[alsa-devel] [PATCH 00/12] ASoC: SOF: stability fixes
No new functionality here but fixes to the IPC, memory allocation, workqueues, control, runtime_pm and HDaudio support. These issues were identified during the integration/productization of new platforms.
I added a 'Fixes' tag to make sure backports/stable can pick up these patches, in case it's too late for them to land in 5.2
The next batch for SOF will be a set of new capabilities for trace, IPC test, Hdaudio support, they will be submitted after additional tests and once these fixes are reviewed/merged.
Thanks!
Amadeusz Sławiński (1): ALSA: hdac: fix memory release for SST and SOF drivers
Guennadi Liakhovetski (1): ASoC: SOF: ipc: fix a race, leading to IPC timeouts
Keyon Jie (1): ASoC: SOF: control: correct the copy size for bytes kcontrol put
Libin Yang (2): ASoC: SOF: pcm: clear hw_params_upon_resume flag correctly ASoC: SOF: Intel: hda-codec: fix memory allocation
Pierre-Louis Bossart (2): ASoC: SOF: core: fix error handling with the probe workqueue ASoC: SOF: pcm: remove warning - initialize workqueue on open
Ranjani Sridharan (3): ASoC: SOF: core: remove DSP after unregistering machine driver ASoC: SOF: core: remove snd_soc_unregister_component in case of error ASoC: SOF: pcm: remove runtime PM calls during pcm open/close
Zhu Yingjiang (2): ASoC: SOF: Intel: hda: fix the hda init chip ASoC: SOF: Intel: hda: use the defined ppcap functions
sound/hda/ext/hdac_ext_bus.c | 1 - sound/soc/sof/control.c | 9 +-- sound/soc/sof/core.c | 29 +++++++-- sound/soc/sof/intel/bdw.c | 11 ++-- sound/soc/sof/intel/byt.c | 12 ++-- sound/soc/sof/intel/cnl.c | 6 ++ sound/soc/sof/intel/hda-codec.c | 6 +- sound/soc/sof/intel/hda-ctrl.c | 102 +++++++++++++++++++++++++++--- sound/soc/sof/intel/hda-ipc.c | 19 +++++- sound/soc/sof/intel/hda.c | 109 +++++++------------------------- sound/soc/sof/ipc.c | 13 ---- sound/soc/sof/pcm.c | 37 ++--------- 12 files changed, 188 insertions(+), 166 deletions(-)
From: Ranjani Sridharan ranjani.sridharan@linux.intel.com
snd_sof_remove() disables the DSP and unmaps the DSP BAR. Removing topology after disabling the DSP results in a kernel panic while unloading the pipeline widget. This is because pipeline widget unload attempts to power down the core it is scheduled on by accessing the DSP registers.
So, the suggested fix here is to unregister the machine driver first to remove the topology and then disable the DSP to avoid the situation described above.
Note that the kernel panic only happens in cases where the HDaudio link is not managed by the hdac library, e.g. no codec or when HDMI is not supported. When the hdac library is used, snd_sof_remove() calls snd_hdac_ext_bus_device_remove() to remove the codec which unregisters the component driver thereby also removing the topology before the DSP is disabled.
Fixes: c16211d6226 ("ASoC: SOF: Add Sound Open Firmware driver core") Signed-off-by: Ranjani Sridharan ranjani.sridharan@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/core.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/sound/soc/sof/core.c b/sound/soc/sof/core.c index 32105e0fabe8..0bc4a8472c10 100644 --- a/sound/soc/sof/core.c +++ b/sound/soc/sof/core.c @@ -484,7 +484,6 @@ int snd_sof_device_remove(struct device *dev) snd_sof_ipc_free(sdev); snd_sof_free_debug(sdev); snd_sof_free_trace(sdev); - snd_sof_remove(sdev);
/* * Unregister machine driver. This will unbind the snd_card which @@ -494,6 +493,14 @@ int snd_sof_device_remove(struct device *dev) if (!IS_ERR_OR_NULL(pdata->pdev_mach)) platform_device_unregister(pdata->pdev_mach);
+ /* + * Unregistering the machine driver results in unloading the topology. + * Some widgets, ex: scheduler, attempt to power down the core they are + * scheduled on, when they are unloaded. Therefore, the DSP must be + * removed only after the topology has been unloaded. + */ + snd_sof_remove(sdev); + /* release firmware */ release_firmware(pdata->fw); pdata->fw = NULL;
From: Ranjani Sridharan ranjani.sridharan@linux.intel.com
No need to call snd_soc_unregister_component in case of error because the component device is resource-managed.
Fixes: c16211d6226 ("ASoC: SOF: Add Sound Open Firmware driver core") Signed-off-by: Ranjani Sridharan ranjani.sridharan@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/core.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/sound/soc/sof/core.c b/sound/soc/sof/core.c index 0bc4a8472c10..693ad83bffc9 100644 --- a/sound/soc/sof/core.c +++ b/sound/soc/sof/core.c @@ -382,7 +382,7 @@ static int sof_probe_continue(struct snd_sof_dev *sdev)
if (IS_ERR(plat_data->pdev_mach)) { ret = PTR_ERR(plat_data->pdev_mach); - goto comp_err; + goto fw_run_err; }
dev_dbg(sdev->dev, "created machine %s\n", @@ -393,8 +393,6 @@ static int sof_probe_continue(struct snd_sof_dev *sdev)
return 0;
-comp_err: - snd_soc_unregister_component(sdev->dev); fw_run_err: snd_sof_fw_unload(sdev); fw_load_err:
In some configurations, it's a requirement to split the probe in two, with a second part handled in a workqueue (e.g. for HDMI support which depends on the DRM modules).
SOF already handles these configurations but the error flow is incorrect. When an error occurs in the workqueue, the probe has technically already completed. If we release the resources on errors, this generates kernel oops/use-after-free when the resources are released a second time on module removal.
GitHub issue: https://github.com/thesofproject/linux/issues/945 Fixes: c16211d6226 ("ASoC: SOF: Add Sound Open Firmware driver core") Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/core.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+)
diff --git a/sound/soc/sof/core.c b/sound/soc/sof/core.c index 693ad83bffc9..5beda47cdf9f 100644 --- a/sound/soc/sof/core.c +++ b/sound/soc/sof/core.c @@ -393,6 +393,7 @@ static int sof_probe_continue(struct snd_sof_dev *sdev)
return 0;
+#if !IS_ENABLED(CONFIG_SND_SOC_SOF_PROBE_WORK_QUEUE) fw_run_err: snd_sof_fw_unload(sdev); fw_load_err: @@ -401,6 +402,21 @@ static int sof_probe_continue(struct snd_sof_dev *sdev) snd_sof_free_debug(sdev); dbg_err: snd_sof_remove(sdev); +#else + + /* + * when the probe_continue is handled in a work queue, the + * probe does not fail so we don't release resources here. + * They will be released with an explicit call to + * snd_sof_device_remove() when the PCI/ACPI device is removed + */ + +fw_run_err: +fw_load_err: +ipc_err: +dbg_err: + +#endif
return ret; }
From: Ranjani Sridharan ranjani.sridharan@linux.intel.com
pm_runtime_get_sync()/pm_runtime_put_autosuspend() calls are already invoked by the ASoC core in soc_pcm_open() and soc_pcm_close(). So the SOF component driver does not need to call them again.
