( +static struct dma_async_tx_descriptor *pl08x_prep_slave_sg( struct dma_chan *chan, struct scatterlist *sgl, unsigned int sg_len, enum dma_data_direction direction, unsigned long flags) @@ -1453,6 +1356,7 @@ struct dma_async_tx_descriptor *pl08x_prep_slave_sg( struct pl08x_dma_chan *plchan = to_pl08x_chan(chan); struct pl08x_driver_data *pl08x = plchan->host; struct pl08x_txd *txd; + u8 src_buses, dst_buses; int ret; /* @@ -1467,14 +1371,12 @@ struct dma_async_tx_descriptor *pl08x_prep_slave_sg( dev_dbg(&pl08x->adev->dev, "%s prepare transaction of %d bytes from %s\n", __func__, sgl->length, plchan->name); - txd = kzalloc(sizeof(struct pl08x_txd), GFP_NOWAIT); + txd = pl08x_get_txd(plchan, flags); if (!txd) { dev_err(&pl08x->adev->dev, "%s no txd\n", __func__); return NULL; } - dma_async_tx_descriptor_init(&txd->tx, chan); - if (direction != plchan->runtime_direction) dev_err(&pl08x->adev->dev, "%s DMA setup does not match " "the direction configured for the PrimeCell\n", @@ -1486,37 +1388,47 @@ struct dma_async_tx_descriptor *pl08x_prep_slave_sg( * channel target address dynamically at runtime. */ txd->direction = direction; + txd->len = sgl->length; + + txd->cctl = plchan->cd->cctl & + ~(PL080_CONTROL_SRC_AHB2 | PL080_CONTROL_DST_AHB2 | + PL080_CONTROL_SRC_INCR | PL080_CONTROL_DST_INCR | + PL080_CONTROL_PROT_MASK); + + /* Access the cell in privileged mode, non-bufferable, non-cacheable */ + txd->cctl |= PL080_CONTROL_PROT_SYS; + if (direction == DMA_TO_DEVICE) { - txd->srcbus.addr = sgl->dma_address; + txd->ccfg |= PL080_FLOW_MEM2PER << PL080_CONFIG_FLOW_CONTROL_SHIFT; + txd->cctl |= PL080_CONTROL_SRC_INCR; + txd->src_addr = sgl->dma_address; if (plchan->runtime_addr) - txd->dstbus.addr = plchan->runtime_addr; + txd->dst_addr = plchan->runtime_addr; else - txd->dstbus.addr = plchan->cd->addr; + txd->dst_addr = plchan->cd->addr; + src_buses = pl08x->mem_buses; + dst_buses = plchan->cd->periph_buses; } else if (direction == DMA_FROM_DEVICE) { + txd->ccfg |= PL080_FLOW_PER2MEM << PL080_CONFIG_FLOW_CONTROL_SHIFT; + txd->cctl |= PL080_CONTROL_DST_INCR; if (plchan->runtime_addr) - txd->srcbus.addr = plchan->runtime_addr; + txd->src_addr = plchan->runtime_addr; else - txd->srcbus.addr = plchan->cd->addr; - txd->dstbus.addr = sgl->dma_address; + txd->src_addr = plchan->cd->addr; + txd->dst_addr = sgl->dma_address; + src_buses = plchan->cd->periph_buses; + dst_buses = pl08x->mem_buses; } else { dev_err(&pl08x->adev->dev, "%s direction unsupported\n", __func__); return NULL; } - txd->cd = plchan->cd; - txd->tx.tx_submit = pl08x_tx_submit; - txd->tx.callback = NULL; - txd->tx.callback_param = NULL; - txd->len = sgl->length; - INIT_LIST_HEAD(&txd->node); + + txd->cctl |= pl08x_select_bus(pl08x, src_buses, dst_buses); ret = pl08x_prep_channel_resources(plchan, txd); if (ret) return NULL; - /* - * NB: the channel lock is held at this point so tx_submit() - * must be called in direct succession. - */ return &txd->tx; } @@ -1531,10 +1443,8 @@ static int pl08x_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd, /* Controls applicable to inactive channels */ if (cmd == DMA_SLAVE_CONFIG) { - dma_set_runtime_config(chan, - (struct dma_slave_config *) - arg); - return 0; + return dma_set_runtime_config(chan, + (struct dma_slave_config *)arg); } /* @@ -1552,22 +1462,14 @@ static int pl08x_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd, plchan->state = PL08X_CHAN_IDLE; if (plchan->phychan) { - pl08x_stop_phy_chan(plchan->phychan); + pl08x_terminate_phy_chan(pl08x, plchan->phychan); /* * Mark physical channel as free and free any slave * signal */ - if ((plchan->phychan->signal >= 0) && - pl08x->pd->put_signal) { - pl08x->pd->put_signal(plchan); - plchan->phychan->signal = -1; - } - pl08x_put_phy_channel(pl08x, plchan->phychan); - plchan->phychan = NULL; + release_phy_channel(plchan); } - /* Stop any pending tasklet */ - tasklet_disable(&plchan->tasklet); /* Dequeue jobs and free LLIs */ if (plchan->at) { pl08x_free_txd(pl08x, plchan->at); @@ -1609,10 +1511,9 @@ bool pl08x_filter_id(struct dma_chan *chan, void *chan_id) /* * Just check that the device is there and active - * TODO: turn this bit on/off depending on the number of - * physical channels actually used, if it is zero... well - * shut it off. That will save some power. Cut the clock - * at the same time. + * TODO: turn this bit on/off depending on the number of physical channels + * actually used, if it is zero... well shut it off. That will save some + * power. Cut the clock at the same time. */ static void pl08x_ensure_on(struct pl08x_driver_data *pl08x) { @@ -1620,78 +1521,66 @@ static void pl08x_ensure_on(struct pl08x_driver_data *pl08x) val = readl(pl08x->base + PL080_CONFIG); val &= ~(PL080_CONFIG_M2_BE | PL080_CONFIG_M1_BE | PL080_CONFIG_ENABLE); - /* We implictly clear bit 1 and that means little-endian mode */ + /* We implicitly clear bit 1 and that means little-endian mode */ val |= PL080_CONFIG_ENABLE; writel(val, pl08x->base + PL080_CONFIG); } +static void pl08x_unmap_buffers(struct pl08x_txd *txd) +{ + struct device *dev = txd->tx.chan->device->dev; + + if (!(txd->tx.flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + if (txd->tx.flags & DMA_COMPL_SRC_UNMAP_SINGLE) + dma_unmap_single(dev, txd->src_addr, txd->len, + DMA_TO_DEVICE); + else + dma_unmap_page(dev, txd->src_addr, txd->len, + DMA_TO_DEVICE); + } + if (!(txd->tx.flags & DMA_COMPL_SKIP_DEST_UNMAP)) { + if (txd->tx.flags & DMA_COMPL_DEST_UNMAP_SINGLE) + dma_unmap_single(dev, txd->dst_addr, txd->len, + DMA_FROM_DEVICE); + else + dma_unmap_page(dev, txd->dst_addr, txd->len, + DMA_FROM_DEVICE); + } +} + static void pl08x_tasklet(unsigned long data) { struct pl08x_dma_chan *plchan = (struct pl08x_dma_chan *) data; - struct pl08x_phy_chan *phychan = plchan->phychan; struct pl08x_driver_data *pl08x = plchan->host; + struct pl08x_txd *txd; + unsigned long flags; - if (!plchan) - BUG(); - - spin_lock(&plchan->lock); - - if (plchan->at) { - dma_async_tx_callback callback = - plchan->at->tx.callback; - void *callback_param = - plchan->at->tx.callback_param; - - /* - * Update last completed - */ - plchan->lc = - (plchan->at->tx.cookie); - - /* - * Callback to signal completion - */ - if (callback) - callback(callback_param); + spin_lock_irqsave(&plchan->lock, flags); - /* - * Device callbacks should NOT clear - * the current transaction on the channel - * Linus: sometimes they should? - */ - if (!plchan->at) - BUG(); + txd = plchan->at; + plchan->at = NULL; - /* - * Free the descriptor if it's not for a device - * using a circular buffer - */ - if (!plchan->at->cd->circular_buffer) { - pl08x_free_txd(pl08x, plchan->at); - plchan->at = NULL; - } - /* - * else descriptor for circular - * buffers only freed when - * client has disabled dma - */ + if (txd) { + /* Update last completed */ + plchan->lc = txd->tx.cookie; } - /* - * If a new descriptor is queued, set it up - * plchan->at is NULL here - */ - if (!list_empty(&plchan->desc_list)) { + + /* If a new descriptor is queued, set it up plchan->at is NULL here */ + if (!list_empty(&plchan->pend_list)) { struct pl08x_txd *next; - next = list_first_entry(&plchan->desc_list, + next = list_first_entry(&plchan->pend_list, struct pl08x_txd, node); list_del(&next->node); - plchan->at = next; - /* Configure the physical channel for the next txd */ - pl08x_config_phychan_for_txd(plchan); - pl08x_set_cregs(pl08x, plchan->phychan); - pl08x_enable_phy_chan(pl08x, plchan->phychan); + + pl08x_start_txd(plchan, next); + } else if (plchan->phychan_hold) { + /* + * This channel is still in use - we have a new txd being + * prepared and will soon be queued. Don't give up the + * physical channel. + */ } else { struct pl08x_dma_chan *waiting = NULL; @@ -1699,20 +1588,14 @@ static void pl08x_tasklet(unsigned long data) * No more jobs, so free up the physical channel * Free any allocated signal on slave transfers too */ - if ((phychan->signal >= 0) && pl08x->pd->put_signal) { - pl08x->pd->put_signal(plchan); - phychan->signal = -1; - } - pl08x_put_phy_channel(pl08x, phychan); - plchan->phychan = NULL; + release_phy_channel(plchan); plchan->state = PL08X_CHAN_IDLE; /* - * And NOW before anyone else can grab that free:d - * up physical channel, see if there is some memcpy - * pending that seriously needs to start because of - * being stacked up while we were choking the - * physical channels with data. + * And NOW before anyone else can grab that free:d up + * physical channel, see if there is some memcpy pending + * that seriously needs to start because of being stacked + * up while we were choking the physical channels with data. */ list_for_each_entry(waiting, &pl08x->memcpy.channels, chan.device_node) { @@ -1724,6 +1607,7 @@ static void pl08x_tasklet(unsigned long data) ret = prep_phy_channel(waiting, waiting->waiting); BUG_ON(ret); + waiting->phychan_hold--; waiting->state = PL08X_CHAN_RUNNING; waiting->waiting = NULL; pl08x_issue_pending(&waiting->chan); @@ -1732,7 +1616,25 @@ static void pl08x_tasklet(unsigned long data) } } - spin_unlock(&plchan->lock); + spin_unlock_irqrestore(&plchan->lock, flags); + + if (txd) { + dma_async_tx_callback callback = txd->tx.callback; + void *callback_param = txd->tx.callback_param; + + /* Don't try to unmap buffers on slave channels */ + if (!plchan->slave) + pl08x_unmap_buffers(txd); + + /* Free the descriptor */ + spin_lock_irqsave(&plchan->lock, flags); + pl08x_free_txd(pl08x, txd); + spin_unlock_irqrestore(&plchan->lock, flags); + + /* Callback to signal completion */ + if (callback) + callback(callback_param); + } } static irqreturn_t pl08x_irq(int irq, void *dev) @@ -1744,9 +1646,7 @@ static irqreturn_t pl08x_irq(int irq, void *dev) val = readl(pl08x->base + PL080_ERR_STATUS); if (val) { - /* - * An error interrupt (on one or more channels) - */ + /* An error interrupt (on one or more channels) */ dev_err(&pl08x->adev->dev, "%s error interrupt, register value 0x%08x\n", __func__, val); @@ -1770,9 +1670,7 @@ static irqreturn_t pl08x_irq(int irq, void *dev) mask |= (1 << i); } } - /* - * Clear only the terminal interrupts on channels we processed - */ + /* Clear only the terminal interrupts on channels we processed */ writel(mask, pl08x->base + PL080_TC_CLEAR); return mask ? IRQ_HANDLED : IRQ_NONE; @@ -1791,6 +1689,7 @@ static int pl08x_dma_init_virtual_channels(struct pl08x_driver_data *pl08x, int i; INIT_LIST_HEAD(&dmadev->channels); + /* * Register as many many memcpy as we have physical channels, * we won't always be able to use all but the code will have @@ -1819,16 +1718,23 @@ static int pl08x_dma_init_virtual_channels(struct pl08x_driver_data *pl08x, return -ENOMEM; } } + if (chan->cd->circular_buffer) { + dev_err(&pl08x->adev->dev, + "channel %s: circular buffers not supported\n", + chan->name); + kfree(chan); + continue; + } dev_info(&pl08x->adev->dev, "initialize virtual channel \"%s\"\n", chan->name); chan->chan.device = dmadev; - atomic_set(&chan->last_issued, 0); - chan->lc = atomic_read(&chan->last_issued); + chan->chan.cookie = 0; + chan->lc = 0; spin_lock_init(&chan->lock); - INIT_LIST_HEAD(&chan->desc_list); + INIT_LIST_HEAD(&chan->pend_list); tasklet_init(&chan->tasklet, pl08x_tasklet, (unsigned long) chan); @@ -1898,7 +1804,7 @@ static int pl08x_debugfs_show(struct seq_file *s, void *data) seq_printf(s, "CHANNEL:\tSTATE:\n"); seq_printf(s, "--------\t------\n"); list_for_each_entry(chan, &pl08x->memcpy.channels, chan.device_node) { - seq_printf(s, "%s\t\t\%s\n", chan->name, + seq_printf(s, "%s\t\t%s\n", chan->name, pl08x_state_str(chan->state)); } @@ -1906,7 +1812,7 @@ static int pl08x_debugfs_show(struct seq_file *s, void *data) seq_printf(s, "CHANNEL:\tSTATE:\n"); seq_printf(s, "--------\t------\n"); list_for_each_entry(chan, &pl08x->slave.channels, chan.device_node) { - seq_printf(s, "%s\t\t\%s\n", chan->name, + seq_printf(s, "%s\t\t%s\n", chan->name, pl08x_state_str(chan->state)); } @@ -1942,7 +1848,7 @@ static inline void init_pl08x_debugfs(struct pl08x_driver_data *pl08x) static int pl08x_probe(struct amba_device *adev, struct amba_id *id) { struct pl08x_driver_data *pl08x; - struct vendor_data *vd = id->data; + const struct vendor_data *vd = id->data; int ret = 0; int i; @@ -1990,6 +1896,14 @@ static int pl08x_probe(struct amba_device *adev, struct amba_id *id) pl08x->adev = adev; pl08x->vd = vd; + /* By default, AHB1 only. If dualmaster, from platform */ + pl08x->lli_buses = PL08X_AHB1; + pl08x->mem_buses = PL08X_AHB1; + if (pl08x->vd->dualmaster) { + pl08x->lli_buses = pl08x->pd->lli_buses; + pl08x->mem_buses = pl08x->pd->mem_buses; + } + /* A DMA memory pool for LLIs, align on 1-byte boundary */ pl08x->pool = dma_pool_create(DRIVER_NAME, &pl08x->adev->dev, PL08X_LLI_TSFR_SIZE, PL08X_ALIGN, 0); @@ -2009,14 +1923,12 @@ static int pl08x_probe(struct amba_device *adev, struct amba_id *id) /* Turn on the PL08x */ pl08x_ensure_on(pl08x); - /* - * Attach the interrupt handler - */ + /* Attach the interrupt handler */ writel(0x000000FF, pl08x->base + PL080_ERR_CLEAR); writel(0x000000FF, pl08x->base + PL080_TC_CLEAR); ret = request_irq(adev->irq[0], pl08x_irq, IRQF_DISABLED, - vd->name, pl08x); + DRIVER_NAME, pl08x); if (ret) { dev_err(&adev->dev, "%s failed to request interrupt %d\n", __func__, adev->irq[0]); @@ -2087,8 +1999,9 @@ static int pl08x_probe(struct amba_device *adev, struct amba_id *id) amba_set_drvdata(adev, pl08x); init_pl08x_debugfs(pl08x); - dev_info(&pl08x->adev->dev, "ARM(R) %s DMA block initialized @%08x\n", - vd->name, adev->res.start); + dev_info(&pl08x->adev->dev, "DMA: PL%03x rev%u at 0x%08llx irq %d\n", + amba_part(adev), amba_rev(adev), + (unsigned long long)adev->res.start, adev->irq[0]); return 0; out_no_slave_reg: @@ -2115,13 +2028,11 @@ out_no_pl08x: /* PL080 has 8 channels and the PL080 have just 2 */ static struct vendor_data vendor_pl080 = { - .name = "PL080", .channels = 8, .dualmaster = true, }; static struct vendor_data vendor_pl081 = { - .name = "PL081", .channels = 2, .dualmaster = false, }; @@ -2160,7 +2071,7 @@ static int __init pl08x_init(void) retval = amba_driver_register(&pl08x_amba_driver); if (retval) printk(KERN_WARNING DRIVER_NAME - "failed to register as an amba device (%d)\n", + "failed to register as an AMBA device (%d)\n", retval); return retval; } diff --git a/drivers/dma/at_hdmac.c b/drivers/dma/at_hdmac.c index a0f3e6a06e06..3d7d705f026f 100644 --- a/drivers/dma/at_hdmac.c +++ b/drivers/dma/at_hdmac.c @@ -253,7 +253,7 @@ atc_chain_complete(struct at_dma_chan *atchan, struct at_desc *desc) /* move myself to free_list */ list_move(&desc->desc_node, &atchan->free_list); - /* unmap dma addresses */ + /* unmap dma addresses (not on slave channels) */ if (!atchan->chan_common.private) { struct device *parent = chan2parent(&atchan->chan_common); if (!