1
0
Fork 0
alistair23-linux/drivers/dma/imx-sdma.c

1683 lines
42 KiB
C
Raw Normal View History

/*
* drivers/dma/imx-sdma.c
*
* This file contains a driver for the Freescale Smart DMA engine
*
* Copyright 2010 Sascha Hauer, Pengutronix <s.hauer@pengutronix.de>
*
* Based on code from Freescale:
*
* Copyright 2004-2009 Freescale Semiconductor, Inc. All Rights Reserved.
*
* The code contained herein is licensed under the GNU General Public
* License. You may obtain a copy of the GNU General Public License
* Version 2 or later at the following locations:
*
* http://www.opensource.org/licenses/gpl-license.html
* http://www.gnu.org/copyleft/gpl.html
*/
#include <linux/init.h>
#include <linux/module.h>
#include <linux/types.h>
#include <linux/bitops.h>
#include <linux/mm.h>
#include <linux/interrupt.h>
#include <linux/clk.h>
#include <linux/delay.h>
#include <linux/sched.h>
#include <linux/semaphore.h>
#include <linux/spinlock.h>
#include <linux/device.h>
#include <linux/dma-mapping.h>
#include <linux/firmware.h>
#include <linux/slab.h>
#include <linux/platform_device.h>
#include <linux/dmaengine.h>
#include <linux/of.h>
#include <linux/of_device.h>
#include <linux/of_dma.h>
#include <asm/irq.h>
#include <linux/platform_data/dma-imx-sdma.h>
#include <linux/platform_data/dma-imx.h>
#include "dmaengine.h"
/* SDMA registers */
#define SDMA_H_C0PTR 0x000
#define SDMA_H_INTR 0x004
#define SDMA_H_STATSTOP 0x008
#define SDMA_H_START 0x00c
#define SDMA_H_EVTOVR 0x010
#define SDMA_H_DSPOVR 0x014
#define SDMA_H_HOSTOVR 0x018
#define SDMA_H_EVTPEND 0x01c
#define SDMA_H_DSPENBL 0x020
#define SDMA_H_RESET 0x024
#define SDMA_H_EVTERR 0x028
#define SDMA_H_INTRMSK 0x02c
#define SDMA_H_PSW 0x030
#define SDMA_H_EVTERRDBG 0x034
#define SDMA_H_CONFIG 0x038
#define SDMA_ONCE_ENB 0x040
#define SDMA_ONCE_DATA 0x044
#define SDMA_ONCE_INSTR 0x048
#define SDMA_ONCE_STAT 0x04c
#define SDMA_ONCE_CMD 0x050
#define SDMA_EVT_MIRROR 0x054
#define SDMA_ILLINSTADDR 0x058
#define SDMA_CHN0ADDR 0x05c
#define SDMA_ONCE_RTB 0x060
#define SDMA_XTRIG_CONF1 0x070
#define SDMA_XTRIG_CONF2 0x074
#define SDMA_CHNENBL0_IMX35 0x200
#define SDMA_CHNENBL0_IMX31 0x080
#define SDMA_CHNPRI_0 0x100
/*
* Buffer descriptor status values.
*/
#define BD_DONE 0x01
#define BD_WRAP 0x02
#define BD_CONT 0x04
#define BD_INTR 0x08
#define BD_RROR 0x10
#define BD_LAST 0x20
#define BD_EXTD 0x80
/*
* Data Node descriptor status values.
*/
#define DND_END_OF_FRAME 0x80
#define DND_END_OF_XFER 0x40
#define DND_DONE 0x20
#define DND_UNUSED 0x01
/*
* IPCV2 descriptor status values.
*/
#define BD_IPCV2_END_OF_FRAME 0x40
#define IPCV2_MAX_NODES 50
/*
* Error bit set in the CCB status field by the SDMA,
* in setbd routine, in case of a transfer error
*/
#define DATA_ERROR 0x10000000
/*
* Buffer descriptor commands.
*/
#define C0_ADDR 0x01
#define C0_LOAD 0x02
#define C0_DUMP 0x03
#define C0_SETCTX 0x07
#define C0_GETCTX 0x03
#define C0_SETDM 0x01
#define C0_SETPM 0x04
#define C0_GETDM 0x02
#define C0_GETPM 0x08
/*
* Change endianness indicator in the BD command field
*/
#define CHANGE_ENDIANNESS 0x80
/*
* Mode/Count of data node descriptors - IPCv2
*/
struct sdma_mode_count {
u32 count : 16; /* size of the buffer pointed by this BD */
u32 status : 8; /* E,R,I,C,W,D status bits stored here */
u32 command : 8; /* command mostlky used for channel 0 */
};
/*
* Buffer descriptor
*/
struct sdma_buffer_descriptor {
struct sdma_mode_count mode;
u32 buffer_addr; /* address of the buffer described */
u32 ext_buffer_addr; /* extended buffer address */
} __attribute__ ((packed));
/**
* struct sdma_channel_control - Channel control Block
*
* @current_bd_ptr current buffer descriptor processed
* @base_bd_ptr first element of buffer descriptor array
* @unused padding. The SDMA engine expects an array of 128 byte
* control blocks
*/
struct sdma_channel_control {
u32 current_bd_ptr;
u32 base_bd_ptr;
u32 unused[2];
} __attribute__ ((packed));
/**
* struct sdma_state_registers - SDMA context for a channel
*
* @pc: program counter
* @t: test bit: status of arithmetic & test instruction
* @rpc: return program counter
* @sf: source fault while loading data
* @spc: loop start program counter
* @df: destination fault while storing data
* @epc: loop end program counter
* @lm: loop mode
*/
struct sdma_state_registers {
u32 pc :14;
u32 unused1: 1;
u32 t : 1;
u32 rpc :14;
u32 unused0: 1;
u32 sf : 1;
u32 spc :14;
u32 unused2: 1;
u32 df : 1;
u32 epc :14;
u32 lm : 2;
} __attribute__ ((packed));
/**
* struct sdma_context_data - sdma context specific to a channel
*
* @channel_state: channel state bits
* @gReg: general registers
* @mda: burst dma destination address register
* @msa: burst dma source address register
* @ms: burst dma status register
* @md: burst dma data register
* @pda: peripheral dma destination address register
* @psa: peripheral dma source address register
* @ps: peripheral dma status register
* @pd: peripheral dma data register
* @ca: CRC polynomial register
* @cs: CRC accumulator register
* @dda: dedicated core destination address register
* @dsa: dedicated core source address register
* @ds: dedicated core status register
* @dd: dedicated core data register
*/
struct sdma_context_data {
struct sdma_state_registers channel_state;
u32 gReg[8];
u32 mda;
u32 msa;
u32 ms;
u32 md;
u32 pda;
u32 psa;
u32 ps;
u32 pd;
u32 ca;
u32 cs;
u32 dda;
u32 dsa;
u32 ds;
u32 dd;
u32 scratch0;
u32 scratch1;
u32 scratch2;
u32 scratch3;
u32 scratch4;
u32 scratch5;
u32 scratch6;
u32 scratch7;
} __attribute__ ((packed));
#define NUM_BD (int)(PAGE_SIZE / sizeof(struct sdma_buffer_descriptor))
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 + 1
* @direction transfer type. Needed for setting SDMA script
* @peripheral_type Peripheral type. Needed for setting SDMA script
* @event_id0 aka dma request line
* @event_id1 for channels that use 2 events
* @word_size peripheral access size
* @buf_tail ID of the buffer that was processed
* @num_bd max NUM_BD. number of descriptors currently handling
*/
struct sdma_channel {
struct sdma_engine *sdma;
unsigned int channel;
enum dma_transfer_direction direction;
enum sdma_peripheral_type peripheral_type;
unsigned int event_id0;
unsigned int event_id1;
enum dma_slave_buswidth word_size;
unsigned int buf_tail;
unsigned int num_bd;
