diff mbox series

[06/18] ASoC: SOF: Intel: hda-mlink: add structures to parse ALT links

Message ID 20230327112931.23411-7-peter.ujfalusi@linux.intel.com (mailing list archive)
State Superseded
Headers show
Series ASoC: SOF: Intel: hda-mlink: HDaudio multi-link extension update | expand

Commit Message

Peter Ujfalusi March 27, 2023, 11:29 a.m. UTC
From: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>

Extend hdac_ext_link to store information needed for ALT
links. Follow-up patches will include more functional patches for
power-up and down.

Note that this patch suggests the use of an 'eml_lock' to serialize
access to shared registers. SoundWire-specific sequence require the
lock to be taken at a higher level, as a result the helpers added in
follow-up patches will provide 'unlocked' versions when needed.

Also note that the low-level sequences with the 'hdaml_' prefix are
taken directly from the hardware specifications - naming conventions
included. The code will be split in two, with locking and linked-list
management handled separately to avoid mixing required hardware setup
and Linux-based resource management.

Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
---
 sound/soc/sof/intel/hda-mlink.c | 217 +++++++++++++++++++++++++++++++-
 1 file changed, 214 insertions(+), 3 deletions(-)

Comments

Takashi Iwai March 30, 2023, 3:51 p.m. UTC | #1
On Mon, 27 Mar 2023 13:29:19 +0200,
Peter Ujfalusi wrote:
> 
>  int hda_bus_ml_get_capabilities(struct hdac_bus *bus)
>  {
> -	if (bus->mlcap)
> -		return snd_hdac_ext_bus_get_ml_capabilities(bus);
> +	u32 link_count;
> +	int ret;
> +	int i;
> +
> +	if (!bus->mlcap)
> +		return 0;
> +
> +	link_count = readl(bus->mlcap + AZX_REG_ML_MLCD) + 1;
> +
> +	dev_dbg(bus->dev, "HDAudio Multi-Link count: %d\n", link_count);
> +
> +	for (i = 0; i < link_count; i++) {
> +		ret = hda_ml_alloc_h2link(bus, i);
> +		if (ret < 0) {
> +			hda_bus_ml_free(bus);
> +			return ret;
> +		}
> +	}
>  	return 0;

This makes that each call of hda_bus_ml_get_capabilities() adds the
h2link entries blindly.  If the driver calls it multiple times
mistakenly (the function name sounds as if it's just a helper to query
the capability bits), it'll lead to doubly entries.  Maybe adding some
check would be safer, IMO.


thanks,

Takashi
Pierre-Louis Bossart March 30, 2023, 4:09 p.m. UTC | #2
On 3/30/23 10:51, Takashi Iwai wrote:
> On Mon, 27 Mar 2023 13:29:19 +0200,
> Peter Ujfalusi wrote:
>>
>>  int hda_bus_ml_get_capabilities(struct hdac_bus *bus)
>>  {
>> -	if (bus->mlcap)
>> -		return snd_hdac_ext_bus_get_ml_capabilities(bus);
>> +	u32 link_count;
>> +	int ret;
>> +	int i;
>> +
>> +	if (!bus->mlcap)
>> +		return 0;
>> +
>> +	link_count = readl(bus->mlcap + AZX_REG_ML_MLCD) + 1;
>> +
>> +	dev_dbg(bus->dev, "HDAudio Multi-Link count: %d\n", link_count);
>> +
>> +	for (i = 0; i < link_count; i++) {
>> +		ret = hda_ml_alloc_h2link(bus, i);
>> +		if (ret < 0) {
>> +			hda_bus_ml_free(bus);
>> +			return ret;
>> +		}
>> +	}
>>  	return 0;
> 
> This makes that each call of hda_bus_ml_get_capabilities() adds the
> h2link entries blindly.  If the driver calls it multiple times
> mistakenly (the function name sounds as if it's just a helper to query
> the capability bits), it'll lead to doubly entries.  Maybe adding some
> check would be safer, IMO.