(txd->flags & DMA_COMPL_SKIP_DEST_UNMAP)) { @@ -583,7 +583,6 @@ atc_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src, desc->lli.ctrlb = ctrlb; desc->txd.cookie = 0; - async_tx_ack(&desc->txd); if (!first) { first = desc; @@ -604,7 +603,7 @@ atc_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src, /* set end-of-link to the last link descriptor of list*/ set_desc_eol(desc); - desc->txd.flags = flags; /* client is in control of this ack */ + first->txd.flags = flags; /* client is in control of this ack */ return &first->txd; @@ -670,7 +669,7 @@ atc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, if (!desc) goto err_desc_get; - mem = sg_phys(sg); + mem = sg_dma_address(sg); len = sg_dma_len(sg); mem_width = 2; if (unlikely(mem & 3 || len & 3)) @@ -712,7 +711,7 @@ atc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, if (!desc) goto err_desc_get; - mem = sg_phys(sg); + mem = sg_dma_address(sg); len = sg_dma_len(sg); mem_width = 2; if (unlikely(mem & 3 || len & 3)) @@ -722,7 +721,7 @@ atc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, desc->lli.daddr = mem; desc->lli.ctrla = ctrla | ATC_DST_WIDTH(mem_width) - | len >> mem_width; + | len >> reg_width; desc->lli.ctrlb = ctrlb; if (!first) { @@ -749,8 +748,8 @@ atc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, first->txd.cookie = -EBUSY; first->len = total_len; - /* last link descriptor of list is responsible of flags */ - prev->txd.flags = flags; /* client is in control of this ack */ + /* first link descriptor of list is responsible of flags */ + first->txd.flags = flags; /* client is in control of this ack */ return &first->txd; @@ -854,11 +853,11 @@ static void atc_issue_pending(struct dma_chan *chan) dev_vdbg(chan2dev(chan), "issue_pending\n"); + spin_lock_bh(&atchan->lock); if (!atc_chan_is_enabled(atchan)) { - spin_lock_bh(&atchan->lock); atc_advance_work(atchan); - spin_unlock_bh(&atchan->lock); } + spin_unlock_bh(&atchan->lock); } /** @@ -1210,7 +1209,7 @@ static int __init at_dma_init(void) { return platform_driver_probe(&at_dma_driver, at_dma_probe); } -module_init(at_dma_init); +subsys_initcall(at_dma_init); static void __exit at_dma_exit(void) { diff --git a/drivers/dma/fsldma.c b/drivers/dma/fsldma.c index 286c3ac6bdcc..4de947a450fc 100644 --- a/drivers/dma/fsldma.c +++ b/drivers/dma/fsldma.c @@ -1,7 +1,7 @@ /* * Freescale MPC85xx, MPC83xx DMA Engine support * - * Copyright (C) 2007 Freescale Semiconductor, Inc. All rights reserved. + * Copyright (C) 2007-2010 Freescale Semiconductor, Inc. All rights reserved. * * Author: * Zhang Wei <wei.zhang@freescale.com>, Jul 2007 @@ -50,9 +50,11 @@ static void dma_init(struct fsldma_chan *chan) * EIE - Error interrupt enable * EOSIE - End of segments interrupt enable (basic mode) * EOLNIE - End of links interrupt enable + * BWC - Bandwidth sharing among channels */ - DMA_OUT(chan, &chan->regs->mr, FSL_DMA_MR_EIE - | FSL_DMA_MR_EOLNIE | FSL_DMA_MR_EOSIE, 32); + DMA_OUT(chan, &chan->regs->mr, FSL_DMA_MR_BWC + | FSL_DMA_MR_EIE | FSL_DMA_MR_EOLNIE + | FSL_DMA_MR_EOSIE, 32); break; case FSL_DMA_IP_83XX: /* Set the channel to below modes: @@ -1322,6 +1324,8 @@ static int __devinit fsldma_of_probe(struct platform_device *op, fdev->common.device_control = fsl_dma_device_control; fdev->common.dev = &op->dev; + dma_set_mask(&(op->dev), DMA_BIT_MASK(36)); + dev_set_drvdata(&op->dev, fdev); /* diff --git a/drivers/dma/fsldma.h b/drivers/dma/fsldma.h index cb4d6ff51597..ba9f403c0fbe 100644 --- a/drivers/dma/fsldma.h +++ b/drivers/dma/fsldma.h @@ -1,5 +1,5 @@ /* - * Copyright (C) 2007 Freescale Semiconductor, Inc. All rights reserved. + * Copyright (C) 2007-2010 Freescale Semiconductor, Inc. All rights reserved. * * Author: * Zhang Wei <wei.zhang@freescale.com>, Jul 2007 @@ -36,6 +36,13 @@ #define FSL_DMA_MR_DAHE 0x00002000 #define FSL_DMA_MR_SAHE 0x00001000 +/* + * Bandwidth/pause control determines how many bytes a given + * channel is allowed to transfer before the DMA engine pauses + * the current channel and switches to the next channel + */ +#define FSL_DMA_MR_BWC 0x08000000 + /* Special MR definition for MPC8349 */ #define FSL_DMA_MR_EOTIE 0x00000080 #define FSL_DMA_MR_PRC_RM 0x00000800 diff --git a/drivers/dma/imx-dma.c b/drivers/dma/imx-dma.c index f629e4961af5..e18eaabe92b9 100644 --- a/drivers/dma/imx-dma.c +++ b/drivers/dma/imx-dma.c @@ -49,6 +49,7 @@ struct imxdma_channel { struct imxdma_engine { struct device *dev; + struct device_dma_parameters dma_parms; struct dma_device dma_device; struct imxdma_channel channel[MAX_DMA_CHANNELS]; }; @@ -242,6 +243,21 @@ static struct dma_async_tx_descriptor *imxdma_prep_slave_sg( else dmamode = DMA_MODE_WRITE; + switch (imxdmac->word_size) { + case DMA_SLAVE_BUSWIDTH_4_BYTES: + if (sgl->length & 3 || sgl->dma_address & 3) + return NULL; + break; + case DMA_SLAVE_BUSWIDTH_2_BYTES: + if (sgl->length & 1 || sgl->dma_address & 1) + return NULL; + break; + case DMA_SLAVE_BUSWIDTH_1_BYTE: + break; + default: + return NULL; + } + ret = imx_dma_setup_sg(imxdmac->imxdma_channel, sgl, sg_len, dma_length, imxdmac->per_address, dmamode); if (ret) @@ -329,6 +345,9 @@ static int __init imxdma_probe(struct platform_device *pdev) INIT_LIST_HEAD(&imxdma->dma_device.channels); + dma_cap_set(DMA_SLAVE, imxdma->dma_device.cap_mask); + dma_cap_set(DMA_CYCLIC, imxdma->dma_device.cap_mask); + /* Initialize channel parameters */ for (i = 0; i < MAX_DMA_CHANNELS; i++) { struct imxdma_channel *imxdmac = &imxdma->channel[i]; @@ -346,11 +365,7 @@ static int __init imxdma_probe(struct platform_device *pdev) imxdmac->imxdma = imxdma; spin_lock_init(&imxdmac->lock); - dma_cap_set(DMA_SLAVE, imxdma->dma_device.cap_mask); - dma_cap_set(DMA_CYCLIC, imxdma->dma_device.cap_mask); - imxdmac->chan.device = &imxdma->dma_device; - imxdmac->chan.chan_id = i; imxdmac->channel = i; /* Add the channel to the DMAC list */ @@ -370,6 +385,9 @@ static int __init imxdma_probe(struct platform_device *pdev) platform_set_drvdata(pdev, imxdma); + imxdma->dma_device.dev->dma_parms = &imxdma->dma_parms; + dma_set_max_seg_size(imxdma->dma_device.dev, 0xffffff); + ret = dma_async_device_register(&imxdma->dma_device); if (ret) { dev_err(&pdev->dev, "unable to register\n"); @@ -379,7 +397,7 @@ static int __init imxdma_probe(struct platform_device *pdev) return 0; err_init: - while (i-- >= 0) { + while (--i >= 0) { struct imxdma_channel *imxdmac = &imxdma->channel[i]; imx_dma_free(imxdmac->imxdma_channel); } diff --git a/drivers/dma/imx-sdma.c b/drivers/dma/imx-sdma.c index 0834323a0599..b6d1455fa936 100644 --- a/drivers/dma/imx-sdma.c +++ b/drivers/dma/imx-sdma.c @@ -230,7 +230,7 @@ struct sdma_engine; * struct sdma_channel - housekeeping for a SDMA channel * * @sdma pointer to the SDMA engine for this channel - * @channel the channel number, matches dmaengine chan_id + * @channel the channel number, matches dmaengine chan_id + 1 * @direction transfer type. Needed for setting SDMA script * @peripheral_type Peripheral type. Needed for setting SDMA script * @event_id0 aka dma request line @@ -273,50 +273,6 @@ struct sdma_channel { #define MXC_SDMA_MIN_PRIORITY 1 #define MXC_SDMA_MAX_PRIORITY 7 -/** - * struct sdma_script_start_addrs - SDMA script start pointers - * - * start addresses of the different functions in the physical - * address space of the SDMA engine. - */ -struct sdma_script_start_addrs { - u32 ap_2_ap_addr; - u32 ap_2_bp_addr; - u32 ap_2_ap_fixed_addr; - u32 bp_2_ap_addr; - u32 loopback_on_dsp_side_addr; - u32 mcu_interrupt_only_addr; - u32 firi_2_per_addr; - u32 firi_2_mcu_addr; - u32 per_2_firi_addr; - u32 mcu_2_firi_addr; - u32 uart_2_per_addr; - u32 uart_2_mcu_addr; - u32 per_2_app_addr; - u32 mcu_2_app_addr; - u32 per_2_per_addr; - u32 uartsh_2_per_addr; - u32 uartsh_2_mcu_addr; - u32 per_2_shp_addr; - u32 mcu_2_shp_addr; - u32 ata_2_mcu_addr; - u32 mcu_2_ata_addr; - u32 app_2_per_addr; - u32 app_2_mcu_addr; - u32 shp_2_per_addr; - u32 shp_2_mcu_addr; - u32 mshc_2_mcu_addr; - u32 mcu_2_mshc_addr; - u32 spdif_2_mcu_addr; - u32 mcu_2_spdif_addr; - u32 asrc_2_mcu_addr; - u32 ext_mem_2_ipu_addr; - u32 descrambler_addr; - u32 dptc_dvfs_addr; - u32 utra_addr; - u32 ram_code_start_addr; -}; - #define SDMA_FIRMWARE_MAGIC 0x414d4453 /** @@ -345,6 +301,7 @@ struct sdma_firmware_header { struct sdma_engine { struct device *dev; + struct device_dma_parameters dma_parms; struct sdma_channel channel[MAX_DMA_CHANNELS]; struct sdma_channel_control *channel_control; void __iomem *regs; @@ -493,7 +450,7 @@ static void sdma_handle_channel_loop(struct sdma_channel *sdmac) if (bd->mode.status & BD_RROR) sdmac->status = DMA_ERROR; else - sdmac->status = DMA_SUCCESS; + sdmac->status = DMA_IN_PROGRESS; bd->mode.status |= BD_DONE; sdmac->buf_tail++; @@ -814,15 +771,15 @@ static void sdma_enable_channel(struct sdma_engine *sdma, int channel) __raw_writel(1 << channel, sdma->regs + SDMA_H_START); } -static dma_cookie_t sdma_assign_cookie(struct sdma_channel *sdma) +static dma_cookie_t sdma_assign_cookie(struct sdma_channel *sdmac) { - dma_cookie_t cookie = sdma->chan.cookie; + dma_cookie_t cookie = sdmac->chan.cookie; if (++cookie < 0) cookie = 1; - sdma->chan.cookie = cookie; - sdma->desc.cookie = cookie; + sdmac->chan.cookie = cookie; + sdmac->desc.cookie = cookie; return cookie; } @@ -842,7 +799,7 @@ static dma_cookie_t sdma_tx_submit(struct dma_async_tx_descriptor *tx) cookie = sdma_assign_cookie(sdmac); - sdma_enable_channel(sdma, tx->chan->chan_id); + sdma_enable_channel(sdma, sdmac->channel); spin_unlock_irq(&sdmac->lock); @@ -855,10 +812,6 @@ static int sdma_alloc_chan_resources(struct dma_chan *chan) struct imx_dma_data *data = chan->private; int prio, ret; - /* No need to execute this for internal channel 0 */ - if (chan->chan_id == 0) - return 0; - if (!data) return -EINVAL; @@ -923,7 +876,7 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( struct sdma_channel *sdmac = to_sdma_chan(chan); struct sdma_engine *sdma = sdmac->sdma; int ret, i, count; - int channel = chan->chan_id; + int channel = sdmac->channel; struct scatterlist *sg; if (sdmac->status == DMA_IN_PROGRESS) @@ -951,7 +904,7 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( struct sdma_buffer_descriptor *bd = &sdmac->bd[i]; int param; - bd->buffer_addr = sgl->dma_address; + bd->buffer_addr = sg->dma_address; count = sg->length; @@ -968,22 +921,33 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( ret = -EINVAL; goto err_out; } - if (sdmac->word_size == DMA_SLAVE_BUSWIDTH_4_BYTES) + + switch (sdmac->word_size) { + case DMA_SLAVE_BUSWIDTH_4_BYTES: bd->mode.command = 0; - else - bd->mode.command = sdmac->word_size; + if (count & 3 || sg->dma_address & 3) + return NULL; + break; + case DMA_SLAVE_BUSWIDTH_2_BYTES: + bd->mode.command = 2; + if (count & 1 || sg->dma_address & 1) + return NULL; + break; + case DMA_SLAVE_BUSWIDTH_1_BYTE: + bd->mode.command = 1; + break; + default: + return NULL; + } param = BD_DONE | BD_EXTD | BD_CONT; - if (sdmac->flags & IMX_DMA_SG_LOOP) { + if (i + 1 == sg_len) { param |= BD_INTR; - if (i + 1 == sg_len) - param |= BD_WRAP; + param |= BD_LAST; + param &= ~BD_CONT; } - if (i + 1 == sg_len) - param |= BD_INTR; - dev_dbg(sdma->dev, "entry %d: count: %d dma: 0x%08x %s%s\n", i, count, sg->dma_address, param & BD_WRAP ? "wrap" : "", @@ -997,6 +961,7 @@ static struct dma_async_tx_descriptor *sdma_prep_slave_sg( return &sdmac->desc; err_out: + sdmac->status = DMA_ERROR; return NULL; } @@ -1007,7 +972,7 @@ static struct dma_async_tx_descriptor *sdma_prep_dma_cyclic( struct sdma_channel *sdmac = to_sdma_chan(chan); struct sdma_engine *sdma = sdmac->sdma; int num_periods = buf_len / period_len; - int channel = chan->chan_id; + int channel = sdmac->channel; int ret, i = 0, buf = 0; dev_dbg(sdma->dev, "%s channel: %d\n", __func__, channel); @@ -1110,14 +1075,12 @@ static enum dma_status sdma_tx_status(struct dma_chan *chan, { struct sdma_channel *sdmac = to_sdma_chan(chan); dma_cookie_t last_used; - enum dma_status ret; last_used = chan->cookie; - ret = dma_async_is_complete(cookie, sdmac->last_completed, last_used); dma_set_tx_state(txstate, sdmac->last_completed, last_used, 0); - return ret; + return sdmac->status; } static void sdma_issue_pending(struct dma_chan *chan) @@ -1127,8 +1090,74 @@ static void sdma_issue_pending(struct dma_chan *chan) */ } -static int __init sdma_init(struct sdma_engine *sdma, - void *ram_code, int ram_code_size) +#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1 34 + +static void sdma_add_scripts(struct sdma_engine *sdma, + const struct sdma_script_start_addrs *addr) +{ + s32 *addr_arr = (u32 *)addr; + s32 *saddr_arr = (u32 *)sdma->script_addrs; + int i; + + for (i = 0; i < SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1; i++) + if (addr_arr[i] > 0) + saddr_arr[i] = addr_arr[i]; +} + +static int __init sdma_get_firmware(struct sdma_engine *sdma, + const char *cpu_name, int to_version) +{ + const struct firmware *fw; + char *fwname; + const struct sdma_firmware_header *header; + int ret; + const struct sdma_script_start_addrs *addr; + unsigned short *ram_code; + + fwname = kasprintf(GFP_KERNEL, "sdma-%s-to%d.bin", cpu_name, to_version); + if (!fwname) + return -ENOMEM; + + ret = request_firmware(&fw, fwname, sdma->dev); + if (ret) { + kfree(fwname); + return ret; + } + kfree(fwname); + + if (fw->size < sizeof(*header)) + goto err_firmware; + + header = (struct sdma_firmware_header *)fw->data; + + if (header->magic != SDMA_FIRMWARE_MAGIC) + goto err_firmware; + if (header->ram_code_start + header->ram_code_size > fw->size) + goto err_firmware; + + addr = (void *)header + header->script_addrs_start; + ram_code = (void *)header + header->ram_code_start; + + clk_enable(sdma->clk); + /* download the RAM image for SDMA */ + sdma_load_script(sdma, ram_code, + header->ram_code_size, + addr->ram_code_start_addr); + clk_disable(sdma->clk); + + sdma_add_scripts(sdma, addr); + + dev_info(sdma->dev, "loaded firmware %d.%d\n", + header->version_major, + header->version_minor); + +err_firmware: + release_firmware(fw); + + return ret; +} + +static int __init sdma_init(struct sdma_engine *sdma) { int i, ret; dma_addr_t ccb_phys; @@ -1192,11 +1221,6 @@ static int __init sdma_init(struct sdma_engine *sdma, __raw_writel(ccb_phys, sdma->regs + SDMA_H_C0PTR); - /* download the RAM image for SDMA */ - sdma_load_script(sdma, ram_code, - ram_code_size, - sdma->script_addrs->ram_code_start_addr); - /* Set bits of CONFIG register with given context switching mode */ __raw_writel(SDMA_H_CONFIG_CSM, sdma->regs + SDMA_H_CONFIG); @@ -1216,16 +1240,10 @@ err_dma_alloc: static int __init sdma_probe(struct platform_device *pdev) { int ret; - const struct firmware *fw; - const struct sdma_firmware_header *header; - const struct sdma_script_start_addrs *addr; int irq; - unsigned short *ram_code; struct resource *iores; struct sdma_platform_data *pdata = pdev->dev.platform_data; - char *fwname; int i; - dma_cap_mask_t mask; struct sdma_engine *sdma; sdma = kzalloc(sizeof(*sdma), GFP_KERNEL); @@ -1262,41 +1280,15 @@ static int __init sdma_probe(struct platform_device *pdev) if (ret) goto err_request_irq; - fwname = kasprintf(GFP_KERNEL, "sdma-%s-to%d.bin", - pdata->cpu_name, pdata->to_version); - if (!fwname) { - ret = -ENOMEM; - goto err_cputype; - } - - ret = request_firmware(&fw, fwname, &pdev->dev); - if (ret) { - dev_err(&pdev->dev, "request firmware \"%s\" failed with %d\n", - fwname, ret); - kfree(fwname); - goto err_cputype; - } - kfree(fwname); - - if (fw->size < sizeof(*header)) - goto err_firmware; - - header = (struct sdma_firmware_header *)fw->data; - - if (header->magic != SDMA_FIRMWARE_MAGIC) - goto err_firmware; - if (header->ram_code_start + header->ram_code_size > fw->size) - goto err_firmware; - - addr = (void *)header + header->script_addrs_start; - ram_code = (void *)header + header->ram_code_start; - sdma->script_addrs = kmalloc(sizeof(*addr), GFP_KERNEL); + sdma->script_addrs = kzalloc(sizeof(*sdma->script_addrs), GFP_KERNEL); if (!sdma->script_addrs) - goto err_firmware; - memcpy(sdma->script_addrs, addr, sizeof(*addr)); + goto err_alloc; sdma->version = pdata->sdma_version; + dma_cap_set(DMA_SLAVE, sdma->dma_device.cap_mask); + dma_cap_set(DMA_CYCLIC, sdma->dma_device.cap_mask); + INIT_LIST_HEAD(&sdma->dma_device.channels); /* Initialize channel parameters */ for (i = 0; i < MAX_DMA_CHANNELS; i++) { @@ -1305,21 +1297,28 @@ static int __init sdma_probe(struct platform_device *pdev) sdmac->sdma = sdma; spin_lock_init(&sdmac->lock); - dma_cap_set(DMA_SLAVE, sdma->dma_device.cap_mask); - dma_cap_set(DMA_CYCLIC, sdma->dma_device.cap_mask); - sdmac->chan.device = &sdma->dma_device; - sdmac->chan.chan_id = i; sdmac->channel = i; - /* Add the channel to the DMAC list */ - list_add_tail(&sdmac->chan.device_node, &sdma->dma_device.channels); + /* + * Add the channel to the DMAC list. Do not add channel 0 though + * because we need it internally in the SDMA driver. This also means + * that channel 0 in dmaengine counting matches sdma channel 1. + */ + if (i) + list_add_tail(&sdmac->chan.device_node, + &sdma->dma_device.channels); } - ret = sdma_init(sdma, ram_code, header->ram_code_size); + ret = sdma_init(sdma); if (ret) goto err_init; + if (pdata->script_addrs) + sdma_add_scripts(sdma, pdata->script_addrs); + + sdma_get_firmware(sdma, pdata->cpu_name, pdata->to_version); + sdma->dma_device.dev = &pdev->dev; sdma->dma_device.device_alloc_chan_resources = sdma_alloc_chan_resources; @@ -1329,6 +1328,8 @@ static int __init sdma_probe(struct platform_device *pdev) sdma->dma_device.device_prep_dma_cyclic = sdma_prep_dma_cyclic; sdma->dma_device.device_control = sdma_control; sdma->dma_device.device_issue_pending = sdma_issue_pending; + sdma->dma_device.dev->dma_parms = &sdma->dma_parms; + dma_set_max_seg_size(sdma->dma_device.dev, 65535); ret = dma_async_device_register(&sdma->dma_device); if (ret) { @@ -1336,26 +1337,13 @@ static int __init sdma_probe(struct platform_device *pdev) goto err_init; } - dev_info(&pdev->dev, "initialized (firmware %d.