| Message ID | 20231128165638.757665-2-krzysztof.kozlowski@linaro.org (mailing list archive) |
|---|---|
| State | Not Applicable |
| Headers | show |
| Series | [1/2] ASoC: qcom: Add helper for allocating Soundwire stream runtime | expand |
On 28/11/2023 17:56, Krzysztof Kozlowski wrote: > Currently the Qualcomm Soundwire controller in its DAI startup op > allocates the Soundwire stream runtime. This works fine for existing > designs, but has limitations for stream runtimes with multiple > controllers, like upcoming Qualcomm X1E80100 SoC with four WSA8840 > speakers on two Soundwire controllers. > > When two Soundwire controllers are added to sound card codecs, Soundwire > startup() is called twice, one for each Soundwire controller, and second > execution overwrites what was set before. During shutdown() this causes > double free. > > It is expected to have only one Soundwire stream runtime, thus it should > be allocated from SoC soundcard context startup(), not from each > Soundwire startup(). Such way will properly handle both cases: one and > two Soundwire controllers in the stream runtime. > > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> > > --- > > This is an entirely different approach than my previous try here: > https://lore.kernel.org/all/20231025144601.268645-1-krzysztof.kozlowski@linaro.org/ ... and I forgot to thank you Pierre-Louis for patient explanation of the case in my previous try. Your review was much appreciated! Thank you. Best regards, Krzysztof
On 11/28/23 10:56, Krzysztof Kozlowski wrote: > Currently the Qualcomm Soundwire controller in its DAI startup op > allocates the Soundwire stream runtime. This works fine for existing > designs, but has limitations for stream runtimes with multiple > controllers, like upcoming Qualcomm X1E80100 SoC with four WSA8840 > speakers on two Soundwire controllers. > > When two Soundwire controllers are added to sound card codecs, Soundwire > startup() is called twice, one for each Soundwire controller, and second > execution overwrites what was set before. During shutdown() this causes > double free. > > It is expected to have only one Soundwire stream runtime, thus it should > be allocated from SoC soundcard context startup(), not from each > Soundwire startup(). Such way will properly handle both cases: one and > two Soundwire controllers in the stream runtime. > > Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> > > --- > > This is an entirely different approach than my previous try here: > https://lore.kernel.org/all/20231025144601.268645-1-krzysztof.kozlowski@linaro.org/ LGTM Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
On 11/28/23 10:59, Krzysztof Kozlowski wrote: > On 28/11/2023 17:56, Krzysztof Kozlowski wrote: >> Currently the Qualcomm Soundwire controller in its DAI startup op >> allocates the Soundwire stream runtime. This works fine for existing >> designs, but has limitations for stream runtimes with multiple >> controllers, like upcoming Qualcomm X1E80100 SoC with four WSA8840 >> speakers on two Soundwire controllers. >> >> When two Soundwire controllers are added to sound card codecs, Soundwire >> startup() is called twice, one for each Soundwire controller, and second >> execution overwrites what was set before. During shutdown() this causes >> double free. >> >> It is expected to have only one Soundwire stream runtime, thus it should >> be allocated from SoC soundcard context startup(), not from each >> Soundwire startup(). Such way will properly handle both cases: one and >> two Soundwire controllers in the stream runtime. >> >> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> >> >> --- >> >> This is an entirely different approach than my previous try here: >> https://lore.kernel.org/all/20231025144601.268645-1-krzysztof.kozlowski@linaro.org/ > > ... and I forgot to thank you Pierre-Louis for patient explanation of > the case in my previous try. Your review was much appreciated! You're welcome. It's good if we have multiple platforms using the 'stream' concept in similar ways.
diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c index f42c83c390ff..ac9176f714bf 100644 --- a/drivers/soundwire/qcom.c +++ b/drivers/soundwire/qcom.c @@ -1291,10 +1291,7 @@ static int qcom_swrm_startup(struct snd_pcm_substream *substream, struct snd_soc_dai *dai) { struct qcom_swrm_ctrl *ctrl = dev_get_drvdata(dai->dev); - struct snd_soc_pcm_runtime *rtd = substream->private_data; - struct sdw_stream_runtime *sruntime; - struct snd_soc_dai *codec_dai; - int ret, i; + int ret; ret = pm_runtime_get_sync(ctrl->dev); if (ret < 0 && ret != -EACCES) { @@ -1305,33 +1302,7 @@ static int qcom_swrm_startup(struct snd_pcm_substream *substream, return ret; } - sruntime = sdw_alloc_stream(dai->name); - if (!sruntime) { - ret = -ENOMEM; - goto err_alloc; - } - - ctrl->sruntime[dai->id] = sruntime; - - for_each_rtd_codec_dais(rtd, i, codec_dai) { - ret = snd_soc_dai_set_stream(codec_dai, sruntime, - substream->stream); - if (ret < 0 && ret != -ENOTSUPP) { - dev_err(dai->dev, "Failed to set sdw stream on %s\n", - codec_dai->name); - goto err_set_stream; - } - } - return 0; - -err_set_stream: - sdw_release_stream(sruntime); -err_alloc: - pm_runtime_mark_last_busy(ctrl->dev); - pm_runtime_put_autosuspend(ctrl->dev); - - return ret; } static void qcom_swrm_shutdown(struct snd_pcm_substream *substream, @@ -1340,8 +1311,6 @@ static void qcom_swrm_shutdown(struct snd_pcm_substream *substream, struct qcom_swrm_ctrl *ctrl = dev_get_drvdata(dai->dev); swrm_wait_for_wr_fifo_done(ctrl); - sdw_release_stream(ctrl->sruntime[dai->id]); - ctrl->sruntime[dai->id] = NULL; pm_runtime_mark_last_busy(ctrl->dev); pm_runtime_put_autosuspend(ctrl->dev); diff --git a/sound/soc/qcom/sc8280xp.c b/sound/soc/qcom/sc8280xp.c index d93b18f07be5..7c20b25ba3de 100644 --- a/sound/soc/qcom/sc8280xp.c +++ b/sound/soc/qcom/sc8280xp.c @@ -31,6 +31,17 @@ static int sc8280xp_snd_init(struct snd_soc_pcm_runtime *rtd) return qcom_snd_wcd_jack_setup(rtd, &data->jack, &data->jack_setup); } +static void sc8280xp_snd_shutdown(struct snd_pcm_substream *substream) +{ + struct snd_soc_pcm_runtime *rtd = substream->private_data; + struct snd_soc_dai *cpu_dai = snd_soc_rtd_to_cpu(rtd, 0); + struct sc8280xp_snd_data *pdata = snd_soc_card_get_drvdata(rtd->card); + struct sdw_stream_runtime *sruntime = pdata->sruntime[cpu_dai->id]; + + pdata->sruntime[cpu_dai->id] = NULL; + sdw_release_stream(sruntime); +} + static int sc8280xp_be_hw_params_fixup(struct snd_soc_pcm_runtime *rtd, struct snd_pcm_hw_params *params) { @@ -91,6 +102,8 @@ static int sc8280xp_snd_hw_free(struct snd_pcm_substream *substream) } static const struct snd_soc_ops sc8280xp_be_ops = { + .startup = qcom_snd_sdw_startup, + .shutdown = sc8280xp_snd_shutdown, .hw_params = sc8280xp_snd_hw_params, .hw_free = sc8280xp_snd_hw_free, .prepare = sc8280xp_snd_prepare, diff --git a/sound/soc/qcom/sm8250.c b/sound/soc/qcom/sm8250.c index 9cc869fd70ac..f298167c2a23 100644 --- a/sound/soc/qcom/sm8250.c +++ b/sound/soc/qcom/sm8250.c @@ -66,7 +66,19 @@ static int sm8250_snd_startup(struct snd_pcm_substream *substream) default: break; } - return 0; + + return qcom_snd_sdw_startup(substream); +} + +static void sm2450_snd_shutdown(struct snd_pcm_substream *substream) +{ + struct snd_soc_pcm_runtime *rtd = substream->private_data; + struct snd_soc_dai *cpu_dai = snd_soc_rtd_to_cpu(rtd, 0); + struct sm8250_snd_data *data = snd_soc_card_get_drvdata(rtd->card); + struct sdw_stream_runtime *sruntime = data->sruntime[cpu_dai->id]; + + data->sruntime[cpu_dai->id] = NULL; + sdw_release_stream(sruntime); } static int sm8250_snd_hw_params(struct snd_pcm_substream *substream, @@ -103,6 +115,7 @@ static int sm8250_snd_hw_free(struct snd_pcm_substream *substream) static const struct snd_soc_ops sm8250_be_ops = { .startup = sm8250_snd_startup, + .shutdown = sm2450_snd_shutdown, .hw_params = sm8250_snd_hw_params, .hw_free = sm8250_snd_hw_free, .prepare = sm8250_snd_prepare,
Currently the Qualcomm Soundwire controller in its DAI startup op allocates the Soundwire stream runtime. This works fine for existing designs, but has limitations for stream runtimes with multiple controllers, like upcoming Qualcomm X1E80100 SoC with four WSA8840 speakers on two Soundwire controllers. When two Soundwire controllers are added to sound card codecs, Soundwire startup() is called twice, one for each Soundwire controller, and second execution overwrites what was set before. During shutdown() this causes double free. It is expected to have only one Soundwire stream runtime, thus it should be allocated from SoC soundcard context startup(), not from each Soundwire startup(). Such way will properly handle both cases: one and two Soundwire controllers in the stream runtime. Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> --- This is an entirely different approach than my previous try here: https://lore.kernel.org/all/20231025144601.268645-1-krzysztof.kozlowski@linaro.org/ --- drivers/soundwire/qcom.c | 33 +-------------------------------- sound/soc/qcom/sc8280xp.c | 13 +++++++++++++ sound/soc/qcom/sm8250.c | 15 ++++++++++++++- 3 files changed, 28 insertions(+), 33 deletions(-)