// SPDX-License-Identifier: GPL-2.0
// Copyright (C) 2019 Spreadtrum Communications Inc.

#include <linux/dma-mapping.h>
#include <linux/dmaengine.h>
#include <linux/dma/sprd-dma.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <sound/pcm.h>
#include <sound/pcm_params.h>
#include <sound/soc.h>
#include <sound/compress_driver.h>

#include "sprd-pcm-dma.h"

#define SPRD_COMPR_DMA_CHANS            2

/* Default values if userspace does not set */
#define SPRD_COMPR_MIN_FRAGMENT_SIZE    SZ_8K
#define SPRD_COMPR_MAX_FRAGMENT_SIZE    SZ_128K
#define SPRD_COMPR_MIN_NUM_FRAGMENTS    4
#define SPRD_COMPR_MAX_NUM_FRAGMENTS    64

/* DSP FIFO size */
#define SPRD_COMPR_MCDT_EMPTY_WMK       0
#define SPRD_COMPR_MCDT_FIFO_SIZE       512

/* Stage 0 IRAM buffer size definition */
#define SPRD_COMPR_IRAM_BUF_SIZE        SZ_32K
#define SPRD_COMPR_IRAM_INFO_SIZE       (sizeof(struct sprd_compr_playinfo))
#define SPRD_COMPR_IRAM_LINKLIST_SIZE   (1024 - SPRD_COMPR_IRAM_INFO_SIZE)
#define SPRD_COMPR_IRAM_SIZE            (SPRD_COMPR_IRAM_BUF_SIZE + \
                                         SPRD_COMPR_IRAM_INFO_SIZE + \
                                         SPRD_COMPR_IRAM_LINKLIST_SIZE)

/* Stage 1 DDR buffer size definition */
#define SPRD_COMPR_AREA_BUF_SIZE        SZ_2M
#define SPRD_COMPR_AREA_LINKLIST_SIZE   1024
#define SPRD_COMPR_AREA_SIZE            (SPRD_COMPR_AREA_BUF_SIZE + \
                                         SPRD_COMPR_AREA_LINKLIST_SIZE)

struct sprd_compr_dma {
        struct dma_chan *chan;
        struct dma_async_tx_descriptor *desc;
        dma_cookie_t cookie;
        dma_addr_t phys;
        void *virt;
        int trans_len;
};

/*
 * The Spreadtrum Audio compress offload mode will use 2-stage DMA transfer to
 * save power. That means we can request 2 dma channels, one for source channel,
 * and another one for destination channel. Once the source channel's transaction
 * is done, it will trigger the destination channel's transaction automatically
 * by hardware signal.
 *
 * For 2-stage DMA transfer, we can allocate 2 buffers: IRAM buffer (always
 * power-on) and DDR buffer. The source channel will transfer data from IRAM
 * buffer to the DSP fifo to decoding/encoding, once IRAM buffer is empty by
 * transferring done, the destination channel will start to transfer data from
 * DDR buffer to IRAM buffer.
 *
 * Since the DSP fifo is only 512B, IRAM buffer is allocated by 32K, and DDR
 * buffer is larger to 2M. That means only the IRAM 32k data is transferred
 * done, we can wake up the AP system to transfer data from DDR to IRAM, and
 * other time the AP system can be suspended to save power.
 */
struct sprd_compr_stream {
        struct snd_compr_stream *cstream;
        struct sprd_compr_ops *compr_ops;
        struct sprd_compr_dma dma[SPRD_COMPR_DMA_CHANS];

        /* DMA engine channel number */
        int num_channels;

        /* Stage 0 IRAM buffer */
        struct snd_dma_buffer iram_buffer;
        /* Stage 1 DDR buffer */
        struct snd_dma_buffer compr_buffer;

        /* DSP play information IRAM buffer */
        dma_addr_t info_phys;
        void *info_area;
        int info_size;