%d)\n", - header->version_major, - header->version_minor); - - /* request channel 0. This is an internal control channel - * to the SDMA engine and not available to clients. - */ - dma_cap_zero(mask); - dma_cap_set(DMA_SLAVE, mask); - dma_request_channel(mask, NULL, NULL); - - release_firmware(fw); + dev_info(sdma->dev, "initialized\n"); return 0; err_init: kfree(sdma->script_addrs); -err_firmware: - release_firmware(fw); -err_cputype: +err_alloc: free_irq(irq, sdma); err_request_irq: iounmap(sdma->regs); @@ -1366,7 +1354,7 @@ err_clk: err_request_region: err_irq: kfree(sdma); - return 0; + return ret; } static int __exit sdma_remove(struct platform_device *pdev) @@ -1385,7 +1373,7 @@ static int __init sdma_module_init(void) { return platform_driver_probe(&sdma_driver, sdma_probe); } -subsys_initcall(sdma_module_init); +module_init(sdma_module_init); MODULE_AUTHOR("Sascha Hauer, Pengutronix <s.hauer@pengutronix.de>"); MODULE_DESCRIPTION("i.MX SDMA driver"); diff --git a/drivers/dma/intel_mid_dma.c b/drivers/dma/intel_mid_dma.c index 338bc4eed1f3..798f46a4590d 100644 --- a/drivers/dma/intel_mid_dma.c +++ b/drivers/dma/intel_mid_dma.c @@ -664,11 +664,20 @@ static struct dma_async_tx_descriptor *intel_mid_dma_prep_memcpy( /*calculate CTL_LO*/ ctl_lo.ctl_lo = 0; ctl_lo.ctlx.int_en = 1; - ctl_lo.ctlx.dst_tr_width = mids->dma_slave.dst_addr_width; - ctl_lo.ctlx.src_tr_width = mids->dma_slave.src_addr_width; ctl_lo.ctlx.dst_msize = mids->dma_slave.src_maxburst; ctl_lo.ctlx.src_msize = mids->dma_slave.dst_maxburst; + /* + * Here we need some translation from "enum dma_slave_buswidth" + * to the format for our dma controller + * standard intel_mid_dmac's format + * 1 Byte 0b000 + * 2 Bytes 0b001 + * 4 Bytes 0b010 + */ + ctl_lo.ctlx.dst_tr_width = mids->dma_slave.dst_addr_width / 2; + ctl_lo.ctlx.src_tr_width = mids->dma_slave.src_addr_width / 2; + if (mids->cfg_mode == LNW_DMA_MEM_TO_MEM) { ctl_lo.ctlx.tt_fc = 0; ctl_lo.ctlx.sinc = 0; @@ -746,8 +755,18 @@ static struct dma_async_tx_descriptor *intel_mid_dma_prep_slave_sg( BUG_ON(!mids); if (!midc->dma->pimr_mask) { - pr_debug("MDMA: SG list is not supported by this controller\n"); - return NULL; + /* We can still handle sg list with only one item */ + if (sg_len == 1) { + txd = intel_mid_dma_prep_memcpy(chan, + mids->dma_slave.dst_addr, + mids->dma_slave.src_addr, + sgl->length, + flags); + return txd; + } else { + pr_warn("MDMA: SG list is not supported by this controller\n"); + return NULL; + } } pr_debug("MDMA: SG Length = %d, direction = %d, Flags = %#lx\n", @@ -758,6 +777,7 @@ static struct dma_async_tx_descriptor *intel_mid_dma_prep_slave_sg( pr_err("MDMA: Prep memcpy failed\n"); return NULL; } + desc = to_intel_mid_dma_desc(txd); desc->dirn = direction; ctl_lo.ctl_lo = desc->ctl_lo; @@ -1021,11 +1041,6 @@ static irqreturn_t intel_mid_dma_interrupt(int irq, void *data) /*DMA Interrupt*/ pr_debug("MDMA:Got an interrupt on irq %d\n", irq); - if (!mid) { - pr_err("ERR_MDMA:null pointer mid\n"); - return -EINVAL; - } - pr_debug("MDMA: Status %x, Mask %x\n", tfr_status, mid->intr_mask); tfr_status &= mid->intr_mask; if (tfr_status) { @@ -1060,8 +1075,8 @@ static irqreturn_t intel_mid_dma_interrupt2(int irq, void *data) * mid_setup_dma - Setup the DMA controller * @pdev: Controller PCI device structure * - * Initilize the DMA controller, channels, registers with DMA engine, - * ISR. Initilize DMA controller channels. + * Initialize the DMA controller, channels, registers with DMA engine, + * ISR. Initialize DMA controller channels. */ static int mid_setup_dma(struct pci_dev *pdev) { @@ -1075,7 +1090,6 @@ static int mid_setup_dma(struct pci_dev *pdev) if (NULL == dma->dma_pool) { pr_err("ERR_MDMA:pci_pool_create failed\n"); err = -ENOMEM; - kfree(dma); goto err_dma_pool; } @@ -1186,7 +1200,6 @@ err_engine: free_irq(pdev->irq, dma); err_irq: pci_pool_destroy(dma->dma_pool); - kfree(dma); err_dma_pool: pr_err("ERR_MDMA:setup_dma failed: %d\n", err); return err; @@ -1219,7 +1232,7 @@ static void middma_shutdown(struct pci_dev *pdev) * @pdev: Controller PCI device structure * @id: pci device id structure * - * Initilize the PCI device, map BARs, query driver data. + * Initialize the PCI device, map BARs, query driver data. * Call setup_dma to complete contoller and chan initilzation */ static int __devinit intel_mid_dma_probe(struct pci_dev *pdev, @@ -1413,7 +1426,7 @@ static const struct dev_pm_ops intel_mid_dma_pm = { .runtime_idle = dma_runtime_idle, }; -static struct pci_driver intel_mid_dma_pci = { +static struct pci_driver intel_mid_dma_pci_driver = { .name = "Intel MID DMA", .id_table = intel_mid_dma_ids, .probe = intel_mid_dma_probe, @@ -1431,13 +1444,13 @@ static int __init intel_mid_dma_init(void) { pr_debug("INFO_MDMA: LNW DMA Driver Version %s\n", INTEL_MID_DMA_DRIVER_VERSION); - return pci_register_driver(&intel_mid_dma_pci); + return pci_register_driver(&intel_mid_dma_pci_driver); } fs_initcall(intel_mid_dma_init); static void __exit intel_mid_dma_exit(void) { - pci_unregister_driver(&intel_mid_dma_pci); + pci_unregister_driver(&intel_mid_dma_pci_driver); } module_exit(intel_mid_dma_exit); diff --git a/drivers/dma/ioat/Makefile b/drivers/dma/ioat/Makefile index 8997d3fb9051..0ff7270af25b 100644 --- a/drivers/dma/ioat/Makefile +++ b/drivers/dma/ioat/Makefile @@ -1,2 +1,2 @@ obj-$(CONFIG_INTEL_IOATDMA) += ioatdma.o -ioatdma-objs := pci.o dma.o dma_v2.o dma_v3.o dca.o +ioatdma-y := pci.o dma.o dma_v2.o dma_v3.o dca.o diff --git a/drivers/dma/iop-adma.c b/drivers/dma/iop-adma.c index 161c452923b8..c6b01f535b29 100644 --- a/drivers/dma/iop-adma.c +++ b/drivers/dma/iop-adma.c @@ -1261,7 +1261,7 @@ out: return err; } -#ifdef CONFIG_MD_RAID6_PQ +#ifdef CONFIG_RAID6_PQ static int __devinit iop_adma_pq_zero_sum_self_test(struct iop_adma_device *device) { @@ -1584,7 +1584,7 @@ static int __devinit iop_adma_probe(struct platform_device *pdev) if (dma_has_cap(DMA_PQ, dma_dev->cap_mask) && dma_has_cap(DMA_PQ_VAL, dma_dev->cap_mask)) { - #ifdef CONFIG_MD_RAID6_PQ + #ifdef CONFIG_RAID6_PQ ret = iop_adma_pq_zero_sum_self_test(adev); dev_dbg(&pdev->dev, "pq self test returned %d\n", ret); #else diff --git a/drivers/dma/ipu/ipu_idmac.c b/drivers/dma/ipu/ipu_idmac.c index cb26ee9773d6..c1a125e7d1df 100644 --- a/drivers/dma/ipu/ipu_idmac.c +++ b/drivers/dma/ipu/ipu_idmac.c @@ -1145,29 +1145,6 @@ static int ipu_disable_channel(struct idmac *idmac, struct idmac_channel *ichan, reg = idmac_read_icreg(ipu, IDMAC_CHA_EN); idmac_write_icreg(ipu, reg & ~chan_mask, IDMAC_CHA_EN); - /* - * Problem (observed with channel DMAIC_7): after enabling the channel - * and initialising buffers, there comes an interrupt with current still - * pointing at buffer 0, whereas it should use buffer 0 first and only - * generate an interrupt when it is done, then current should already - * point to buffer 1. This spurious interrupt also comes on channel - * DMASDC_0. With DMAIC_7 normally, is we just leave the ISR after the - * first interrupt, there comes the second with current correctly - * pointing to buffer 1 this time. But sometimes this second interrupt - * doesn't come and the channel hangs. Clearing BUFx_RDY when disabling - * the channel seems to prevent the channel from hanging, but it doesn't - * prevent the spurious interrupt. This might also be unsafe. Think - * about the IDMAC controller trying to switch to a buffer, when we - * clear the ready bit, and re-enable it a moment later. - */ - reg = idmac_read_ipureg(ipu, IPU_CHA_BUF0_RDY); - idmac_write_ipureg(ipu, 0, IPU_CHA_BUF0_RDY); - idmac_write_ipureg(ipu, reg & ~(1UL << channel), IPU_CHA_BUF0_RDY); - - reg = idmac_read_ipureg(ipu, IPU_CHA_BUF1_RDY); - idmac_write_ipureg(ipu, 0, IPU_CHA_BUF1_RDY); - idmac_write_ipureg(ipu, reg & ~(1UL << channel), IPU_CHA_BUF1_RDY); - spin_unlock_irqrestore(&ipu->lock, flags); return 0; @@ -1246,33 +1223,6 @@ static irqreturn_t idmac_interrupt(int irq, void *dev_id) /* Other interrupts do not interfere with this channel */ spin_lock(&ichan->lock); - if (unlikely(chan_id != IDMAC_SDC_0 && chan_id != IDMAC_SDC_1 && - ((curbuf >> chan_id) & 1) == ichan->active_buffer && - !list_is_last(ichan->queue.next, &ichan->queue))) { - int i = 100; - - /* This doesn't help. See comment in ipu_disable_channel() */ - while (--i) { - curbuf = idmac_read_ipureg(&ipu_data, IPU_CHA_CUR_BUF); - if (((curbuf >> chan_id) & 1) != ichan->active_buffer) - break; - cpu_relax(); - } - - if (!i) { - spin_unlock(&ichan->lock); - dev_dbg(dev, - "IRQ on active buffer on channel %x, active " - "%d, ready %x, %x, current %x!\n", chan_id, - ichan->active_buffer, ready0, ready1, curbuf); - return IRQ_NONE; - } else - dev_dbg(dev, - "Buffer deactivated on channel %x, active " - "%d, ready %x, %x, current %x, rest %d!\n", chan_id, - ichan->active_buffer, ready0, ready1, curbuf, i); - } - if (unlikely((ichan->active_buffer && (ready1 >> chan_id) & 1) || (!ichan->active_buffer && (ready0 >> chan_id) & 1) )) { diff --git a/drivers/dma/mpc512x_dma.c b/drivers/dma/mpc512x_dma.c index 4e9cbf300594..59c270192ccc 100644 --- a/drivers/dma/mpc512x_dma.c +++ b/drivers/dma/mpc512x_dma.c @@ -1,6 +1,7 @@ /* * Copyright (C) Freescale Semicondutor, Inc. 2007, 2008. * Copyright (C) Semihalf 2009 + * Copyright (C) Ilya Yanok, Emcraft Systems 2010 * * Written by Piotr Ziecik <kosmo@semihalf.com>. Hardware description * (defines, structures and comments) was taken from MPC5121 DMA driver @@ -70,6 +71,8 @@ #define MPC_DMA_DMAES_SBE (1 << 1) #define MPC_DMA_DMAES_DBE (1 << 0) +#define MPC_DMA_DMAGPOR_SNOOP_ENABLE (1 << 6) + #define MPC_DMA_TSIZE_1 0x00 #define MPC_DMA_TSIZE_2 0x01 #define MPC_DMA_TSIZE_4 0x02 @@ -104,7 +107,10 @@ struct __attribute__ ((__packed__)) mpc_dma_regs { /* 0x30 */ u32 dmahrsh; /* DMA hw request status high(ch63~32) */ u32 dmahrsl; /* DMA hardware request status low(ch31~0) */ - u32 dmaihsa; /* DMA interrupt high select AXE(ch63~32) */ + union { + u32 dmaihsa; /* DMA interrupt high select AXE(ch63~32) */ + u32 dmagpor; /* (General purpose register on MPC8308) */ + }; u32 dmailsa; /* DMA interrupt low select AXE(ch31~0) */ /* 0x40 ~ 0xff */ u32 reserve0[48]; /* Reserved */ @@ -195,7 +201,9 @@ struct mpc_dma { struct mpc_dma_regs __iomem *regs; struct mpc_dma_tcd __iomem *tcd; int irq; + int irq2; uint error_status; + int is_mpc8308; /* Lock for error_status field in this structure */ spinlock_t error_status_lock; @@ -252,11 +260,13 @@ static void mpc_dma_execute(struct mpc_dma_chan *mchan) prev = mdesc; } - prev->tcd->start = 0; prev->tcd->int_maj = 1; /* Send first descriptor in chain into hardware */ memcpy_toio(&mdma->tcd[cid], first->tcd, sizeof(struct mpc_dma_tcd)); + + if (first != prev) + mdma->tcd[cid].e_sg = 1; out_8(&mdma->regs->dmassrt, cid); } @@ -274,6 +284,9 @@ static void mpc_dma_irq_process(struct mpc_dma *mdma, u32 is, u32 es, int off) spin_lock(&mchan->lock); + out_8(&mdma->regs->dmacint, ch + off); + out_8(&mdma->regs->dmacerr, ch + off); + /* Check error status */ if (es & (1 << ch)) list_for_each_entry(mdesc, &mchan->active, node) @@ -302,36 +315,68 @@ static irqreturn_t mpc_dma_irq(int irq, void *data) spin_unlock(&mdma->error_status_lock); /* Handle interrupt on each channel */ - mpc_dma_irq_process(mdma, in_be32(&mdma->regs->dmainth), + if (mdma->dma.chancnt > 32) { + mpc_dma_irq_process(mdma, in_be32(&mdma->regs->dmainth), in_be32(&mdma->regs->dmaerrh), 32); + } mpc_dma_irq_process(mdma, in_be32(&mdma->regs->dmaintl), in_be32(&mdma->regs->dmaerrl), 0); - /* Ack interrupt on all channels */ - out_be32(&mdma->regs->dmainth, 0xFFFFFFFF); - out_be32(&mdma->regs->dmaintl, 0xFFFFFFFF); - out_be32(&mdma->regs->dmaerrh, 0xFFFFFFFF); - out_be32(&mdma->regs->dmaerrl, 0xFFFFFFFF); - /* Schedule tasklet */ tasklet_schedule(&mdma->tasklet); return IRQ_HANDLED; } -/* DMA Tasklet */ -static void mpc_dma_tasklet(unsigned long data) +/* proccess completed descriptors */ +static void mpc_dma_process_completed(struct mpc_dma *mdma) { - struct mpc_dma *mdma = (void *)data; dma_cookie_t last_cookie = 0; struct mpc_dma_chan *mchan; struct mpc_dma_desc *mdesc; struct dma_async_tx_descriptor *desc; unsigned long flags; LIST_HEAD(list); - uint es; int i; + for (i = 0; i < mdma->dma.chancnt; i++) { + mchan = &mdma->channels[i]; + + /* Get all completed descriptors */ + spin_lock_irqsave(&mchan->lock, flags); + if (!list_empty(&mchan->completed)) + list_splice_tail_init(&mchan->completed, &list); + spin_unlock_irqrestore(&mchan->lock, flags); + + if (list_empty(&list)) + continue; + + /* Execute callbacks and run dependencies */ + list_for_each_entry(mdesc, &list, node) { + desc = &mdesc->desc; + + if (desc->callback) + desc->callback(desc->callback_param); + + last_cookie = desc->cookie; + dma_run_dependencies(desc); + } + + /* Free descriptors */ + spin_lock_irqsave(&mchan->lock, flags); + list_splice_tail_init(&list, &mchan->free); + mchan->completed_cookie = last_cookie; + spin_unlock_irqrestore(&mchan->lock, flags); + } +} + +/* DMA Tasklet */ +static void mpc_dma_tasklet(unsigned long data) +{ + struct mpc_dma *mdma = (void *)data; + unsigned long flags; + uint es; + spin_lock_irqsave(&mdma->error_status_lock, flags); es = mdma->error_status; mdma->error_status = 0; @@ -370,35 +415,7 @@ static void mpc_dma_tasklet(unsigned long data) dev_err(mdma->dma.dev, "- Destination Bus Error\n"); } - for (i = 0; i < mdma->dma.chancnt; i++) { - mchan = &mdma->channels[i]; - - /* Get all completed descriptors */ - spin_lock_irqsave(&mchan->lock, flags); - if (!list_empty(&mchan->completed)) - list_splice_tail_init(&mchan->completed, &list); - spin_unlock_irqrestore(&mchan->lock, flags); - - if (list_empty(&list)) - continue; - - /* Execute callbacks and run dependencies */ - list_for_each_entry(mdesc, &list, node) { - desc = &mdesc->desc; - - if (desc->callback) - desc->callback(desc->callback_param); - - last_cookie = desc->cookie; - dma_run_dependencies(desc); - } - - /* Free descriptors */ - spin_lock_irqsave(&mchan->lock, flags); - list_splice_tail_init(&list, &mchan->free); - mchan->completed_cookie = last_cookie; - spin_unlock_irqrestore(&mchan->lock, flags); - } + mpc_dma_process_completed(mdma); } /* Submit descriptor to hardware */ @@ -563,6 +580,7 @@ static struct dma_async_tx_descriptor * mpc_dma_prep_memcpy(struct dma_chan *chan, dma_addr_t dst, dma_addr_t src, size_t len, unsigned long flags) { + struct mpc_dma *mdma = dma_chan_to_mpc_dma(chan); struct mpc_dma_chan *mchan = dma_chan_to_mpc_dma_chan(chan); struct mpc_dma_desc *mdesc = NULL; struct mpc_dma_tcd *tcd; @@ -577,8 +595,11 @@ mpc_dma_prep_memcpy(struct dma_chan *chan, dma_addr_t dst, dma_addr_t src, } spin_unlock_irqrestore(&mchan->lock, iflags); - if (!mdesc) + if (!mdesc) { + /* try to free completed descriptors */ + mpc_dma_process_completed(mdma); return NULL; + } mdesc->error = 0; tcd = mdesc->tcd; @@ -591,7 +612,8 @@ mpc_dma_prep_memcpy(struct dma_chan *chan, dma_addr_t dst, dma_addr_t src, tcd->dsize = MPC_DMA_TSIZE_32; tcd->soff = 32; tcd->doff = 32; - } else if (IS_ALIGNED(src | dst | len, 16)) { + } else if (!mdma->is_mpc8308 && IS_ALIGNED(src | dst | len, 16)) { + /* MPC8308 doesn't support 16 byte transfers */ tcd->ssize = MPC_DMA_TSIZE_16; tcd->dsize = MPC_DMA_TSIZE_16; tcd->soff = 16; @@ -651,6 +673,15 @@ static int __devinit mpc_dma_probe(struct platform_device *op, return -EINVAL; } + if (of_device_is_compatible(dn, "fsl,mpc8308-dma")) { + mdma->is_mpc8308 = 1; + mdma->irq2 = irq_of_parse_and_map(dn, 1); + if (mdma->irq2 == NO_IRQ) { + dev_err(dev, "Error mapping IRQ!\n"); + return -EINVAL; + } + } + retval = of_address_to_resource(dn, 0, &res); if (retval) { dev_err(dev, "Error parsing memory region!\n"); @@ -681,11 +712,23 @@ static int __devinit mpc_dma_probe(struct platform_device *op, return -EINVAL; } + if (mdma->is_mpc8308) { + retval = devm_request_irq(dev, mdma->irq2, &mpc_dma_irq, 0, + DRV_NAME, mdma); + if (retval) { + dev_err(dev, "Error requesting IRQ2!\n"); + return -EINVAL; + } + } + spin_lock_init(&mdma->error_status_lock); dma = &mdma->dma; dma->dev = dev; - dma->chancnt = MPC_DMA_CHANNELS; + if (!mdma->is_mpc8308) + dma->chancnt = MPC_DMA_CHANNELS; + else + dma->chancnt = 16; /* MPC8308 DMA has only 16 channels */ dma->device_alloc_chan_resources = mpc_dma_alloc_chan_resources; dma->device_free_chan_resources = mpc_dma_free_chan_resources; dma->device_issue_pending = mpc_dma_issue_pending; @@ -721,26 +764,40 @@ static int __devinit mpc_dma_probe(struct platform_device *op, * - Round-robin group arbitration, * - Round-robin channel arbitration. */ - out_be32(&mdma->regs->dmacr, MPC_DMA_DMACR_EDCG | - MPC_DMA_DMACR_ERGA | MPC_DMA_DMACR_ERCA); - - /* Disable hardware DMA requests */ - out_be32(&mdma->regs->dmaerqh, 0); - out_be32(&mdma->regs->dmaerql, 0); - - /* Disable error interrupts */ - out_be32(&mdma->regs->dmaeeih, 0); - out_be32(&mdma->regs->dmaeeil, 0); - - /* Clear interrupts status */ - out_be32(&mdma->regs->dmainth, 0xFFFFFFFF); - out_be32(&mdma->regs->dmaintl, 0xFFFFFFFF); - out_be32(&mdma->regs->dmaerrh, 0xFFFFFFFF); - out_be32(&mdma->regs->dmaerrl, 0xFFFFFFFF); - - /* Route interrupts to IPIC */ - out_be32(&mdma->regs->dmaihsa, 0); - out_be32(&mdma->regs->dmailsa, 0); + if (!mdma->is_mpc8308) { + out_be32(&mdma->regs->dmacr, MPC_DMA_DMACR_EDCG | + MPC_DMA_DMACR_ERGA | MPC_DMA_DMACR_ERCA); + + /* Disable hardware DMA requests */ + out_be32(&mdma->regs->dmaerqh, 0); + out_be32(&mdma->regs->dmaerql, 0); + + /* Disable error interrupts */ + out_be32(&mdma->regs->dmaeeih, 0); + out_be32(&mdma->regs->dmaeeil, 0); + + /* Clear interrupts status */ + out_be32(&mdma->regs->dmainth, 0xFFFFFFFF); + out_be32(&mdma->regs->dmaintl, 0xFFFFFFFF); + out_be32(&mdma->regs->dmaerrh, 0xFFFFFFFF); + out_be32(&mdma->regs->dmaerrl, 0xFFFFFFFF); + + /* Route interrupts to IPIC */ + out_be32(&mdma->regs->dmaihsa, 0); + out_be32(&mdma->regs->dmailsa, 0); + } else { + /* MPC8308 has 16 channels and lacks some registers */ + out_be32(&mdma->regs->dmacr, MPC_DMA_DMACR_ERCA); + + /* enable snooping */ + out_be32(&mdma->regs->dmagpor, MPC_DMA_DMAGPOR_SNOOP_ENABLE); + /* Disable error interrupts */ + out_be32(&mdma->regs->dmaeeil, 0); + + /* Clear interrupts status */ + out_be32(&mdma->regs->dmaintl, 0xFFFF); + out_be32(&mdma->regs->dmaerrl, 0xFFFF); + } /* Register DMA engine */ dev_set_drvdata(dev, mdma); diff --git a/drivers/dma/mv_xor.c b/drivers/dma/mv_xor.c index 411d5bf50fc4..a25f5f61e0e0 100644 --- a/drivers/dma/mv_xor.c +++ b/drivers/dma/mv_xor.c @@ -449,7 +449,7 @@ mv_xor_slot_cleanup(struct mv_xor_chan *mv_chan) static void mv_xor_tasklet(unsigned long data) { struct mv_xor_chan *chan = (struct mv_xor_chan *) data; - __mv_xor_slot_cleanup(chan); + mv_xor_slot_cleanup(chan); } static struct mv_xor_desc_slot * diff --git a/drivers/dma/pch_dma.c b/drivers/dma/pch_dma.c index 92b679024fed..1c38418ae61f 100644 --- a/drivers/dma/pch_dma.c +++ b/drivers/dma/pch_dma.c @@ -1,6 +1,7 @@ /* * Topcliff PCH DMA controller driver * Copyright (c) 2010 Intel Corporation + * Copyright (C) 2011 OKI SEMICONDUCTOR CO., LTD. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 as @@ -259,11 +260,6 @@ static void pdc_dostart(struct pch_dma_chan *pd_chan, struct pch_dma_desc* desc) return; } - channel_writel(pd_chan, DEV_ADDR, desc->regs.dev_addr); - channel_writel(pd_chan, MEM_ADDR, desc->regs.mem_addr); - channel_writel(pd_chan, SIZE, desc->regs.size); - channel_writel(pd_chan, NEXT, desc->regs.next); - dev_dbg(chan2dev(&pd_chan->chan), "chan %d -> dev_addr: %x\n", pd_chan->chan.chan_id, desc->regs.dev_addr); dev_dbg(chan2dev(&pd_chan->chan), "chan %d -> mem_addr: %x\n", @@ -273,10 +269,16 @@ static void pdc_dostart(struct pch_dma_chan *pd_chan, struct pch_dma_desc* desc) dev_dbg(chan2dev(&pd_chan->chan), "chan %d -> next: %x\n", pd_chan->chan.chan_id, desc->regs.next); - if (list_empty(&desc->tx_list)) + if (list_empty(&desc->tx_list)) { + channel_writel(pd_chan, DEV_ADDR, desc->regs.dev_addr); + channel_writel(pd_chan, MEM_ADDR, desc->regs.mem_addr); + channel_writel(pd_chan, SIZE, desc->regs.size); + channel_writel(pd_chan, NEXT, desc->regs.next); pdc_set_mode(&pd_chan->chan, DMA_CTL0_ONESHOT); - else + } else { + channel_writel(pd_chan, NEXT, desc->txd.phys); pdc_set_mode(&pd_chan->chan, DMA_CTL0_SG); + } val = dma_readl(pd, CTL2); val |= 1 << (DMA_CTL2_START_SHIFT_BITS + pd_chan->chan.chan_id); @@ -920,12 +922,19 @@ static void __devexit pch_dma_remove(struct pci_dev *pdev) } /* PCI Device ID of DMA device */ -#define PCI_DEVICE_ID_PCH_DMA_8CH 0x8810 -#define PCI_DEVICE_ID_PCH_DMA_4CH 0x8815 +#define PCI_VENDOR_ID_ROHM 0x10DB +#define PCI_DEVICE_ID_EG20T_PCH_DMA_8CH 0x8810 +#define PCI_DEVICE_ID_EG20T_PCH_DMA_4CH 0x8815 +#define PCI_DEVICE_ID_ML7213_DMA1_8CH 0x8026 +#define PCI_DEVICE_ID_ML7213_DMA2_8CH 0x802B +#define PCI_DEVICE_ID_ML7213_DMA3_4CH 0x8034 static const struct pci_device_id pch_dma_id_table[] = { - { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_PCH_DMA_8CH), 8 }, - { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_PCH_DMA_4CH), 4 }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_EG20T_PCH_DMA_8CH), 8 }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_EG20T_PCH_DMA_4CH), 4 }, + { PCI_VDEVICE(ROHM, PCI_DEVICE_ID_ML7213_DMA1_8CH), 8}, /* UART Video */ + { PCI_VDEVICE(ROHM, PCI_DEVICE_ID_ML7213_DMA2_8CH), 8}, /* PCMIF SPI */ + { PCI_VDEVICE(ROHM, PCI_DEVICE_ID_ML7213_DMA3_4CH), 4}, /* FPGA */ { 0, }, }; @@ -953,6 +962,7 @@ static void __exit pch_dma_exit(void) module_init(pch_dma_init); module_exit(pch_dma_exit); -MODULE_DESCRIPTION("Topcliff PCH DMA controller driver"); +MODULE_DESCRIPTION("Intel EG20T PCH / OKI SEMICONDUCTOR ML7213 IOH " + "DMA controller driver"); MODULE_AUTHOR("Yong Wang <yong.y.wang@intel.com>"); MODULE_LICENSE("GPL v2"); diff --git a/drivers/dma/ppc4xx/adma.c b/drivers/dma/ppc4xx/adma.c index 0d58a4a4487f..cef584533ee8 100644 --- a/drivers/dma/ppc4xx/adma.c +++ b/drivers/dma/ppc4xx/adma.c @@ -4449,9 +4449,8 @@ static int __devinit ppc440spe_adma_probe(struct platform_device *ofdev, if (!request_mem_region(res.start, resource_size(&res), dev_driver_string(&ofdev->dev))) { - dev_err(&ofdev->dev, "failed to request memory region " - "(0x%016llx-0x%016llx)\n", - (u64)res.start, (u64)res.end); + dev_err(&ofdev->dev, "failed to request memory region %pR\n", + &res); initcode = PPC_ADMA_INIT_MEMREG; ret = -EBUSY; goto out; diff --git a/drivers/dma/shdma.c b/drivers/dma/shdma.c index eb6b54dbb806..28720d3103c4 100644 --- a/drivers/dma/shdma.c +++ b/drivers/dma/shdma.c @@ -27,7 +27,10 @@ #include <linux/platform_device.h> #include <linux/pm_runtime.h> #include <linux/sh_dma.h> - +#include <linux/notifier.h> +#include <linux/kdebug.h> +#include <linux/spinlock.h> +#include <linux/rculist.h> #include "shdma.h" /* DMA descriptor control */ @@ -43,6 +46,13 @@ enum sh_dmae_desc_status { /* Default MEMCPY transfer size = 2^2 = 4 bytes */ #define LOG2_DEFAULT_XFER_SIZE 2 +/* + * Used for write-side mutual exclusion for the global device list, + * read-side synchronization by way of RCU. + */ +static DEFINE_SPINLOCK(sh_dmae_lock); +static LIST_HEAD(sh_dmae_devices); + /* A bitmask with bits enough for enum sh_dmae_slave_chan_id */ static unsigned long sh_dmae_slave_used[BITS_TO_LONGS(SH_DMA_SLAVE_NUMBER)]; @@ -817,10 +827,9 @@ static irqreturn_t sh_dmae_interrupt(int irq, void *data) return ret; } -#if defined(CONFIG_CPU_SH4) || defined(CONFIG_ARCH_SHMOBILE) -static irqreturn_t sh_dmae_err(int irq, void *data) +static unsigned int sh_dmae_reset(struct sh_dmae_device *shdev) { - struct sh_dmae_device *shdev = (struct sh_dmae_device *)data; + unsigned int handled = 0; int i; /* halt the dma controller */ @@ -829,25 +838,35 @@ static irqreturn_t sh_dmae_err(int irq, void *data) /* We cannot detect, which channel caused the error, have to reset all */ for (i = 0; i < SH_DMAC_MAX_CHANNELS; i++) { struct sh_dmae_chan *sh_chan = shdev->chan[i]; - if (sh_chan) { - struct sh_desc *desc; - /* Stop the channel */ - dmae_halt(sh_chan); - /* Complete all */ - list_for_each_entry(desc, &sh_chan->ld_queue, node) { - struct dma_async_tx_descriptor *tx = &desc->async_tx; - desc->mark = DESC_IDLE; - if (tx->callback) - tx->callback(tx->callback_param); - } - list_splice_init(&sh_chan->ld_queue, &sh_chan->ld_free); + struct sh_desc *desc; + + if (!sh_chan) + continue; + + /* Stop the channel */ + dmae_halt(sh_chan); + + /* Complete all */ + list_for_each_entry(desc, &sh_chan->ld_queue, node) { + struct dma_async_tx_descriptor *tx = &desc->async_tx; + desc->mark = DESC_IDLE; + if (tx->callback) + tx->callback(tx->callback_param); } + + list_splice_init(&sh_chan->ld_queue, &sh_chan->ld_free); + handled++; } + sh_dmae_rst(shdev); - return IRQ_HANDLED; + return !!handled; +} + +static irqreturn_t sh_dmae_err(int irq, void *data) +{ + return IRQ_RETVAL(sh_dmae_reset(data)); } -#endif static void dmae_do_tasklet(unsigned long data) { @@ -876,6 +895,60 @@ static void dmae_do_tasklet(unsigned long data) sh_dmae_chan_ld_cleanup(sh_chan, false); } +static bool sh_dmae_nmi_notify(struct sh_dmae_device *shdev) +{ + unsigned int handled; + + /* Fast path out if NMIF is not asserted for this controller */ + if ((dmaor_read(shdev) & DMAOR_NMIF) == 0) + return false; + + handled = sh_dmae_reset(shdev); + if (handled) + return true; + + return false; +} + +static int sh_dmae_nmi_handler(struct notifier_block *self, + unsigned long cmd, void *data) +{ + struct sh_dmae_device *shdev; + int ret = NOTIFY_DONE; + bool triggered; + + /* + * Only concern ourselves with NMI events. + * + * Normally we would check the die chain value, but as this needs + * to be architecture independent, check for NMI context instead. + */ + if (!in_nmi()) + return NOTIFY_DONE; + + rcu_read_lock(); + list_for_each_entry_rcu(shdev, &sh_dmae_devices, node) { + /* + * Only stop if one of the controllers has NMIF asserted, + * we do not want to interfere with regular address error + * handling or NMI events that don't concern the DMACs. + */ + triggered = sh_dmae_nmi_notify(shdev); + if (triggered == true) + ret = NOTIFY_OK; + } + rcu_read_unlock(); + + return ret; +} + +static struct notifier_block sh_dmae_nmi_notifier __read_mostly = { + .notifier_call = sh_dmae_nmi_handler, + + /* Run before NMI debug handler and KGDB */ + .priority = 1, +}; + static int __devinit sh_dmae_chan_probe(struct sh_dmae_device *shdev, int id, int irq, unsigned long flags) { @@ -967,6 +1040,7 @@ static int __init sh_dmae_probe(struct platform_device *pdev) struct sh_dmae_pdata *pdata = pdev->dev.platform_data; unsigned long irqflags = IRQF_DISABLED, chan_flag[SH_DMAC_MAX_CHANNELS] = {}; + unsigned long flags; int errirq, chan_irq[SH_DMAC_MAX_CHANNELS]; int err, i, irq_cnt = 0, irqres = 0; struct sh_dmae_device *shdev; @@ -1032,6 +1106,10 @@ static int __init sh_dmae_probe(struct platform_device *pdev) pm_runtime_enable(&pdev->dev); pm_runtime_get_sync(&pdev->dev); + spin_lock_irqsave(&sh_dmae_lock, flags); + list_add_tail_rcu(&shdev->node, &sh_dmae_devices); + spin_unlock_irqrestore(&sh_dmae_lock, flags); + /* reset dma controller */ err = sh_dmae_rst(shdev); if (err) @@ -1135,6 +1213,10 @@ eirqres: eirq_err: #endif rst_err: + spin_lock_irqsave(&sh_dmae_lock, flags); + list_del_rcu(&shdev->node); + spin_unlock_irqrestore(&sh_dmae_lock, flags); + pm_runtime_put(&pdev->dev); if (dmars) iounmap(shdev->dmars); @@ -1155,6 +1237,7 @@ static int __exit sh_dmae_remove(struct platform_device *pdev) { struct sh_dmae_device *shdev = platform_get_drvdata(pdev); struct resource *res; + unsigned long flags; int errirq = platform_get_irq(pdev, 0); dma_async_device_unregister(&shdev->common); @@ -1162,6 +1245,10 @@ static int __exit sh_dmae_remove(struct platform_device *pdev) if (errirq > 0) free_irq(errirq, shdev); + spin_lock_irqsave(&sh_dmae_lock, flags); + list_del_rcu(&shdev->node); + spin_unlock_irqrestore(&sh_dmae_lock, flags); + /* channel data remove */ sh_dmae_chan_remove(shdev); @@ -1200,6 +1287,11 @@ static struct platform_driver sh_dmae_driver = { static int __init sh_dmae_init(void) { + /* Wire up NMI handling */ + int err = register_die_notifier(&sh_dmae_nmi_notifier); + if (err) + return err; + return platform_driver_probe(&sh_dmae_driver, sh_dmae_probe); } module_init(sh_dmae_init); @@ -1207,9 +1299,12 @@ module_init(sh_dmae_init); static void __exit sh_dmae_exit(void) { platform_driver_unregister(&sh_dmae_driver); + + unregister_die_notifier(&sh_dmae_nmi_notifier); } module_exit(sh_dmae_exit); MODULE_AUTHOR("Nobuhiro Iwamatsu <iwamatsu.nobuhiro@renesas.com>"); MODULE_DESCRIPTION("Renesas SH DMA Engine driver"); MODULE_LICENSE("GPL"); +MODULE_ALIAS("platform:sh-dma-engine"); diff --git a/drivers/dma/shdma.h b/drivers/dma/shdma.h index 4021275a0a43..52e4fb173805 100644 --- a/drivers/dma/shdma.h +++ b/drivers/dma/shdma.h @@ -43,6 +43,7 @@ struct sh_dmae_device { struct dma_device common; struct sh_dmae_chan *chan[SH_DMAC_MAX_CHANNELS]; struct sh_dmae_pdata *pdata; + struct list_head node; u32 __iomem *chan_reg; u16 __iomem *dmars; }; diff --git a/drivers/dma/ste_dma40.c b/drivers/dma/ste_dma40.c index fab68a553205..af955de035f4 100644 --- a/drivers/dma/ste_dma40.c +++ b/drivers/dma/ste_dma40.c @@ -1,5 +1,6 @@ /* - * Copyright (C) ST-Ericsson SA 2007-2010 + * Copyright (C) Ericsson AB 2007-2008 + * Copyright (C) ST-Ericsson SA 2008-2010 * Author: Per Forlin <per.forlin@stericsson.com> for ST-Ericsson * Author: Jonas Aaberg <jonas.aberg@stericsson.com> for ST-Ericsson * License terms: GNU General Public License (GPL) version 2 @@ -67,6 +68,7 @@ enum d40_command { * @base: Pointer to memory area when the pre_alloc_lli's are not large * enough, IE bigger than the most common case, 1 dst and 1 src. NULL if * pre_alloc_lli is used. + * @dma_addr: DMA address, if mapped * @size: The size in bytes of the memory at base or the size of pre_alloc_lli. * @pre_alloc_lli: Pre allocated area for the most common case of transfers, * one buffer to one buffer. @@ -74,6 +76,7 @@ enum d40_command { struct d40_lli_pool { void *base; int size; + dma_addr_t dma_addr; /* Space for dst and src, plus an extra for padding */ u8 pre_alloc_lli[3 * sizeof(struct d40_phy_lli)]; }; @@ -93,7 +96,6 @@ struct d40_lli_pool { * during a transfer. * @node: List entry. * @is_in_client_list: true if the client owns this descriptor. - * @is_hw_linked: true if this job will automatically be continued for * the previous one. * * This descriptor is used for both logical and physical transfers. @@ -113,7 +115,7 @@ struct d40_desc { struct list_head node; bool is_in_client_list; - bool is_hw_linked; + bool cyclic; }; /** @@ -129,6 +131,7 @@ struct d40_desc { */ struct d40_lcla_pool { void *base; + dma_addr_t dma_addr; void *base_unaligned; int pages; spinlock_t lock; @@ -302,9 +305,37 @@ struct d40_reg_val { unsigned int val; }; -static int d40_pool_lli_alloc(struct d40_desc *d40d, - int lli_len, bool is_log) +static struct device *chan2dev(struct d40_chan *d40c) { + return &d40c->chan.dev->device; +} + +static bool chan_is_physical(struct d40_chan *chan) +{ + return chan->log_num == D40_PHY_CHAN; +} + +static bool chan_is_logical(struct d40_chan *chan) +{ + return !chan_is_physical(chan); +} + +static void __iomem *chan_base(struct d40_chan *chan) +{ + return chan->base->virtbase + D40_DREG_PCBASE + + chan->phy_chan->num * D40_DREG_PCDELTA; +} + +#define d40_err(dev, format, arg...) \ + dev_err(dev, "[%s] " format, __func__, ## arg) + +#define chan_err(d40c, format, arg...) \ + d40_err(chan2dev(d40c), format, ## arg) + +static int d40_pool_lli_alloc(struct d40_chan *d40c, struct d40_desc *d40d, + int lli_len) +{ + bool is_log = chan_is_logical(d40c); u32 align; void *base; @@ -318,7 +349,7 @@ static int d40_pool_lli_alloc(struct d40_desc *d40d, d40d->lli_pool.size = sizeof(d40d->lli_pool.pre_alloc_lli); d40d->lli_pool.base = NULL; } else { - d40d->lli_pool.size = ALIGN(lli_len * 2 * align, align); + d40d->lli_pool.size = lli_len * 2 * align; base = kmalloc(d40d->lli_pool.size + align, GFP_NOWAIT); d40d->lli_pool.base = base; @@ -328,22 +359,37 @@ static int d40_pool_lli_alloc(struct d40_desc *d40d, } if (is_log) { - d40d->lli_log.src = PTR_ALIGN((struct d40_log_lli *) base, - align); - d40d->lli_log.dst = PTR_ALIGN(d40d->lli_log.src + lli_len, - align); + d40d->lli_log.src = PTR_ALIGN(base, align); + d40d->lli_log.dst = d40d->lli_log.src + lli_len; + + d40d->lli_pool.dma_addr = 0; } else { - d40d->lli_phy.src = PTR_ALIGN((struct d40_phy_lli *)base, - align); - d40d->lli_phy.dst = PTR_ALIGN(d40d->lli_phy.src + lli_len, - align); + d40d->lli_phy.src = PTR_ALIGN(base, align); + d40d->lli_phy.dst = d40d->lli_phy.src + lli_len; + + d40d->lli_pool.dma_addr = dma_map_single(d40c->base->dev, + d40d->lli_phy.src, + d40d->lli_pool.size, + DMA_TO_DEVICE); + + if (dma_mapping_error(d40c->base->dev, + d40d->lli_pool.dma_addr)) { + kfree(d40d->lli_pool.base); + d40d->lli_pool.base = NULL; + d40d->lli_pool.dma_addr = 0; + return -ENOMEM; + } } return 0; } -static void d40_pool_lli_free(struct d40_desc *d40d) +static void d40_pool_lli_free(struct d40_chan *d40c, struct d40_desc *d40d) { + if (d40d->lli_pool.dma_addr) + dma_unmap_single(d40c->base->dev, d40d->lli_pool.dma_addr, + d40d->lli_pool.size, DMA_TO_DEVICE); + kfree(d40d->lli_pool.base); d40d->lli_pool.base = NULL; d40d->lli_pool.size = 0; @@ -390,7 +436,7 @@ static int d40_lcla_free_all(struct d40_chan *d40c, int i; int ret = -EINVAL; - if (d40c->log_num == D40_PHY_CHAN) + if (chan_is_physical(d40c)) return 0; spin_lock_irqsave(&d40c->base->lcla_pool.lock, flags); @@ -429,7 +475,7 @@ static struct d40_desc *d40_desc_get(struct d40_chan *d40c) list_for_each_entry_safe(d, _d, &d40c->client, node) if (async_tx_test_ack(&d->txd)) { - d40_pool_lli_free(d); + d40_pool_lli_free(d40c, d); d40_desc_remove(d); desc = d; memset(desc, 0, sizeof(*desc)); @@ -449,6 +495,7 @@ static struct d40_desc *d40_desc_get(struct d40_chan *d40c) static void d40_desc_free(struct d40_chan *d40c, struct d40_desc *d40d) { + d40_pool_lli_free(d40c, d40d); d40_lcla_free_all(d40c, d40d); kmem_cache_free(d40c->base->desc_slab, d40d); } @@ -458,57 +505,128 @@ static void d40_desc_submit(struct d40_chan *d40c, struct d40_desc *desc) list_add_tail(&desc->node, &d40c->active); } -static void d40_desc_load(struct d40_chan *d40c, struct d40_desc *d40d) +static void d40_phy_lli_load(struct d40_chan *chan, struct d40_desc *desc) { - int curr_lcla = -EINVAL, next_lcla; + struct d40_phy_lli *lli_dst = desc->lli_phy.dst; + struct d40_phy_lli *lli_src = desc->lli_phy.src; + void __iomem *base = chan_base(chan); + + writel(lli_src->reg_cfg, base + D40_CHAN_REG_SSCFG); + writel(lli_src->reg_elt, base + D40_CHAN_REG_SSELT); + writel(lli_src->reg_ptr, base + D40_CHAN_REG_SSPTR); + writel(lli_src->reg_lnk, base + D40_CHAN_REG_SSLNK); + + writel(lli_dst->reg_cfg, base + D40_CHAN_REG_SDCFG); + writel(lli_dst->reg_elt, base + D40_CHAN_REG_SDELT); + writel(lli_dst->reg_ptr, base + D40_CHAN_REG_SDPTR); + writel(lli_dst->reg_lnk, base + D40_CHAN_REG_SDLNK); +} - if (d40c->log_num == D40_PHY_CHAN) { - d40_phy_lli_write(d40c->base->virtbase, - d40c->phy_chan->num, - d40d->lli_phy.dst, - d40d->lli_phy.src); - d40d->lli_current = d40d->lli_len; - } else { +static void d40_log_lli_to_lcxa(struct d40_chan *chan, struct d40_desc *desc) +{ + struct d40_lcla_pool *pool = &chan->base->lcla_pool; + struct d40_log_lli_bidir *lli = &desc->lli_log; + int lli_current = desc->lli_current; + int lli_len = desc->lli_len; + bool cyclic = desc->cyclic; + int curr_lcla = -EINVAL; + int first_lcla = 0; + bool linkback; - if ((d40d->lli_len - d40d->lli_current) > 1) - curr_lcla = d40_lcla_alloc_one(d40c, d40d); + /* + * We may have partially running cyclic transfers, in case we did't get + * enough LCLA entries. + */ + linkback = cyclic && lli_current == 0; - d40_log_lli_lcpa_write(d40c->lcpa, - &d40d->lli_log.dst[d40d->lli_current], - &d40d->lli_log.src[d40d->lli_current], - curr_lcla); + /* + * For linkback, we need one LCLA even with only one link, because we + * can't link back to the one in LCPA space + */ + if (linkback || (lli_len - lli_current > 1)) { + curr_lcla = d40_lcla_alloc_one(chan, desc); + first_lcla = curr_lcla; + } - d40d->lli_current++; - for (; d40d->lli_current < d40d->lli_len; d40d->lli_current++) { - struct d40_log_lli *lcla; + /* + * For linkback, we normally load the LCPA in the loop since we need to + * link it to the second LCLA and not the first. However, if we + * couldn't even get a first LCLA, then we have to run in LCPA and + * reload manually. + */ + if (!linkback || curr_lcla == -EINVAL) { + unsigned int flags = 0; - if (d40d->lli_current + 1 < d40d->lli_len) - next_lcla = d40_lcla_alloc_one(d40c, d40d); - else - next_lcla = -EINVAL; + if (curr_lcla == -EINVAL) + flags |= LLI_TERM_INT; - lcla = d40c->base->lcla_pool.base + - d40c->phy_chan->num * 1024 + - 8 * curr_lcla * 2; + d40_log_lli_lcpa_write(chan->lcpa, + &lli->dst[lli_current], + &lli->src[lli_current], + curr_lcla, + flags); + lli_current++; + } - d40_log_lli_lcla_write(lcla, - &d40d->lli_log.dst[d40d->lli_current], - &d40d->lli_log.src[d40d->lli_current], - next_lcla); + if (curr_lcla < 0) + goto out; - (void) dma_map_single(d40c->base->dev, lcla, - 2 * sizeof(struct d40_log_lli), - DMA_TO_DEVICE); + for (; lli_current < lli_len; lli_current++) { + unsigned int lcla_offset = chan->phy_chan->num * 1024 + + 8 * curr_lcla * 2; + struct d40_log_lli *lcla = pool->base + lcla_offset; + unsigned int flags = 0; + int next_lcla; - curr_lcla = next_lcla; + if (lli_current + 1 < lli_len) + next_lcla = d40_lcla_alloc_one(chan, desc); + else + next_lcla = linkback ? first_lcla : -EINVAL; - if (curr_lcla == -EINVAL) { - d40d->lli_current++; - break; - } + if (cyclic || next_lcla == -EINVAL) + flags |= LLI_TERM_INT; + if (linkback && curr_lcla == first_lcla) { + /* First link goes in both LCPA and LCLA */ + d40_log_lli_lcpa_write(chan->lcpa, + &lli->dst[lli_current], + &lli->src[lli_current], + next_lcla, flags); + } + + /* + * One unused LCLA in the cyclic case if the very first + * next_lcla fails... + */ + d40_log_lli_lcla_write(lcla, + &lli->dst[lli_current], + &lli->src[lli_current], + next_lcla, flags); + + dma_sync_single_range_for_device(chan->base->dev, + pool->dma_addr, lcla_offset, + 2 * sizeof(struct d40_log_lli), + DMA_TO_DEVICE); + + curr_lcla = next_lcla; + + if (curr_lcla == -EINVAL || curr_lcla == first_lcla) { + lli_current++; + break; } } + +out: + desc->lli_current = lli_current; +} + +static void d40_desc_load(struct d40_chan *d40c, struct d40_desc *d40d) +{ + if (chan_is_physical(d40c)) { + d40_phy_lli_load(d40c, d40d); + d40d->lli_current = d40d->lli_len; + } else + d40_log_lli_to_lcxa(d40c, d40d); } static struct d40_desc *d40_first_active_get(struct d40_chan *d40c) @@ -542,20 +660,66 @@ static struct d40_desc *d40_first_queued(struct d40_chan *d40c) return d; } -static struct d40_desc *d40_last_queued(struct d40_chan *d40c) +static int d40_psize_2_burst_size(bool is_log, int psize) { - struct d40_desc *d; + if (is_log) { + if (psize == STEDMA40_PSIZE_LOG_1) + return 1; + } else { + if (psize == STEDMA40_PSIZE_PHY_1) + return 1; + } - if (list_empty(&d40c->queue)) - return NULL; - list_for_each_entry(d, &d40c->queue, node) - if (list_is_last(&d->node, &d40c->queue)) - break; - return d; + return 2 << psize; } -/* Support functions for logical channels */ +/* + * The dma only supports transmitting packages up to + * STEDMA40_MAX_SEG_SIZE << data_width. Calculate the total number of + * dma elements required to send the entire sg list + */ +static int d40_size_2_dmalen(int size, u32 data_width1, u32 data_width2) +{ + int dmalen; + u32 max_w = max(data_width1, data_width2); + u32 min_w = min(data_width1, data_width2); + u32 seg_max = ALIGN(STEDMA40_MAX_SEG_SIZE << min_w, 1 << max_w); + if (seg_max > STEDMA40_MAX_SEG_SIZE) + seg_max -= (1 << max_w); + + if (!IS_ALIGNED(size, 1 << max_w)) + return -EINVAL; + + if (size <= seg_max) + dmalen = 1; + else { + dmalen = size / seg_max; + if (dmalen * seg_max < size) + dmalen++; + } + return dmalen; +} + +static int d40_sg_2_dmalen(struct scatterlist *sgl, int sg_len, + u32 data_width1, u32 data_width2) +{ + struct scatterlist *sg; + int i; + int len = 0; + int ret; + + for_each_sg(sgl, sg, sg_len, i) { + ret = d40_size_2_dmalen(sg_dma_len(sg), + data_width1, data_width2); + if (ret < 0) + return ret; + len += ret; + } + return len; +} + +/* Support functions for logical channels */ static int d40_channel_execute_command(struct d40_chan *d40c, enum d40_command command) @@ -607,9 +771,9 @@ static int d40_channel_execute_command(struct d40_chan *d40c, } if (i == D40_SUSPEND_MAX_IT) { - dev_err(&d40c->chan.dev->device, - "[%s]: unable to suspend the chl %d (log: %d) status %x\n", - __func__, d40c->phy_chan->num, d40c->log_num, + chan_err(d40c, + "unable to suspend the chl %d (log: %d) status %x\n", + d40c->phy_chan->num, d40c->log_num, status); dump_stack(); ret = -EBUSY; @@ -642,17 +806,45 @@ static void d40_term_all(struct d40_chan *d40c) d40c->busy = false; } +static void __d40_config_set_event(struct d40_chan *d40c, bool enable, + u32 event, int reg) +{ + void __iomem *addr = chan_base(d40c) + reg; + int tries; + + if (!enable) { + writel((D40_DEACTIVATE_EVENTLINE << D40_EVENTLINE_POS(event)) + | ~D40_EVENTLINE_MASK(event), addr); + return; + } + + /* + * The hardware sometimes doesn't register the enable when src and dst + * event lines are active on the same logical channel. Retry to ensure + * it does. Usually only one retry is sufficient. + */ + tries = 100; + while (--tries) { + writel((D40_ACTIVATE_EVENTLINE << D40_EVENTLINE_POS(event)) + | ~D40_EVENTLINE_MASK(event), addr); + + if (readl(addr) & D40_EVENTLINE_MASK(event)) + break; + } + + if (tries != 99) + dev_dbg(chan2dev(d40c), + "[%s] workaround enable S%cLNK (%d tries)\n", + __func__, reg == D40_CHAN_REG_SSLNK ? 'S' : 'D', + 100 - tries); + + WARN_ON(!tries); +} + static void d40_config_set_event(struct d40_chan *d40c, bool do_enable) { - u32 val; unsigned long flags; - /* Notice, that disable requires the physical channel to be stopped */ - if (do_enable) - val = D40_ACTIVATE_EVENTLINE; - else - val = D40_DEACTIVATE_EVENTLINE; - spin_lock_irqsave(&d40c->phy_chan->lock, flags); /* Enable event line connected to device (or memcpy) */ @@ -660,20 +852,15 @@ static void d40_config_set_event(struct d40_chan *d40c, bool do_enable) (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_PERIPH)) { u32 event = D40_TYPE_TO_EVENT(d40c->dma_cfg.src_dev_type); - writel((val << D40_EVENTLINE_POS(event)) | - ~D40_EVENTLINE_MASK(event), - d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SSLNK); + __d40_config_set_event(d40c, do_enable, event, + D40_CHAN_REG_SSLNK); } + if (d40c->dma_cfg.dir != STEDMA40_PERIPH_TO_MEM) { u32 event = D40_TYPE_TO_EVENT(d40c->dma_cfg.dst_dev_type); - writel((val << D40_EVENTLINE_POS(event)) | - ~D40_EVENTLINE_MASK(event), - d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SDLNK); + __d40_config_set_event(d40c, do_enable, event, + D40_CHAN_REG_SDLNK); } spin_unlock_irqrestore(&d40c->phy_chan->lock, flags); @@ -681,15 +868,12 @@ static void d40_config_set_event(struct d40_chan *d40c, bool do_enable) static u32 d40_chan_has_events(struct d40_chan *d40c) { + void __iomem *chanbase = chan_base(d40c); u32 val; - val = readl(d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SSLNK); + val = readl(chanbase + D40_CHAN_REG_SSLNK); + val |= readl(chanbase + D40_CHAN_REG_SDLNK); - val |= readl(d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SDLNK); return val; } @@ -712,7 +896,7 @@ static u32 d40_get_prmo(struct d40_chan *d40c) = D40_DREG_PRMO_LCHAN_SRC_LOG_DST_LOG, }; - if (d40c->log_num == D40_PHY_CHAN) + if (chan_is_physical(d40c)) return phy_map[d40c->dma_cfg.mode_opt]; else return log_map[d40c->dma_cfg.mode_opt]; @@ -726,7 +910,7 @@ static void d40_config_write(struct d40_chan *d40c) /* Odd addresses are even addresses + 4 */ addr_base = (d40c->phy_chan->num % 2) * 4; /* Setup channel mode to logical or physical */ - var = ((u32)(d40c->log_num != D40_PHY_CHAN) + 1) << + var = ((u32)(chan_is_logical(d40c)) + 1) << D40_CHAN_POS(d40c->phy_chan->num); writel(var, d40c->base->virtbase + D40_DREG_PRMSE + addr_base); @@ -735,30 +919,18 @@ static void d40_config_write(struct d40_chan *d40c) writel(var, d40c->base->virtbase + D40_DREG_PRMOE + addr_base); - if (d40c->log_num != D40_PHY_CHAN) { + if (chan_is_logical(d40c)) { + int lidx = (d40c->phy_chan->num << D40_SREG_ELEM_LOG_LIDX_POS) + & D40_SREG_ELEM_LOG_LIDX_MASK; + void __iomem *chanbase = chan_base(d40c); + /* Set default config for CFG reg */ - writel(d40c->src_def_cfg, - d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SSCFG); - writel(d40c->dst_def_cfg, - d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SDCFG); + writel(d40c->src_def_cfg, chanbase + D40_CHAN_REG_SSCFG); + writel(d40c->dst_def_cfg, chanbase + D40_CHAN_REG_SDCFG); /* Set LIDX for lcla */ - writel((d40c->phy_chan->num << D40_SREG_ELEM_LOG_LIDX_POS) & - D40_SREG_ELEM_LOG_LIDX_MASK, - d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SDELT); - - writel((d40c->phy_chan->num << D40_SREG_ELEM_LOG_LIDX_POS) & - D40_SREG_ELEM_LOG_LIDX_MASK, - d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SSELT); - + writel(lidx, chanbase + D40_CHAN_REG_SSELT); + writel(lidx, chanbase + D40_CHAN_REG_SDELT); } } @@ -766,15 +938,15 @@ static u32 d40_residue(struct d40_chan *d40c) { u32 num_elt; - if (d40c->log_num != D40_PHY_CHAN) + if (chan_is_logical(d40c)) num_elt = (readl(&d40c->lcpa->lcsp2) & D40_MEM_LCSP2_ECNT_MASK) >> D40_MEM_LCSP2_ECNT_POS; - else - num_elt = (readl(d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SDELT) & - D40_SREG_ELEM_PHY_ECNT_MASK) >> - D40_SREG_ELEM_PHY_ECNT_POS; + else { + u32 val = readl(chan_base(d40c) + D40_CHAN_REG_SDELT); + num_elt = (val & D40_SREG_ELEM_PHY_ECNT_MASK) + >> D40_SREG_ELEM_PHY_ECNT_POS; + } + return num_elt * (1 << d40c->dma_cfg.dst_info.data_width); } @@ -782,20 +954,17 @@ static bool d40_tx_is_linked(struct d40_chan *d40c) { bool is_link; - if (d40c->log_num != D40_PHY_CHAN) + if (chan_is_logical(d40c)) is_link = readl(&d40c->lcpa->lcsp3) & D40_MEM_LCSP3_DLOS_MASK; else - is_link = readl(d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SDLNK) & - D40_SREG_LNK_PHYS_LNK_MASK; + is_link = readl(chan_base(d40c) + D40_CHAN_REG_SDLNK) + & D40_SREG_LNK_PHYS_LNK_MASK; + return is_link; } -static int d40_pause(struct dma_chan *chan) +static int d40_pause(struct d40_chan *d40c) { - struct d40_chan *d40c = - container_of(chan, struct d40_chan, chan); int res = 0; unsigned long flags; @@ -806,7 +975,7 @@ static int d40_pause(struct dma_chan *chan) res = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ); if (res == 0) { - if (d40c->log_num != D40_PHY_CHAN) { + if (chan_is_logical(d40c)) { d40_config_set_event(d40c, false); /* Resume the other logical channels if any */ if (d40_chan_has_events(d40c)) @@ -819,10 +988,8 @@ static int d40_pause(struct dma_chan *chan) return res; } -static int d40_resume(struct dma_chan *chan) +static int d40_resume(struct d40_chan *d40c) { - struct d40_chan *d40c = - container_of(chan, struct d40_chan, chan); int res = 0; unsigned long flags; @@ -832,7 +999,7 @@ static int d40_resume(struct dma_chan *chan) spin_lock_irqsave(&d40c->lock, flags); if (d40c->base->rev == 0) - if (d40c->log_num != D40_PHY_CHAN) { + if (chan_is_logical(d40c)) { res = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ); goto no_suspend; @@ -841,7 +1008,7 @@ static int d40_resume(struct dma_chan *chan) /* If bytes left to transfer or linked tx resume job */ if (d40_residue(d40c) || d40_tx_is_linked(d40c)) { - if (d40c->log_num != D40_PHY_CHAN) + if (chan_is_logical(d40c)) d40_config_set_event(d40c, true); res = d40_channel_execute_command(d40c, D40_DMA_RUN); @@ -852,75 +1019,20 @@ no_suspend: return res; } -static void d40_tx_submit_log(struct d40_chan *d40c, struct d40_desc *d40d) -{ - /* TODO: Write */ -} - -static void d40_tx_submit_phy(struct d40_chan *d40c, struct d40_desc *d40d) +static int d40_terminate_all(struct d40_chan *chan) { - struct d40_desc *d40d_prev = NULL; - int i; - u32 val; - - if (!list_empty(&d40c->queue)) - d40d_prev = d40_last_queued(d40c); - else if (!list_empty(&d40c->active)) - d40d_prev = d40_first_active_get(d40c); - - if (!d40d_prev) - return; - - /* Here we try to join this job with previous jobs */ - val = readl(d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SSLNK); - - /* Figure out which link we're currently transmitting */ - for (i = 0; i < d40d_prev->lli_len; i++) - if (val == d40d_prev->lli_phy.src[i].reg_lnk) - break; - - val = readl(d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SSELT) >> D40_SREG_ELEM_LOG_ECNT_POS; - - if (i == (d40d_prev->lli_len - 1) && val > 0) { - /* Change the current one */ - writel(virt_to_phys(d40d->lli_phy.src), - d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SSLNK); - writel(virt_to_phys(d40d->lli_phy.dst), - d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SDLNK); - - d40d->is_hw_linked = true; - - } else if (i < d40d_prev->lli_len) { - (void) dma_unmap_single(d40c->base->dev, - virt_to_phys(d40d_prev->lli_phy.src), - d40d_prev->lli_pool.size, - DMA_TO_DEVICE); + unsigned long flags; + int ret = 0; - /* Keep the settings */ - val = d40d_prev->lli_phy.src[d40d_prev->lli_len - 1].reg_lnk & - ~D40_SREG_LNK_PHYS_LNK_MASK; - d40d_prev->lli_phy.src[d40d_prev->lli_len - 1].reg_lnk = - val | virt_to_phys(d40d->lli_phy.src); + ret = d40_pause(chan); + if (!ret && chan_is_physical(chan)) + ret = d40_channel_execute_command(chan, D40_DMA_STOP); - val = d40d_prev->lli_phy.dst[d40d_prev->lli_len - 1].reg_lnk & - ~D40_SREG_LNK_PHYS_LNK_MASK; - d40d_prev->lli_phy.dst[d40d_prev->lli_len - 1].reg_lnk = - val | virt_to_phys(d40d->lli_phy.dst); + spin_lock_irqsave(&chan->lock, flags); + d40_term_all(chan); + spin_unlock_irqrestore(&chan->lock, flags); - (void) dma_map_single(d40c->base->dev, - d40d_prev->lli_phy.src, - d40d_prev->lli_pool.size, - DMA_TO_DEVICE); - d40d->is_hw_linked = true; - } + return ret; } static dma_cookie_t d40_tx_submit(struct dma_async_tx_descriptor *tx) @@ -931,8 +1043,6 @@ static dma_cookie_t d40_tx_submit(struct dma_async_tx_descriptor *tx) struct d40_desc *d40d = container_of(tx, struct d40_desc, txd); unsigned long flags; - (void) d40_pause(&d40c->chan); - spin_lock_irqsave(&d40c->lock, flags); d40c->chan.cookie++; @@ -942,17 +1052,10 @@ static dma_cookie_t d40_tx_submit(struct dma_async_tx_descriptor *tx) d40d->txd.cookie = d40c->chan.cookie; - if (d40c->log_num == D40_PHY_CHAN) - d40_tx_submit_phy(d40c, d40d); - else - d40_tx_submit_log(d40c, d40d); - d40_desc_queue(d40c, d40d); spin_unlock_irqrestore(&d40c->lock, flags); - (void) d40_resume(&d40c->chan); - return tx->cookie; } @@ -961,7 +1064,7 @@ static int d40_start(struct d40_chan *d40c) if (d40c->base->rev == 0) { int err; - if (d40c->log_num != D40_PHY_CHAN) { + if (chan_is_logical(d40c)) { err = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ); if (err) @@ -969,7 +1072,7 @@ static int d40_start(struct d40_chan *d40c) } } - if (d40c->log_num != D40_PHY_CHAN) + if (chan_is_logical(d40c)) d40_config_set_event(d40c, true); return d40_channel_execute_command(d40c, D40_DMA_RUN); @@ -992,21 +1095,14 @@ static struct d40_desc *d40_queue_start(struct d40_chan *d40c) /* Add to active queue */ d40_desc_submit(d40c, d40d); - /* - * If this job is already linked in hw, - * do not submit it. - */ - - if (!d40d->is_hw_linked) { - /* Initiate DMA job */ - d40_desc_load(d40c, d40d); + /* Initiate DMA job */ + d40_desc_load(d40c, d40d); - /* Start dma job */ - err = d40_start(d40c); + /* Start dma job */ + err = d40_start(d40c); - if (err) - return NULL; - } + if (err) + return NULL; } return d40d; @@ -1023,17 +1119,36 @@ static void dma_tc_handle(struct d40_chan *d40c) if (d40d == NULL) return; - d40_lcla_free_all(d40c, d40d); + if (d40d->cyclic) { + /* + * If this was a paritially loaded list, we need to reloaded + * it, and only when the list is completed. We need to check + * for done because the interrupt will hit for every link, and + * not just the last one. + */ + if (d40d->lli_current < d40d->lli_len + && !d40_tx_is_linked(d40c) + && !d40_residue(d40c)) { + d40_lcla_free_all(d40c, d40d); + d40_desc_load(d40c, d40d); + (void) d40_start(d40c); - if (d40d->lli_current < d40d->lli_len) { - d40_desc_load(d40c, d40d); - /* Start dma job */ - (void) d40_start(d40c); - return; - } + if (d40d->lli_current == d40d->lli_len) + d40d->lli_current = 0; + } + } else { + d40_lcla_free_all(d40c, d40d); - if (d40_queue_start(d40c) == NULL) - d40c->busy = false; + if (d40d->lli_current < d40d->lli_len) { + d40_desc_load(d40c, d40d); + /* Start dma job */ + (void) d40_start(d40c); + return; + } + + if (d40_queue_start(d40c) == NULL) + d40c->busy = false; + } d40c->pending_tx++; tasklet_schedule(&d40c->tasklet); @@ -1052,11 +1167,11 @@ static void dma_tasklet(unsigned long data) /* Get first active entry from list */ d40d = d40_first_active_get(d40c); - if (d40d == NULL) goto err; - d40c->completed = d40d->txd.cookie; + if (!d40d->cyclic) + d40c->completed = d40d->txd.cookie; /* * If terminating a channel pending_tx is set to zero. @@ -1071,16 +1186,18 @@ static void dma_tasklet(unsigned long data) callback = d40d->txd.callback; callback_param = d40d->txd.callback_param; - if (async_tx_test_ack(&d40d->txd)) { - d40_pool_lli_free(d40d); - d40_desc_remove(d40d); - d40_desc_free(d40c, d40d); - } else { - if (!d40d->is_in_client_list) { + if (!d40d->cyclic) { + if (async_tx_test_ack(&d40d->txd)) { + d40_pool_lli_free(d40c, d40d); d40_desc_remove(d40d); - d40_lcla_free_all(d40c, d40d); - list_add_tail(&d40d->node, &d40c->client); - d40d->is_in_client_list = true; + d40_desc_free(d40c, d40d); + } else { + if (!d40d->is_in_client_list) { + d40_desc_remove(d40d); + d40_lcla_free_all(d40c, d40d); + list_add_tail(&d40d->node, &d40c->client); + d40d->is_in_client_list = true; + } } } @@ -1157,9 +1274,8 @@ static irqreturn_t d40_handle_interrupt(int irq, void *data) if (!il[row].is_error) dma_tc_handle(d40c); else - dev_err(base->dev, - "[%s] IRQ chan: %ld offset %d idx %d\n", - __func__, chan, il[row].offset, idx); + d40_err(base->dev, "IRQ chan: %ld offset %d idx %d\n", + chan, il[row].offset, idx); spin_unlock(&d40c->lock); } @@ -1178,8 +1294,7 @@ static int d40_validate_conf(struct d40_chan *d40c, bool is_log = conf->mode == STEDMA40_MODE_LOGICAL; if (!conf->dir) { - dev_err(&d40c->chan.dev->device, "[%s] Invalid direction.