Interesting comment, I didn't think of that one.

This function is currently called in the probe and indirectly via
hda_init_caps(). I think the driver framework guarantees the probe is
only called once, doesn't it?

we can also rename this function to make it clearer if there are any
suggestions, but the name is aligned with previous implementations of
the snd_hdac_ext stuff.
Takashi Iwai March 30, 2023, 6:07 p.m. UTC | #3
On Thu, 30 Mar 2023 18:09:36 +0200,
Pierre-Louis Bossart wrote:
> 
> 
> 
> On 3/30/23 10:51, Takashi Iwai wrote:
> > On Mon, 27 Mar 2023 13:29:19 +0200,
> > Peter Ujfalusi wrote:
> >>
> >>  int hda_bus_ml_get_capabilities(struct hdac_bus *bus)
> >>  {
> >> -	if (bus->mlcap)
> >> -		return snd_hdac_ext_bus_get_ml_capabilities(bus);
> >> +	u32 link_count;
> >> +	int ret;
> >> +	int i;
> >> +
> >> +	if (!bus->mlcap)
> >> +		return 0;
> >> +
> >> +	link_count = readl(bus->mlcap + AZX_REG_ML_MLCD) + 1;
> >> +
> >> +	dev_dbg(bus->dev, "HDAudio Multi-Link count: %d\n", link_count);
> >> +
> >> +	for (i = 0; i < link_count; i++) {
> >> +		ret = hda_ml_alloc_h2link(bus, i);
> >> +		if (ret < 0) {
> >> +			hda_bus_ml_free(bus);
> >> +			return ret;
> >> +		}
> >> +	}
> >>  	return 0;
> > 
> > This makes that each call of hda_bus_ml_get_capabilities() adds the
> > h2link entries blindly.  If the driver calls it multiple times
> > mistakenly (the function name sounds as if it's just a helper to query
> > the capability bits), it'll lead to doubly entries.  Maybe adding some
> > check would be safer, IMO.
> 
> Interesting comment, I didn't think of that one.
> 
> This function is currently called in the probe and indirectly via
> hda_init_caps(). I think the driver framework guarantees the probe is
> only called once, doesn't it?
> 
> we can also rename this function to make it clearer if there are any
> suggestions, but the name is aligned with previous implementations of
> the snd_hdac_ext stuff.

Yeah, naming it as an init function would avoid the misuse.


Takashi
Pierre-Louis Bossart March 31, 2023, 3:05 a.m. UTC | #4
>>>>  int hda_bus_ml_get_capabilities(struct hdac_bus *bus)
>>>>  {
>>>> -	if (bus->mlcap)
>>>> -		return snd_hdac_ext_bus_get_ml_capabilities(bus);
>>>> +	u32 link_count;
>>>> +	int ret;
>>>> +	int i;
>>>> +
>>>> +	if (!bus->mlcap)
>>>> +		return 0;
>>>> +
>>>> +	link_count = readl(bus->mlcap + AZX_REG_ML_MLCD) + 1;
>>>> +
>>>> +	dev_dbg(bus->dev, "HDAudio Multi-Link count: %d\n", link_count);
>>>> +
>>>> +	for (i = 0; i < link_count; i++) {
>>>> +		ret = hda_ml_alloc_h2link(bus, i);
>>>> +		if (ret < 0) {
>>>> +			hda_bus_ml_free(bus);
>>>> +			return ret;
>>>> +		}
>>>> +	}
>>>>  	return 0;
>>>
>>> This makes that each call of hda_bus_ml_get_capabilities() adds the
>>> h2link entries blindly.  If the driver calls it multiple times
>>> mistakenly (the function name sounds as if it's just a helper to query
>>> the capability bits), it'll lead to doubly entries.  Maybe adding some
>>> check would be safer, IMO.
>>
>> Interesting comment, I didn't think of that one.
>>
>> This function is currently called in the probe and indirectly via
>> hda_init_caps(). I think the driver framework guarantees the probe is
>> only called once, doesn't it?
>>
>> we can also rename this function to make it clearer if there are any
>> suggestions, but the name is aligned with previous implementations of
>> the snd_hdac_ext stuff.
> 
> Yeah, naming it as an init function would avoid the misuse.