        /* Data size copied to IRAM buffer */
        int copied_total;
        /* Total received data size from userspace */
        int received_total;
        /* Stage 0 IRAM buffer received data size */
        int received_stage0;
        /* Stage 1 DDR buffer received data size */
        int received_stage1;
        /* Stage 1 DDR buffer pointer */
        int stage1_pointer;
};

static int sprd_platform_compr_trigger(struct snd_compr_stream *cstream,
                                       int cmd);

static void sprd_platform_compr_drain_notify(void *arg)
{
        struct snd_compr_stream *cstream = arg;
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct sprd_compr_stream *stream = runtime->private_data;

        memset(stream->info_area, 0, sizeof(struct sprd_compr_playinfo));

        snd_compr_drain_notify(cstream);
}

static void sprd_platform_compr_dma_complete(void *data)
{
        struct snd_compr_stream *cstream = data;
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct sprd_compr_stream *stream = runtime->private_data;
        struct sprd_compr_dma *dma = &stream->dma[1];

        /* Update data size copied to IRAM buffer */
        stream->copied_total += dma->trans_len;
        if (stream->copied_total > stream->received_total)
                stream->copied_total = stream->received_total;

        snd_compr_fragment_elapsed(cstream);
}

static int sprd_platform_compr_dma_config(struct snd_compr_stream *cstream,
                                          struct snd_compr_params *params,
                                          int channel)
{
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct sprd_compr_stream *stream = runtime->private_data;
        struct snd_soc_pcm_runtime *rtd = cstream->private_data;
        struct snd_soc_component *component =
                snd_soc_rtdcom_lookup(rtd, DRV_NAME);
        struct device *dev = component->dev;
        struct sprd_compr_data *data = snd_soc_dai_get_drvdata(rtd->cpu_dai);
        struct sprd_pcm_dma_params *dma_params = data->dma_params;
        struct sprd_compr_dma *dma = &stream->dma[channel];
        struct dma_slave_config config = { };
        struct sprd_dma_linklist link = { };
        enum dma_transfer_direction dir;
        struct scatterlist *sg, *sgt;
        enum dma_slave_buswidth bus_width;
        int period, period_cnt, sg_num = 2;
        dma_addr_t src_addr, dst_addr;
        unsigned long flags;
        int ret, j;

        if (!dma_params) {
                dev_err(dev, "no dma parameters setting\n");
                return -EINVAL;
        }

        dma->chan = dma_request_slave_channel(dev,
                                              dma_params->chan_name[channel]);
        if (!dma->chan) {
                dev_err(dev, "failed to request dma channel\n");
                return -ENODEV;
        }

        sgt = sg = devm_kcalloc(dev, sg_num, sizeof(*sg), GFP_KERNEL);
        if (!sg) {
                ret = -ENOMEM;
                goto sg_err;
        }

        switch (channel) {
        case 0:
                bus_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
                period = (SPRD_COMPR_MCDT_FIFO_SIZE - SPRD_COMPR_MCDT_EMPTY_WMK) * 4;
                period_cnt = params->buffer.fragment_size / period;
                src_addr = stream->iram_buffer.addr;
                dst_addr = dma_params->dev_phys[channel];
                flags = SPRD_DMA_FLAGS(SPRD_DMA_SRC_CHN1,
                                       SPRD_DMA_TRANS_DONE_TRG,
                                       SPRD_DMA_FRAG_REQ,
                                       SPRD_DMA_TRANS_INT);
                break;

        case 1:
                bus_width = DMA_SLAVE_BUSWIDTH_2_BYTES;
                period = params->buffer.fragment_size;
                period_cnt = params->buffer.fragments;
                src_addr = stream->compr_buffer.addr;
                dst_addr = stream->iram_buffer.addr;
                flags = SPRD_DMA_FLAGS(SPRD_DMA_DST_CHN1,
                                       SPRD_DMA_TRANS_DONE_TRG,
                                       SPRD_DMA_FRAG_REQ,
                                       SPRD_DMA_TRANS_INT);
                break;

        default:
                ret = -EINVAL;
                goto config_err;
        }

        dma->trans_len = period * period_cnt;

        config.src_maxburst = period;
        config.src_addr_width = bus_width;
        config.dst_addr_width = bus_width;
        if (cstream->direction == SND_COMPRESS_PLAYBACK) {
                config.src_addr = src_addr;
                config.dst_addr = dst_addr;
                dir = DMA_MEM_TO_DEV;
        } else {
                config.src_addr = dst_addr;
                config.dst_addr = src_addr;
                dir = DMA_DEV_TO_MEM;
        }

        sg_init_table(sgt, sg_num);
        for (j = 0; j < sg_num; j++, sgt++) {
                sg_dma_len(sgt) = dma->trans_len;
                sg_dma_address(sgt) = dst_addr;
        }

