[PATCH v5 16/21] ASoC: qdsp6: audioreach: add module configuration command helpers
Srinivas Kandagatla
srinivas.kandagatla at linaro.org
Mon Sep 6 18:29:34 CEST 2021
thanks Pierre for taking time to review the patches.
On 03/09/2021 16:13, Pierre-Louis Bossart wrote:
>
>
> On 9/3/21 6:20 AM, Srinivas Kandagatla wrote:
>> Audioreach module configuration helpers, which will be used by
>> the q6apm-dai driver.
>>
>> Also add Kconfig and Makefile changes as now all the code for audioreach
>> and q6apm are in at this point.
>
> separate patches?
>
>
Sure I will do that in next spin.
>> +int audioreach_graph_send_cmd_sync(struct q6apm_graph *graph,
>> + struct gpr_pkt *pkt, uint32_t rsp_opcode)
>> +{
>> +
>> + struct device *dev = graph->dev;
>> + struct gpr_hdr *hdr = &pkt->hdr;
>> + int rc;
>> +
>> + mutex_lock(&graph->cmd_lock);
>> + graph->result.opcode = 0;
>> + graph->result.status = 0;
>> +
>> + rc = gpr_send_port_pkt(graph->port, pkt);
>> + if (rc < 0)
>> + goto err;
>> +
>> + if (rsp_opcode)
>> + rc = wait_event_timeout(graph->cmd_wait,
>> + (graph->result.opcode == hdr->opcode) ||
>> + (graph->result.opcode == rsp_opcode),
>> + 5 * HZ);
>> + else
>> + rc = wait_event_timeout(graph->cmd_wait,
>> + (graph->result.opcode == hdr->opcode),
>> + 5 * HZ);
>> +
>> + if (!rc) {
>> + dev_err(dev, "CMD timeout for [%x] opcode\n", hdr->opcode);
>> + rc = -ETIMEDOUT;
>> + } else if (graph->result.status > 0) {
>> + dev_err(dev, "DSP returned error[%x] %x\n", hdr->opcode,
>> + graph->result.status);
>> + rc = -EINVAL;
>> + } else {
>> + dev_err(dev, "DSP returned [%x]\n", graph->result.status);
>> + rc = 0;
>> + }
>> +
>> +err:
>> + mutex_unlock(&graph->cmd_lock);
>> + return rc;
>> +}
>> +EXPORT_SYMBOL_GPL(audioreach_graph_send_cmd_sync);
>
> I'm pretty sure I've seen this code before...
>
> This looks almost identical to
>
> int q6apm_send_cmd_sync(struct q6apm *apm, struct gpr_pkt *pkt,
> uint32_t rsp_opcode)
>
> Can this be refactored?
I did this in next version.
>
> If not, at the very least make sure the errors messages are unique...
>
>> +static int audioreach_codec_dma_set_media_format(struct q6apm_graph *graph,
>> + struct audioreach_module *module,
>> + struct audioreach_module_config *cfg)
>
> maybe explain what a 'codec dma' is?
Qualcomm LPASS (Low Power Audio Subsystem) has digital part of codec
integrated into it, and connected to Analog part of codec using SoundWire.
>
>> +{
>> + struct apm_module_param_data *param_data;
>> + struct apm_codec_dma_module_intf_cfg *intf_cfg;
>> + struct apm_module_hw_ep_mf_cfg *hw_cfg;
>> + struct apm_module_frame_size_factor_cfg *fs_cfg;
>> + struct apm_module_hw_ep_power_mode_cfg *pm_cfg;
>> + int ic_sz, ep_sz, fs_sz, pm_sz, dl_sz;
>> + int rc, payload_size;
>> + struct gpr_pkt *pkt;
>> + void *p;
>> +
>> +
...
>> +
>> +static int audioreach_i2s_set_media_format(struct q6apm_graph *graph,
>> + struct audioreach_module *module,
>> + struct audioreach_module_config *cfg)
>> +{
>> + struct apm_module_frame_size_factor_cfg *fs_cfg;
>> + struct apm_module_param_data *param_data;
>> + struct apm_i2s_module_intf_cfg *intf_cfg;
>> + struct apm_module_hw_ep_mf_cfg *hw_cfg;
>> + int ic_sz, ep_sz, fs_sz;
>> + int rc, payload_size;
>> + struct gpr_pkt *pkt;
>> + void *p;
>> +
>> + ic_sz = APM_I2S_INTF_CFG_PSIZE;
>> + ep_sz = APM_HW_EP_CFG_PSIZE;
>> + fs_sz = APM_FS_CFG_PSIZE;
>> +
>> + payload_size = ic_sz + ep_sz + fs_sz;
>> +
>> + p = audioreach_alloc_apm_cmd_pkt(payload_size, APM_CMD_SET_CFG, 0);
>> + if (IS_ERR(p))
>> + return -ENOMEM;
>> +
>> + pkt = p;
>> + p = p + GPR_HDR_SIZE + APM_CMD_HDR_SIZE;
>> + intf_cfg = p;
>> +
>> + param_data = &intf_cfg->param_data;
>> + param_data->module_instance_id = module->instance_id;
>> + param_data->error_code = 0;
>> + param_data->param_id = PARAM_ID_I2S_INTF_CFG;
>> + param_data->param_size = ic_sz - APM_MODULE_PARAM_DATA_SIZE;
>> +
>> + intf_cfg->cfg.intf_idx = module->hw_interface_idx;
>> + intf_cfg->cfg.sd_line_idx = module->sd_line_idx;
>> +
>> + switch (cfg->fmt & SND_SOC_DAIFMT_MASTER_MASK) {
>> + case SND_SOC_DAIFMT_CBS_CFS:
>
> CBC_CFC
>
>> + intf_cfg->cfg.ws_src = CONFIG_I2S_WS_SRC_INTERNAL;
>> + break;
>> + case SND_SOC_DAIFMT_CBM_CFM:
>
> CBP_CFP
Sure I was not aware of these new formats, will use them.