\n", - __func__); + chan_err(d40c, "Invalid direction.\n"); res = -EINVAL; } @@ -1187,46 +1302,40 @@ static int d40_validate_conf(struct d40_chan *d40c, d40c->base->plat_data->dev_tx[conf->dst_dev_type] == 0 && d40c->runtime_addr == 0) { - dev_err(&d40c->chan.dev->device, - "[%s] Invalid TX channel address (%d)\n", - __func__, conf->dst_dev_type); + chan_err(d40c, "Invalid TX channel address (%d)\n", + conf->dst_dev_type); res = -EINVAL; } if (conf->src_dev_type != STEDMA40_DEV_SRC_MEMORY && d40c->base->plat_data->dev_rx[conf->src_dev_type] == 0 && d40c->runtime_addr == 0) { - dev_err(&d40c->chan.dev->device, - "[%s] Invalid RX channel address (%d)\n", - __func__, conf->src_dev_type); + chan_err(d40c, "Invalid RX channel address (%d)\n", + conf->src_dev_type); res = -EINVAL; } if (conf->dir == STEDMA40_MEM_TO_PERIPH && dst_event_group == STEDMA40_DEV_DST_MEMORY) { - dev_err(&d40c->chan.dev->device, "[%s] Invalid dst\n", - __func__); + chan_err(d40c, "Invalid dst\n"); res = -EINVAL; } if (conf->dir == STEDMA40_PERIPH_TO_MEM && src_event_group == STEDMA40_DEV_SRC_MEMORY) { - dev_err(&d40c->chan.dev->device, "[%s] Invalid src\n", - __func__); + chan_err(d40c, "Invalid src\n"); res = -EINVAL; } if (src_event_group == STEDMA40_DEV_SRC_MEMORY && dst_event_group == STEDMA40_DEV_DST_MEMORY && is_log) { - dev_err(&d40c->chan.dev->device, - "[%s] No event line\n", __func__); + chan_err(d40c, "No event line\n"); res = -EINVAL; } if (conf->dir == STEDMA40_PERIPH_TO_PERIPH && (src_event_group != dst_event_group)) { - dev_err(&d40c->chan.dev->device, - "[%s] Invalid event group\n", __func__); + chan_err(d40c, "Invalid event group\n"); res = -EINVAL; } @@ -1235,9 +1344,20 @@ static int d40_validate_conf(struct d40_chan *d40c, * DMAC HW supports it. Will be added to this driver, * in case any dma client requires it. */ - dev_err(&d40c->chan.dev->device, - "[%s] periph to periph not supported\n", - __func__); + chan_err(d40c, "periph to periph not supported\n"); + res = -EINVAL; + } + + if (d40_psize_2_burst_size(is_log, conf->src_info.psize) * + (1 << conf->src_info.data_width) != + d40_psize_2_burst_size(is_log, conf->dst_info.psize) * + (1 << conf->dst_info.data_width)) { + /* + * The DMAC hardware only supports + * src (burst x width) == dst (burst x width) + */ + + chan_err(d40c, "src (burst x width) != dst (burst x width)\n"); res = -EINVAL; } @@ -1440,8 +1560,7 @@ static int d40_config_memcpy(struct d40_chan *d40c) dma_has_cap(DMA_SLAVE, cap)) { d40c->dma_cfg = *d40c->base->plat_data->memcpy_conf_phy; } else { - dev_err(&d40c->chan.dev->device, "[%s] No memcpy\n", - __func__); + chan_err(d40c, "No memcpy\n"); return -EINVAL; } @@ -1466,21 +1585,19 @@ static int d40_free_dma(struct d40_chan *d40c) /* Release client owned descriptors */ if (!list_empty(&d40c->client)) list_for_each_entry_safe(d, _d, &d40c->client, node) { - d40_pool_lli_free(d); + d40_pool_lli_free(d40c, d); d40_desc_remove(d); d40_desc_free(d40c, d); } if (phy == NULL) { - dev_err(&d40c->chan.dev->device, "[%s] phy == null\n", - __func__); + chan_err(d40c, "phy == null\n"); return -EINVAL; } if (phy->allocated_src == D40_ALLOC_FREE && phy->allocated_dst == D40_ALLOC_FREE) { - dev_err(&d40c->chan.dev->device, "[%s] channel already free\n", - __func__); + chan_err(d40c, "channel already free\n"); return -EINVAL; } @@ -1492,19 +1609,17 @@ static int d40_free_dma(struct d40_chan *d40c) event = D40_TYPE_TO_EVENT(d40c->dma_cfg.src_dev_type); is_src = true; } else { - dev_err(&d40c->chan.dev->device, - "[%s] Unknown direction\n", __func__); + chan_err(d40c, "Unknown direction\n"); return -EINVAL; } res = d40_channel_execute_command(d40c, D40_DMA_SUSPEND_REQ); if (res) { - dev_err(&d40c->chan.dev->device, "[%s] suspend failed\n", - __func__); + chan_err(d40c, "suspend failed\n"); return res; } - if (d40c->log_num != D40_PHY_CHAN) { + if (chan_is_logical(d40c)) { /* Release logical channel, deactivate the event line */ d40_config_set_event(d40c, false); @@ -1520,9 +1635,8 @@ static int d40_free_dma(struct d40_chan *d40c) res = d40_channel_execute_command(d40c, D40_DMA_RUN); if (res) { - dev_err(&d40c->chan.dev->device, - "[%s] Executing RUN command\n", - __func__); + chan_err(d40c, + "Executing RUN command\n"); return res; } } @@ -1535,8 +1649,7 @@ static int d40_free_dma(struct d40_chan *d40c) /* Release physical channel */ res = d40_channel_execute_command(d40c, D40_DMA_STOP); if (res) { - dev_err(&d40c->chan.dev->device, - "[%s] Failed to stop channel\n", __func__); + chan_err(d40c, "Failed to stop channel\n"); return res; } d40c->phy_chan = NULL; @@ -1548,6 +1661,7 @@ static int d40_free_dma(struct d40_chan *d40c) static bool d40_is_paused(struct d40_chan *d40c) { + void __iomem *chanbase = chan_base(d40c); bool is_paused = false; unsigned long flags; void __iomem *active_reg; @@ -1556,7 +1670,7 @@ static bool d40_is_paused(struct d40_chan *d40c) spin_lock_irqsave(&d40c->lock, flags); - if (d40c->log_num == D40_PHY_CHAN) { + if (chan_is_physical(d40c)) { if (d40c->phy_chan->num % 2 == 0) active_reg = d40c->base->virtbase + D40_DREG_ACTIVE; else @@ -1574,17 +1688,12 @@ static bool d40_is_paused(struct d40_chan *d40c) if (d40c->dma_cfg.dir == STEDMA40_MEM_TO_PERIPH || d40c->dma_cfg.dir == STEDMA40_MEM_TO_MEM) { event = D40_TYPE_TO_EVENT(d40c->dma_cfg.dst_dev_type); - status = readl(d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SDLNK); + status = readl(chanbase + D40_CHAN_REG_SDLNK); } else if (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_MEM) { event = D40_TYPE_TO_EVENT(d40c->dma_cfg.src_dev_type); - status = readl(d40c->base->virtbase + D40_DREG_PCBASE + - d40c->phy_chan->num * D40_DREG_PCDELTA + - D40_CHAN_REG_SSLNK); + status = readl(chanbase + D40_CHAN_REG_SSLNK); } else { - dev_err(&d40c->chan.dev->device, - "[%s] Unknown direction\n", __func__); + chan_err(d40c, "Unknown direction\n"); goto _exit; } @@ -1614,102 +1723,184 @@ static u32 stedma40_residue(struct dma_chan *chan) return bytes_left; } -struct dma_async_tx_descriptor *stedma40_memcpy_sg(struct dma_chan *chan, - struct scatterlist *sgl_dst, - struct scatterlist *sgl_src, - unsigned int sgl_len, - unsigned long dma_flags) +static int +d40_prep_sg_log(struct d40_chan *chan, struct d40_desc *desc, + struct scatterlist *sg_src, struct scatterlist *sg_dst, + unsigned int sg_len, dma_addr_t src_dev_addr, + dma_addr_t dst_dev_addr) { - int res; - struct d40_desc *d40d; - struct d40_chan *d40c = container_of(chan, struct d40_chan, - chan); - unsigned long flags; + struct stedma40_chan_cfg *cfg = &chan->dma_cfg; + struct stedma40_half_channel_info *src_info = &cfg->src_info; + struct stedma40_half_channel_info *dst_info = &cfg->dst_info; + int ret; - if (d40c->phy_chan == NULL) { - dev_err(&d40c->chan.dev->device, - "[%s] Unallocated channel.\n", __func__); - return ERR_PTR(-EINVAL); - } + ret = d40_log_sg_to_lli(sg_src, sg_len, + src_dev_addr, + desc->lli_log.src, + chan->log_def.lcsp1, + src_info->data_width, + dst_info->data_width); - spin_lock_irqsave(&d40c->lock, flags); - d40d = d40_desc_get(d40c); + ret = d40_log_sg_to_lli(sg_dst, sg_len, + dst_dev_addr, + desc->lli_log.dst, + chan->log_def.lcsp3, + dst_info->data_width, + src_info->data_width); - if (d40d == NULL) + return ret < 0 ? ret : 0; +} + +static int +d40_prep_sg_phy(struct d40_chan *chan, struct d40_desc *desc, + struct scatterlist *sg_src, struct scatterlist *sg_dst, + unsigned int sg_len, dma_addr_t src_dev_addr, + dma_addr_t dst_dev_addr) +{ + struct stedma40_chan_cfg *cfg = &chan->dma_cfg; + struct stedma40_half_channel_info *src_info = &cfg->src_info; + struct stedma40_half_channel_info *dst_info = &cfg->dst_info; + unsigned long flags = 0; + int ret; + + if (desc->cyclic) + flags |= LLI_CYCLIC | LLI_TERM_INT; + + ret = d40_phy_sg_to_lli(sg_src, sg_len, src_dev_addr, + desc->lli_phy.src, + virt_to_phys(desc->lli_phy.src), + chan->src_def_cfg, + src_info, dst_info, flags); + + ret = d40_phy_sg_to_lli(sg_dst, sg_len, dst_dev_addr, + desc->lli_phy.dst, + virt_to_phys(desc->lli_phy.dst), + chan->dst_def_cfg, + dst_info, src_info, flags); + + dma_sync_single_for_device(chan->base->dev, desc->lli_pool.dma_addr, + desc->lli_pool.size, DMA_TO_DEVICE); + + return ret < 0 ? ret : 0; +} + + +static struct d40_desc * +d40_prep_desc(struct d40_chan *chan, struct scatterlist *sg, + unsigned int sg_len, unsigned long dma_flags) +{ + struct stedma40_chan_cfg *cfg = &chan->dma_cfg; + struct d40_desc *desc; + int ret; + + desc = d40_desc_get(chan); + if (!desc) + return NULL; + + desc->lli_len = d40_sg_2_dmalen(sg, sg_len, cfg->src_info.data_width, + cfg->dst_info.data_width); + if (desc->lli_len < 0) { + chan_err(chan, "Unaligned size\n"); goto err; + } - d40d->lli_len = sgl_len; - d40d->lli_current = 0; - d40d->txd.flags = dma_flags; + ret = d40_pool_lli_alloc(chan, desc, desc->lli_len); + if (ret < 0) { + chan_err(chan, "Could not allocate lli\n"); + goto err; + } - if (d40c->log_num != D40_PHY_CHAN) { - if (d40_pool_lli_alloc(d40d, sgl_len, true) < 0) { - dev_err(&d40c->chan.dev->device, - "[%s] Out of memory\n", __func__); - goto err; - } + desc->lli_current = 0; + desc->txd.flags = dma_flags; + desc->txd.tx_submit = d40_tx_submit; - (void) d40_log_sg_to_lli(sgl_src, - sgl_len, - d40d->lli_log.src, - d40c->log_def.lcsp1, - d40c->dma_cfg.src_info.data_width); - - (void) d40_log_sg_to_lli(sgl_dst, - sgl_len, - d40d->lli_log.dst, - d40c->log_def.lcsp3, - d40c->dma_cfg.dst_info.data_width); - } else { - if (d40_pool_lli_alloc(d40d, sgl_len, false) < 0) { - dev_err(&d40c->chan.dev->device, - "[%s] Out of memory\n", __func__); - goto err; - } + dma_async_tx_descriptor_init(&desc->txd, &chan->chan); + + return desc; + +err: + d40_desc_free(chan, desc); + return NULL; +} + +static dma_addr_t +d40_get_dev_addr(struct d40_chan *chan, enum dma_data_direction direction) +{ + struct stedma40_platform_data *plat = chan->base->plat_data; + struct stedma40_chan_cfg *cfg = &chan->dma_cfg; + dma_addr_t addr; - res = d40_phy_sg_to_lli(sgl_src, - sgl_len, - 0, - d40d->lli_phy.src, - virt_to_phys(d40d->lli_phy.src), - d40c->src_def_cfg, - d40c->dma_cfg.src_info.data_width, - d40c->dma_cfg.src_info.psize); + if (chan->runtime_addr) + return chan->runtime_addr; - if (res < 0) - goto err; + if (direction == DMA_FROM_DEVICE) + addr = plat->dev_rx[cfg->src_dev_type]; + else if (direction == DMA_TO_DEVICE) + addr = plat->dev_tx[cfg->dst_dev_type]; - res = d40_phy_sg_to_lli(sgl_dst, - sgl_len, - 0, - d40d->lli_phy.dst, - virt_to_phys(d40d->lli_phy.dst), - d40c->dst_def_cfg, - d40c->dma_cfg.dst_info.data_width, - d40c->dma_cfg.dst_info.psize); + return addr; +} - if (res < 0) - goto err; +static struct dma_async_tx_descriptor * +d40_prep_sg(struct dma_chan *dchan, struct scatterlist *sg_src, + struct scatterlist *sg_dst, unsigned int sg_len, + enum dma_data_direction direction, unsigned long dma_flags) +{ + struct d40_chan *chan = container_of(dchan, struct d40_chan, chan); + dma_addr_t src_dev_addr = 0; + dma_addr_t dst_dev_addr = 0; + struct d40_desc *desc; + unsigned long flags; + int ret; - (void) dma_map_single(d40c->base->dev, d40d->lli_phy.src, - d40d->lli_pool.size, DMA_TO_DEVICE); + if (!chan->phy_chan) { + chan_err(chan, "Cannot prepare unallocated channel\n"); + return NULL; } - dma_async_tx_descriptor_init(&d40d->txd, chan); - d40d->txd.tx_submit = d40_tx_submit; + spin_lock_irqsave(&chan->lock, flags); - spin_unlock_irqrestore(&d40c->lock, flags); + desc = d40_prep_desc(chan, sg_src, sg_len, dma_flags); + if (desc == NULL) + goto err; + + if (sg_next(&sg_src[sg_len - 1]) == sg_src) + desc->cyclic = true; + + if (direction != DMA_NONE) { + dma_addr_t dev_addr = d40_get_dev_addr(chan, direction); + + if (direction == DMA_FROM_DEVICE) + src_dev_addr = dev_addr; + else if (direction == DMA_TO_DEVICE) + dst_dev_addr = dev_addr; + } + + if (chan_is_logical(chan)) + ret = d40_prep_sg_log(chan, desc, sg_src, sg_dst, + sg_len, src_dev_addr, dst_dev_addr); + else + ret = d40_prep_sg_phy(chan, desc, sg_src, sg_dst, + sg_len, src_dev_addr, dst_dev_addr); + + if (ret) { + chan_err(chan, "Failed to prepare %s sg job: %d\n", + chan_is_logical(chan) ? "log" : "phy", ret); + goto err; + } + + spin_unlock_irqrestore(&chan->lock, flags); + + return &desc->txd; - return &d40d->txd; err: - if (d40d) - d40_desc_free(d40c, d40d); - spin_unlock_irqrestore(&d40c->lock, flags); + if (desc) + d40_desc_free(chan, desc); + spin_unlock_irqrestore(&chan->lock, flags); return NULL; } -EXPORT_SYMBOL(stedma40_memcpy_sg); bool stedma40_filter(struct dma_chan *chan, void *data) { @@ -1732,6 +1923,38 @@ bool stedma40_filter(struct dma_chan *chan, void *data) } EXPORT_SYMBOL(stedma40_filter); +static void __d40_set_prio_rt(struct d40_chan *d40c, int dev_type, bool src) +{ + bool realtime = d40c->dma_cfg.realtime; + bool highprio = d40c->dma_cfg.high_priority; + u32 prioreg = highprio ? D40_DREG_PSEG1 : D40_DREG_PCEG1; + u32 rtreg = realtime ? D40_DREG_RSEG1 : D40_DREG_RCEG1; + u32 event = D40_TYPE_TO_EVENT(dev_type); + u32 group = D40_TYPE_TO_GROUP(dev_type); + u32 bit = 1 << event; + + /* Destination event lines are stored in the upper halfword */ + if (!src) + bit <<= 16; + + writel(bit, d40c->base->virtbase + prioreg + group * 4); + writel(bit, d40c->base->virtbase + rtreg + group * 4); +} + +static void d40_set_prio_realtime(struct d40_chan *d40c) +{ + if (d40c->base->rev < 3) + return; + + if ((d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_MEM) || + (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_PERIPH)) + __d40_set_prio_rt(d40c, d40c->dma_cfg.src_dev_type, true); + + if ((d40c->dma_cfg.dir == STEDMA40_MEM_TO_PERIPH) || + (d40c->dma_cfg.dir == STEDMA40_PERIPH_TO_PERIPH)) + __d40_set_prio_rt(d40c, d40c->dma_cfg.dst_dev_type, false); +} + /* DMA ENGINE functions */ static int d40_alloc_chan_resources(struct dma_chan *chan) { @@ -1748,9 +1971,7 @@ static int d40_alloc_chan_resources(struct dma_chan *chan) if (!d40c->configured) { err = d40_config_memcpy(d40c); if (err) { - dev_err(&d40c->chan.dev->device, - "[%s] Failed to configure memcpy channel\n", - __func__); + chan_err(d40c, "Failed to configure memcpy channel\n"); goto fail; } } @@ -1758,16 +1979,17 @@ static int d40_alloc_chan_resources(struct dma_chan *chan) err = d40_allocate_channel(d40c); if (err) { - dev_err(&d40c->chan.dev->device, - "[%s] Failed to allocate channel\n", __func__); + chan_err(d40c, "Failed to allocate channel\n"); goto fail; } /* Fill in basic CFG register values */ d40_phy_cfg(&d40c->dma_cfg, &d40c->src_def_cfg, - &d40c->dst_def_cfg, d40c->log_num != D40_PHY_CHAN); + &d40c->dst_def_cfg, chan_is_logical(d40c)); + + d40_set_prio_realtime(d40c); - if (d40c->log_num != D40_PHY_CHAN) { + if (chan_is_logical(d40c)) { d40_log_cfg(&d40c->dma_cfg, &d40c->log_def.lcsp1, &d40c->log_def.lcsp3); @@ -1800,8 +2022,7 @@ static void d40_free_chan_resources(struct dma_chan *chan) unsigned long flags; if (d40c->phy_chan == NULL) { - dev_err(&d40c->chan.dev->device, - "[%s] Cannot free unallocated channel\n", __func__); + chan_err(d40c, "Cannot free unallocated channel\n"); return; } @@ -1811,8 +2032,7 @@ static void d40_free_chan_resources(struct dma_chan *chan) err = d40_free_dma(d40c); if (err) - dev_err(&d40c->chan.dev->device, - "[%s] Failed to free channel\n", __func__); + chan_err(d40c, "Failed to free channel\n"); spin_unlock_irqrestore(&d40c->lock, flags); } @@ -1822,226 +2042,31 @@ static struct dma_async_tx_descriptor *d40_prep_memcpy(struct dma_chan *chan, size_t size, unsigned long dma_flags) { - struct d40_desc *d40d; - struct d40_chan *d40c = container_of(chan, struct d40_chan, - chan); - unsigned long flags; - int err = 0; + struct scatterlist dst_sg; + struct scatterlist src_sg; - if (d40c->phy_chan == NULL) { - dev_err(&d40c->chan.dev->device, - "[%s] Channel is not allocated.