        /*
         * Configure the link-list address for the DMA engine link-list
         * mode.
         */
        link.virt_addr = (unsigned long)dma->virt;
        link.phy_addr = dma->phys;

        ret = dmaengine_slave_config(dma->chan, &config);
        if (ret) {
                dev_err(dev,
                        "failed to set slave configuration: %d\n", ret);
                goto config_err;
        }

        /*
         * We configure the DMA request mode, interrupt mode, channel
         * mode and channel trigger mode by the flags.
         */
        dma->desc = dma->chan->device->device_prep_slave_sg(dma->chan, sg,
                                                            sg_num, dir,
                                                            flags, &link);
        if (!dma->desc) {
                dev_err(dev, "failed to prepare slave sg\n");
                ret = -ENOMEM;
                goto config_err;
        }

        /* Only channel 1 transfer can wake up the AP system. */
        if (!params->no_wake_mode && channel == 1) {
                dma->desc->callback = sprd_platform_compr_dma_complete;
                dma->desc->callback_param = cstream;
        }

        devm_kfree(dev, sg);

        return 0;

config_err:
        devm_kfree(dev, sg);
sg_err:
        dma_release_channel(dma->chan);
        return ret;
}

static int sprd_platform_compr_set_params(struct snd_compr_stream *cstream,
                                          struct snd_compr_params *params)
{
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct sprd_compr_stream *stream = runtime->private_data;
        struct snd_soc_pcm_runtime *rtd = cstream->private_data;
        struct snd_soc_component *component =
                snd_soc_rtdcom_lookup(rtd, DRV_NAME);
        struct device *dev = component->dev;
        struct sprd_compr_params compr_params = { };
        int ret;

        /*
         * Configure the DMA engine 2-stage transfer mode. Channel 1 set as the
         * destination channel, and channel 0 set as the source channel, that
         * means once the source channel's transaction is done, it will trigger
         * the destination channel's transaction automatically.
         */
        ret = sprd_platform_compr_dma_config(cstream, params, 1);
        if (ret) {
                dev_err(dev, "failed to config stage 1 DMA: %d\n", ret);
                return ret;
        }

        ret = sprd_platform_compr_dma_config(cstream, params, 0);
        if (ret) {
                dev_err(dev, "failed to config stage 0 DMA: %d\n", ret);
                goto config_err;
        }

        compr_params.direction = cstream->direction;
        compr_params.sample_rate = params->codec.sample_rate;
        compr_params.channels = stream->num_channels;
        compr_params.info_phys = stream->info_phys;
        compr_params.info_size = stream->info_size;
        compr_params.rate = params->codec.bit_rate;
        compr_params.format = params->codec.id;

        ret = stream->compr_ops->set_params(cstream->direction, &compr_params);
        if (ret) {
                dev_err(dev, "failed to set parameters: %d\n", ret);
                goto params_err;
        }

        return 0;

params_err:
        dma_release_channel(stream->dma[0].chan);
config_err:
        dma_release_channel(stream->dma[1].chan);
        return ret;
}

static int sprd_platform_compr_open(struct snd_compr_stream *cstream)
{
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct snd_soc_pcm_runtime *rtd = cstream->private_data;
        struct snd_soc_component *component =
                snd_soc_rtdcom_lookup(rtd, DRV_NAME);
        struct device *dev = component->dev;
        struct sprd_compr_data *data = snd_soc_dai_get_drvdata(rtd->cpu_dai);
        struct sprd_compr_stream *stream;
        struct sprd_compr_callback cb;
        int stream_id = cstream->direction, ret;

        ret = dma_coerce_mask_and_coherent(dev, DMA_BIT_MASK(32));
        if (ret)
                return ret;

        stream = devm_kzalloc(dev, sizeof(*stream), GFP_KERNEL);
        if (!stream)
                return -ENOMEM;

        stream->cstream = cstream;
        stream->num_channels = 2;
        stream->compr_ops = data->ops;

