diff options
author | Pavel Pisa <ppisa@pikron.com> | 2006-04-02 16:58:37 +0100 |
---|---|---|
committer | Russell King <rmk+kernel@arm.linux.org.uk> | 2006-04-02 16:58:37 +0100 |
commit | 999331af456bf6fc1520ea7b68b6a3dbb4af8ff6 (patch) | |
tree | d7610740f58f942fcb46e07bb790dfd785e21508 /arch/arm/mach-imx/dma.c | |
parent | 7ba01f9728a9f1cd1a3e1e2d5206f76061182675 (diff) |
[ARM] 3444/1: i.MX: Scatter-gather DMA emulation for i.MX/MX1
Patch from Pavel Pisa
This patch contains simplified set of changes to add scatter-gather
emulation capability into MX1 DMA support. The result should
be still usable for next combination of DMA transfers
Statter-Gather/linear/2D/FIFO to linear/2D/FIFO and
linear/2D/FIFO to Statter-Gather/2D/FIFO
The patch corrects channel priority allocation to be compatible
with MX1 hardware implementation.
Previous code has not been adapted from its PXA original.
Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
Acked-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Diffstat (limited to 'arch/arm/mach-imx/dma.c')
-rw-r--r-- | arch/arm/mach-imx/dma.c | 511 |
1 files changed, 448 insertions, 63 deletions
diff --git a/arch/arm/mach-imx/dma.c b/arch/arm/mach-imx/dma.c index 71a59e196166..4ca51dcf13ac 100644 --- a/arch/arm/mach-imx/dma.c +++ b/arch/arm/mach-imx/dma.c @@ -7,11 +7,18 @@ * it under the terms of the GNU General Public License version 2 as * published by the Free Software Foundation. * - * 03/03/2004 Sascha Hauer <sascha@saschahauer.de> + * 2004-03-03 Sascha Hauer <sascha@saschahauer.de> * initial version heavily inspired by * linux/arch/arm/mach-pxa/dma.c + * + * 2005-04-17 Pavel Pisa <pisa@cmp.felk.cvut.cz> + * Changed to support scatter gather DMA + * by taking Russell's code from RiscPC + * */ +#undef DEBUG + #include <linux/module.h> #include <linux/init.h> #include <linux/kernel.h> @@ -22,69 +29,368 @@ #include <asm/irq.h> #include <asm/hardware.h> #include <asm/dma.h> +#include <asm/arch/imx-dma.h> + +struct imx_dma_channel imx_dma_channels[IMX_DMA_CHANNELS]; + +/* + * imx_dma_sg_next - prepare next chunk for scatter-gather DMA emulation + * @dma_ch: i.MX DMA channel number + * @lastcount: number of bytes transferred during last transfer + * + * Functions prepares DMA controller for next sg data chunk transfer. + * The @lastcount argument informs function about number of bytes transferred + * during last block. Zero value can be used for @lastcount to setup DMA + * for the first chunk. + */ +static inline int imx_dma_sg_next(imx_dmach_t dma_ch, unsigned int lastcount) +{ + struct imx_dma_channel *imxdma = &imx_dma_channels[dma_ch]; + unsigned int nextcount; + unsigned int nextaddr; + + if (!imxdma->name) { + printk(KERN_CRIT "%s: called for not allocated channel %d\n", + __FUNCTION__, dma_ch); + return 0; + } + + imxdma->resbytes -= lastcount; + + if (!imxdma->sg) { + pr_debug("imxdma%d: no sg data\n", dma_ch); + return 0; + } + + imxdma->sgbc += lastcount; + if ((imxdma->sgbc >= imxdma->sg->length) || !imxdma->resbytes) { + if ((imxdma->sgcount <= 1) || !imxdma->resbytes) { + pr_debug("imxdma%d: sg transfer limit reached\n", + dma_ch); + imxdma->sgcount=0; + imxdma->sg = NULL; + return 0; + } else { + imxdma->sgcount--; + imxdma->sg++; + imxdma->sgbc = 0; + } + } + nextcount = imxdma->sg->length - imxdma->sgbc; + nextaddr = imxdma->sg->dma_address + imxdma->sgbc; -static struct dma_channel { - char *name; - void (*irq_handler) (int, void *, struct pt_regs *); - void (*err_handler) (int, void *, struct pt_regs *); - void *data; -} dma_channels[11]; + if(imxdma->resbytes < nextcount) + nextcount = imxdma->resbytes; -/* set err_handler to NULL to have the standard