Fixes: 868bd00f495 ("ASoC: SOF: Add PCM operations support") Signed-off-by: Ranjani Sridharan ranjani.sridharan@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/pcm.c | 29 ++--------------------------- 1 file changed, 2 insertions(+), 27 deletions(-)
diff --git a/sound/soc/sof/pcm.c b/sound/soc/sof/pcm.c index 649968841dad..4f536c0de0a5 100644 --- a/sound/soc/sof/pcm.c +++ b/sound/soc/sof/pcm.c @@ -416,7 +416,6 @@ static int sof_pcm_open(struct snd_pcm_substream *substream) struct snd_sof_pcm *spcm; struct snd_soc_tplg_stream_caps *caps; int ret; - int err;
/* nothing to do for BE */ if (rtd->dai_link->no_pcm) @@ -434,14 +433,6 @@ static int sof_pcm_open(struct snd_pcm_substream *substream)
caps = &spcm->pcm.caps[substream->stream];
- ret = pm_runtime_get_sync(sdev->dev); - if (ret < 0) { - dev_err(sdev->dev, "error: pcm open failed to resume %d\n", - ret); - pm_runtime_put_noidle(sdev->dev); - return ret; - } - /* set any runtime constraints based on topology */ snd_pcm_hw_constraint_step(substream->runtime, 0, SNDRV_PCM_HW_PARAM_BUFFER_BYTES, @@ -485,17 +476,8 @@ static int sof_pcm_open(struct snd_pcm_substream *substream) spcm->stream[substream->stream].substream = substream;
ret = snd_sof_pcm_platform_open(sdev, substream); - if (ret < 0) { - dev_err(sdev->dev, "error: pcm open failed %d\n", - ret); - - pm_runtime_mark_last_busy(sdev->dev); - - err = pm_runtime_put_autosuspend(sdev->dev); - if (err < 0) - dev_err(sdev->dev, "error: pcm close failed to idle %d\n", - err); - } + if (ret < 0) + dev_err(sdev->dev, "error: pcm open failed %d\n", ret);
return ret; } @@ -530,13 +512,6 @@ static int sof_pcm_close(struct snd_pcm_substream *substream) */ }
- pm_runtime_mark_last_busy(sdev->dev); - - err = pm_runtime_put_autosuspend(sdev->dev); - if (err < 0) - dev_err(sdev->dev, "error: pcm close failed to idle %d\n", - err); - return 0; }
From: Libin Yang libin.yang@intel.com
sof_pcm_hw_params() can only be called once to setup the FW hw_params. So after calling sof_pcm_hw_params(), hw_params_upon_resume flag must be cleared to avoid multiple invoking sof_pcm_hw_params() by prepare.
For example, after resume, there is an xrun happened, prepare() will be called. As the hw_params_upon_resume flag is not cleared, sof_pcm_hw_params() will be called and this will cause IPC timeout.
This patch fixes such issues.
Fixes: 868bd00f495 ("ASoC: SOF: Add PCM operations support") Signed-off-by: Libin Yang libin.yang@intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/pcm.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/sound/soc/sof/pcm.c b/sound/soc/sof/pcm.c index 4f536c0de0a5..e94892053690 100644 --- a/sound/soc/sof/pcm.c +++ b/sound/soc/sof/pcm.c @@ -214,6 +214,9 @@ static int sof_pcm_hw_params(struct snd_pcm_substream *substream, INIT_WORK(&spcm->stream[substream->stream].period_elapsed_work, sof_pcm_period_elapsed_work);
+ /* clear hw_params_upon_resume flag */ + spcm->hw_params_upon_resume[substream->stream] = 0; + return ret; }
@@ -428,9 +431,6 @@ static int sof_pcm_open(struct snd_pcm_substream *substream) dev_dbg(sdev->dev, "pcm: open stream %d dir %d\n", spcm->pcm.pcm_id, substream->stream);
- /* clear hw_params_upon_resume flag */ - spcm->hw_params_upon_resume[substream->stream] = 0; - caps = &spcm->pcm.caps[substream->stream];
/* set any runtime constraints based on topology */
If the SOF hw_params() fail, typically with an IPC error thrown by the firmware, the period_elapsed workqueue is not initialized, but we still cancel it in hw_free(), which results in a kernel warning.
Move the initialization to the .open callback. Tested on Broadwell (Samus) and IceLake.
Fixes: e2803e610ae ("ASoC: SOF: PCM: add period_elapsed work to fix race condition in interrupt context")
GitHub issue: https://github.com/thesofproject/linux/issues/932 Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/pcm.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/sound/soc/sof/pcm.c b/sound/soc/sof/pcm.c index e94892053690..6dc5f97be0bc 100644 --- a/sound/soc/sof/pcm.c +++ b/sound/soc/sof/pcm.c @@ -211,9 +211,6 @@ static int sof_pcm_hw_params(struct snd_pcm_substream *substream, /* save pcm hw_params */ memcpy(&spcm->params[substream->stream], params, sizeof(*params));
- INIT_WORK(&spcm->stream[substream->stream].period_elapsed_work, - sof_pcm_period_elapsed_work); - /* clear hw_params_upon_resume flag */ spcm->hw_params_upon_resume[substream->stream] = 0;
@@ -431,6 +428,9 @@ static int sof_pcm_open(struct snd_pcm_substream *substream) dev_dbg(sdev->dev, "pcm: open stream %d dir %d\n", spcm->pcm.pcm_id, substream->stream);
+ INIT_WORK(&spcm->stream[substream->stream].period_elapsed_work, + sof_pcm_period_elapsed_work); + caps = &spcm->pcm.caps[substream->stream];
/* set any runtime constraints based on topology */
From: Keyon Jie yang.jie@linux.intel.com
The size for the bytes kcontrol should include the abi header, that is, data->size + sizeof(*data), it is also aligned with get method after this change.