        /*
         * Allocate the stage 0 IRAM buffer size, including the DMA 0
         * link-list size and play information of DSP address size.
         */
        ret = snd_dma_alloc_pages(SNDRV_DMA_TYPE_DEV_IRAM, dev,
                                  SPRD_COMPR_IRAM_SIZE, &stream->iram_buffer);
        if (ret < 0)
                goto err_iram;

        /* Use to save link-list configuration for DMA 0. */
        stream->dma[0].virt = stream->iram_buffer.area + SPRD_COMPR_IRAM_SIZE;
        stream->dma[0].phys = stream->iram_buffer.addr + SPRD_COMPR_IRAM_SIZE;

        /* Use to update the current data offset of DSP. */
        stream->info_phys = stream->iram_buffer.addr + SPRD_COMPR_IRAM_SIZE +
                SPRD_COMPR_IRAM_LINKLIST_SIZE;
        stream->info_area = stream->iram_buffer.area + SPRD_COMPR_IRAM_SIZE +
                SPRD_COMPR_IRAM_LINKLIST_SIZE;
        stream->info_size = SPRD_COMPR_IRAM_INFO_SIZE;

        /*
         * Allocate the stage 1 DDR buffer size, including the DMA 1 link-list
         * size.
         */
        ret = snd_dma_alloc_pages(SNDRV_DMA_TYPE_DEV, dev,
                                  SPRD_COMPR_AREA_SIZE, &stream->compr_buffer);
        if (ret < 0)
                goto err_compr;

        /* Use to save link-list configuration for DMA 1. */
        stream->dma[1].virt = stream->compr_buffer.area + SPRD_COMPR_AREA_SIZE;
        stream->dma[1].phys = stream->compr_buffer.addr + SPRD_COMPR_AREA_SIZE;

        cb.drain_notify = sprd_platform_compr_drain_notify;
        cb.drain_data = cstream;
        ret = stream->compr_ops->open(stream_id, &cb);
        if (ret) {
                dev_err(dev, "failed to open compress platform: %d\n", ret);
                goto err_open;
        }

        runtime->private_data = stream;
        return 0;

err_open:
        snd_dma_free_pages(&stream->compr_buffer);
err_compr:
        snd_dma_free_pages(&stream->iram_buffer);
err_iram:
        devm_kfree(dev, stream);

        return ret;
}

static int sprd_platform_compr_free(struct snd_compr_stream *cstream)
{
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct sprd_compr_stream *stream = runtime->private_data;
        struct snd_soc_pcm_runtime *rtd = cstream->private_data;
        struct snd_soc_component *component =
                snd_soc_rtdcom_lookup(rtd, DRV_NAME);
        struct device *dev = component->dev;
        int stream_id = cstream->direction, i;

        for (i = 0; i < stream->num_channels; i++) {
                struct sprd_compr_dma *dma = &stream->dma[i];

                if (dma->chan) {
                        dma_release_channel(dma->chan);
                        dma->chan = NULL;
                }
        }

        snd_dma_free_pages(&stream->compr_buffer);
        snd_dma_free_pages(&stream->iram_buffer);

        stream->compr_ops->close(stream_id);

        devm_kfree(dev, stream);
        return 0;
}

static int sprd_platform_compr_trigger(struct snd_compr_stream *cstream,
                                       int cmd)
{
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct sprd_compr_stream *stream = runtime->private_data;
        struct snd_soc_pcm_runtime *rtd = cstream->private_data;
        struct snd_soc_component *component =
                snd_soc_rtdcom_lookup(rtd, DRV_NAME);
        struct device *dev = component->dev;
        int channels = stream->num_channels, ret = 0, i;
        int stream_id = cstream->direction;

        if (cstream->direction != SND_COMPRESS_PLAYBACK) {
                dev_err(dev, "unsupported compress direction\n");
                return -EINVAL;
        }

        switch (cmd) {
        case SNDRV_PCM_TRIGGER_START:
                for (i = channels - 1; i >= 0; i--) {
                        struct sprd_compr_dma *dma = &stream->dma[i];