info-only error handler */ + if ((imxdma->dma_mode & DMA_MODE_MASK) == DMA_MODE_READ) + DAR(dma_ch) = nextaddr; + else + SAR(dma_ch) = nextaddr; + + CNTR(dma_ch) = nextcount; + pr_debug("imxdma%d: next sg chunk dst 0x%08x, src 0x%08x, size 0x%08x\n", + dma_ch, DAR(dma_ch), SAR(dma_ch), CNTR(dma_ch)); + + return nextcount; +} + +/* + * imx_dma_setup_sg_base - scatter-gather DMA emulation + * @dma_ch: i.MX DMA channel number + * @sg: pointer to the scatter-gather list/vector + * @sgcount: scatter-gather list hungs count + * + * Functions sets up i.MX DMA state for emulated scatter-gather transfer + * and sets up channel registers to be ready for the first chunk + */ +static int +imx_dma_setup_sg_base(imx_dmach_t dma_ch, + struct scatterlist *sg, unsigned int sgcount) +{ + struct imx_dma_channel *imxdma = &imx_dma_channels[dma_ch]; + + imxdma->sg = sg; + imxdma->sgcount = sgcount; + imxdma->sgbc = 0; + return imx_dma_sg_next(dma_ch, 0); +} + +/** + * imx_dma_setup_single - setup i.MX DMA channel for linear memory to/from device transfer + * @dma_ch: i.MX DMA channel number + * @dma_address: the DMA/physical memory address of the linear data block + * to transfer + * @dma_length: length of the data block in bytes + * @dev_addr: physical device port address + * @dmamode: DMA transfer mode, %DMA_MODE_READ from the device to the memory + * or %DMA_MODE_WRITE from memory to the device + * + * The function setups DMA channel source and destination addresses for transfer + * specified by provided parameters. The scatter-gather emulation is disabled, + * because linear data block + * form the physical address range is transfered. + * Return value: if incorrect parameters are provided -%EINVAL. + * Zero indicates success. + */ int -imx_request_dma(char *name, imx_dma_prio prio, - void (*irq_handler) (int, void *, struct pt_regs *), - void (*err_handler) (int, void *, struct pt_regs *), void *data) +imx_dma_setup_single(imx_dmach_t dma_ch, dma_addr_t dma_address, + unsigned int dma_length, unsigned int dev_addr, + dmamode_t dmamode) { - unsigned long flags; - int i, found = 0; + struct imx_dma_channel *imxdma = &imx_dma_channels[dma_ch]; - /* basic sanity checks */ - if (!name || !irq_handler) + imxdma->sg = NULL; + imxdma->sgcount = 0; + imxdma->dma_mode = dmamode; + imxdma->resbytes = dma_length; + + if (!dma_address) { + printk(KERN_ERR "imxdma%d: imx_dma_setup_single null address\n", + dma_ch); return -EINVAL; + } - local_irq_save(flags); + if (!dma_length) { + printk(KERN_ERR "imxdma%d: imx_dma_setup_single zero length\n", + dma_ch); + return -EINVAL; + } - /* try grabbing a DMA channel with the requested priority */ - for (i = prio; i < prio + (prio == DMA_PRIO_LOW) ? 8 : 4; i++) { - if (!dma_channels[i].name) { - found = 1; - break; - } + if ((dmamode & DMA_MODE_MASK) == DMA_MODE_READ) { + pr_debug("imxdma%d: mx_dma_setup_single2dev dma_addressg=0x%08x dma_length=%d dev_addr=0x%08x for read\n", + dma_ch, (unsigned int)dma_address, dma_length, + dev_addr); + SAR(dma_ch) = dev_addr; + DAR(dma_ch) = (unsigned int)dma_address; + } else if ((dmamode & DMA_MODE_MASK) == DMA_MODE_WRITE) { + pr_debug("imxdma%d: mx_dma_setup_single2dev dma_addressg=0x%08x dma_length=%d dev_addr=0x%08x for write\n", + dma_ch, (unsigned int)dma_address, dma_length, + dev_addr); + SAR(dma_ch) = (unsigned int)dma_address; + DAR(dma_ch) = dev_addr; + } else { + printk(KERN_ERR "imxdma%d: imx_dma_setup_single bad dmamode\n", + dma_ch); + return -EINVAL; } - if (!found) { - /* requested prio group is full, try hier priorities */ - for (i = prio - 1; i >= 0; i--) { - if (!dma_channels[i].name) { - found = 1; - break; - } - } + CNTR(dma_ch) = dma_length; + + return 