Update imx-sdma cyclic handling to report residue I received a report this morning from one of the Novena developers that the behaviour of the iMX6 ASoC codec driver (using imx-pcm-dma.c) was sub-optimal under high system load. While there are issues relating to system load remaining, upon reviewing the ASoC imx-pcm-dma.c driver, it was noticed that it not using the residue support, because SDMA doesn't support it. This has the effect that SDMA has to make multiple calls into the ASoC and ALSA code, one for each period. Since ALSA's snd_pcm_elapsed() does not need to be called multiple times and it is entirely sufficient to call it once to update ALSA with the current buffer position via the pointer method, we can do better here. We can also avoid stopping the DMA entirely, just like real cyclic DMA implementations behave. While this means that we replay some old samples, this is a nicer behaviour than having audio stop and restart. The changes to achieve this are relatively minor - imx-sdma.c can track where the DMA is to the nearest descriptor boundary - it does this already when deciding how many callbacks to issue. In doing this, buf_tail always points at the descriptor which will complete next. The residue is defined by the bytes remaining to the end of the buffer, when the buffer is viewed as a single block of memory [start...end]. So, when we start out, there's a full buffer worth of residue, and this counts down as we approach the end of the buffer, eventually becoming zero at the end, before returning to the full buffer worth when we wrap back to the start. Moving the walking of the descriptors into the interrupt handler means that we can update the BD_DONE flag at interrupt time, thus avoiding a delayed tasklet stopping the cyclic DMA. This means that the residue can be calculated from (total descriptors - buf_tail) * descriptor size. This is what the change below does. We update imx-pcm-dma.c to remove the NO_RESIDUE flag since we now provide the residue. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Shawn Guo <shawn.guo@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2014-06-25 06:00:33 -06:00
unsigned int period_len;
struct sdma_buffer_descriptor *bd;
dma_addr_t bd_phys;
unsigned int pc_from_device, pc_to_device;
unsigned long flags;
dma_addr_t per_address;
unsigned long event_mask[2];
unsigned long watermark_level;
u32 shp_addr, per_addr;
struct dma_chan chan;
spinlock_t lock;
struct dma_async_tx_descriptor desc;
enum dma_status status;
unsigned int chn_count;
unsigned int chn_real_count;
struct tasklet_struct tasklet;
struct imx_dma_data data;
};
#define IMX_DMA_SG_LOOP BIT(0)
#define MAX_DMA_CHANNELS 32
#define MXC_SDMA_DEFAULT_PRIORITY 1
#define MXC_SDMA_MIN_PRIORITY 1
#define MXC_SDMA_MAX_PRIORITY 7
#define SDMA_FIRMWARE_MAGIC 0x414d4453
/**
* struct sdma_firmware_header - Layout of the firmware image
*
* @magic "SDMA"
* @version_major increased whenever layout of struct sdma_script_start_addrs
* changes.
* @version_minor firmware minor version (for binary compatible changes)
* @script_addrs_start offset of struct sdma_script_start_addrs in this image
* @num_script_addrs Number of script addresses in this image
* @ram_code_start offset of SDMA ram image in this firmware image
* @ram_code_size size of SDMA ram image
* @script_addrs Stores the start address of the SDMA scripts
* (in SDMA memory space)
*/
struct sdma_firmware_header {
u32 magic;
u32 version_major;
u32 version_minor;
u32 script_addrs_start;
u32 num_script_addrs;
u32 ram_code_start;
u32 ram_code_size;
};
struct sdma_driver_data {
int chnenbl0;
int num_events;
struct sdma_script_start_addrs *script_addrs;
};
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;
struct sdma_context_data *context;
dma_addr_t context_phys;
struct dma_device dma_device;
struct clk *clk_ipg;
struct clk *clk_ahb;
spinlock_t channel_0_lock;
u32 script_number;
struct sdma_script_start_addrs *script_addrs;
const struct sdma_driver_data *drvdata;
};
static struct sdma_driver_data sdma_imx31 = {
.chnenbl0 = SDMA_CHNENBL0_IMX31,
.num_events = 32,
};
static struct sdma_script_start_addrs sdma_script_imx25 = {
.ap_2_ap_addr = 729,
.uart_2_mcu_addr = 904,
.per_2_app_addr = 1255,
.mcu_2_app_addr = 834,
.uartsh_2_mcu_addr = 1120,
.per_2_shp_addr = 1329,
.mcu_2_shp_addr = 1048,
.ata_2_mcu_addr = 1560,
.mcu_2_ata_addr = 1479,
.app_2_per_addr = 1189,
.app_2_mcu_addr = 770,
.shp_2_per_addr = 1407,
.shp_2_mcu_addr = 979,
};
static struct sdma_driver_data sdma_imx25 = {
.chnenbl0 = SDMA_CHNENBL0_IMX35,
.num_events = 48,
.script_addrs = &sdma_script_imx25,
};
static struct sdma_driver_data sdma_imx35 = {
.chnenbl0 = SDMA_CHNENBL0_IMX35,
.num_events = 48,
};
static struct sdma_script_start_addrs sdma_script_imx51 = {
.ap_2_ap_addr = 642,
.uart_2_mcu_addr = 817,
.mcu_2_app_addr = 747,
.mcu_2_shp_addr = 961,
.ata_2_mcu_addr = 1473,
.mcu_2_ata_addr = 1392,
.app_2_per_addr = 1033,
.app_2_mcu_addr = 683,
.shp_2_per_addr = 1251,
.shp_2_mcu_addr = 892,
};
static struct sdma_driver_data sdma_imx51 = {
.chnenbl0 = SDMA_CHNENBL0_IMX35,
.num_events = 48,
.script_addrs = &sdma_script_imx51,
};
static struct sdma_script_start_addrs sdma_script_imx53 = {
.ap_2_ap_addr = 642,
.app_2_mcu_addr = 683,
.mcu_2_app_addr = 747,
.uart_2_mcu_addr = 817,
.shp_2_mcu_addr = 891,
.mcu_2_shp_addr = 960,
.uartsh_2_mcu_addr = 1032,
.spdif_2_mcu_addr = 1100,
.mcu_2_spdif_addr = 1134,
.firi_2_mcu_addr = 1193,
.mcu_2_firi_addr = 1290,
};
static struct sdma_driver_data sdma_imx53 = {
.chnenbl0 = SDMA_CHNENBL0_IMX35,
.num_events = 48,
.script_addrs = &sdma_script_imx53,
};
static struct sdma_script_start_addrs sdma_script_imx6q = {
.ap_2_ap_addr = 642,
.uart_2_mcu_addr = 817,
.mcu_2_app_addr = 747,
.per_2_per_addr = 6331,
.uartsh_2_mcu_addr = 1032,
.mcu_2_shp_addr = 960,
.app_2_mcu_addr = 683,
.shp_2_mcu_addr = 891,
.spdif_2_mcu_addr = 1100,
.mcu_2_spdif_addr = 1134,
};
static struct sdma_driver_data sdma_imx6q = {
.chnenbl0 = SDMA_CHNENBL0_IMX35,
.num_events = 48,
.script_addrs = &sdma_script_imx6q,
};
static struct platform_device_id sdma_devtypes[] = {
{
.name = "imx25-sdma",
.driver_data = (unsigned long)&sdma_imx25,
}, {
.name = "imx31-sdma",
.driver_data = (unsigned long)&sdma_imx31,
}, {
.name = "imx35-sdma",
.driver_data = (unsigned long)&sdma_imx35,
}, {
.name = "imx51-sdma",
.driver_data = (unsigned long)&sdma_imx51,
}, {
.name = "imx53-sdma",
.driver_data = (unsigned long)&sdma_imx53,
}, {