                        if (!dma->desc)
                                continue;

                        dma->cookie = dmaengine_submit(dma->desc);
                        ret = dma_submit_error(dma->cookie);
                        if (ret) {
                                dev_err(dev, "failed to submit request: %d\n",
                                        ret);
                                return ret;
                        }
                }

                for (i = channels - 1; i >= 0; i--) {
                        struct sprd_compr_dma *dma = &stream->dma[i];

                        if (dma->chan)
                                dma_async_issue_pending(dma->chan);
                }

                ret = stream->compr_ops->start(stream_id);
                break;

        case SNDRV_PCM_TRIGGER_STOP:
                for (i = channels - 1; i >= 0; i--) {
                        struct sprd_compr_dma *dma = &stream->dma[i];

                        if (dma->chan)
                                dmaengine_terminate_async(dma->chan);
                }

                stream->copied_total = 0;
                stream->stage1_pointer  = 0;
                stream->received_total = 0;
                stream->received_stage0 = 0;
                stream->received_stage1 = 0;

                ret = stream->compr_ops->stop(stream_id);
                break;

        case SNDRV_PCM_TRIGGER_SUSPEND:
        case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
                for (i = channels - 1; i >= 0; i--) {
                        struct sprd_compr_dma *dma = &stream->dma[i];

                        if (dma->chan)
                                dmaengine_pause(dma->chan);
                }

                ret = stream->compr_ops->pause(stream_id);
                break;

        case SNDRV_PCM_TRIGGER_RESUME:
        case SNDRV_PCM_TRIGGER_PAUSE_RELEASE:
                for (i = channels - 1; i >= 0; i--) {
                        struct sprd_compr_dma *dma = &stream->dma[i];

                        if (dma->chan)
                                dmaengine_resume(dma->chan);
                }

                ret = stream->compr_ops->pause_release(stream_id);
                break;

        case SND_COMPR_TRIGGER_PARTIAL_DRAIN:
        case SND_COMPR_TRIGGER_DRAIN:
                ret = stream->compr_ops->drain(stream->received_total);
                break;

        default:
                ret = -EINVAL;
                break;
        }

        return ret;
}

static int sprd_platform_compr_pointer(struct snd_compr_stream *cstream,
                                       struct snd_compr_tstamp *tstamp)
{
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct sprd_compr_stream *stream = runtime->private_data;
        struct sprd_compr_playinfo *info =
                (struct sprd_compr_playinfo *)stream->info_area;

        tstamp->copied_total = stream->copied_total;
        tstamp->pcm_io_frames = info->current_data_offset;

        return 0;
}

static int sprd_platform_compr_copy(struct snd_compr_stream *cstream,
                                    char __user *buf, size_t count)
{
        struct snd_compr_runtime *runtime = cstream->runtime;
        struct sprd_compr_stream *stream = runtime->private_data;
        int avail_bytes, data_count = count;
        void *dst;

        /*
         * We usually set fragment size as 32K, and the stage 0 IRAM buffer
         * size is 32K too. So if now the received data size of the stage 0
         * IRAM buffer is less than 32K, that means we have some available
         * spaces for the stage 0 IRAM buffer.
         */
        if (stream->received_stage0 < runtime->fragment_size) {
                avail_bytes = runtime->fragment_size - stream->received_stage0;
                dst = stream->iram_buffer.area + stream->received_stage0;

                if (avail_bytes >= data_count) {
                        /*
                         * Copy data to the stage 0 IRAM buffer directly if
                         * spaces are enough.
                         */
                        if (copy_from_user(dst, buf, data_count))
                                return -EFAULT;