\n", __func__); - return ERR_PTR(-EINVAL); - } + sg_init_table(&dst_sg, 1); + sg_init_table(&src_sg, 1); - spin_lock_irqsave(&d40c->lock, flags); - d40d = d40_desc_get(d40c); + sg_dma_address(&dst_sg) = dst; + sg_dma_address(&src_sg) = src; - if (d40d == NULL) { - dev_err(&d40c->chan.dev->device, - "[%s] Descriptor is NULL\n", __func__); - goto err; - } - - d40d->txd.flags = dma_flags; - - dma_async_tx_descriptor_init(&d40d->txd, chan); - - d40d->txd.tx_submit = d40_tx_submit; - - if (d40c->log_num != D40_PHY_CHAN) { - - if (d40_pool_lli_alloc(d40d, 1, true) < 0) { - dev_err(&d40c->chan.dev->device, - "[%s] Out of memory\n", __func__); - goto err; - } - d40d->lli_len = 1; - d40d->lli_current = 0; - - d40_log_fill_lli(d40d->lli_log.src, - src, - size, - d40c->log_def.lcsp1, - d40c->dma_cfg.src_info.data_width, - true); - - d40_log_fill_lli(d40d->lli_log.dst, - dst, - size, - d40c->log_def.lcsp3, - d40c->dma_cfg.dst_info.data_width, - true); - - } else { - - if (d40_pool_lli_alloc(d40d, 1, false) < 0) { - dev_err(&d40c->chan.dev->device, - "[%s] Out of memory\n", __func__); - goto err; - } - - err = d40_phy_fill_lli(d40d->lli_phy.src, - src, - size, - d40c->dma_cfg.src_info.psize, - 0, - d40c->src_def_cfg, - true, - d40c->dma_cfg.src_info.data_width, - false); - if (err) - goto err_fill_lli; - - err = d40_phy_fill_lli(d40d->lli_phy.dst, - dst, - size, - d40c->dma_cfg.dst_info.psize, - 0, - d40c->dst_def_cfg, - true, - d40c->dma_cfg.dst_info.data_width, - false); + sg_dma_len(&dst_sg) = size; + sg_dma_len(&src_sg) = size; - if (err) - goto err_fill_lli; - - (void) dma_map_single(d40c->base->dev, d40d->lli_phy.src, - d40d->lli_pool.size, DMA_TO_DEVICE); - } - - spin_unlock_irqrestore(&d40c->lock, flags); - return &d40d->txd; - -err_fill_lli: - dev_err(&d40c->chan.dev->device, - "[%s] Failed filling in PHY LLI\n", __func__); -err: - if (d40d) - d40_desc_free(d40c, d40d); - spin_unlock_irqrestore(&d40c->lock, flags); - return NULL; + return d40_prep_sg(chan, &src_sg, &dst_sg, 1, DMA_NONE, dma_flags); } static struct dma_async_tx_descriptor * -d40_prep_sg(struct dma_chan *chan, - struct scatterlist *dst_sg, unsigned int dst_nents, - struct scatterlist *src_sg, unsigned int src_nents, - unsigned long dma_flags) +d40_prep_memcpy_sg(struct dma_chan *chan, + struct scatterlist *dst_sg, unsigned int dst_nents, + struct scatterlist *src_sg, unsigned int src_nents, + unsigned long dma_flags) { if (dst_nents != src_nents) return NULL; - return stedma40_memcpy_sg(chan, dst_sg, src_sg, dst_nents, dma_flags); -} - -static int d40_prep_slave_sg_log(struct d40_desc *d40d, - struct d40_chan *d40c, - struct scatterlist *sgl, - unsigned int sg_len, - enum dma_data_direction direction, - unsigned long dma_flags) -{ - dma_addr_t dev_addr = 0; - int total_size; - - if (d40_pool_lli_alloc(d40d, sg_len, true) < 0) { - dev_err(&d40c->chan.dev->device, - "[%s] Out of memory\n", __func__); - return -ENOMEM; - } - - d40d->lli_len = sg_len; - d40d->lli_current = 0; - - if (direction == DMA_FROM_DEVICE) - if (d40c->runtime_addr) - dev_addr = d40c->runtime_addr; - else - dev_addr = d40c->base->plat_data->dev_rx[d40c->dma_cfg.src_dev_type]; - else if (direction == DMA_TO_DEVICE) - if (d40c->runtime_addr) - dev_addr = d40c->runtime_addr; - else - dev_addr = d40c->base->plat_data->dev_tx[d40c->dma_cfg.dst_dev_type]; - - else - return -EINVAL; - - total_size = d40_log_sg_to_dev(sgl, sg_len, - &d40d->lli_log, - &d40c->log_def, - d40c->dma_cfg.src_info.data_width, - d40c->dma_cfg.dst_info.data_width, - direction, - dev_addr); - - if (total_size < 0) - return -EINVAL; - - return 0; -} - -static int d40_prep_slave_sg_phy(struct d40_desc *d40d, - struct d40_chan *d40c, - struct scatterlist *sgl, - unsigned int sgl_len, - enum dma_data_direction direction, - unsigned long dma_flags) -{ - dma_addr_t src_dev_addr; - dma_addr_t dst_dev_addr; - int res; - - if (d40_pool_lli_alloc(d40d, sgl_len, false) < 0) { - dev_err(&d40c->chan.dev->device, - "[%s] Out of memory\n", __func__); - return -ENOMEM; - } - - d40d->lli_len = sgl_len; - d40d->lli_current = 0; - - if (direction == DMA_FROM_DEVICE) { - dst_dev_addr = 0; - if (d40c->runtime_addr) - src_dev_addr = d40c->runtime_addr; - else - src_dev_addr = d40c->base->plat_data->dev_rx[d40c->dma_cfg.src_dev_type]; - } else if (direction == DMA_TO_DEVICE) { - if (d40c->runtime_addr) - dst_dev_addr = d40c->runtime_addr; - else - dst_dev_addr = d40c->base->plat_data->dev_tx[d40c->dma_cfg.dst_dev_type]; - src_dev_addr = 0; - } else - return -EINVAL; - - res = d40_phy_sg_to_lli(sgl, - sgl_len, - src_dev_addr, - d40d->lli_phy.src, - virt_to_phys(d40d->lli_phy.src), - d40c->src_def_cfg, - d40c->dma_cfg.src_info.data_width, - d40c->dma_cfg.src_info.psize); - if (res < 0) - return res; - - res = d40_phy_sg_to_lli(sgl, - sgl_len, - dst_dev_addr, - d40d->lli_phy.dst, - virt_to_phys(d40d->lli_phy.dst), - d40c->dst_def_cfg, - d40c->dma_cfg.dst_info.data_width, - d40c->dma_cfg.dst_info.psize); - if (res < 0) - return res; - - (void) dma_map_single(d40c->base->dev, d40d->lli_phy.src, - d40d->lli_pool.size, DMA_TO_DEVICE); - return 0; + return d40_prep_sg(chan, src_sg, dst_sg, src_nents, DMA_NONE, dma_flags); } static struct dma_async_tx_descriptor *d40_prep_slave_sg(struct dma_chan *chan, @@ -2050,52 +2075,40 @@ static struct dma_async_tx_descriptor *d40_prep_slave_sg(struct dma_chan *chan, enum dma_data_direction direction, unsigned long dma_flags) { - struct d40_desc *d40d; - struct d40_chan *d40c = container_of(chan, struct d40_chan, - chan); - unsigned long flags; - int err; - - if (d40c->phy_chan == NULL) { - dev_err(&d40c->chan.dev->device, - "[%s] Cannot prepare unallocated channel\n", __func__); - return ERR_PTR(-EINVAL); - } + if (direction != DMA_FROM_DEVICE && direction != DMA_TO_DEVICE) + return NULL; - spin_lock_irqsave(&d40c->lock, flags); - d40d = d40_desc_get(d40c); + return d40_prep_sg(chan, sgl, sgl, sg_len, direction, dma_flags); +} - if (d40d == NULL) - goto err; +static struct dma_async_tx_descriptor * +dma40_prep_dma_cyclic(struct dma_chan *chan, dma_addr_t dma_addr, + size_t buf_len, size_t period_len, + enum dma_data_direction direction) +{ + unsigned int periods = buf_len / period_len; + struct dma_async_tx_descriptor *txd; + struct scatterlist *sg; + int i; - if (d40c->log_num != D40_PHY_CHAN) - err = d40_prep_slave_sg_log(d40d, d40c, sgl, sg_len, - direction, dma_flags); - else - err = d40_prep_slave_sg_phy(d40d, d40c, sgl, sg_len, - direction, dma_flags); - if (err) { - dev_err(&d40c->chan.dev->device, - "[%s] Failed to prepare %s slave sg job: %d\n", - __func__, - d40c->log_num != D40_PHY_CHAN ? "log" : "phy", err); - goto err; + sg = kcalloc(periods + 1, sizeof(struct scatterlist), GFP_KERNEL); + for (i = 0; i < periods; i++) { + sg_dma_address(&sg[i]) = dma_addr; + sg_dma_len(&sg[i]) = period_len; + dma_addr += period_len; } - d40d->txd.flags = dma_flags; - - dma_async_tx_descriptor_init(&d40d->txd, chan); + sg[periods].offset = 0; + sg[periods].length = 0; + sg[periods].page_link = + ((unsigned long)sg | 0x01) & ~0x02; - d40d->txd.tx_submit = d40_tx_submit; + txd = d40_prep_sg(chan, sg, sg, periods, direction, + DMA_PREP_INTERRUPT); - spin_unlock_irqrestore(&d40c->lock, flags); - return &d40d->txd; + kfree(sg); -err: - if (d40d) - d40_desc_free(d40c, d40d); - spin_unlock_irqrestore(&d40c->lock, flags); - return NULL; + return txd; } static enum dma_status d40_tx_status(struct dma_chan *chan, @@ -2108,9 +2121,7 @@ static enum dma_status d40_tx_status(struct dma_chan *chan, int ret; if (d40c->phy_chan == NULL) { - dev_err(&d40c->chan.dev->device, - "[%s] Cannot read status of unallocated channel\n", - __func__); + chan_err(d40c, "Cannot read status of unallocated channel\n"); return -EINVAL; } @@ -2134,8 +2145,7 @@ static void d40_issue_pending(struct dma_chan *chan) unsigned long flags; if (d40c->phy_chan == NULL) { - dev_err(&d40c->chan.dev->device, - "[%s] Channel is not allocated!\n", __func__); + chan_err(d40c, "Channel is not allocated!\n"); return; } @@ -2228,7 +2238,7 @@ static void d40_set_runtime_config(struct dma_chan *chan, return; } - if (d40c->log_num != D40_PHY_CHAN) { + if (chan_is_logical(d40c)) { if (config_maxburst >= 16) psize = STEDMA40_PSIZE_LOG_16; else if (config_maxburst >= 8) @@ -2244,6 +2254,8 @@ static void d40_set_runtime_config(struct dma_chan *chan, psize = STEDMA40_PSIZE_PHY_8; else if (config_maxburst >= 4) psize = STEDMA40_PSIZE_PHY_4; + else if (config_maxburst >= 2) + psize = STEDMA40_PSIZE_PHY_2; else psize = STEDMA40_PSIZE_PHY_1; } @@ -2259,7 +2271,7 @@ static void d40_set_runtime_config(struct dma_chan *chan, cfg->dst_info.flow_ctrl = STEDMA40_NO_FLOW_CTRL; /* Fill in register values */ - if (d40c->log_num != D40_PHY_CHAN) + if (chan_is_logical(d40c)) d40_log_cfg(cfg, &d40c->log_def.lcsp1, &d40c->log_def.lcsp3); else d40_phy_cfg(cfg, &d40c->src_def_cfg, @@ -2280,25 +2292,20 @@ static void d40_set_runtime_config(struct dma_chan *chan, static int d40_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd, unsigned long arg) { - unsigned long flags; struct d40_chan *d40c = container_of(chan, struct d40_chan, chan); if (d40c->phy_chan == NULL) { - dev_err(&d40c->chan.dev->device, - "[%s] Channel is not allocated!\n", __func__); + chan_err(d40c, "Channel is not allocated!\n"); return -EINVAL; } switch (cmd) { case DMA_TERMINATE_ALL: - spin_lock_irqsave(&d40c->lock, flags); - d40_term_all(d40c); - spin_unlock_irqrestore(&d40c->lock, flags); - return 0; + return d40_terminate_all(d40c); case DMA_PAUSE: - return d40_pause(chan); + return d40_pause(d40c); case DMA_RESUME: - return d40_resume(chan); + return d40_resume(d40c); case DMA_SLAVE_CONFIG: d40_set_runtime_config(chan, (struct dma_slave_config *) arg); @@ -2343,6 +2350,35 @@ static void __init d40_chan_init(struct d40_base *base, struct dma_device *dma, } } +static void d40_ops_init(struct d40_base *base, struct dma_device *dev) +{ + if (dma_has_cap(DMA_SLAVE, dev->cap_mask)) + dev->device_prep_slave_sg = d40_prep_slave_sg; + + if (dma_has_cap(DMA_MEMCPY, dev->cap_mask)) { + dev->device_prep_dma_memcpy = d40_prep_memcpy; + + /* + * This controller can only access address at even + * 32bit boundaries, i.e. 2^2 + */ + dev->copy_align = 2; + } + + if (dma_has_cap(DMA_SG, dev->cap_mask)) + dev->device_prep_dma_sg = d40_prep_memcpy_sg; + + if (dma_has_cap(DMA_CYCLIC, dev->cap_mask)) + dev->device_prep_dma_cyclic = dma40_prep_dma_cyclic; + + dev->device_alloc_chan_resources = d40_alloc_chan_resources; + dev->device_free_chan_resources = d40_free_chan_resources; + dev->device_issue_pending = d40_issue_pending; + dev->device_tx_status = d40_tx_status; + dev->device_control = d40_control; + dev->dev = base->dev; +} + static int __init d40_dmaengine_init(struct d40_base *base, int num_reserved_chans) { @@ -2353,23 +2389,14 @@ static int __init d40_dmaengine_init(struct d40_base *base, dma_cap_zero(base->dma_slave.cap_mask); dma_cap_set(DMA_SLAVE, base->dma_slave.cap_mask); + dma_cap_set(DMA_CYCLIC, base->dma_slave.cap_mask); - base->dma_slave.device_alloc_chan_resources = d40_alloc_chan_resources; - base->dma_slave.device_free_chan_resources = d40_free_chan_resources; - base->dma_slave.device_prep_dma_memcpy = d40_prep_memcpy; - base->dma_slave.device_prep_dma_sg = d40_prep_sg; - base->dma_slave.device_prep_slave_sg = d40_prep_slave_sg; - base->dma_slave.device_tx_status = d40_tx_status; - base->dma_slave.device_issue_pending = d40_issue_pending; - base->dma_slave.device_control = d40_control; - base->dma_slave.dev = base->dev; + d40_ops_init(base, &base->dma_slave); err = dma_async_device_register(&base->dma_slave); if (err) { - dev_err(base->dev, - "[%s] Failed to register slave channels\n", - __func__); + d40_err(base->dev, "Failed to register slave channels\n"); goto failure1; } @@ -2378,29 +2405,15 @@ static int __init d40_dmaengine_init(struct d40_base *base, dma_cap_zero(base->dma_memcpy.cap_mask); dma_cap_set(DMA_MEMCPY, base->dma_memcpy.cap_mask); - dma_cap_set(DMA_SG, base->dma_slave.cap_mask); - - base->dma_memcpy.device_alloc_chan_resources = d40_alloc_chan_resources; - base->dma_memcpy.device_free_chan_resources = d40_free_chan_resources; - base->dma_memcpy.device_prep_dma_memcpy = d40_prep_memcpy; - base->dma_slave.device_prep_dma_sg = d40_prep_sg; - base->dma_memcpy.device_prep_slave_sg = d40_prep_slave_sg; - base->dma_memcpy.device_tx_status = d40_tx_status; - base->dma_memcpy.device_issue_pending = d40_issue_pending; - base->dma_memcpy.device_control = d40_control; - base->dma_memcpy.dev = base->dev; - /* - * This controller can only access address at even - * 32bit boundaries, i.e. 2^2 - */ - base->dma_memcpy.copy_align = 2; + dma_cap_set(DMA_SG, base->dma_memcpy.cap_mask); + + d40_ops_init(base, &base->dma_memcpy); err = dma_async_device_register(&base->dma_memcpy); if (err) { - dev_err(base->dev, - "[%s] Failed to regsiter memcpy only channels\n", - __func__); + d40_err(base->dev, + "Failed to regsiter memcpy only channels\n"); goto failure2; } @@ -2410,24 +2423,15 @@ static int __init d40_dmaengine_init(struct d40_base *base, dma_cap_zero(base->dma_both.cap_mask); dma_cap_set(DMA_SLAVE, base->dma_both.cap_mask); dma_cap_set(DMA_MEMCPY, base->dma_both.cap_mask); - dma_cap_set(DMA_SG, base->dma_slave.cap_mask); - - base->dma_both.device_alloc_chan_resources = d40_alloc_chan_resources; - base->dma_both.device_free_chan_resources = d40_free_chan_resources; - base->dma_both.device_prep_dma_memcpy = d40_prep_memcpy; - base->dma_slave.device_prep_dma_sg = d40_prep_sg; - base->dma_both.device_prep_slave_sg = d40_prep_slave_sg; - base->dma_both.device_tx_status = d40_tx_status; - base->dma_both.device_issue_pending = d40_issue_pending; - base->dma_both.device_control = d40_control; - base->dma_both.dev = base->dev; - base->dma_both.copy_align = 2; + dma_cap_set(DMA_SG, base->dma_both.cap_mask); + dma_cap_set(DMA_CYCLIC, base->dma_slave.cap_mask); + + d40_ops_init(base, &base->dma_both); err = dma_async_device_register(&base->dma_both); if (err) { - dev_err(base->dev, - "[%s] Failed to register logical and physical capable channels\n", - __func__); + d40_err(base->dev, + "Failed to register logical and physical capable channels\n"); goto failure3; } return 0; @@ -2503,9 +2507,10 @@ static struct d40_base * __init d40_hw_detect_init(struct platform_device *pdev) { .reg = D40_DREG_PERIPHID1, .val = 0x0000}, /* * D40_DREG_PERIPHID2 Depends on HW revision: - * MOP500/HREF ED has 0x0008, + * DB8500ed has 0x0008, * ? has 0x0018, - * HREF V1 has 0x0028 + * DB8500v1 has 0x0028 + * DB8500v2 has 0x0038 */ { .reg = D40_DREG_PERIPHID3, .val = 0x0000}, @@ -2529,8 +2534,7 @@ static struct d40_base * __init d40_hw_detect_init(struct platform_device *pdev) clk = clk_get(&pdev->dev, NULL); if (IS_ERR(clk)) { - dev_err(&pdev->dev, "[%s] No matching clock found\n", - __func__); + d40_err(&pdev->dev, "No matching clock found\n"); goto failure; } @@ -2553,9 +2557,8 @@ static struct d40_base * __init d40_hw_detect_init(struct platform_device *pdev) for (i = 0; i < ARRAY_SIZE(dma_id_regs); i++) { if (dma_id_regs[i].val != readl(virtbase + dma_id_regs[i].reg)) { - dev_err(&pdev->dev, - "[%s] Unknown hardware! Expected 0x%x at 0x%x but got 0x%x\n", - __func__, + d40_err(&pdev->dev, + "Unknown hardware! Expected 0x%x at 0x%x but got 0x%x\n", dma_id_regs[i].val, dma_id_regs[i].reg, readl(virtbase + dma_id_regs[i].reg)); @@ -2568,9 +2571,8 @@ static struct d40_base * __init d40_hw_detect_init(struct platform_device *pdev) if ((val & D40_DREG_PERIPHID2_DESIGNER_MASK) != D40_HW_DESIGNER) { - dev_err(&pdev->dev, - "[%s] Unknown designer! Got %x wanted %x\n", - __func__, val & D40_DREG_PERIPHID2_DESIGNER_MASK, + d40_err(&pdev->dev, "Unknown designer! Got %x wanted %x\n", + val & D40_DREG_PERIPHID2_DESIGNER_MASK, D40_HW_DESIGNER); goto failure; } @@ -2600,7 +2602,7 @@ static struct d40_base * __init d40_hw_detect_init(struct platform_device *pdev) sizeof(struct d40_chan), GFP_KERNEL); if (base == NULL) { - dev_err(&pdev->dev, "[%s] Out of memory\n", __func__); + d40_err(&pdev->dev, "Out of memory\n"); goto failure; } @@ -2747,6 +2749,7 @@ static void __init d40_hw_init(struct d40_base *base) static int __init d40_lcla_allocate(struct d40_base *base) { + struct d40_lcla_pool *pool = &base->lcla_pool; unsigned long *page_list; int i, j; int ret = 0; @@ -2772,9 +2775,8 @@ static int __init d40_lcla_allocate(struct d40_base *base) base->lcla_pool.pages); if (!page_list[i]) { - dev_err(base->dev, - "[%s] Failed to allocate %d pages.