.name = "imx6q-sdma",
.driver_data = (unsigned long)&sdma_imx6q,
}, {
/* sentinel */
}
};
MODULE_DEVICE_TABLE(platform, sdma_devtypes);
static const struct of_device_id sdma_dt_ids[] = {
{ .compatible = "fsl,imx6q-sdma", .data = &sdma_imx6q, },
{ .compatible = "fsl,imx53-sdma", .data = &sdma_imx53, },
{ .compatible = "fsl,imx51-sdma", .data = &sdma_imx51, },
{ .compatible = "fsl,imx35-sdma", .data = &sdma_imx35, },
{ .compatible = "fsl,imx31-sdma", .data = &sdma_imx31, },
{ .compatible = "fsl,imx25-sdma", .data = &sdma_imx25, },
{ /* sentinel */ }
};
MODULE_DEVICE_TABLE(of, sdma_dt_ids);
#define SDMA_H_CONFIG_DSPDMA BIT(12) /* indicates if the DSPDMA is used */
#define SDMA_H_CONFIG_RTD_PINS BIT(11) /* indicates if Real-Time Debug pins are enabled */
#define SDMA_H_CONFIG_ACR BIT(4) /* indicates if AHB freq /core freq = 2 or 1 */
#define SDMA_H_CONFIG_CSM (3) /* indicates which context switch mode is selected*/
static inline u32 chnenbl_ofs(struct sdma_engine *sdma, unsigned int event)
{
u32 chnenbl0 = sdma->drvdata->chnenbl0;
return chnenbl0 + event * 4;
}
static int sdma_config_ownership(struct sdma_channel *sdmac,
bool event_override, bool mcu_override, bool dsp_override)
{
struct sdma_engine *sdma = sdmac->sdma;
int channel = sdmac->channel;
unsigned long evt, mcu, dsp;
if (event_override && mcu_override && dsp_override)
return -EINVAL;
evt = readl_relaxed(sdma->regs + SDMA_H_EVTOVR);
mcu = readl_relaxed(sdma->regs + SDMA_H_HOSTOVR);
dsp = readl_relaxed(sdma->regs + SDMA_H_DSPOVR);
if (dsp_override)
__clear_bit(channel, &dsp);
else
__set_bit(channel, &dsp);
if (event_override)
__clear_bit(channel, &evt);
else
__set_bit(channel, &evt);
if (mcu_override)
__clear_bit(channel, &mcu);
else
__set_bit(channel, &mcu);
writel_relaxed(evt, sdma->regs + SDMA_H_EVTOVR);
writel_relaxed(mcu, sdma->regs + SDMA_H_HOSTOVR);
writel_relaxed(dsp, sdma->regs + SDMA_H_DSPOVR);
return 0;
}
static void sdma_enable_channel(struct sdma_engine *sdma, int channel)
{
writel(BIT(channel), sdma->regs + SDMA_H_START);
}
/*
* sdma_run_channel0 - run a channel and wait till it's done
*/
static int sdma_run_channel0(struct sdma_engine *sdma)
{
int ret;
unsigned long timeout = 500;
sdma_enable_channel(sdma, 0);
while (!(ret = readl_relaxed(sdma->regs + SDMA_H_INTR) & 1)) {
if (timeout-- <= 0)
break;
udelay(1);
}
if (ret) {
/* Clear the interrupt status */
writel_relaxed(ret, sdma->regs + SDMA_H_INTR);
} else {
dev_err(sdma->dev, "Timeout waiting for CH0 ready\n");
}
return ret ? 0 : -ETIMEDOUT;
}
static int sdma_load_script(struct sdma_engine *sdma, void *buf, int size,
u32 address)
{
struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd;
void *buf_virt;
dma_addr_t buf_phys;
int ret;
unsigned long flags;
buf_virt = dma_alloc_coherent(NULL,
size,
&buf_phys, GFP_KERNEL);
if (!buf_virt) {
return -ENOMEM;
}
spin_lock_irqsave(&sdma->channel_0_lock, flags);
bd0->mode.command = C0_SETPM;
bd0->mode.status = BD_DONE | BD_INTR | BD_WRAP | BD_EXTD;
bd0->mode.count = size / 2;
bd0->buffer_addr = buf_phys;
bd0->ext_buffer_addr = address;
memcpy(buf_virt, buf, size);
ret = sdma_run_channel0(sdma);
spin_unlock_irqrestore(&sdma->channel_0_lock, flags);
dma_free_coherent(NULL, size, buf_virt, buf_phys);
return ret;
}
static void sdma_event_enable(struct sdma_channel *sdmac, unsigned int event)
{
struct sdma_engine *sdma = sdmac->sdma;
int channel = sdmac->channel;
unsigned long val;
u32 chnenbl = chnenbl_ofs(sdma, event);
val = readl_relaxed(sdma->regs + chnenbl);
__set_bit(channel, &val);
writel_relaxed(val, sdma->regs + chnenbl);
}
static void sdma_event_disable(struct sdma_channel *sdmac, unsigned int event)
{
struct sdma_engine *sdma = sdmac->sdma;
int channel = sdmac->channel;
u32 chnenbl = chnenbl_ofs(sdma, event);
unsigned long val;
val = readl_relaxed(sdma->regs + chnenbl);
__clear_bit(channel, &val);
writel_relaxed(val, sdma->regs + chnenbl);
}
static void sdma_handle_channel_loop(struct sdma_channel *sdmac)
Update imx-sdma cyclic handling to report residue I received a report this morning from one of the Novena developers that the behaviour of the iMX6 ASoC codec driver (using imx-pcm-dma.c) was sub-optimal under high system load. While there are issues relating to system load remaining, upon reviewing the ASoC imx-pcm-dma.c driver, it was noticed that it not using the residue support, because SDMA doesn't support it. This has the effect that SDMA has to make multiple calls into the ASoC and ALSA code, one for each period. Since ALSA's snd_pcm_elapsed() does not need to be called multiple times and it is entirely sufficient to call it once to update ALSA with the current buffer position via the pointer method, we can do better here. We can also avoid stopping the DMA entirely, just like real cyclic DMA implementations behave. While this means that we replay some old samples, this is a nicer behaviour than having audio stop and restart. The changes to achieve this are relatively minor - imx-sdma.c can track where the DMA is to the nearest descriptor boundary - it does this already when deciding how many callbacks to issue. In doing this, buf_tail always points at the descriptor which will complete next. The residue is defined by the bytes remaining to the end of the buffer, when the buffer is viewed as a single block of memory [start...end]. So, when we start out, there's a full buffer worth of residue, and this counts down as we approach the end of the buffer, eventually becoming zero at the end, before returning to the full buffer worth when we wrap back to the start. Moving the walking of the descriptors into the interrupt handler means that we can update the BD_DONE flag at interrupt time, thus avoiding a delayed tasklet stopping the cyclic DMA. This means that the residue can be calculated from (total descriptors - buf_tail) * descriptor size. This is what the change below does. We update imx-pcm-dma.c to remove the NO_RESIDUE flag since we now provide the residue. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Shawn Guo <shawn.guo@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2014-06-25 06:00:33 -06:00
{
if (sdmac->desc.callback)
sdmac->desc.callback(sdmac->desc.callback_param);
}
static void sdma_update_channel_loop(struct sdma_channel *sdmac)
{
struct sdma_buffer_descriptor *bd;
/*
* loop mode. Iterate over descriptors, re-setup them and
* call callback function.