>
>> + /* CPU is slave */
>> + intf_cfg->cfg.ws_src = CONFIG_I2S_WS_SRC_EXTERNAL;
>> + break;
>> + default:
>> + break;
>> + }
>> +
>> +int q6apm_map_memory_regions(struct q6apm_graph *graph,
>> + unsigned int dir, phys_addr_t phys,
>> + size_t period_sz, unsigned int periods)
>> +{
>> + struct audioreach_graph_data *data;
>> + struct audio_buffer *buf;
>> + unsigned long flags;
>> + int cnt;
>> + int rc;
>> +
>> + if (dir == SNDRV_PCM_STREAM_PLAYBACK)
>> + data = &graph->rx_data;
>> + else
>> + data = &graph->tx_data;
>> +
>> + spin_lock_irqsave(&graph->lock, flags);
>> +
>> + if (data->buf) {
>> + dev_err(graph->dev, "Buffer already allocated\n");
>> + spin_unlock_irqrestore(&graph->lock, flags);
>> + return 0;
>> + }
>> +
>> + buf = kzalloc(((sizeof(struct audio_buffer)) * periods), GFP_ATOMIC);
>
> Why GFP_ATOMIC?
Its under spinlock.
>
>> + if (!buf) {
>> + spin_unlock_irqrestore(&graph->lock, flags);
>> + return -ENOMEM;
>> + }
>> +
>> + if (dir == SNDRV_PCM_STREAM_PLAYBACK)
>> + data = &graph->rx_data;
>> + else
>> + data = &graph->tx_data;
>> +
>> + data->buf = buf;
>> +
>> + buf[0].phys = phys;
>> + buf[0].size = period_sz;
>> +
>> + for (cnt = 1; cnt < periods; cnt++) {
>> + if (period_sz > 0) {
>> + buf[cnt].phys = buf[0].phys + (cnt * period_sz);
>> + buf[cnt].size = period_sz;
>> + }
>> + }
>> + data->num_periods = periods;
>> +
>> + spin_unlock_irqrestore(&graph->lock, flags);
>> +
>> + rc = audioreach_map_memory_regions(graph, dir, period_sz,
>> + periods, 1);
>> + if (rc < 0) {
>> + dev_err(graph->dev, "Memory_map_regions failed\n");
>> + audioreach_graph_free_buf(graph);
>> + }
>> +
>> + return rc;
>> +}
>> +EXPORT_SYMBOL_GPL(q6apm_map_memory_regions);
>
>> +int q6apm_write_async(struct q6apm_graph *graph, uint32_t len, uint32_t msw_ts,
>> + uint32_t lsw_ts, uint32_t wflags)
>> +{
>> + struct gpr_pkt *pkt;
>> + void *p;
>> + int rc, payload_size, iid;
>> + struct apm_data_cmd_wr_sh_mem_ep_data_buffer_v2 *write;
>> + struct audio_buffer *ab;
>> + unsigned long flags;
>> +
>> + payload_size = sizeof(*write);
>> +
>> + iid = q6apm_graph_get_rx_shmem_module_iid(graph);
>> + p = audioreach_alloc_pkt(payload_size,
>> + DATA_CMD_WR_SH_MEM_EP_DATA_BUFFER_V2,
>> + graph->rx_data.dsp_buf | (len << APM_WRITE_TOKEN_LEN_SHIFT),
>> + graph->port->id, iid);
>> + if (IS_ERR(p))
>> + return -ENOMEM;
>> +
>> + pkt = p;
>> + p = p + GPR_HDR_SIZE;
>> + write = p;
>> +
>> + spin_lock_irqsave(&graph->lock, flags);
>> + ab = &graph->rx_data.buf[graph->rx_data.dsp_buf];
>> +
>> + write->buf_addr_lsw = lower_32_bits(ab->phys);
>> + write->buf_addr_msw = upper_32_bits(ab->phys);
>> + write->buf_size = len;
>> + write->timestamp_lsw = lsw_ts;
>> + write->timestamp_msw = msw_ts;
>> + write->mem_map_handle = graph->rx_data.mem_map_handle;
>> +
>> + if (wflags == NO_TIMESTAMP)
>> + write->flags = 0;
>> + else
>> + write->flags = 0x80000000;
>
> BIT(31) ?
This should be
#define WR_SH_MEM_EP_BIT_MASK_TIMESTAMP_VALID_FLAG 0x80000000
>
> it's also odd to define NO_TIMESTAMP, it's usually when you want a
> timestamp that you ask for it...
Thats right, I can try to cleanup some logic around this, It might be
that I have inherited some of this logic from old dsp drivers :-)
>
>> +
>> + graph->rx_data.dsp_buf++;
>> +
>> + if (graph->rx_data.dsp_buf >= graph->rx_data.num_periods)
>> + graph->rx_data.dsp_buf = 0;
>> +
>> + spin_unlock_irqrestore(&graph->lock, flags);
>> +
>> + rc = gpr_send_port_pkt(graph->port, pkt);
>> +
>> + kfree(pkt);
>> +
>> + return rc;
>> +}
>> +EXPORT_SYMBOL_GPL(q6apm_write_async);
>
More information about the Alsa-devel
mailing list