\n", - __func__, base->lcla_pool.pages); + d40_err(base->dev, "Failed to allocate %d pages.\n", + base->lcla_pool.pages); for (j = 0; j < i; j++) free_pages(page_list[j], base->lcla_pool.pages); @@ -2812,6 +2814,15 @@ static int __init d40_lcla_allocate(struct d40_base *base) LCLA_ALIGNMENT); } + pool->dma_addr = dma_map_single(base->dev, pool->base, + SZ_1K * base->num_phy_chans, + DMA_TO_DEVICE); + if (dma_mapping_error(base->dev, pool->dma_addr)) { + pool->dma_addr = 0; + ret = -ENOMEM; + goto failure; + } + writel(virt_to_phys(base->lcla_pool.base), base->virtbase + D40_DREG_LCLA); failure: @@ -2844,9 +2855,7 @@ static int __init d40_probe(struct platform_device *pdev) res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "lcpa"); if (!res) { ret = -ENOENT; - dev_err(&pdev->dev, - "[%s] No \"lcpa\" memory resource\n", - __func__); + d40_err(&pdev->dev, "No \"lcpa\" memory resource\n"); goto failure; } base->lcpa_size = resource_size(res); @@ -2855,9 +2864,9 @@ static int __init d40_probe(struct platform_device *pdev) if (request_mem_region(res->start, resource_size(res), D40_NAME " I/O lcpa") == NULL) { ret = -EBUSY; - dev_err(&pdev->dev, - "[%s] Failed to request LCPA region 0x%x-0x%x\n", - __func__, res->start, res->end); + d40_err(&pdev->dev, + "Failed to request LCPA region 0x%x-0x%x\n", + res->start, res->end); goto failure; } @@ -2873,16 +2882,13 @@ static int __init d40_probe(struct platform_device *pdev) base->lcpa_base = ioremap(res->start, resource_size(res)); if (!base->lcpa_base) { ret = -ENOMEM; - dev_err(&pdev->dev, - "[%s] Failed to ioremap LCPA region\n", - __func__); + d40_err(&pdev->dev, "Failed to ioremap LCPA region\n"); goto failure; } ret = d40_lcla_allocate(base); if (ret) { - dev_err(&pdev->dev, "[%s] Failed to allocate LCLA area\n", - __func__); + d40_err(&pdev->dev, "Failed to allocate LCLA area\n"); goto failure; } @@ -2891,9 +2897,8 @@ static int __init d40_probe(struct platform_device *pdev) base->irq = platform_get_irq(pdev, 0); ret = request_irq(base->irq, d40_handle_interrupt, 0, D40_NAME, base); - if (ret) { - dev_err(&pdev->dev, "[%s] No IRQ defined\n", __func__); + d40_err(&pdev->dev, "No IRQ defined\n"); goto failure; } @@ -2912,6 +2917,12 @@ failure: kmem_cache_destroy(base->desc_slab); if (base->virtbase) iounmap(base->virtbase); + + if (base->lcla_pool.dma_addr) + dma_unmap_single(base->dev, base->lcla_pool.dma_addr, + SZ_1K * base->num_phy_chans, + DMA_TO_DEVICE); + if (!base->lcla_pool.base_unaligned && base->lcla_pool.base) free_pages((unsigned long)base->lcla_pool.base, base->lcla_pool.pages); @@ -2936,7 +2947,7 @@ failure: kfree(base); } - dev_err(&pdev->dev, "[%s] probe failed\n", __func__); + d40_err(&pdev->dev, "probe failed\n"); return ret; } @@ -2947,7 +2958,7 @@ static struct platform_driver d40_driver = { }, }; -int __init stedma40_init(void) +static int __init stedma40_init(void) { return platform_driver_probe(&d40_driver, d40_probe); } diff --git a/drivers/dma/ste_dma40_ll.c b/drivers/dma/ste_dma40_ll.c index 8557cb88b255..cad9e1daedff 100644 --- a/drivers/dma/ste_dma40_ll.c +++ b/drivers/dma/ste_dma40_ll.c @@ -1,6 +1,6 @@ /* * Copyright (C) ST-Ericsson SA 2007-2010 - * Author: Per Friden <per.friden@stericsson.com> for ST-Ericsson + * Author: Per Forlin <per.forlin@stericsson.com> for ST-Ericsson * Author: Jonas Aaberg <jonas.aberg@stericsson.com> for ST-Ericsson * License terms: GNU General Public License (GPL) version 2 */ @@ -122,16 +122,18 @@ void d40_phy_cfg(struct stedma40_chan_cfg *cfg, *dst_cfg = dst; } -int d40_phy_fill_lli(struct d40_phy_lli *lli, - dma_addr_t data, - u32 data_size, - int psize, - dma_addr_t next_lli, - u32 reg_cfg, - bool term_int, - u32 data_width, - bool is_device) +static int d40_phy_fill_lli(struct d40_phy_lli *lli, + dma_addr_t data, + u32 data_size, + dma_addr_t next_lli, + u32 reg_cfg, + struct stedma40_half_channel_info *info, + unsigned int flags) { + bool addr_inc = flags & LLI_ADDR_INC; + bool term_int = flags & LLI_TERM_INT; + unsigned int data_width = info->data_width; + int psize = info->psize; int num_elems; if (psize == STEDMA40_PSIZE_PHY_1) @@ -139,13 +141,6 @@ int d40_phy_fill_lli(struct d40_phy_lli *lli, else num_elems = 2 << psize; - /* - * Size is 16bit. data_width is 8, 16, 32 or 64 bit - * Block large than 64 KiB must be split. - */ - if (data_size > (0xffff << data_width)) - return -EINVAL; - /* Must be aligned */ if (!IS_ALIGNED(data, 0x1 << data_width)) return -EINVAL; @@ -161,7 +156,7 @@ int d40_phy_fill_lli(struct d40_phy_lli *lli, * Distance to next element sized entry. * Usually the size of the element unless you want gaps. */ - if (!is_device) + if (addr_inc) lli->reg_elt |= (0x1 << data_width) << D40_SREG_ELEM_PHY_EIDX_POS; @@ -187,97 +182,137 @@ int d40_phy_fill_lli(struct d40_phy_lli *lli, return 0; } +static int d40_seg_size(int size, int data_width1, int data_width2) +{ + u32 max_w = max(data_width1, data_width2); + u32 min_w = min(data_width1, data_width2); + u32 seg_max = ALIGN(STEDMA40_MAX_SEG_SIZE << min_w, 1 << max_w); + + if (seg_max > STEDMA40_MAX_SEG_SIZE) + seg_max -= (1 << max_w); + + if (size <= seg_max) + return size; + + if (size <= 2 * seg_max) + return ALIGN(size / 2, 1 << max_w); + + return seg_max; +} + +static struct d40_phy_lli * +d40_phy_buf_to_lli(struct d40_phy_lli *lli, dma_addr_t addr, u32 size, + dma_addr_t lli_phys, dma_addr_t first_phys, u32 reg_cfg, + struct stedma40_half_channel_info *info, + struct stedma40_half_channel_info *otherinfo, + unsigned long flags) +{ + bool lastlink = flags & LLI_LAST_LINK; + bool addr_inc = flags & LLI_ADDR_INC; + bool term_int = flags & LLI_TERM_INT; + bool cyclic = flags & LLI_CYCLIC; + int err; + dma_addr_t next = lli_phys; + int size_rest = size; + int size_seg = 0; + + /* + * This piece may be split up based on d40_seg_size(); we only want the + * term int on the last part. + */ + if (term_int) + flags &= ~LLI_TERM_INT; + + do { + size_seg = d40_seg_size(size_rest, info->data_width, + otherinfo->data_width); + size_rest -= size_seg; + + if (size_rest == 0 && term_int) + flags |= LLI_TERM_INT; + + if (size_rest == 0 && lastlink) + next = cyclic ? first_phys : 0; + else + next = ALIGN(next + sizeof(struct d40_phy_lli), + D40_LLI_ALIGN); + + err = d40_phy_fill_lli(lli, addr, size_seg, next, + reg_cfg, info, flags); + + if (err) + goto err; + + lli++; + if (addr_inc) + addr += size_seg; + } while (size_rest); + + return lli; + + err: + return NULL; +} + int d40_phy_sg_to_lli(struct scatterlist *sg, int sg_len, dma_addr_t target, - struct d40_phy_lli *lli, + struct d40_phy_lli *lli_sg, dma_addr_t lli_phys, u32 reg_cfg, - u32 data_width, - int psize) + struct stedma40_half_channel_info *info, + struct stedma40_half_channel_info *otherinfo, + unsigned long flags) { int total_size = 0; int i; struct scatterlist *current_sg = sg; - dma_addr_t next_lli_phys; - dma_addr_t dst; - int err = 0; + struct d40_phy_lli *lli = lli_sg; + dma_addr_t l_phys = lli_phys; + + if (!target) + flags |= LLI_ADDR_INC; for_each_sg(sg, current_sg, sg_len, i) { + dma_addr_t sg_addr = sg_dma_address(current_sg); + unsigned int len = sg_dma_len(current_sg); + dma_addr_t dst = target ?: sg_addr; total_size += sg_dma_len(current_sg); - /* If this scatter list entry is the last one, no next link */ - if (sg_len - 1 == i) - next_lli_phys = 0; - else - next_lli_phys = ALIGN(lli_phys + (i + 1) * - sizeof(struct d40_phy_lli), - D40_LLI_ALIGN); + if (i == sg_len - 1) + flags |= LLI_TERM_INT | LLI_LAST_LINK; - if (target) - dst = target; - else - dst = sg_phys(current_sg); - - err = d40_phy_fill_lli(&lli[i], - dst, - sg_dma_len(current_sg), - psize, - next_lli_phys, - reg_cfg, - !next_lli_phys, - data_width, - target == dst); - if (err) - goto err; + l_phys = ALIGN(lli_phys + (lli - lli_sg) * + sizeof(struct d40_phy_lli), D40_LLI_ALIGN); + + lli = d40_phy_buf_to_lli(lli, dst, len, l_phys, lli_phys, + reg_cfg, info, otherinfo, flags); + + if (lli == NULL) + return -EINVAL; } return total_size; -err: - return err; } -void d40_phy_lli_write(void __iomem *virtbase, - u32 phy_chan_num, - struct d40_phy_lli *lli_dst, - struct d40_phy_lli *lli_src) -{ - - writel(lli_src->reg_cfg, virtbase + D40_DREG_PCBASE + - phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SSCFG); - writel(lli_src->reg_elt, virtbase + D40_DREG_PCBASE + - phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SSELT); - writel(lli_src->reg_ptr, virtbase + D40_DREG_PCBASE + - phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SSPTR); - writel(lli_src->reg_lnk, virtbase + D40_DREG_PCBASE + - phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SSLNK); - - writel(lli_dst->reg_cfg, virtbase + D40_DREG_PCBASE + - phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SDCFG); - writel(lli_dst->reg_elt, virtbase + D40_DREG_PCBASE + - phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SDELT); - writel(lli_dst->reg_ptr, virtbase + D40_DREG_PCBASE + - phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SDPTR); - writel(lli_dst->reg_lnk, virtbase + D40_DREG_PCBASE + - phy_chan_num * D40_DREG_PCDELTA + D40_CHAN_REG_SDLNK); - -} - /* DMA logical lli operations */ static void d40_log_lli_link(struct d40_log_lli *lli_dst, struct d40_log_lli *lli_src, - int next) + int next, unsigned int flags) { + bool interrupt = flags & LLI_TERM_INT; u32 slos = 0; u32 dlos = 0; if (next != -EINVAL) { slos = next * 2; dlos = next * 2 + 1; - } else { + } + + if (interrupt) { lli_dst->lcsp13 |= D40_MEM_LCSP1_SCFG_TIM_MASK; lli_dst->lcsp13 |= D40_MEM_LCSP3_DTCP_MASK; } @@ -292,9 +327,9 @@ static void d40_log_lli_link(struct d40_log_lli *lli_dst, void d40_log_lli_lcpa_write(struct d40_log_lli_full *lcpa, struct d40_log_lli *lli_dst, struct d40_log_lli *lli_src, - int next) + int next, unsigned int flags) { - d40_log_lli_link(lli_dst, lli_src, next); + d40_log_lli_link(lli_dst, lli_src, next, flags); writel(lli_src->lcsp02, &lcpa[0].lcsp0); writel(lli_src->lcsp13, &lcpa[0].lcsp1); @@ -305,9 +340,9 @@ void d40_log_lli_lcpa_write(struct d40_log_lli_full *lcpa, void d40_log_lli_lcla_write(struct d40_log_lli *lcla, struct d40_log_lli *lli_dst, struct d40_log_lli *lli_src, - int next) + int next, unsigned int flags) { - d40_log_lli_link(lli_dst, lli_src, next); + d40_log_lli_link(lli_dst, lli_src, next, flags); writel(lli_src->lcsp02, &lcla[0].lcsp02); writel(lli_src->lcsp13, &lcla[0].lcsp13); @@ -315,17 +350,22 @@ void d40_log_lli_lcla_write(struct d40_log_lli *lcla, writel(lli_dst->lcsp13, &lcla[1].lcsp13); } -void d40_log_fill_lli(struct d40_log_lli *lli, - dma_addr_t data, u32 data_size, - u32 reg_cfg, - u32 data_width, - bool addr_inc) +static void d40_log_fill_lli(struct d40_log_lli *lli, + dma_addr_t data, u32 data_size, + u32 reg_cfg, + u32 data_width, + unsigned int flags) { + bool addr_inc = flags & LLI_ADDR_INC; + lli->lcsp13 = reg_cfg; /* The number of elements to transfer */ lli->lcsp02 = ((data_size >> data_width) << D40_MEM_LCSP0_ECNT_POS) & D40_MEM_LCSP0_ECNT_MASK; + + BUG_ON((data_size >> data_width) > STEDMA40_MAX_SEG_SIZE); + /* 16 LSBs address of the current element */ lli->lcsp02 |= data & D40_MEM_LCSP0_SPTR_MASK; /* 16 MSBs address of the current element */ @@ -336,67 +376,65 @@ void d40_log_fill_lli(struct d40_log_lli *lli, } -int d40_log_sg_to_dev(struct scatterlist *sg, - int sg_len, - struct d40_log_lli_bidir *lli, - struct d40_def_lcsp *lcsp, - u32 src_data_width, - u32 dst_data_width, - enum dma_data_direction direction, - dma_addr_t dev_addr) +static struct d40_log_lli *d40_log_buf_to_lli(struct d40_log_lli *lli_sg, + dma_addr_t addr, + int size, + u32 lcsp13, /* src or dst*/ + u32 data_width1, + u32 data_width2, + unsigned int flags) { - int total_size = 0; - struct scatterlist *current_sg = sg; - int i; - - for_each_sg(sg, current_sg, sg_len, i) { - total_size += sg_dma_len(current_sg); - - if (direction == DMA_TO_DEVICE) { - d40_log_fill_lli(&lli->src[i], - sg_phys(current_sg), - sg_dma_len(current_sg), - lcsp->lcsp1, src_data_width, - true); - d40_log_fill_lli(&lli->dst[i], - dev_addr, - sg_dma_len(current_sg), - lcsp->lcsp3, dst_data_width, - false); - } else { - d40_log_fill_lli(&lli->dst[i], - sg_phys(current_sg), - sg_dma_len(current_sg), - lcsp->lcsp3, dst_data_width, - true); - d40_log_fill_lli(&lli->src[i], - dev_addr, - sg_dma_len(current_sg), - lcsp->lcsp1, src_data_width, - false); - } - } - return total_size; + bool addr_inc = flags & LLI_ADDR_INC; + struct d40_log_lli *lli = lli_sg; + int size_rest = size; + int size_seg = 0; + + do { + size_seg = d40_seg_size(size_rest, data_width1, data_width2); + size_rest -= size_seg; + + d40_log_fill_lli(lli, + addr, + size_seg, + lcsp13, data_width1, + flags); + if (addr_inc) + addr += size_seg; + lli++; + } while (size_rest); + + return lli; } int d40_log_sg_to_lli(struct scatterlist *sg, int sg_len, + dma_addr_t dev_addr, struct d40_log_lli *lli_sg, u32 lcsp13, /* src or dst*/ - u32 data_width) + u32 data_width1, u32 data_width2) { int total_size = 0; struct scatterlist *current_sg = sg; int i; + struct d40_log_lli *lli = lli_sg; + unsigned long flags = 0; + + if (!dev_addr) + flags |= LLI_ADDR_INC; for_each_sg(sg, current_sg, sg_len, i) { + dma_addr_t sg_addr = sg_dma_address(current_sg); + unsigned int len = sg_dma_len(current_sg); + dma_addr_t addr = dev_addr ?: sg_addr; + total_size += sg_dma_len(current_sg); - d40_log_fill_lli(&lli_sg[i], - sg_phys(current_sg), - sg_dma_len(current_sg), - lcsp13, data_width, - true); + lli = d40_log_buf_to_lli(lli, addr, len, + lcsp13, + data_width1, + data_width2, + flags); } + return total_size; } diff --git a/drivers/dma/ste_dma40_ll.h b/drivers/dma/ste_dma40_ll.h index 9e419b907544..195ee65ee7f3 100644 --- a/drivers/dma/ste_dma40_ll.h +++ b/drivers/dma/ste_dma40_ll.h @@ -163,6 +163,22 @@ #define D40_DREG_LCEIS1 0x0B4 #define D40_DREG_LCEIS2 0x0B8 #define D40_DREG_LCEIS3 0x0BC +#define D40_DREG_PSEG1 0x110 +#define D40_DREG_PSEG2 0x114 +#define D40_DREG_PSEG3 0x118 +#define D40_DREG_PSEG4 0x11C +#define D40_DREG_PCEG1 0x120 +#define D40_DREG_PCEG2 0x124 +#define D40_DREG_PCEG3 0x128 +#define D40_DREG_PCEG4 0x12C +#define D40_DREG_RSEG1 0x130 +#define D40_DREG_RSEG2 0x134 +#define D40_DREG_RSEG3 0x138 +#define D40_DREG_RSEG4 0x13C +#define D40_DREG_RCEG1 0x140 +#define D40_DREG_RCEG2 0x144 +#define D40_DREG_RCEG3 0x148 +#define D40_DREG_RCEG4 0x14C #define D40_DREG_STFU 0xFC8 #define D40_DREG_ICFG 0xFCC #define D40_DREG_PERIPHID0 0xFE0 @@ -277,6 +293,13 @@ struct d40_def_lcsp { /* Physical channels */ +enum d40_lli_flags { + LLI_ADDR_INC = 1 << 0, + LLI_TERM_INT = 1 << 1, + LLI_CYCLIC = 1 << 2, + LLI_LAST_LINK = 1 << 3, +}; + void d40_phy_cfg(struct stedma40_chan_cfg *cfg, u32 *src_cfg, u32 *dst_cfg, @@ -292,56 +315,27 @@ int d40_phy_sg_to_lli(struct scatterlist *sg, struct d40_phy_lli *lli, dma_addr_t lli_phys, u32 reg_cfg, - u32 data_width, - int psize); - -int d40_phy_fill_lli(struct d40_phy_lli *lli, - dma_addr_t data, - u32 data_size, - int psize, - dma_addr_t next_lli, - u32 reg_cfg, - bool term_int, - u32 data_width, - bool is_device); - -void d40_phy_lli_write(void __iomem *virtbase, - u32 phy_chan_num, - struct d40_phy_lli *lli_dst, - struct d40_phy_lli *lli_src); + struct stedma40_half_channel_info *info, + struct stedma40_half_channel_info *otherinfo, + unsigned long flags); /* Logical channels */ -void d40_log_fill_lli(struct d40_log_lli *lli, - dma_addr_t data, - u32 data_size, - u32 reg_cfg, - u32 data_width, - bool addr_inc); - -int d40_log_sg_to_dev(struct scatterlist *sg, - int sg_len, - struct d40_log_lli_bidir *lli, - struct d40_def_lcsp *lcsp, - u32 src_data_width, - u32 dst_data_width, - enum dma_data_direction direction, - dma_addr_t dev_addr); - int d40_log_sg_to_lli(struct scatterlist *sg, int sg_len, + dma_addr_t dev_addr, struct d40_log_lli *lli_sg, u32 lcsp13, /* src or dst*/ - u32 data_width); + u32 data_width1, u32 data_width2); void d40_log_lli_lcpa_write(struct d40_log_lli_full *lcpa, struct d40_log_lli *lli_dst, struct d40_log_lli *lli_src, - int next); + int next, unsigned int flags); void d40_log_lli_lcla_write(struct d40_log_lli *lcla, struct d40_log_lli *lli_dst, struct d40_log_lli *lli_src, - int next); + int next, unsigned int flags); #endif /* STE_DMA40_LLI_H */ |