*/
while (1) {
bd = &sdmac->bd[sdmac->buf_tail];
if (bd->mode.status & BD_DONE)
break;
if (bd->mode.status & BD_RROR)
sdmac->status = DMA_ERROR;
bd->mode.status |= BD_DONE;
sdmac->buf_tail++;
sdmac->buf_tail %= sdmac->num_bd;
}
}
static void mxc_sdma_handle_channel_normal(struct sdma_channel *sdmac)
{
struct sdma_buffer_descriptor *bd;
int i, error = 0;
sdmac->chn_real_count = 0;
/*
* non loop mode. Iterate over all descriptors, collect
* errors and call callback function
*/
for (i = 0; i < sdmac->num_bd; i++) {
bd = &sdmac->bd[i];
if (bd->mode.status & (BD_DONE | BD_RROR))
error = -EIO;
sdmac->chn_real_count += bd->mode.count;
}
if (error)
sdmac->status = DMA_ERROR;
else
sdmac->status = DMA_COMPLETE;
dma_cookie_complete(&sdmac->desc);
if (sdmac->desc.callback)
sdmac->desc.callback(sdmac->desc.callback_param);
}
static void sdma_tasklet(unsigned long data)
{
struct sdma_channel *sdmac = (struct sdma_channel *) data;
if (sdmac->flags & IMX_DMA_SG_LOOP)
sdma_handle_channel_loop(sdmac);
else
mxc_sdma_handle_channel_normal(sdmac);
}
static irqreturn_t sdma_int_handler(int irq, void *dev_id)
{
struct sdma_engine *sdma = dev_id;
unsigned long stat;
stat = readl_relaxed(sdma->regs + SDMA_H_INTR);
/* not interested in channel 0 interrupts */
stat &= ~1;
writel_relaxed(stat, sdma->regs + SDMA_H_INTR);
while (stat) {
int channel = fls(stat) - 1;
struct sdma_channel *sdmac = &sdma->channel[channel];
Update imx-sdma cyclic handling to report residue I received a report this morning from one of the Novena developers that the behaviour of the iMX6 ASoC codec driver (using imx-pcm-dma.c) was sub-optimal under high system load. While there are issues relating to system load remaining, upon reviewing the ASoC imx-pcm-dma.c driver, it was noticed that it not using the residue support, because SDMA doesn't support it. This has the effect that SDMA has to make multiple calls into the ASoC and ALSA code, one for each period. Since ALSA's snd_pcm_elapsed() does not need to be called multiple times and it is entirely sufficient to call it once to update ALSA with the current buffer position via the pointer method, we can do better here. We can also avoid stopping the DMA entirely, just like real cyclic DMA implementations behave. While this means that we replay some old samples, this is a nicer behaviour than having audio stop and restart. The changes to achieve this are relatively minor - imx-sdma.c can track where the DMA is to the nearest descriptor boundary - it does this already when deciding how many callbacks to issue. In doing this, buf_tail always points at the descriptor which will complete next. The residue is defined by the bytes remaining to the end of the buffer, when the buffer is viewed as a single block of memory [start...end]. So, when we start out, there's a full buffer worth of residue, and this counts down as we approach the end of the buffer, eventually becoming zero at the end, before returning to the full buffer worth when we wrap back to the start. Moving the walking of the descriptors into the interrupt handler means that we can update the BD_DONE flag at interrupt time, thus avoiding a delayed tasklet stopping the cyclic DMA. This means that the residue can be calculated from (total descriptors - buf_tail) * descriptor size. This is what the change below does. We update imx-pcm-dma.c to remove the NO_RESIDUE flag since we now provide the residue. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Shawn Guo <shawn.guo@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2014-06-25 06:00:33 -06:00
if (sdmac->flags & IMX_DMA_SG_LOOP)
sdma_update_channel_loop(sdmac);
tasklet_schedule(&sdmac->tasklet);
__clear_bit(channel, &stat);
}
return IRQ_HANDLED;
}
/*
* sets the pc of SDMA script according to the peripheral type
*/
static void sdma_get_pc(struct sdma_channel *sdmac,
enum sdma_peripheral_type peripheral_type)
{
struct sdma_engine *sdma = sdmac->sdma;
int per_2_emi = 0, emi_2_per = 0;
/*
* These are needed once we start to support transfers between
* two peripherals or memory-to-memory transfers
*/
int per_2_per = 0, emi_2_emi = 0;
sdmac->pc_from_device = 0;
sdmac->pc_to_device = 0;
switch (peripheral_type) {
case IMX_DMATYPE_MEMORY:
emi_2_emi = sdma->script_addrs->ap_2_ap_addr;
break;
case IMX_DMATYPE_DSP:
emi_2_per = sdma->script_addrs->bp_2_ap_addr;
per_2_emi = sdma->script_addrs->ap_2_bp_addr;
break;
case IMX_DMATYPE_FIRI:
per_2_emi = sdma->script_addrs->firi_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_firi_addr;
break;
case IMX_DMATYPE_UART:
per_2_emi = sdma->script_addrs->uart_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_app_addr;
break;
case IMX_DMATYPE_UART_SP:
per_2_emi = sdma->script_addrs->uartsh_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_shp_addr;
break;
case IMX_DMATYPE_ATA:
per_2_emi = sdma->script_addrs->ata_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_ata_addr;
break;
case IMX_DMATYPE_CSPI:
case IMX_DMATYPE_EXT:
case IMX_DMATYPE_SSI:
per_2_emi = sdma->script_addrs->app_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_app_addr;
break;
case IMX_DMATYPE_SSI_DUAL:
per_2_emi = sdma->script_addrs->ssish_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_ssish_addr;
break;
case IMX_DMATYPE_SSI_SP:
case IMX_DMATYPE_MMC:
case IMX_DMATYPE_SDHC:
case IMX_DMATYPE_CSPI_SP:
case IMX_DMATYPE_ESAI:
case IMX_DMATYPE_MSHC_SP:
per_2_emi = sdma->script_addrs->shp_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_shp_addr;
break;
case IMX_DMATYPE_ASRC:
per_2_emi = sdma->script_addrs->asrc_2_mcu_addr;
emi_2_per = sdma->script_addrs->asrc_2_mcu_addr;
per_2_per = sdma->script_addrs->per_2_per_addr;
break;
case IMX_DMATYPE_ASRC_SP:
per_2_emi = sdma->script_addrs->shp_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_shp_addr;
per_2_per = sdma->script_addrs->per_2_per_addr;
break;
case IMX_DMATYPE_MSHC:
per_2_emi = sdma->script_addrs->mshc_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_mshc_addr;
break;
case IMX_DMATYPE_CCM:
per_2_emi = sdma->script_addrs->dptc_dvfs_addr;
break;
case IMX_DMATYPE_SPDIF:
per_2_emi = sdma->script_addrs->spdif_2_mcu_addr;
emi_2_per = sdma->script_addrs->mcu_2_spdif_addr;
break;
case IMX_DMATYPE_IPU_MEMORY:
emi_2_per = sdma->script_addrs->ext_mem_2_ipu_addr;
break;
default:
break;
}
sdmac->pc_from_device = per_2_emi;
sdmac->pc_to_device = emi_2_per;
}
static int sdma_load_context(struct sdma_channel *sdmac)
{
struct sdma_engine *sdma = sdmac->sdma;
int channel = sdmac->channel;
int load_address;
struct sdma_context_data *context = sdma->context;
struct sdma_buffer_descriptor *bd0 = sdma->channel[0].bd;