                        stream->received_stage0 += data_count;
                        stream->copied_total += data_count;
                        goto copy_done;
                } else {
                        /*
                         * If the data count is larger than the available spaces
                         * of the the stage 0 IRAM buffer, we should copy one
                         * partial data to the stage 0 IRAM buffer, and copy
                         * the left to the stage 1 DDR buffer.
                         */
                        if (copy_from_user(dst, buf, avail_bytes))
                                return -EFAULT;

                        data_count -= avail_bytes;
                        stream->received_stage0 += avail_bytes;
                        stream->copied_total += avail_bytes;
                        buf += avail_bytes;
                }
        }

        /*
         * Copy data to the stage 1 DDR buffer if no spaces for the stage 0 IRAM
         * buffer.
         */
        dst = stream->compr_buffer.area + stream->stage1_pointer;
        if (data_count < stream->compr_buffer.bytes - stream->stage1_pointer) {
                if (copy_from_user(dst, buf, data_count))
                        return -EFAULT;

                stream->stage1_pointer += data_count;
        } else {
                avail_bytes = stream->compr_buffer.bytes - stream->stage1_pointer;

                if (copy_from_user(dst, buf, avail_bytes))
                        return -EFAULT;

                if (copy_from_user(stream->compr_buffer.area, buf + avail_bytes,
                                   data_count - avail_bytes))
                        return -EFAULT;

                stream->stage1_pointer = data_count - avail_bytes;
        }

        stream->received_stage1 += data_count;

copy_done:
        /* Update the copied data size. */
        stream->received_total += count;
        return count;
}

static int sprd_platform_compr_get_caps(struct snd_compr_stream *cstream,
                                        struct snd_compr_caps *caps)
{
        caps->direction = cstream->direction;
        caps->min_fragment_size = SPRD_COMPR_MIN_FRAGMENT_SIZE;
        caps->max_fragment_size = SPRD_COMPR_MAX_FRAGMENT_SIZE;
        caps->min_fragments = SPRD_COMPR_MIN_NUM_FRAGMENTS;
        caps->max_fragments = SPRD_COMPR_MAX_NUM_FRAGMENTS;
        caps->num_codecs = 2;
        caps->codecs[0] = SND_AUDIOCODEC_MP3;
        caps->codecs[1] = SND_AUDIOCODEC_AAC;

        return 0;
}

static int
sprd_platform_compr_get_codec_caps(struct snd_compr_stream *cstream,
                                   struct snd_compr_codec_caps *codec)
{
        switch (codec->codec) {
        case SND_AUDIOCODEC_MP3:
                codec->num_descriptors = 2;
                codec->descriptor[0].max_ch = 2;
                codec->descriptor[0].bit_rate[0] = 320;
                codec->descriptor[0].bit_rate[1] = 128;
                codec->descriptor[0].num_bitrates = 2;
                codec->descriptor[0].profiles = 0;
                codec->descriptor[0].modes = SND_AUDIOCHANMODE_MP3_STEREO;
                codec->descriptor[0].formats = 0;
                break;

        case SND_AUDIOCODEC_AAC:
                codec->num_descriptors = 2;
                codec->descriptor[1].max_ch = 2;
                codec->descriptor[1].bit_rate[0] = 320;
                codec->descriptor[1].bit_rate[1] = 128;
                codec->descriptor[1].num_bitrates = 2;
                codec->descriptor[1].profiles = 0;
                codec->descriptor[1].modes = 0;
                codec->descriptor[1].formats = 0;
                break;

        default:
                return -EINVAL;
        }

        return 0;
}

const struct snd_compr_ops sprd_platform_compr_ops = {
        .open = sprd_platform_compr_open,
        .free = sprd_platform_compr_free,
        .set_params = sprd_platform_compr_set_params,
        .trigger = sprd_platform_compr_trigger,
        .pointer = sprd_platform_compr_pointer,
        .copy = sprd_platform_compr_copy,
        .get_caps = sprd_platform_compr_get_caps,
        .get_codec_caps = sprd_platform_compr_get_codec_caps,
};

MODULE_DESCRIPTION("Spreadtrum ASoC Compress Platform Driver");
MODULE_LICENSE("GPL v2");
MODULE_ALIAS("platform:compress-platform");