Sure thing, I'll rename it as hda_bus_ml_init().

Thanks for the feedback.
diff mbox series

Patch

diff --git a/sound/soc/sof/intel/hda-mlink.c b/sound/soc/sof/intel/hda-mlink.c
index fbf86f2350fb..cc944311a671 100644
--- a/sound/soc/sof/intel/hda-mlink.c
+++ b/sound/soc/sof/intel/hda-mlink.c
@@ -14,14 +14,221 @@ 
 #include <sound/hda_register.h>
 #include <sound/hda-mlink.h>
 
+#include <linux/bitfield.h>
 #include <linux/module.h>
 
 #if IS_ENABLED(CONFIG_SND_SOC_SOF_HDA_MLINK)
 
+/**
+ * struct hdac_ext2_link - HDAudio extended+alternate link
+ *
+ * @hext_link:		hdac_ext_link
+ * @alt:		flag set for alternate extended links
+ * @intc:		boolean for interrupt capable
+ * @ofls:		boolean for offload support
+ * @lss:		boolean for link synchronization capabilities
+ * @slcount:		sublink count
+ * @elid:		extended link ID (AZX_REG_ML_LEPTR_ID_ defines)
+ * @elver:		extended link version
+ * @leptr:		extended link pointer
+ * @eml_lock:		mutual exclusion to access shared registers e.g. CPA/SPA bits
+ * in LCTL register
+ * @base_ptr:		pointer to shim/ip/shim_vs space
+ * @instance_offset:	offset between each of @slcount instances managed by link
+ * @shim_offset:	offset to SHIM register base
+ * @ip_offset:		offset to IP register base
+ * @shim_vs_offset:	offset to vendor-specific (VS) SHIM base
+ */
+struct hdac_ext2_link {
+	struct hdac_ext_link hext_link;
+
+	/* read directly from LCAP register */
+	bool alt;
+	bool intc;
+	bool ofls;
+	bool lss;
+	int slcount;
+	int elid;
+	int elver;
+	u32 leptr;
+
+	struct mutex eml_lock; /* prevent concurrent access to e.g. CPA/SPA */
+
+	/* internal values computed from LCAP contents */
+	void __iomem *base_ptr;
+	u32 instance_offset;
+	u32 shim_offset;
+	u32 ip_offset;
+	u32 shim_vs_offset;
+};
+
+#define hdac_ext_link_to_ext2(h) container_of(h, struct hdac_ext2_link, hext_link)
+
+#define AZX_REG_SDW_INSTANCE_OFFSET			0x8000
+#define AZX_REG_SDW_SHIM_OFFSET				0x0
+#define AZX_REG_SDW_IP_OFFSET				0x100
+#define AZX_REG_SDW_VS_SHIM_OFFSET			0x6000
+
+/* only one instance supported */
+#define AZX_REG_INTEL_DMIC_SHIM_OFFSET			0x0
+#define AZX_REG_INTEL_DMIC_IP_OFFSET			0x100
+#define AZX_REG_INTEL_DMIC_VS_SHIM_OFFSET		0x6000
+
+#define AZX_REG_INTEL_SSP_INSTANCE_OFFSET		0x1000
+#define AZX_REG_INTEL_SSP_SHIM_OFFSET			0x0
+#define AZX_REG_INTEL_SSP_IP_OFFSET			0x100
+#define AZX_REG_INTEL_SSP_VS_SHIM_OFFSET		0xC00
+
+/* only one instance supported */
+#define AZX_REG_INTEL_UAOL_SHIM_OFFSET			0x0
+#define AZX_REG_INTEL_UAOL_IP_OFFSET			0x100
+#define AZX_REG_INTEL_UAOL_VS_SHIM_OFFSET		0xC00