Fixes: c3078f53970 ("ASoC: SOF: Add Sound Open Firmware KControl support") Signed-off-by: Keyon Jie yang.jie@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/control.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/sound/soc/sof/control.c b/sound/soc/sof/control.c index 11762c4580f1..84e2cbfbbcbb 100644 --- a/sound/soc/sof/control.c +++ b/sound/soc/sof/control.c @@ -349,6 +349,7 @@ int snd_sof_bytes_put(struct snd_kcontrol *kcontrol, struct snd_sof_dev *sdev = scontrol->sdev; struct sof_ipc_ctrl_data *cdata = scontrol->control_data; struct sof_abi_hdr *data = cdata->data; + size_t size = data->size + sizeof(*data); int ret, err;
if (be->max > sizeof(ucontrol->value.bytes.data)) { @@ -358,10 +359,10 @@ int snd_sof_bytes_put(struct snd_kcontrol *kcontrol, return -EINVAL; }
- if (data->size > be->max) { + if (size > be->max) { dev_err_ratelimited(sdev->dev, - "error: size too big %d bytes max is %d\n", - data->size, be->max); + "error: size too big %zu bytes max is %d\n", + size, be->max); return -EINVAL; }
@@ -375,7 +376,7 @@ int snd_sof_bytes_put(struct snd_kcontrol *kcontrol, }
/* copy from kcontrol */ - memcpy(data, ucontrol->value.bytes.data, data->size); + memcpy(data, ucontrol->value.bytes.data, size);
/* notify DSP of byte control updates */ snd_sof_ipc_set_get_comp_data(sdev->ipc, scontrol,
From: Guennadi Liakhovetski guennadi.liakhovetski@linux.intel.com
Currently on all supported platforms the IPC IRQ thread first signals the sender when an IPC response is received from the DSP, then unmasks the IPC interrupt. Those actions are performed without holding any locks, so the thread can be interrupted between them. IPC timeouts have been observed in such scenarios: if the sender is woken up and it proceeds with sending the next message without unmasking the IPC interrupt, it can miss the next response. This patch takes a spin-lock to prevent the IRQ thread from being preempted at that point. It also makes sure, that the next IPC transmission by the host cannot take place before the IRQ thread has finished updating all the required IPC registers.
Fixes: 53e0c72d98b ("ASoC: SOF: Add support for IPC IO between DSP and Host") Signed-off-by: Guennadi Liakhovetski guennadi.liakhovetski@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/intel/bdw.c | 11 ++++++----- sound/soc/sof/intel/byt.c | 12 +++++++----- sound/soc/sof/intel/cnl.c | 6 ++++++ sound/soc/sof/intel/hda-ipc.c | 19 ++++++++++++++++--- sound/soc/sof/ipc.c | 13 ------------- 5 files changed, 35 insertions(+), 26 deletions(-)
diff --git a/sound/soc/sof/intel/bdw.c b/sound/soc/sof/intel/bdw.c index 065cb868bdfa..9dfdc02192b7 100644 --- a/sound/soc/sof/intel/bdw.c +++ b/sound/soc/sof/intel/bdw.c @@ -278,11 +278,15 @@ static irqreturn_t bdw_irq_thread(int irq, void *context) /* reply message from DSP */ if (ipcx & SHIM_IPCX_DONE && !(imrx & SHIM_IMRX_DONE)) { + unsigned long flags; + /* Mask Done interrupt before return */ snd_sof_dsp_update_bits_unlocked(sdev, BDW_DSP_BAR, SHIM_IMRX, SHIM_IMRX_DONE, SHIM_IMRX_DONE);
+ spin_lock_irqsave(&sdev->ipc_lock, flags); + /* * handle immediate reply from DSP core. If the msg is * found, set done bit in cmd_done which is called at the @@ -294,6 +298,8 @@ static irqreturn_t bdw_irq_thread(int irq, void *context) snd_sof_ipc_reply(sdev, ipcx);
bdw_dsp_done(sdev); + + spin_unlock_irqrestore(&sdev->ipc_lock, flags); }
ipcd = snd_sof_dsp_read(sdev, BDW_DSP_BAR, SHIM_IPCD); @@ -485,7 +491,6 @@ static void bdw_get_reply(struct snd_sof_dev *sdev) { struct snd_sof_ipc_msg *msg = sdev->msg; struct sof_ipc_reply reply; - unsigned long flags; int ret = 0;
/* @@ -501,8 +506,6 @@ static void bdw_get_reply(struct snd_sof_dev *sdev) /* get reply */ sof_mailbox_read(sdev, sdev->host_box.offset, &reply, sizeof(reply));
- spin_lock_irqsave(&sdev->ipc_lock, flags); - if (reply.error < 0) { memcpy(msg->reply_data, &reply, sizeof(reply)); ret = reply.error; @@ -521,8 +524,6 @@ static void bdw_get_reply(struct snd_sof_dev *sdev) }
msg->reply_error = ret; - - spin_unlock_irqrestore(&sdev->ipc_lock, flags); }
static void bdw_host_done(struct snd_sof_dev *sdev) diff --git a/sound/soc/sof/intel/byt.c b/sound/soc/sof/intel/byt.c index 7bf9143d3106..5a11a104110b 100644 --- a/sound/soc/sof/intel/byt.c +++ b/sound/soc/sof/intel/byt.c @@ -324,11 +324,16 @@ static irqreturn_t byt_irq_thread(int irq, void *context) /* reply message from DSP */ if (ipcx & SHIM_BYT_IPCX_DONE && !(imrx & SHIM_IMRX_DONE)) { + unsigned long flags; + /* Mask Done interrupt before first */ snd_sof_dsp_update_bits64_unlocked(sdev, BYT_DSP_BAR, SHIM_IMRX, SHIM_IMRX_DONE, SHIM_IMRX_DONE); + + spin_lock_irqsave(&sdev->ipc_lock, flags); + /* * handle immediate reply from DSP core. If the msg is * found, set done bit in cmd_done which is called at the @@ -340,6 +345,8 @@ static irqreturn_t byt_irq_thread(int irq, void *context) snd_sof_ipc_reply(sdev, ipcx);
byt_dsp_done(sdev); + + spin_unlock_irqrestore(&sdev->ipc_lock, flags); }
/* new message from DSP */ @@ -383,7 +390,6 @@ static void byt_get_reply(struct snd_sof_dev *sdev) { struct snd_sof_ipc_msg *msg = sdev->msg; struct sof_ipc_reply reply; - unsigned long flags; int ret = 0;
/* @@ -399,8 +405,6 @@ static void byt_get_reply(struct snd_sof_dev *sdev) /* get reply */ sof_mailbox_read(sdev, sdev->host_box.offset, &reply, sizeof(reply));
- spin_lock_irqsave(&sdev->ipc_lock, flags); - if (reply.error < 0) { memcpy(msg->reply_data, &reply, sizeof(reply)); ret = reply.error; @@ -419,8 +423,6 @@ static void byt_get_reply(struct snd_sof_dev *sdev) }
msg->reply_error = ret; - - spin_unlock_irqrestore(&sdev->ipc_lock, flags); }