int ret;
unsigned long flags;
if (sdmac->direction == DMA_DEV_TO_MEM) {
load_address = sdmac->pc_from_device;
} else {
load_address = sdmac->pc_to_device;
}
if (load_address < 0)
return load_address;
dev_dbg(sdma->dev, "load_address = %d\n", load_address);
dev_dbg(sdma->dev, "wml = 0x%08x\n", (u32)sdmac->watermark_level);
dev_dbg(sdma->dev, "shp_addr = 0x%08x\n", sdmac->shp_addr);
dev_dbg(sdma->dev, "per_addr = 0x%08x\n", sdmac->per_addr);
dev_dbg(sdma->dev, "event_mask0 = 0x%08x\n", (u32)sdmac->event_mask[0]);
dev_dbg(sdma->dev, "event_mask1 = 0x%08x\n", (u32)sdmac->event_mask[1]);
spin_lock_irqsave(&sdma->channel_0_lock, flags);
memset(context, 0, sizeof(*context));
context->channel_state.pc = load_address;
/* Send by context the event mask,base address for peripheral
* and watermark level
*/
context->gReg[0] = sdmac->event_mask[1];
context->gReg[1] = sdmac->event_mask[0];
context->gReg[2] = sdmac->per_addr;
context->gReg[6] = sdmac->shp_addr;
context->gReg[7] = sdmac->watermark_level;
bd0->mode.command = C0_SETDM;
bd0->mode.status = BD_DONE | BD_INTR | BD_WRAP | BD_EXTD;
bd0->mode.count = sizeof(*context) / 4;
bd0->buffer_addr = sdma->context_phys;
bd0->ext_buffer_addr = 2048 + (sizeof(*context) / 4) * channel;
ret = sdma_run_channel0(sdma);
spin_unlock_irqrestore(&sdma->channel_0_lock, flags);
return ret;
}
static void sdma_disable_channel(struct sdma_channel *sdmac)
{
struct sdma_engine *sdma = sdmac->sdma;
int channel = sdmac->channel;
writel_relaxed(BIT(channel), sdma->regs + SDMA_H_STATSTOP);
sdmac->status = DMA_ERROR;
}
static int sdma_config_channel(struct sdma_channel *sdmac)
{
int ret;
sdma_disable_channel(sdmac);
sdmac->event_mask[0] = 0;
sdmac->event_mask[1] = 0;
sdmac->shp_addr = 0;
sdmac->per_addr = 0;
if (sdmac->event_id0) {
if (sdmac->event_id0 >= sdmac->sdma->drvdata->num_events)
return -EINVAL;
sdma_event_enable(sdmac, sdmac->event_id0);
}
switch (sdmac->peripheral_type) {
case IMX_DMATYPE_DSP:
sdma_config_ownership(sdmac, false, true, true);
break;
case IMX_DMATYPE_MEMORY:
sdma_config_ownership(sdmac, false, true, false);
break;
default:
sdma_config_ownership(sdmac, true, true, false);
break;
}
sdma_get_pc(sdmac, sdmac->peripheral_type);
if ((sdmac->peripheral_type != IMX_DMATYPE_MEMORY) &&
(sdmac->peripheral_type != IMX_DMATYPE_DSP)) {
/* Handle multiple event channels differently */
if (sdmac->event_id1) {
sdmac->event_mask[1] = BIT(sdmac->event_id1 % 32);
if (sdmac->event_id1 > 31)
__set_bit(31, &sdmac->watermark_level);
sdmac->event_mask[0] = BIT(sdmac->event_id0 % 32);
if (sdmac->event_id0 > 31)
__set_bit(30, &sdmac->watermark_level);
} else {
__set_bit(sdmac->event_id0, sdmac->event_mask);
}
/* Watermark Level */
sdmac->watermark_level |= sdmac->watermark_level;
/* Address */
sdmac->shp_addr = sdmac->per_address;
} else {
sdmac->watermark_level = 0; /* FIXME: M3_BASE_ADDRESS */
}
ret = sdma_load_context(sdmac);
return ret;
}
static int sdma_set_channel_priority(struct sdma_channel *sdmac,
unsigned int priority)
{
struct sdma_engine *sdma = sdmac->sdma;
int channel = sdmac->channel;
if (priority < MXC_SDMA_MIN_PRIORITY
|| priority > MXC_SDMA_MAX_PRIORITY) {
return -EINVAL;
}
writel_relaxed(priority, sdma->regs + SDMA_CHNPRI_0 + 4 * channel);
return 0;
}
static int sdma_request_channel(struct sdma_channel *sdmac)
{
struct sdma_engine *sdma = sdmac->sdma;
int channel = sdmac->channel;
int ret = -EBUSY;
sdmac->bd = dma_zalloc_coherent(NULL, PAGE_SIZE, &sdmac->bd_phys,
GFP_KERNEL);
if (!sdmac->bd) {
ret = -ENOMEM;
goto out;
}
sdma->channel_control[channel].base_bd_ptr = sdmac->bd_phys;
sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys;
sdma_set_channel_priority(sdmac, MXC_SDMA_DEFAULT_PRIORITY);
return 0;
out:
return ret;
}
static struct sdma_channel *to_sdma_chan(struct dma_chan *chan)
{
return container_of(chan, struct sdma_channel, chan);
}
static dma_cookie_t sdma_tx_submit(struct dma_async_tx_descriptor *tx)
{
unsigned long flags;
struct sdma_channel *sdmac = to_sdma_chan(tx->chan);
dma_cookie_t cookie;
spin_lock_irqsave(&sdmac->lock, flags);
cookie = dma_cookie_assign(tx);
spin_unlock_irqrestore(&sdmac->lock, flags);
return cookie;
}
static int sdma_alloc_chan_resources(struct dma_chan *chan)
{
struct sdma_channel *sdmac = to_sdma_chan(chan);
struct imx_dma_data *data = chan->private;
int prio, ret;
if (!data)
return -EINVAL;
switch (data->priority) {
case DMA_PRIO_HIGH:
prio = 3;
break;
case DMA_PRIO_MEDIUM:
prio = 2;
break;
case DMA_PRIO_LOW:
default:
prio = 1;
break;
}
sdmac->peripheral_type = data->peripheral_type;
sdmac->event_id0 = data->dma_request;
clk_enable(sdmac->sdma->clk_ipg);
clk_enable(sdmac->sdma->clk_ahb);
ret = sdma_request_channel(sdmac);
if (ret)
return ret;
ret = sdma_set_channel_priority(sdmac, prio);
if (ret)
return ret;
dma_async_tx_descriptor_init(&sdmac->desc, chan);
sdmac->desc.tx_submit = sdma_tx_submit;
/* txd.flags will be overwritten in prep funcs */
sdmac->desc.flags = DMA_CTRL_ACK;
return 0;
}
static void sdma_free_chan_resources(struct dma_chan *chan)
{
struct sdma_channel *sdmac = to_sdma_chan(chan);
struct sdma_engine *sdma = sdmac->sdma;
sdma_disable_channel(sdmac);
if (sdmac->event_id0)
sdma_event_disable(sdmac, sdmac->event_id0);
if (sdmac->event_id1)
sdma_event_disable(sdmac, sdmac->event_id1);
sdmac->event_id0 = 0;
sdmac->event_id1 = 0;
sdma_set_channel_priority(sdmac, 0);
dma_free_coherent(NULL, PAGE_SIZE, sdmac->bd, sdmac->bd_phys);
clk_disable(sdma->clk_ipg);
clk_disable(sdma->clk_ahb);
}
static struct dma_async_tx_descriptor *sdma_prep_slave_sg(
struct dma_chan *chan, struct scatterlist *sgl,
unsigned int sg_len, enum dma_transfer_direction direction,
unsigned long flags, void *context)
{
struct sdma_channel *sdmac = to_sdma_chan(chan);
struct sdma_engine *sdma = sdmac->sdma;
int ret, i, count;
int channel = sdmac->channel;
struct scatterlist *sg;
if (sdmac->status == DMA_IN_PROGRESS)
return NULL;
sdmac->status = DMA_IN_PROGRESS;
sdmac->flags = 0;
sdmac->buf_tail = 0;
dev_dbg(sdma->dev, "setting up %d entries for channel %d.\n",
sg_len, channel);
sdmac->direction = direction;
ret = sdma_load_context(sdmac);
if (ret)
goto err_out;
if (sg_len > NUM_BD) {
dev_err(sdma->dev, "SDMA channel %d: maximum number of sg exceeded: %d > %d\n",
channel, sg_len, NUM_BD);
ret = -EINVAL;
goto err_out;
}
sdmac->chn_count = 0;
for_each_sg(sgl, sg, sg_len, i) {
struct sdma_buffer_descriptor *bd = &sdmac->bd[i];
int param;
bd->buffer_addr = sg->dma_address;
count = sg_dma_len(sg);
if (count > 0xffff) {
dev_err(sdma->dev, "SDMA channel %d: maximum bytes for sg entry exceeded: %d > %d\n",
channel, count, 0xffff);
ret = -EINVAL;