+
+/* HDAML section - this part follows sequences in the hardware specification,
+ * including naming conventions and the use of the hdaml_ prefix.
+ * The code is intentionally minimal with limited dependencies on frameworks or
+ * helpers. Locking and scanning lists is handled at a higher level
+ */
+
+static int hdaml_lnk_enum(struct device *dev, struct hdac_ext2_link *h2link,
+			  void __iomem *ml_addr, int link_idx)
+{
+	struct hdac_ext_link *hlink = &h2link->hext_link;
+	u32 base_offset;
+
+	hlink->lcaps  = readl(ml_addr + AZX_REG_ML_LCAP);
+
+	h2link->alt = FIELD_GET(AZX_ML_HDA_LCAP_ALT, hlink->lcaps);
+
+	/* handle alternate extensions */
+	if (!h2link->alt) {
+		h2link->slcount = 1;
+
+		/*
+		 * LSDIID is initialized by hardware for HDaudio link,
+		 * it needs to be setup by software for alternate links
+		 */
+		hlink->lsdiid = readw(ml_addr + AZX_REG_ML_LSDIID);
+
+		dev_dbg(dev, "Link %d: HDAudio - lsdiid=%d\n",
+			link_idx, hlink->lsdiid);
+
+		return 0;
+	}
+
+	h2link->intc = FIELD_GET(AZX_ML_HDA_LCAP_INTC, hlink->lcaps);
+	h2link->ofls = FIELD_GET(AZX_ML_HDA_LCAP_OFLS, hlink->lcaps);
+	h2link->lss = FIELD_GET(AZX_ML_HDA_LCAP_LSS, hlink->lcaps);
+
+	/* read slcount (increment due to zero-based hardware representation */
+	h2link->slcount = FIELD_GET(AZX_ML_HDA_LCAP_SLCOUNT, hlink->lcaps) + 1;
+	dev_dbg(dev, "Link %d: HDAudio extended - sublink count %d\n",
+		link_idx, h2link->slcount);
+
+	/* find IP ID and offsets */
+	h2link->leptr = readl(hlink->ml_addr + AZX_REG_ML_LEPTR);
+
+	h2link->elid = FIELD_GET(AZX_REG_ML_LEPTR_ID, h2link->leptr);
+
+	base_offset = FIELD_GET(AZX_REG_ML_LEPTR_PTR, h2link->leptr);
+	h2link->base_ptr = hlink->ml_addr + base_offset;
+
+	switch (h2link->elid) {
+	case AZX_REG_ML_LEPTR_ID_SDW:
+		h2link->shim_offset = AZX_REG_SDW_SHIM_OFFSET;
+		h2link->ip_offset = AZX_REG_SDW_IP_OFFSET;
+		h2link->shim_vs_offset = AZX_REG_SDW_VS_SHIM_OFFSET;
+		dev_dbg(dev, "Link %d: HDAudio extended - SoundWire alternate link, leptr.ptr %#x\n",
+			link_idx, base_offset);
+		break;
+	case AZX_REG_ML_LEPTR_ID_INTEL_DMIC:
+		h2link->shim_offset = AZX_REG_INTEL_DMIC_SHIM_OFFSET;
+		h2link->ip_offset = AZX_REG_INTEL_DMIC_IP_OFFSET;
+		h2link->shim_vs_offset = AZX_REG_INTEL_DMIC_VS_SHIM_OFFSET;
+		dev_dbg(dev, "Link %d: HDAudio extended - INTEL DMIC alternate link, leptr.ptr %#x\n",
+			link_idx, base_offset);
+		break;
+	case AZX_REG_ML_LEPTR_ID_INTEL_SSP:
+		h2link->shim_offset = AZX_REG_INTEL_SSP_SHIM_OFFSET;
+		h2link->ip_offset = AZX_REG_INTEL_SSP_IP_OFFSET;