static void byt_host_done(struct snd_sof_dev *sdev) diff --git a/sound/soc/sof/intel/cnl.c b/sound/soc/sof/intel/cnl.c index 08a1a3d3c08d..fd8c499b3ea0 100644 --- a/sound/soc/sof/intel/cnl.c +++ b/sound/soc/sof/intel/cnl.c @@ -50,6 +50,8 @@ static irqreturn_t cnl_ipc_irq_thread(int irq, void *context) /* reply message from DSP */ if (hipcida & CNL_DSP_REG_HIPCIDA_DONE && hipcctl & CNL_DSP_REG_HIPCCTL_DONE) { + unsigned long flags; + hipci = snd_sof_dsp_read(sdev, HDA_DSP_BAR, CNL_DSP_REG_HIPCIDR); msg_ext = hipci & CNL_DSP_REG_HIPCIDR_MSG_MASK; @@ -64,6 +66,8 @@ static irqreturn_t cnl_ipc_irq_thread(int irq, void *context) CNL_DSP_REG_HIPCCTL, CNL_DSP_REG_HIPCCTL_DONE, 0);
+ spin_lock_irqsave(&sdev->ipc_lock, flags); + /* handle immediate reply from DSP core */ hda_dsp_ipc_get_reply(sdev); snd_sof_ipc_reply(sdev, msg); @@ -75,6 +79,8 @@ static irqreturn_t cnl_ipc_irq_thread(int irq, void *context)
cnl_ipc_dsp_done(sdev);
+ spin_unlock_irqrestore(&sdev->ipc_lock, flags); + ret = IRQ_HANDLED; }
diff --git a/sound/soc/sof/intel/hda-ipc.c b/sound/soc/sof/intel/hda-ipc.c index 73ead7070cde..c977a447051d 100644 --- a/sound/soc/sof/intel/hda-ipc.c +++ b/sound/soc/sof/intel/hda-ipc.c @@ -72,7 +72,6 @@ void hda_dsp_ipc_get_reply(struct snd_sof_dev *sdev) struct snd_sof_ipc_msg *msg = sdev->msg; struct sof_ipc_reply reply; struct sof_ipc_cmd_hdr *hdr; - unsigned long flags; int ret = 0;
/* @@ -84,7 +83,6 @@ void hda_dsp_ipc_get_reply(struct snd_sof_dev *sdev) dev_warn(sdev->dev, "unexpected ipc interrupt raised!\n"); return; } - spin_lock_irqsave(&sdev->ipc_lock, flags);
hdr = msg->msg_data; if (hdr->cmd == (SOF_IPC_GLB_PM_MSG | SOF_IPC_PM_CTX_SAVE)) { @@ -123,7 +121,6 @@ void hda_dsp_ipc_get_reply(struct snd_sof_dev *sdev) out: msg->reply_error = ret;
- spin_unlock_irqrestore(&sdev->ipc_lock, flags); }
static bool hda_dsp_ipc_is_sof(uint32_t msg) @@ -158,6 +155,8 @@ irqreturn_t hda_dsp_ipc_irq_thread(int irq, void *context) /* is this a reply message from the DSP */ if (hipcie & HDA_DSP_REG_HIPCIE_DONE && hipcctl & HDA_DSP_REG_HIPCCTL_DONE) { + unsigned long flags; + hipci = snd_sof_dsp_read(sdev, HDA_DSP_BAR, HDA_DSP_REG_HIPCI); msg = hipci & HDA_DSP_REG_HIPCI_MSG_MASK; @@ -172,6 +171,18 @@ irqreturn_t hda_dsp_ipc_irq_thread(int irq, void *context) HDA_DSP_REG_HIPCCTL, HDA_DSP_REG_HIPCCTL_DONE, 0);
+ /* + * Make sure the interrupt thread cannot be preempted between + * waking up the sender and re-enabling the interrupt. Also + * protect against a theoretical race with sof_ipc_tx_message(): + * if the DSP is fast enough to receive an IPC message, reply to + * it, and the host interrupt processing calls this function on + * a different core from the one, where the sending is taking + * place, the message might not yet be marked as expecting a + * reply. + */ + spin_lock_irqsave(&sdev->ipc_lock, flags); + /* handle immediate reply from DSP core - ignore ROM messages */ if (hda_dsp_ipc_is_sof(msg)) { hda_dsp_ipc_get_reply(sdev); @@ -187,6 +198,8 @@ irqreturn_t hda_dsp_ipc_irq_thread(int irq, void *context) /* set the done bit */ hda_dsp_ipc_dsp_done(sdev);
+ spin_unlock_irqrestore(&sdev->ipc_lock, flags); + ret = IRQ_HANDLED; }
diff --git a/sound/soc/sof/ipc.c b/sound/soc/sof/ipc.c index 894e68cbd69d..10304a90cf25 100644 --- a/sound/soc/sof/ipc.c +++ b/sound/soc/sof/ipc.c @@ -308,19 +308,8 @@ EXPORT_SYMBOL(sof_ipc_tx_message); int snd_sof_ipc_reply(struct snd_sof_dev *sdev, u32 msg_id) { struct snd_sof_ipc_msg *msg = &sdev->ipc->msg; - unsigned long flags; - - /* - * Protect against a theoretical race with sof_ipc_tx_message(): if the - * DSP is fast enough to receive an IPC message, reply to it, and the - * host interrupt processing calls this function on a different core - * from the one, where the sending is taking place, the message might - * not yet be marked as expecting a reply. - */ - spin_lock_irqsave(&sdev->ipc_lock, flags);
if (msg->ipc_complete) { - spin_unlock_irqrestore(&sdev->ipc_lock, flags); dev_err(sdev->dev, "error: no reply expected, received 0x%x", msg_id); return -EINVAL; @@ -330,8 +319,6 @@ int snd_sof_ipc_reply(struct snd_sof_dev *sdev, u32 msg_id) msg->ipc_complete = true; wake_up(&msg->waitq);
- spin_unlock_irqrestore(&sdev->ipc_lock, flags); - return 0; } EXPORT_SYMBOL(snd_sof_ipc_reply);
On Sat, 18 May 2019 22:27:00 +0200, Pierre-Louis Bossart wrote:
From: Guennadi Liakhovetski guennadi.liakhovetski@linux.intel.com
Currently on all supported platforms the IPC IRQ thread first signals the sender when an IPC response is received from the DSP, then unmasks the IPC interrupt. Those actions are performed without holding any locks, so the thread can be interrupted between them. IPC timeouts have been observed in such scenarios: if the sender is woken up and it proceeds with sending the next message without unmasking the IPC interrupt, it can miss the next response. This patch takes a spin-lock to prevent the IRQ thread from being preempted at that point. It also makes sure, that the next IPC transmission by the host cannot take place before the IRQ thread has finished updating all the required IPC registers.
Fixes: 53e0c72d98b ("ASoC: SOF: Add support for IPC IO between DSP and Host") Signed-off-by: Guennadi Liakhovetski guennadi.liakhovetski@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com
sound/soc/sof/intel/bdw.c | 11 ++++++----- sound/soc/sof/intel/byt.c | 12 +++++++----- sound/soc/sof/intel/cnl.c | 6 ++++++ sound/soc/sof/intel/hda-ipc.c | 19 ++++++++++++++++--- sound/soc/sof/ipc.c | 13 ------------- 5 files changed, 35 insertions(+), 26 deletions(-)
diff --git a/sound/soc/sof/intel/bdw.c b/sound/soc/sof/intel/bdw.c index 065cb868bdfa..9dfdc02192b7 100644 --- a/sound/soc/sof/intel/bdw.c +++ b/sound/soc/sof/intel/bdw.c @@ -278,11 +278,15 @@ static irqreturn_t bdw_irq_thread(int irq, void *context) /* reply message from DSP */ if (ipcx & SHIM_IPCX_DONE && !(imrx & SHIM_IMRX_DONE)) {
unsigned long flags;- /* Mask Done interrupt before return */ snd_sof_dsp_update_bits_unlocked(sdev, BDW_DSP_BAR, SHIM_IMRX, SHIM_IMRX_DONE, SHIM_IMRX_DONE);
spin_lock_irqsave(&sdev->ipc_lock, flags);
Here is an threaded irq handler, so the irqflag is superfluous. You can use spin_lock_irq() and spin_unlock_irq().