goto err_out;
}
bd->mode.count = count;
sdmac->chn_count += count;
if (sdmac->word_size > DMA_SLAVE_BUSWIDTH_4_BYTES) {
ret = -EINVAL;
goto err_out;
}
switch (sdmac->word_size) {
case DMA_SLAVE_BUSWIDTH_4_BYTES:
bd->mode.command = 0;
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 (i + 1 == sg_len) {
param |= BD_INTR;
param |= BD_LAST;
param &= ~BD_CONT;
}
dev_dbg(sdma->dev, "entry %d: count: %d dma: %#llx %s%s\n",
i, count, (u64)sg->dma_address,
param & BD_WRAP ? "wrap" : "",
param & BD_INTR ? " intr" : "");
bd->mode.status = param;
}
sdmac->num_bd = sg_len;
sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys;
return &sdmac->desc;
err_out:
sdmac->status = DMA_ERROR;
return NULL;
}
static struct dma_async_tx_descriptor *sdma_prep_dma_cyclic(
struct dma_chan *chan, dma_addr_t dma_addr, size_t buf_len,
size_t period_len, enum dma_transfer_direction direction,
unsigned long flags)
{
struct sdma_channel *sdmac = to_sdma_chan(chan);
struct sdma_engine *sdma = sdmac->sdma;
int num_periods = buf_len / period_len;
int channel = sdmac->channel;
int ret, i = 0, buf = 0;
dev_dbg(sdma->dev, "%s channel: %d\n", __func__, channel);
if (sdmac->status == DMA_IN_PROGRESS)
return NULL;
sdmac->status = DMA_IN_PROGRESS;
sdmac->buf_tail = 0;
Update imx-sdma cyclic handling to report residue I received a report this morning from one of the Novena developers that the behaviour of the iMX6 ASoC codec driver (using imx-pcm-dma.c) was sub-optimal under high system load. While there are issues relating to system load remaining, upon reviewing the ASoC imx-pcm-dma.c driver, it was noticed that it not using the residue support, because SDMA doesn't support it. This has the effect that SDMA has to make multiple calls into the ASoC and ALSA code, one for each period. Since ALSA's snd_pcm_elapsed() does not need to be called multiple times and it is entirely sufficient to call it once to update ALSA with the current buffer position via the pointer method, we can do better here. We can also avoid stopping the DMA entirely, just like real cyclic DMA implementations behave. While this means that we replay some old samples, this is a nicer behaviour than having audio stop and restart. The changes to achieve this are relatively minor - imx-sdma.c can track where the DMA is to the nearest descriptor boundary - it does this already when deciding how many callbacks to issue. In doing this, buf_tail always points at the descriptor which will complete next. The residue is defined by the bytes remaining to the end of the buffer, when the buffer is viewed as a single block of memory [start...end]. So, when we start out, there's a full buffer worth of residue, and this counts down as we approach the end of the buffer, eventually becoming zero at the end, before returning to the full buffer worth when we wrap back to the start. Moving the walking of the descriptors into the interrupt handler means that we can update the BD_DONE flag at interrupt time, thus avoiding a delayed tasklet stopping the cyclic DMA. This means that the residue can be calculated from (total descriptors - buf_tail) * descriptor size. This is what the change below does. We update imx-pcm-dma.c to remove the NO_RESIDUE flag since we now provide the residue. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Shawn Guo <shawn.guo@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2014-06-25 06:00:33 -06:00
sdmac->period_len = period_len;
sdmac->flags |= IMX_DMA_SG_LOOP;
sdmac->direction = direction;
ret = sdma_load_context(sdmac);
if (ret)
goto err_out;
if (num_periods > NUM_BD) {
dev_err(sdma->dev, "SDMA channel %d: maximum number of sg exceeded: %d > %d\n",
channel, num_periods, NUM_BD);
goto err_out;
}
if (period_len > 0xffff) {
dev_err(sdma->dev, "SDMA channel %d: maximum period size exceeded: %d > %d\n",
channel, period_len, 0xffff);
goto err_out;
}
while (buf < buf_len) {
struct sdma_buffer_descriptor *bd = &sdmac->bd[i];
int param;
bd->buffer_addr = dma_addr;
bd->mode.count = period_len;
if (sdmac->word_size > DMA_SLAVE_BUSWIDTH_4_BYTES)
goto err_out;
if (sdmac->word_size == DMA_SLAVE_BUSWIDTH_4_BYTES)
bd->mode.command = 0;
else
bd->mode.command = sdmac->word_size;
param = BD_DONE | BD_EXTD | BD_CONT | BD_INTR;
if (i + 1 == num_periods)
param |= BD_WRAP;
dev_dbg(sdma->dev, "entry %d: count: %d dma: %#llx %s%s\n",
i, period_len, (u64)dma_addr,
param & BD_WRAP ? "wrap" : "",
param & BD_INTR ? " intr" : "");
bd->mode.status = param;
dma_addr += period_len;
buf += period_len;
i++;
}
sdmac->num_bd = num_periods;
sdma->channel_control[channel].current_bd_ptr = sdmac->bd_phys;
return &sdmac->desc;
err_out:
sdmac->status = DMA_ERROR;
return NULL;
}
static int sdma_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd,
unsigned long arg)
{
struct sdma_channel *sdmac = to_sdma_chan(chan);
struct dma_slave_config *dmaengine_cfg = (void *)arg;
switch (cmd) {
case DMA_TERMINATE_ALL:
sdma_disable_channel(sdmac);
return 0;
case DMA_SLAVE_CONFIG:
if (dmaengine_cfg->direction == DMA_DEV_TO_MEM) {
sdmac->per_address = dmaengine_cfg->src_addr;
sdmac->watermark_level = dmaengine_cfg->src_maxburst *
dmaengine_cfg->src_addr_width;
sdmac->word_size = dmaengine_cfg->src_addr_width;
} else {
sdmac->per_address = dmaengine_cfg->dst_addr;
sdmac->watermark_level = dmaengine_cfg->dst_maxburst *
dmaengine_cfg->dst_addr_width;
sdmac->word_size = dmaengine_cfg->dst_addr_width;
}
sdmac->direction = dmaengine_cfg->direction;
return sdma_config_channel(sdmac);
default:
return -ENOSYS;
}
return -EINVAL;
}
static enum dma_status sdma_tx_status(struct dma_chan *chan,
dma_cookie_t cookie,
struct dma_tx_state *txstate)
{
struct sdma_channel *sdmac = to_sdma_chan(chan);