0; +} + +/** + * imx_dma_setup_sg - setup i.MX DMA channel SG list to/from device transfer + * @dma_ch: i.MX DMA channel number + * @sg: pointer to the scatter-gather list/vector + * @sgcount: scatter-gather list hungs count + * @dma_length: total length of the transfer request in bytes + * @dev_addr: physical device port address + * @dmamode: DMA transfer mode, %DMA_MODE_READ from the device to the memory + * or %DMA_MODE_WRITE from memory to the device + * + * The function setups DMA channel state and registers to be ready for transfer + * specified by provided parameters. The scatter-gather emulation is set up + * according to the parameters. + * + * The full preparation of the transfer requires setup of more register + * by the caller before imx_dma_enable() can be called. + * + * %BLR(dma_ch) holds transfer burst length in bytes, 0 means 64 bytes + * + * %RSSR(dma_ch) has to be set to the DMA request line source %DMA_REQ_xxx + * + * %CCR(dma_ch) has to specify transfer parameters, the next settings is typical + * for linear or simple scatter-gather transfers if %DMA_MODE_READ is specified + * + * %CCR_DMOD_LINEAR | %CCR_DSIZ_32 | %CCR_SMOD_FIFO | %CCR_SSIZ_x + * + * The typical setup for %DMA_MODE_WRITE is specified by next options combination + * + * %CCR_SMOD_LINEAR | %CCR_SSIZ_32 | %CCR_DMOD_FIFO | %CCR_DSIZ_x + * + * Be carefull there and do not mistakenly mix source and target device + * port sizes constants, they are really different: + * %CCR_SSIZ_8, %CCR_SSIZ_16, %CCR_SSIZ_32, + * %CCR_DSIZ_8, %CCR_DSIZ_16, %CCR_DSIZ_32 + * + * Return value: if incorrect parameters are provided -%EINVAL. + * Zero indicates success. + */ +int +imx_dma_setup_sg(imx_dmach_t dma_ch, + struct scatterlist *sg, unsigned int sgcount, unsigned int dma_length, + unsigned int dev_addr, dmamode_t dmamode) +{ + int res; + struct imx_dma_channel *imxdma = &imx_dma_channels[dma_ch]; + + imxdma->sg = NULL; + imxdma->sgcount = 0; + imxdma->dma_mode = dmamode; + imxdma->resbytes = dma_length; + + if (!sg || !sgcount) { + printk(KERN_ERR "imxdma%d: imx_dma_setup_sg epty sg list\n", + dma_ch); + return -EINVAL; + } + + if (!sg->length) { + printk(KERN_ERR "imxdma%d: imx_dma_setup_sg zero length\n", + dma_ch); + return -EINVAL; } - if (found) { - DIMR &= ~(1 << i); - dma_channels[i].name = name; - dma_channels[i].irq_handler = irq_handler; - dma_channels[i].err_handler = err_handler; - dma_channels[i].data = data; + if ((dmamode & DMA_MODE_MASK) == DMA_MODE_READ) { + pr_debug("imxdma%d: mx_dma_setup_sg2dev sg=%p sgcount=%d total length=%d dev_addr=0x%08x for read\n", + dma_ch, sg, sgcount, dma_length, dev_addr); + SAR(dma_ch) = dev_addr; + } else if ((dmamode & DMA_MODE_MASK) == DMA_MODE_WRITE) { + pr_debug("imxdma%d: mx_dma_setup_sg2dev sg=%p sgcount=%d total length=%d dev_addr=0x%08x for write\n", + dma_ch, sg, sgcount, dma_length, dev_addr); + DAR(dma_ch) = dev_addr; } else { - printk(KERN_WARNING "No more available DMA channels for %s\n", - name); - i = -ENODEV; + printk(KERN_ERR "imxdma%d: imx_dma_setup_sg bad dmamode\n", + dma_ch); + return -EINVAL; + } + + res = imx_dma_setup_sg_base(dma_ch, sg, sgcount); + if (res <= 0) { + printk(KERN_ERR "imxdma%d: no sg chunk ready\n", dma_ch); + return -EINVAL; + } + + return 0; +} + +/** + * imx_dma_setup_handlers - setup i.MX DMA channel end and error notification handlers + * @dma_ch: i.MX DMA channel number + * @irq_handler: the pointer to the function called if the transfer + * ends successfully + * @err_handler: the pointer to the function called if the premature + * end caused by error occurs + * @data: user specified value to be passed to the handlers + */ +int +imx_dma_setup_handlers(imx_dmach_t dma_ch, + void (*irq_handler) (int, void *, struct pt_regs *), + void (*err_handler) (int, void *, struct pt_regs *), + void *data) +{ + struct imx_dma_channel *imxdma = &imx_dma_channels[dma_ch]; + unsigned long flags; + + if (!imxdma->name) { + printk(KERN_CRIT "%s: called for not allocated channel %d\n", + __FUNCTION__, dma_ch); + return -ENODEV; + } + + local_irq_save(flags); + DISR = (1 << dma_ch); + imxdma->irq_handler = irq_handler; + imxdma->err_handler = err_handler; + imxdma->data = data; + local_irq_restore(flags); + return 0; +} + +/** + * imx_dma_enable - function to start i.MX DMA channel operation + * @dma_ch: i.MX DMA channel number + * + * The channel has to be allocated by driver through imx_dma_request() + * or imx_dma_request_by_prio() function. + * The transfer parameters has to be set to the channel registers through + * call of the imx_dma_setup_single() or imx_dma_setup_sg() function + * and registers %BLR(dma_ch), %RSSR(dma_ch) and %CCR(dma_ch) has to + * be set prior this function call by the channel user. + */ +void imx_dma_enable(imx_dmach_t dma_ch) +{ + struct imx_dma_channel *imxdma = &imx_dma_channels[dma_ch]; + unsigned long flags; + + pr_debug("imxdma%d: imx_dma_enable\n", dma_ch); + + if (!imxdma->name) { + printk(KERN_CRIT "%s: called for not allocated channel %d\n", + __FUNCTION__, dma_ch); + return; + } + + local_irq_save(flags); + DISR = (1 << dma_ch); + DIMR &= ~(1 << dma_ch); + CCR(dma_ch) |= CCR_CEN; + local_irq_restore(flags); +} + +/** + * imx_dma_disable - stop, finish i.MX DMA channel operatin + * @dma_ch: i.MX DMA channel number + */ +void imx_dma_disable(imx_dmach_t dma_ch) +{ + unsigned long flags; + + pr_debug("imxdma%d: imx_dma_disable\n", dma_ch); + + local_irq_save(flags); + DIMR |= (1 << dma_ch); + CCR(dma_ch) &= ~CCR_CEN; + DISR = (1 << dma_ch); + local_irq_restore(flags); +} + +/** + * imx_dma_request - request/allocate specified channel number + * @dma_ch: i.MX DMA channel number + * @name: the driver/caller own non-%NULL identification + */ +int imx_dma_request(imx_dmach_t dma_ch, const char *name) +{ + struct imx_dma_channel *imxdma = &imx_dma_channels[dma_ch]; + unsigned long flags; + + /* basic sanity checks */ + if (!name) + return -EINVAL; + + if (dma_ch >= IMX_DMA_CHANNELS) { + printk(KERN_CRIT "%s: called for non-existed channel %d\n", + __FUNCTION__, dma_ch); + return -EINVAL; } + local_irq_save(flags); + if (imxdma->name) { + local_irq_restore(flags); + return -ENODEV; + } + + imxdma->name = name; + imxdma->irq_handler = NULL; + imxdma->err_handler = NULL; + imxdma->data = NULL; + imxdma->sg = NULL; local_irq_restore(flags); - return i; + return 0; } -void -imx_free_dma(int dma_ch) +/** + * imx_dma_free - release previously acquired channel + * @dma_ch: i.MX DMA channel number + */ +void imx_dma_free(imx_dmach_t dma_ch) { unsigned long flags; + struct imx_dma_channel *imxdma = &imx_dma_channels[dma_ch]; - if (!dma_channels[dma_ch].name) { + if (!imxdma->name) { printk(KERN_CRIT "%s: trying to free channel %d which is already freed\n", __FUNCTION__, dma_ch); @@ -92,27 +398,84 @@ imx_free_dma(int dma_ch) } local_irq_save(flags); - DIMR &= ~(1 << dma_ch); - dma_channels[dma_ch].name = NULL; + /* Disable interrupts */ + DIMR |= (1 << dma_ch); + CCR(dma_ch) &= ~CCR_CEN; + imxdma->name = NULL; local_irq_restore(flags); } -static irqreturn_t -dma_err_handler(int irq, void *dev_id, struct pt_regs *regs) +/** + * imx_dma_request_by_prio - find and request some of free channels best suiting requested priority + * @dma_ch: i.MX DMA channel number + * @name: the driver/caller own non-%NULL identification + * @prio: one