diff --git a/sound/soc/sof/intel/byt.c b/sound/soc/sof/intel/byt.c index 7bf9143d3106..5a11a104110b 100644 --- a/sound/soc/sof/intel/byt.c +++ b/sound/soc/sof/intel/byt.c @@ -324,11 +324,16 @@ static irqreturn_t byt_irq_thread(int irq, void *context) /* reply message from DSP */ if (ipcx & SHIM_BYT_IPCX_DONE && !(imrx & SHIM_IMRX_DONE)) {
unsigned long flags;- /* Mask Done interrupt before first */ snd_sof_dsp_update_bits64_unlocked(sdev, BYT_DSP_BAR, SHIM_IMRX, SHIM_IMRX_DONE, SHIM_IMRX_DONE);
BTW, is this usage of _unlocked() version safe? The previous one also contained that, and I wonder why _unlocked variant must be used here.
thanks,
Takashi
Hi Takashi,
On Sun, May 19, 2019 at 09:20:55AM +0200, Takashi Iwai wrote:
On Sat, 18 May 2019 22:27:00 +0200, Pierre-Louis Bossart wrote:
From: Guennadi Liakhovetski guennadi.liakhovetski@linux.intel.com
Currently on all supported platforms the IPC IRQ thread first signals the sender when an IPC response is received from the DSP, then unmasks the IPC interrupt. Those actions are performed without holding any locks, so the thread can be interrupted between them. IPC timeouts have been observed in such scenarios: if the sender is woken up and it proceeds with sending the next message without unmasking the IPC interrupt, it can miss the next response. This patch takes a spin-lock to prevent the IRQ thread from being preempted at that point. It also makes sure, that the next IPC transmission by the host cannot take place before the IRQ thread has finished updating all the required IPC registers.
Fixes: 53e0c72d98b ("ASoC: SOF: Add support for IPC IO between DSP and Host") Signed-off-by: Guennadi Liakhovetski guennadi.liakhovetski@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com
sound/soc/sof/intel/bdw.c | 11 ++++++----- sound/soc/sof/intel/byt.c | 12 +++++++----- sound/soc/sof/intel/cnl.c | 6 ++++++ sound/soc/sof/intel/hda-ipc.c | 19 ++++++++++++++++--- sound/soc/sof/ipc.c | 13 ------------- 5 files changed, 35 insertions(+), 26 deletions(-)
diff --git a/sound/soc/sof/intel/bdw.c b/sound/soc/sof/intel/bdw.c index 065cb868bdfa..9dfdc02192b7 100644 --- a/sound/soc/sof/intel/bdw.c +++ b/sound/soc/sof/intel/bdw.c @@ -278,11 +278,15 @@ static irqreturn_t bdw_irq_thread(int irq, void *context) /* reply message from DSP */ if (ipcx & SHIM_IPCX_DONE && !(imrx & SHIM_IMRX_DONE)) {
unsigned long flags;- /* Mask Done interrupt before return */ snd_sof_dsp_update_bits_unlocked(sdev, BDW_DSP_BAR, SHIM_IMRX, SHIM_IMRX_DONE, SHIM_IMRX_DONE);
spin_lock_irqsave(&sdev->ipc_lock, flags);Here is an threaded irq handler, so the irqflag is superfluous. You can use spin_lock_irq() and spin_unlock_irq().
Oh, sure, thanks for catching this! That was a blind moving of the spin-lock from lower-level functions. I'll send an updated version to Pierre, unless you want to apply it immediately, in which case I can send it to you now.
diff --git a/sound/soc/sof/intel/byt.c b/sound/soc/sof/intel/byt.c index 7bf9143d3106..5a11a104110b 100644 --- a/sound/soc/sof/intel/byt.c +++ b/sound/soc/sof/intel/byt.c @@ -324,11 +324,16 @@ static irqreturn_t byt_irq_thread(int irq, void *context) /* reply message from DSP */ if (ipcx & SHIM_BYT_IPCX_DONE && !(imrx & SHIM_IMRX_DONE)) {
unsigned long flags;- /* Mask Done interrupt before first */ snd_sof_dsp_update_bits64_unlocked(sdev, BYT_DSP_BAR, SHIM_IMRX, SHIM_IMRX_DONE, SHIM_IMRX_DONE);
BTW, is this usage of _unlocked() version safe? The previous one also contained that, and I wonder why _unlocked variant must be used here.
Some of those uses seem rather fragile to me too, but it looks like they *should* be safe in normal cases. There seem to be a potential problem on Broadwell, where an ISR is first configured, which uses *_unlocked() to access the SHIM_IMRX register, and then bdw_set_dsp_D0() is called, which also touches that register. So, it's relying on the fact, that no IPC interrupts will occur until probing is completed. This should also normally be the case, but if for some reason such an interrupt does trigger at that time, access to that register can be messed up. This should be reviewed and possibly refined separately.
Thanks Guennadi
thanks,
Takashi
diff --git a/sound/soc/sof/intel/bdw.c b/sound/soc/sof/intel/bdw.c index 065cb868bdfa..9dfdc02192b7 100644 --- a/sound/soc/sof/intel/bdw.c +++ b/sound/soc/sof/intel/bdw.c @@ -278,11 +278,15 @@ static irqreturn_t bdw_irq_thread(int irq, void *context) /* reply message from DSP */ if (ipcx & SHIM_IPCX_DONE && !(imrx & SHIM_IMRX_DONE)) {
unsigned long flags;- /* Mask Done interrupt before return */ snd_sof_dsp_update_bits_unlocked(sdev, BDW_DSP_BAR, SHIM_IMRX, SHIM_IMRX_DONE, SHIM_IMRX_DONE);
spin_lock_irqsave(&sdev->ipc_lock, flags);Here is an threaded irq handler, so the irqflag is superfluous. You can use spin_lock_irq() and spin_unlock_irq().
Oh, sure, thanks for catching this! That was a blind moving of the spin-lock from lower-level functions. I'll send an updated version to Pierre, unless you want to apply it immediately, in which case I can send it to you now.
I can send a v2 with Guennadi's update and Takashi's Reviewed-by tag.
diff --git a/sound/soc/sof/intel/byt.c b/sound/soc/sof/intel/byt.c index 7bf9143d3106..5a11a104110b 100644 --- a/sound/soc/sof/intel/byt.c +++ b/sound/soc/sof/intel/byt.c @@ -324,11 +324,16 @@ static irqreturn_t byt_irq_thread(int irq, void *context) /* reply message from DSP */ if (ipcx & SHIM_BYT_IPCX_DONE && !(imrx & SHIM_IMRX_DONE)) {
unsigned long flags;- /* Mask Done interrupt before first */ snd_sof_dsp_update_bits64_unlocked(sdev, BYT_DSP_BAR, SHIM_IMRX, SHIM_IMRX_DONE, SHIM_IMRX_DONE);
BTW, is this usage of _unlocked() version safe? The previous one also contained that, and I wonder why _unlocked variant must be used here.