Update imx-sdma cyclic handling to report residue I received a report this morning from one of the Novena developers that the behaviour of the iMX6 ASoC codec driver (using imx-pcm-dma.c) was sub-optimal under high system load. While there are issues relating to system load remaining, upon reviewing the ASoC imx-pcm-dma.c driver, it was noticed that it not using the residue support, because SDMA doesn't support it. This has the effect that SDMA has to make multiple calls into the ASoC and ALSA code, one for each period. Since ALSA's snd_pcm_elapsed() does not need to be called multiple times and it is entirely sufficient to call it once to update ALSA with the current buffer position via the pointer method, we can do better here. We can also avoid stopping the DMA entirely, just like real cyclic DMA implementations behave. While this means that we replay some old samples, this is a nicer behaviour than having audio stop and restart. The changes to achieve this are relatively minor - imx-sdma.c can track where the DMA is to the nearest descriptor boundary - it does this already when deciding how many callbacks to issue. In doing this, buf_tail always points at the descriptor which will complete next. The residue is defined by the bytes remaining to the end of the buffer, when the buffer is viewed as a single block of memory [start...end]. So, when we start out, there's a full buffer worth of residue, and this counts down as we approach the end of the buffer, eventually becoming zero at the end, before returning to the full buffer worth when we wrap back to the start. Moving the walking of the descriptors into the interrupt handler means that we can update the BD_DONE flag at interrupt time, thus avoiding a delayed tasklet stopping the cyclic DMA. This means that the residue can be calculated from (total descriptors - buf_tail) * descriptor size. This is what the change below does. We update imx-pcm-dma.c to remove the NO_RESIDUE flag since we now provide the residue. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Shawn Guo <shawn.guo@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2014-06-25 06:00:33 -06:00
u32 residue;
if (sdmac->flags & IMX_DMA_SG_LOOP)
residue = (sdmac->num_bd - sdmac->buf_tail) * sdmac->period_len;
else
residue = sdmac->chn_count - sdmac->chn_real_count;
dma_set_tx_state(txstate, chan->completed_cookie, chan->cookie,
Update imx-sdma cyclic handling to report residue I received a report this morning from one of the Novena developers that the behaviour of the iMX6 ASoC codec driver (using imx-pcm-dma.c) was sub-optimal under high system load. While there are issues relating to system load remaining, upon reviewing the ASoC imx-pcm-dma.c driver, it was noticed that it not using the residue support, because SDMA doesn't support it. This has the effect that SDMA has to make multiple calls into the ASoC and ALSA code, one for each period. Since ALSA's snd_pcm_elapsed() does not need to be called multiple times and it is entirely sufficient to call it once to update ALSA with the current buffer position via the pointer method, we can do better here. We can also avoid stopping the DMA entirely, just like real cyclic DMA implementations behave. While this means that we replay some old samples, this is a nicer behaviour than having audio stop and restart. The changes to achieve this are relatively minor - imx-sdma.c can track where the DMA is to the nearest descriptor boundary - it does this already when deciding how many callbacks to issue. In doing this, buf_tail always points at the descriptor which will complete next. The residue is defined by the bytes remaining to the end of the buffer, when the buffer is viewed as a single block of memory [start...end]. So, when we start out, there's a full buffer worth of residue, and this counts down as we approach the end of the buffer, eventually becoming zero at the end, before returning to the full buffer worth when we wrap back to the start. Moving the walking of the descriptors into the interrupt handler means that we can update the BD_DONE flag at interrupt time, thus avoiding a delayed tasklet stopping the cyclic DMA. This means that the residue can be calculated from (total descriptors - buf_tail) * descriptor size. This is what the change below does. We update imx-pcm-dma.c to remove the NO_RESIDUE flag since we now provide the residue. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Shawn Guo <shawn.guo@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2014-06-25 06:00:33 -06:00
residue);
return sdmac->status;
}
static void sdma_issue_pending(struct dma_chan *chan)
{
struct sdma_channel *sdmac = to_sdma_chan(chan);
struct sdma_engine *sdma = sdmac->sdma;
if (sdmac->status == DMA_IN_PROGRESS)
sdma_enable_channel(sdma, sdmac->channel);
}
#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1 34
#define SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V2 38
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;
/* use the default firmware in ROM if missing external firmware */
if (!sdma->script_number)
sdma->script_number = SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1;
for (i = 0; i < sdma->script_number; i++)
if (addr_arr[i] > 0)
saddr_arr[i] = addr_arr[i];
}
static void sdma_load_firmware(const struct firmware *fw, void *context)
{
struct sdma_engine *sdma = context;
const struct sdma_firmware_header *header;
const struct sdma_script_start_addrs *addr;
unsigned short *ram_code;
if (!fw) {
dev_err(sdma->dev, "firmware not found\n");
return;
}
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;
switch (header->version_major) {
case 1:
sdma->script_number = SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1;
break;
case 2:
sdma->script_number = SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V2;
break;
default:
dev_err(sdma->dev, "unknown firmware version\n");
goto err_firmware;
}
addr = (void *)header + header->script_addrs_start;
ram_code = (void *)header + header->ram_code_start;
clk_enable(sdma->clk_ipg);
clk_enable(sdma->clk_ahb);
/* 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_ipg);
clk_disable(sdma->clk_ahb);
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);
}
static int sdma_get_firmware(struct sdma_engine *sdma,
const char *fw_name)
{
int ret;
ret = request_firmware_nowait(THIS_MODULE,
FW_ACTION_HOTPLUG, fw_name, sdma->dev,
GFP_KERNEL, sdma, sdma_load_firmware);
return ret;
}
static int sdma_init(struct sdma_engine *sdma)
{
int i, ret;
dma_addr_t ccb_phys;
clk_enable(sdma->clk_ipg);
clk_enable(sdma->clk_ahb);
/* Be sure SDMA has not started yet */
writel_relaxed(0, sdma->regs + SDMA_H_C0PTR);
sdma->channel_control = dma_alloc_coherent(NULL,
MAX_DMA_CHANNELS * sizeof (struct sdma_channel_control) +
sizeof(struct sdma_context_data),
&ccb_phys, GFP_KERNEL);
if (!sdma->channel_control) {
ret = -ENOMEM;
goto err_dma_alloc;
}
sdma->context = (void *)sdma->channel_control +
MAX_DMA_CHANNELS * sizeof (struct sdma_channel_control);
sdma->context_phys = ccb_phys +
MAX_DMA_CHANNELS * sizeof (struct sdma_channel_control);
/* Zero-out the CCB structures array just allocated */
memset(sdma->channel_control, 0,
MAX_DMA_CHANNELS * sizeof (struct sdma_channel_control));
/* disable all channels */
for (i = 0; i < sdma->drvdata->num_events; i++)
writel_relaxed(0, sdma->regs + chnenbl_ofs(sdma, i));
/* All channels have priority 0 */
for (i = 0; i < MAX_DMA_CHANNELS; i++)
writel_relaxed(0, sdma->regs + SDMA_CHNPRI_0 + i * 4);
ret = sdma_request_channel(&sdma->channel[0]);
if (ret)