of the hardware distinguished priority level: + * %DMA_PRIO_HIGH, %DMA_PRIO_MEDIUM, %DMA_PRIO_LOW + * + * This function tries to find free channel in the specified priority group + * if the priority cannot be achieved it tries to look for free channel + * in the higher and then even lower priority groups. + * + * Return value: If there is no free channel to allocate, -%ENODEV is returned. + * Zero value indicates successful channel allocation. + */ +int +imx_dma_request_by_prio(imx_dmach_t * pdma_ch, const char *name, + imx_dma_prio prio) +{ + int i; + int best; + + switch (prio) { + case (DMA_PRIO_HIGH): + best = 8; + break; + case (DMA_PRIO_MEDIUM): + best = 4; + break; + case (DMA_PRIO_LOW): + default: + best = 0; + break; + } + + for (i = best; i < IMX_DMA_CHANNELS; i++) { + if (!imx_dma_request(i, name)) { + *pdma_ch = i; + return 0; + } + } + + for (i = best - 1; i >= 0; i--) { + if (!imx_dma_request(i, name)) { + *pdma_ch = i; + return 0; + } + } + + printk(KERN_ERR "%s: no free DMA channel found\n", __FUNCTION__); + + return -ENODEV; +} + +static irqreturn_t dma_err_handler(int irq, void *dev_id, struct pt_regs *regs) { int i, disr = DISR; - struct dma_channel *channel; + struct imx_dma_channel *channel; unsigned int err_mask = DBTOSR | DRTOSR | DSESR | DBOSR; DISR = disr; - for (i = 0; i < 11; i++) { - channel = &dma_channels[i]; + for (i = 0; i < IMX_DMA_CHANNELS; i++) { + channel = &imx_dma_channels[i]; - if ( (err_mask & 1<<i) && channel->name && channel->err_handler) { + if ((err_mask & 1 << i) && channel->name + && channel->err_handler) { channel->err_handler(i, channel->data, regs); continue; } + imx_dma_channels[i].sg = NULL; + if (DBTOSR & (1 << i)) { printk(KERN_WARNING "Burst timeout on channel %d (%s)\n", @@ -141,17 +504,27 @@ dma_err_handler(int irq, void *dev_id, struct pt_regs *regs) return IRQ_HANDLED; } -static irqreturn_t -dma_irq_handler(int irq, void *dev_id, struct pt_regs *regs) +static irqreturn_t dma_irq_handler(int irq, void *dev_id, struct pt_regs *regs) { int i, disr = DISR; + pr_debug("imxdma: dma_irq_handler called, disr=0x%08x\n", + disr); + DISR = disr; - for (i = 0; i < 11; i++) { + for (i = 0; i < IMX_DMA_CHANNELS; i++) { if (disr & (1 << i)) { - struct dma_channel *channel = &dma_channels[i]; - if (channel->name && channel->irq_handler) { - channel->irq_handler(i, channel->data, regs); + struct imx_dma_channel *channel = &imx_dma_channels[i]; + if (channel->name) { + if (imx_dma_sg_next(i, CNTR(i))) { + CCR(i) &= ~CCR_CEN; + mb(); + CCR(i) |= CCR_CEN; + } else { + if (channel->irq_handler) + channel->irq_handler(i, + channel->data, regs); + } } else { /* * IRQ for an unregistered DMA channel: @@ -165,10 +538,10 @@ dma_irq_handler(int irq, void *dev_id, struct pt_regs *regs) return IRQ_HANDLED; } -static int __init -imx_dma_init(void) +static int __init imx_dma_init(void) { int ret; + int i; /* reset DMA module */ DCR = DCR_DRST; @@ -189,15 +562,27 @@ imx_dma_init(void) DCR = DCR_DEN; /* clear all interrupts */ - DISR = 0x3ff; + DISR = (1 << IMX_DMA_CHANNELS) - 1; /* enable interrupts */ - DIMR = 0; + DIMR = (1 << IMX_DMA_CHANNELS) - 1; + + for (i = 0; i < IMX_DMA_CHANNELS; i++) { + imx_dma_channels[i].sg = NULL; + imx_dma_channels[i].dma_num = i; + } return ret; } arch_initcall(imx_dma_init); -EXPORT_SYMBOL(imx_request_dma); -EXPORT_SYMBOL(imx_free_dma); +EXPORT_SYMBOL(imx_dma_setup_single); +EXPORT_SYMBOL(imx_dma_setup_sg); +EXPORT_SYMBOL(imx_dma_setup_handlers); +EXPORT_SYMBOL(imx_dma_enable); +EXPORT_SYMBOL(imx_dma_disable); +EXPORT_SYMBOL(imx_dma_request); +EXPORT_SYMBOL(imx_dma_free); +EXPORT_SYMBOL(imx_dma_request_by_prio); +EXPORT_SYMBOL(imx_dma_channels); |