Some of those uses seem rather fragile to me too, but it looks like they *should* be safe in normal cases. There seem to be a potential problem on Broadwell, where an ISR is first configured, which uses *_unlocked() to access the SHIM_IMRX register, and then bdw_set_dsp_D0() is called, which also touches that register. So, it's relying on the fact, that no IPC interrupts will occur until probing is completed. This should also normally be the case, but if for some reason such an interrupt does trigger at that time, access to that register can be messed up. This should be reviewed and possibly refined separately.
There will be additional IPC hardening fixes that are being tested at the moment, the introduction of HDaudio support seems to have exposed a number of issues we didn't see with I2S/TDM/DMIC interfaces. I included Guennadi's patch in the series since we have evidence it does improve things on the Up2. we can look at the _unlocked parts in the next update with more validation.
From: Zhu Yingjiang yingjiang.zhu@linux.intel.com
re-write hda_init_caps and remove the HDA reset, clean HDA streams and clear interrupt steps in hda_dsp_probe so the HDA init steps will not be called twice if the CONFIG_SND_SOC_SOF_HDA is true.
Fixes: 8a300c8fb17 ("ASoC: SOF: Intel: Add HDA controller for Intel DSP") Signed-off-by: Zhu Yingjiang yingjiang.zhu@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/intel/hda-ctrl.c | 102 ++++++++++++++++++++++++++++++--- sound/soc/sof/intel/hda.c | 99 ++++++-------------------------- 2 files changed, 112 insertions(+), 89 deletions(-)
diff --git a/sound/soc/sof/intel/hda-ctrl.c b/sound/soc/sof/intel/hda-ctrl.c index 2c3645736e1f..07bc123112c9 100644 --- a/sound/soc/sof/intel/hda-ctrl.c +++ b/sound/soc/sof/intel/hda-ctrl.c @@ -161,21 +161,105 @@ int hda_dsp_ctrl_clock_power_gating(struct snd_sof_dev *sdev, bool enable) return 0; }
-#if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA) -/* - * While performing reset, controller may not come back properly and causing - * issues, so recommendation is to set CGCTL.MISCBDCGE to 0 then do reset - * (init chip) and then again set CGCTL.MISCBDCGE to 1 - */ int hda_dsp_ctrl_init_chip(struct snd_sof_dev *sdev, bool full_reset) { struct hdac_bus *bus = sof_to_bus(sdev); - int ret; + struct hdac_stream *stream; + int sd_offset, ret = 0; + + if (bus->chip_init) + return 0;
hda_dsp_ctrl_misc_clock_gating(sdev, false); - ret = snd_hdac_bus_init_chip(bus, full_reset); + + if (full_reset) { + /* clear WAKESTS */ + snd_sof_dsp_update_bits(sdev, HDA_DSP_HDA_BAR, SOF_HDA_WAKESTS, + SOF_HDA_WAKESTS_INT_MASK, + SOF_HDA_WAKESTS_INT_MASK); + + /* reset HDA controller */ + ret = hda_dsp_ctrl_link_reset(sdev, true); + if (ret < 0) { + dev_err(sdev->dev, "error: failed to reset HDA controller\n"); + return ret; + } + + usleep_range(500, 1000); + + /* exit HDA controller reset */ + ret = hda_dsp_ctrl_link_reset(sdev, false); + if (ret < 0) { + dev_err(sdev->dev, "error: failed to exit HDA controller reset\n"); + return ret; + } + + usleep_range(1000, 1200); + } + +#if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA) + /* check to see if controller is ready */ + if (!snd_hdac_chip_readb(bus, GCTL)) { + dev_dbg(bus->dev, "controller not ready!\n"); + return -EBUSY; + } + + /* Accept unsolicited responses */ + snd_hdac_chip_updatel(bus, GCTL, AZX_GCTL_UNSOL, AZX_GCTL_UNSOL); + + /* detect codecs */ + if (!bus->codec_mask) { + bus->codec_mask = snd_hdac_chip_readw(bus, STATESTS); + dev_dbg(bus->dev, "codec_mask = 0x%lx\n", bus->codec_mask); + } +#endif + + /* clear stream status */ + list_for_each_entry(stream, &bus->stream_list, list) { + sd_offset = SOF_STREAM_SD_OFFSET(stream); + snd_sof_dsp_update_bits(sdev, HDA_DSP_HDA_BAR, + sd_offset + + SOF_HDA_ADSP_REG_CL_SD_STS, + SOF_HDA_CL_DMA_SD_INT_MASK, + SOF_HDA_CL_DMA_SD_INT_MASK); + } + + /* clear WAKESTS */ + snd_sof_dsp_update_bits(sdev, HDA_DSP_HDA_BAR, SOF_HDA_WAKESTS, + SOF_HDA_WAKESTS_INT_MASK, + SOF_HDA_WAKESTS_INT_MASK); + +#if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA) + /* clear rirb status */ + snd_hdac_chip_writeb(bus, RIRBSTS, RIRB_INT_MASK); +#endif + + /* clear interrupt status register */ + snd_sof_dsp_write(sdev, HDA_DSP_HDA_BAR, SOF_HDA_INTSTS, + SOF_HDA_INT_CTRL_EN | SOF_HDA_INT_ALL_STREAM); + +#if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA) + /* initialize the codec command I/O */ + snd_hdac_bus_init_cmd_io(bus); +#endif + + /* enable CIE and GIE interrupts */ + snd_sof_dsp_update_bits(sdev, HDA_DSP_HDA_BAR, SOF_HDA_INTCTL, + SOF_HDA_INT_CTRL_EN | SOF_HDA_INT_GLOBAL_EN, + SOF_HDA_INT_CTRL_EN | SOF_HDA_INT_GLOBAL_EN); + +#if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA) + /* program the position buffer */ + if (bus->use_posbuf && bus->posbuf.addr) { + snd_hdac_chip_writel(bus, DPLBASE, (u32)bus->posbuf.addr); + snd_hdac_chip_writel(bus, DPUBASE, + upper_32_bits(bus->posbuf.addr)); + } +#endif + + bus->chip_init = true; + hda_dsp_ctrl_misc_clock_gating(sdev, true);
return ret; } -#endif diff --git a/sound/soc/sof/intel/hda.c b/sound/soc/sof/intel/hda.c index 7e3980a2f7ba..e47f03dc62f0 100644 --- a/sound/soc/sof/intel/hda.c +++ b/sound/soc/sof/intel/hda.c @@ -264,9 +264,12 @@ static const char *fixup_tplg_name(struct snd_sof_dev *sdev, return tplg_filename; }