goto err_dma_alloc;
sdma_config_ownership(&sdma->channel[0], false, true, false);
/* Set Command Channel (Channel Zero) */
writel_relaxed(0x4050, sdma->regs + SDMA_CHN0ADDR);
/* Set bits of CONFIG register but with static context switching */
/* FIXME: Check whether to set ACR bit depending on clock ratios */
writel_relaxed(0, sdma->regs + SDMA_H_CONFIG);
writel_relaxed(ccb_phys, sdma->regs + SDMA_H_C0PTR);
/* Set bits of CONFIG register with given context switching mode */
writel_relaxed(SDMA_H_CONFIG_CSM, sdma->regs + SDMA_H_CONFIG);
/* Initializes channel's priorities */
sdma_set_channel_priority(&sdma->channel[0], 7);
clk_disable(sdma->clk_ipg);
clk_disable(sdma->clk_ahb);
return 0;
err_dma_alloc:
clk_disable(sdma->clk_ipg);
clk_disable(sdma->clk_ahb);
dev_err(sdma->dev, "initialisation failed with %d\n", ret);
return ret;
}
static bool sdma_filter_fn(struct dma_chan *chan, void *fn_param)
{
struct sdma_channel *sdmac = to_sdma_chan(chan);
struct imx_dma_data *data = fn_param;
if (!imx_dma_is_general_purpose(chan))
return false;
sdmac->data = *data;
chan->private = &sdmac->data;
return true;
}
static struct dma_chan *sdma_xlate(struct of_phandle_args *dma_spec,
struct of_dma *ofdma)
{
struct sdma_engine *sdma = ofdma->of_dma_data;
dma_cap_mask_t mask = sdma->dma_device.cap_mask;
struct imx_dma_data data;
if (dma_spec->args_count != 3)
return NULL;
data.dma_request = dma_spec->args[0];
data.peripheral_type = dma_spec->args[1];
data.priority = dma_spec->args[2];
return dma_request_channel(mask, sdma_filter_fn, &data);
}
static int sdma_probe(struct platform_device *pdev)
{
const struct of_device_id *of_id =
of_match_device(sdma_dt_ids, &pdev->dev);
struct device_node *np = pdev->dev.of_node;
const char *fw_name;
int ret;
int irq;
struct resource *iores;
struct sdma_platform_data *pdata = dev_get_platdata(&pdev->dev);
int i;
struct sdma_engine *sdma;
s32 *saddr_arr;
const struct sdma_driver_data *drvdata = NULL;
if (of_id)
drvdata = of_id->data;
else if (pdev->id_entry)
drvdata = (void *)pdev->id_entry->driver_data;
if (!drvdata) {
dev_err(&pdev->dev, "unable to find driver data\n");
return -EINVAL;
}
ret = dma_coerce_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
if (ret)
return ret;
sdma = kzalloc(sizeof(*sdma), GFP_KERNEL);
if (!sdma)
return -ENOMEM;
spin_lock_init(&sdma->channel_0_lock);
sdma->dev = &pdev->dev;
sdma->drvdata = drvdata;
iores = platform_get_resource(pdev, IORESOURCE_MEM, 0);
irq = platform_get_irq(pdev, 0);
if (!iores || irq < 0) {
ret = -EINVAL;
goto err_irq;
}
if (!request_mem_region(iores->start, resource_size(iores), pdev->name)) {
ret = -EBUSY;
goto err_request_region;
}
sdma->clk_ipg = devm_clk_get(&pdev->dev, "ipg");
if (IS_ERR(sdma->clk_ipg)) {
ret = PTR_ERR(sdma->clk_ipg);
goto err_clk;
}
sdma->clk_ahb = devm_clk_get(&pdev->dev, "ahb");
if (IS_ERR(sdma->clk_ahb)) {
ret = PTR_ERR(sdma->clk_ahb);
goto err_clk;
}
clk_prepare(sdma->clk_ipg);
clk_prepare(sdma->clk_ahb);
sdma->regs = ioremap(iores->start, resource_size(iores));
if (!sdma->regs) {
ret = -ENOMEM;
goto err_ioremap;
}
ret = request_irq(irq, sdma_int_handler, 0, "sdma", sdma);
if (ret)
goto err_request_irq;
sdma->script_addrs = kzalloc(sizeof(*sdma->script_addrs), GFP_KERNEL);
if (!sdma->script_addrs) {
ret = -ENOMEM;
goto err_alloc;
}
/* initially no scripts available */
saddr_arr = (s32 *)sdma->script_addrs;
for (i = 0; i < SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V1; i++)
saddr_arr[i] = -EINVAL;
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++) {
struct sdma_channel *sdmac = &sdma->channel[i];
sdmac->sdma = sdma;
spin_lock_init(&sdmac->lock);
sdmac->chan.device = &sdma->dma_device;
dma_cookie_init(&sdmac->chan);
sdmac->channel = i;
tasklet_init(&sdmac->tasklet, sdma_tasklet,
(unsigned long) sdmac);
/*
* 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);
if (ret)
goto err_init;
if (sdma->drvdata->script_addrs)
sdma_add_scripts(sdma, sdma->drvdata->script_addrs);
if (pdata && pdata->script_addrs)
sdma_add_scripts(sdma, pdata->script_addrs);
if (pdata) {
ret = sdma_get_firmware(sdma, pdata->fw_name);
if (ret)
dev_warn(&pdev->dev, "failed to get firmware from platform data\n");
} else {
/*
* Because that device tree does not encode ROM script address,
* the RAM script in firmware is mandatory for device tree
* probe, otherwise it fails.
*/
ret = of_property_read_string(np, "fsl,sdma-ram-script-name",
&fw_name);
if (ret)
dev_warn(&pdev->dev, "failed to get firmware name\n");
else {
ret = sdma_get_firmware(sdma, fw_name);
if (ret)
dev_warn(&pdev->dev, "failed to get firmware from device tree\n");
}
}
sdma->dma_device.dev = &pdev->dev;
sdma->dma_device.device_alloc_chan_resources = sdma_alloc_chan_resources;
sdma->dma_device.device_free_chan_resources = sdma_free_chan_resources;
sdma->dma_device.device_tx_status = sdma_tx_status;
sdma->dma_device.device_prep_slave_sg = sdma_prep_slave_sg;
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);
platform_set_drvdata(pdev, sdma);
ret = dma_async_device_register(&sdma->dma_device);
if (ret) {
dev_err(&pdev->dev, "unable to register\n");
goto err_init;
}
if (np) {
ret = of_dma_controller_register(np, sdma_xlate, sdma);
if (ret) {
dev_err(&pdev->dev, "failed to register controller\n");
goto err_register;
}
}
dev_info(sdma->dev, "initialized\n");
return 0;
err_register:
dma_async_device_unregister(&sdma->dma_device);
err_init:
kfree(sdma->script_addrs);
err_alloc:
free_irq(irq, sdma);
err_request_irq:
iounmap(sdma->regs);
err_ioremap:
err_clk:
release_mem_region(iores->start, resource_size(iores));
err_request_region:
err_irq:
kfree(sdma);
return ret;
}
static int sdma_remove(struct platform_device *pdev)
{
struct sdma_engine *sdma = platform_get_drvdata(pdev);
struct resource *iores = platform_get_resource(pdev, IORESOURCE_MEM, 0);
int irq = platform_get_irq(pdev, 0);
int i;
dma_async_device_unregister(&sdma->dma_device);
kfree(sdma->script_addrs);
free_irq(irq, sdma);
iounmap(sdma->regs);
release_mem_region(iores->start, resource_size(iores));
/* Kill the tasklet */
for (i = 0; i < MAX_DMA_CHANNELS; i++) {
struct sdma_channel *sdmac = &sdma->channel[i];
tasklet_kill(&sdmac->tasklet);
}
kfree(sdma);
platform_set_drvdata(pdev, NULL);
dev_info(&pdev->dev, "Removed...\n");
return 0;
}
static struct platform_driver sdma_driver = {
.driver = {
.name = "imx-sdma",
.of_match_table = sdma_dt_ids,
},
.id_table = sdma_devtypes,
.remove = sdma_remove,
.probe = sdma_probe,
};
module_platform_driver(sdma_driver);
MODULE_AUTHOR("Sascha Hauer, Pengutronix <s.hauer@pengutronix.de>");
MODULE_DESCRIPTION("i.MX SDMA driver");
MODULE_LICENSE("GPL");