+		h2link->shim_vs_offset = AZX_REG_INTEL_SSP_VS_SHIM_OFFSET;
+		dev_dbg(dev, "Link %d: HDAudio extended - INTEL SSP alternate link, leptr.ptr %#x\n",
+			link_idx, base_offset);
+		break;
+	case AZX_REG_ML_LEPTR_ID_INTEL_UAOL:
+		h2link->shim_offset = AZX_REG_INTEL_UAOL_SHIM_OFFSET;
+		h2link->ip_offset = AZX_REG_INTEL_UAOL_IP_OFFSET;
+		h2link->shim_vs_offset = AZX_REG_INTEL_UAOL_VS_SHIM_OFFSET;
+		dev_dbg(dev, "Link %d: HDAudio extended - INTEL UAOL alternate link, leptr.ptr %#x\n",
+			link_idx, base_offset);
+		break;
+	default:
+		dev_err(dev, "Link %d: HDAudio extended - Unsupported alternate link, leptr.id=%#02x value\n",
+			link_idx, h2link->elid);
+		return -EINVAL;
+	}
+	return 0;
+}
+
+/* END HDAML section */
+
+static int hda_ml_alloc_h2link(struct hdac_bus *bus, int index)
+{
+	struct hdac_ext2_link *h2link;
+	struct hdac_ext_link *hlink;
+	int ret;
+
+	h2link  = kzalloc(sizeof(*h2link), GFP_KERNEL);
+	if (!h2link)
+		return -ENOMEM;
+
+	/* basic initialization */
+	hlink = &h2link->hext_link;
+
+	hlink->index = index;
+	hlink->bus = bus;
+	hlink->ml_addr = bus->mlcap + AZX_ML_BASE + (AZX_ML_INTERVAL * index);
+
+	ret = hdaml_lnk_enum(bus->dev, h2link, hlink->ml_addr, index);
+	if (ret < 0) {
+		kfree(h2link);
+		return ret;
+	}
+
+	mutex_init(&h2link->eml_lock);
+
+	list_add_tail(&hlink->list, &bus->hlink_list);
+
+	/*
+	 * HDaudio regular links are powered-on by default, the
+	 * refcount needs to be initialized.
+	 */
+	if (!h2link->alt)
+		hlink->ref_count = 1;
+
+	return 0;
+}
+
 int hda_bus_ml_get_capabilities(struct hdac_bus *bus)
 {
-	if (bus->mlcap)
-		return snd_hdac_ext_bus_get_ml_capabilities(bus);
+	u32 link_count;
+	int ret;
+	int i;
+
+	if (!bus->mlcap)
+		return 0;
+
+	link_count = readl(bus->mlcap + AZX_REG_ML_MLCD) + 1;
+
+	dev_dbg(bus->dev, "HDAudio Multi-Link count: %d\n", link_count);
+
+	for (i = 0; i < link_count; i++) {
+		ret = hda_ml_alloc_h2link(bus, i);
+		if (ret < 0) {
+			hda_bus_ml_free(bus);
+			return ret;
+		}
+	}
 	return 0;
 }
 EXPORT_SYMBOL_NS(hda_bus_ml_get_capabilities, SND_SOC_SOF_HDA_MLINK);
@@ -29,13 +236,17 @@  EXPORT_SYMBOL_NS(hda_bus_ml_get_capabilities, SND_SOC_SOF_HDA_MLINK);
 void hda_bus_ml_free(struct hdac_bus *bus)
 {
 	struct hdac_ext_link *hlink, *_h;
+	struct hdac_ext2_link *h2link;
 
 	if (!bus->mlcap)
 		return;
 
 	list_for_each_entry_safe(hlink, _h, &bus->hlink_list, list) {
 		list_del(&hlink->list);
-		kfree(hlink);
+		h2link = hdac_ext_link_to_ext2(hlink);
+
+		mutex_destroy(&h2link->eml_lock);
+		kfree(h2link);
 	}
 }
 EXPORT_SYMBOL_NS(hda_bus_ml_free, SND_SOC_SOF_HDA_MLINK);