+#endif + static int hda_init_caps(struct snd_sof_dev *sdev) { struct hdac_bus *bus = sof_to_bus(sdev); +#if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA) struct hdac_ext_link *hlink; struct snd_soc_acpi_mach_params *mach_params; struct snd_soc_acpi_mach *hda_mach; @@ -274,8 +277,9 @@ static int hda_init_caps(struct snd_sof_dev *sdev) struct snd_soc_acpi_mach *mach; const char *tplg_filename; int codec_num = 0; - int ret = 0; int i; +#endif + int ret = 0;
device_disable_async_suspend(bus->dev);
@@ -283,6 +287,14 @@ static int hda_init_caps(struct snd_sof_dev *sdev) if (bus->ppcap) dev_dbg(sdev->dev, "PP capability, will probe DSP later.\n");
+ ret = hda_dsp_ctrl_init_chip(sdev, true); + if (ret < 0) { + dev_err(bus->dev, "error: init chip failed with ret: %d\n", + ret); + return ret; + } + +#if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA) if (bus->mlcap) snd_hdac_ext_bus_get_ml_capabilities(bus);
@@ -293,12 +305,6 @@ static int hda_init_caps(struct snd_sof_dev *sdev) return ret; }
- ret = hda_dsp_ctrl_init_chip(sdev, true); - if (ret < 0) { - dev_err(bus->dev, "error: init chip failed with ret: %d\n", ret); - goto out; - } - /* codec detection */ if (!bus->codec_mask) { dev_info(bus->dev, "no hda codecs found!\n"); @@ -339,8 +345,10 @@ static int hda_init_caps(struct snd_sof_dev *sdev) /* use local variable for readability */ tplg_filename = pdata->tplg_filename; tplg_filename = fixup_tplg_name(sdev, tplg_filename); - if (!tplg_filename) - goto out; + if (!tplg_filename) { + hda_codec_i915_exit(sdev); + return ret; + } pdata->tplg_filename = tplg_filename; } } @@ -364,35 +372,10 @@ static int hda_init_caps(struct snd_sof_dev *sdev) */ list_for_each_entry(hlink, &bus->hlink_list, list) snd_hdac_ext_bus_link_put(bus, hlink); - - return 0; - -out: - hda_codec_i915_exit(sdev); - return ret; -} - -#else - -static int hda_init_caps(struct snd_sof_dev *sdev) -{ - /* - * set CGCTL.MISCBDCGE to 0 during reset and set back to 1 - * when reset finished. - * TODO: maybe no need for init_caps? - */ - hda_dsp_ctrl_misc_clock_gating(sdev, 0); - - /* clear WAKESTS */ - snd_sof_dsp_update_bits(sdev, HDA_DSP_HDA_BAR, SOF_HDA_WAKESTS, - SOF_HDA_WAKESTS_INT_MASK, - SOF_HDA_WAKESTS_INT_MASK); - +#endif return 0; }
-#endif - static const struct sof_intel_dsp_desc *get_chip_info(struct snd_sof_pdata *pdata) { @@ -409,9 +392,8 @@ int hda_dsp_probe(struct snd_sof_dev *sdev) struct pci_dev *pci = to_pci_dev(sdev->dev); struct sof_intel_hda_dev *hdev; struct hdac_bus *bus; - struct hdac_stream *stream; const struct sof_intel_dsp_desc *chip; - int sd_offset, ret = 0; + int ret = 0;
/* * detect DSP by checking class/subclass/prog-id information @@ -558,49 +540,6 @@ int hda_dsp_probe(struct snd_sof_dev *sdev) if (ret < 0) goto free_ipc_irq;
- /* reset HDA controller */ - ret = hda_dsp_ctrl_link_reset(sdev, true); - if (ret < 0) { - dev_err(sdev->dev, "error: failed to reset HDA controller\n"); - goto free_ipc_irq; - } - - /* exit HDA controller reset */ - ret = hda_dsp_ctrl_link_reset(sdev, false); - if (ret < 0) { - dev_err(sdev->dev, "error: failed to exit HDA controller reset\n"); - goto free_ipc_irq; - } - - /* clear stream status */ - list_for_each_entry(stream, &bus->stream_list, list) { - sd_offset = SOF_STREAM_SD_OFFSET(stream); - snd_sof_dsp_update_bits(sdev, HDA_DSP_HDA_BAR, - sd_offset + - SOF_HDA_ADSP_REG_CL_SD_STS, - SOF_HDA_CL_DMA_SD_INT_MASK, - SOF_HDA_CL_DMA_SD_INT_MASK); - } - - /* clear WAKESTS */ - snd_sof_dsp_update_bits(sdev, HDA_DSP_HDA_BAR, SOF_HDA_WAKESTS, - SOF_HDA_WAKESTS_INT_MASK, - SOF_HDA_WAKESTS_INT_MASK); - - /* clear interrupt status register */ - snd_sof_dsp_write(sdev, HDA_DSP_HDA_BAR, SOF_HDA_INTSTS, - SOF_HDA_INT_CTRL_EN | SOF_HDA_INT_ALL_STREAM); - - /* enable CIE and GIE interrupts */ - snd_sof_dsp_update_bits(sdev, HDA_DSP_HDA_BAR, SOF_HDA_INTCTL, - SOF_HDA_INT_CTRL_EN | SOF_HDA_INT_GLOBAL_EN, - SOF_HDA_INT_CTRL_EN | SOF_HDA_INT_GLOBAL_EN); - - /* re-enable CGCTL.MISCBDCGE after reset */ - hda_dsp_ctrl_misc_clock_gating(sdev, true); - - device_disable_async_suspend(&pci->dev); - /* enable DSP features */ snd_sof_dsp_update_bits(sdev, HDA_DSP_PP_BAR, SOF_HDA_REG_PP_PPCTL, SOF_HDA_PPCTL_GPROCEN, SOF_HDA_PPCTL_GPROCEN);
From: Zhu Yingjiang yingjiang.zhu@linux.intel.com
There are already defined ppcap and ppcap interrupt functions, use the already defined functions for easy code read.
Fixes: 8a300c8fb17 ("ASoC: SOF: Intel: Add HDA controller for Intel DSP") Signed-off-by: Zhu Yingjiang yingjiang.zhu@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/intel/hda.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-)
diff --git a/sound/soc/sof/intel/hda.c b/sound/soc/sof/intel/hda.c index e47f03dc62f0..7edeee4fc74b 100644 --- a/sound/soc/sof/intel/hda.c +++ b/sound/soc/sof/intel/hda.c @@ -540,13 +540,9 @@ int hda_dsp_probe(struct snd_sof_dev *sdev) if (ret < 0) goto free_ipc_irq;
- /* enable DSP features */ - snd_sof_dsp_update_bits(sdev, HDA_DSP_PP_BAR, SOF_HDA_REG_PP_PPCTL, - SOF_HDA_PPCTL_GPROCEN, SOF_HDA_PPCTL_GPROCEN); - - /* enable DSP IRQ */ - snd_sof_dsp_update_bits(sdev, HDA_DSP_PP_BAR, SOF_HDA_REG_PP_PPCTL, - SOF_HDA_PPCTL_PIE, SOF_HDA_PPCTL_PIE); + /* enable ppcap interrupt */ + hda_dsp_ctrl_ppcap_enable(sdev, true); + hda_dsp_ctrl_ppcap_int_enable(sdev, true);
/* initialize waitq for code loading */ init_waitqueue_head(&sdev->waitq);
From: Amadeusz Sławiński amadeuszx.slawinski@linux.intel.com
During the integration of HDaudio support, we changed the way in which we get hdev in snd_hdac_ext_bus_device_init() to use one preallocated with devm_kzalloc(), however it still left kfree(hdev) in snd_hdac_ext_bus_device_exit(). It leads to oopses when trying to rmmod and modprobe. Fix it, by just removing kfree call.
SOF also uses some of the snd_hdac_ functions for HDAudio support but allocated the memory with kzalloc. A matching fix is provided separately to align all users of the snd_hdac_ library.
Fixes: 6298542fa33b ("ALSA: hdac: remove memory allocation from snd_hdac_ext_bus_device_init") Signed-off-by: Amadeusz Sławiński amadeuszx.slawinski@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/hda/ext/hdac_ext_bus.c | 1 - 1 file changed, 1 deletion(-)
diff --git a/sound/hda/ext/hdac_ext_bus.c b/sound/hda/ext/hdac_ext_bus.c index ec7715c6b0c0..c147ebe542da 100644 --- a/sound/hda/ext/hdac_ext_bus.c +++ b/sound/hda/ext/hdac_ext_bus.c @@ -172,7 +172,6 @@ EXPORT_SYMBOL_GPL(snd_hdac_ext_bus_device_init); void snd_hdac_ext_bus_device_exit(struct hdac_device *hdev) { snd_hdac_device_exit(hdev); - kfree(hdev); } EXPORT_SYMBOL_GPL(snd_hdac_ext_bus_device_exit);
On Sat, 18 May 2019 22:27:03 +0200, Pierre-Louis Bossart wrote:
From: Amadeusz Sławiński amadeuszx.slawinski@linux.intel.com
During the integration of HDaudio support, we changed the way in which we get hdev in snd_hdac_ext_bus_device_init() to use one preallocated with devm_kzalloc(), however it still left kfree(hdev) in snd_hdac_ext_bus_device_exit(). It leads to oopses when trying to rmmod and modprobe. Fix it, by just removing kfree call.
SOF also uses some of the snd_hdac_ functions for HDAudio support but allocated the memory with kzalloc. A matching fix is provided separately to align all users of the snd_hdac_ library.
Fixes: 6298542fa33b ("ALSA: hdac: remove memory allocation from snd_hdac_ext_bus_device_init") Signed-off-by: Amadeusz Sławiński amadeuszx.slawinski@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com
Since this is an individual fix from SOF, I'll take this one via my tree.
thanks,
Takashi
sound/hda/ext/hdac_ext_bus.c | 1 - 1 file changed, 1 deletion(-)
diff --git a/sound/hda/ext/hdac_ext_bus.c b/sound/hda/ext/hdac_ext_bus.c index ec7715c6b0c0..c147ebe542da 100644 --- a/sound/hda/ext/hdac_ext_bus.c +++ b/sound/hda/ext/hdac_ext_bus.c @@ -172,7 +172,6 @@ EXPORT_SYMBOL_GPL(snd_hdac_ext_bus_device_init); void snd_hdac_ext_bus_device_exit(struct hdac_device *hdev) { snd_hdac_device_exit(hdev);
- kfree(hdev);
} EXPORT_SYMBOL_GPL(snd_hdac_ext_bus_device_exit);
-- 2.20.1
Alsa-devel mailing list Alsa-devel@alsa-project.org https://mailman.alsa-project.org/mailman/listinfo/alsa-devel
On Sun, 19 May 2019 09:30:03 +0200, Takashi Iwai wrote:
On Sat, 18 May 2019 22:27:03 +0200, Pierre-Louis Bossart wrote:
From: Amadeusz Sławiński amadeuszx.slawinski@linux.intel.com
During the integration of HDaudio support, we changed the way in which we get hdev in snd_hdac_ext_bus_device_init() to use one preallocated with devm_kzalloc(), however it still left kfree(hdev) in snd_hdac_ext_bus_device_exit(). It leads to oopses when trying to rmmod and modprobe. Fix it, by just removing kfree call.
SOF also uses some of the snd_hdac_ functions for HDAudio support but allocated the memory with kzalloc. A matching fix is provided separately to align all users of the snd_hdac_ library.
Fixes: 6298542fa33b ("ALSA: hdac: remove memory allocation from snd_hdac_ext_bus_device_init") Signed-off-by: Amadeusz Sławiński amadeuszx.slawinski@linux.intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com
Since this is an individual fix from SOF, I'll take this one via my tree.
On the second thought: as the other two patches seem depending on this change as prerequisite, maybe better to leave this together with others. So, Mark feel free to take this with the series, but make sure that they'll go to 5.2-rc.
thanks,
Takashi
From: Libin Yang libin.yang@intel.com
Align all users of the hdac library to use devm_kzalloc.
Note for backports/stable: the patch ("ALSA: hdac: fix memory release for SST and SOF drivers") needs to be applied as well.
Fixes: 5507b8103e26 ("ASoC: SOF: Intel: Add support for HDAudio codecs") Signed-off-by: Libin Yang libin.yang@intel.com Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com --- sound/soc/sof/intel/hda-codec.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/sound/soc/sof/intel/hda-codec.c b/sound/soc/sof/intel/hda-codec.c index b8b37f082309..0d8437b080bf 100644 --- a/sound/soc/sof/intel/hda-codec.c +++ b/sound/soc/sof/intel/hda-codec.c @@ -62,8 +62,7 @@ static int hda_codec_probe(struct snd_sof_dev *sdev, int address) address, resp);
#if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA_AUDIO_CODEC) - /* snd_hdac_ext_bus_device_exit will use kfree to free hdev */ - hda_priv = kzalloc(sizeof(*hda_priv), GFP_KERNEL); + hda_priv = devm_kzalloc(sdev->dev, sizeof(*hda_priv), GFP_KERNEL); if (!hda_priv) return -ENOMEM;
@@ -82,8 +81,7 @@ static int hda_codec_probe(struct snd_sof_dev *sdev, int address)
return 0; #else - /* snd_hdac_ext_bus_device_exit will use kfree to free hdev */ - hdev = kzalloc(sizeof(*hdev), GFP_KERNEL); + hdev = devm_kzalloc(sdev->dev, sizeof(*hdev), GFP_KERNEL); if (!hdev) return -ENOMEM;
On Sat, 18 May 2019 22:26:52 +0200, Pierre-Louis Bossart wrote:
No new functionality here but fixes to the IPC, memory allocation, workqueues, control, runtime_pm and HDaudio support. These issues were identified during the integration/productization of new platforms.
I added a 'Fixes' tag to make sure backports/stable can pick up these patches, in case it's too late for them to land in 5.2
The next batch for SOF will be a set of new capabilities for trace, IPC test, Hdaudio support, they will be submitted after additional tests and once these fixes are reviewed/merged.
I already gave just one comment about the minor irqflag, and with that one, feel free to take my r-b tag for the whole series: Reviewed-by: Takashi Iwai tiwai@suse.de
thanks,
Takashi
participants (3)
-
Guennadi Liakhovetski -
Pierre-Louis Bossart -
Takashi Iwai