Skip to contentSkip to conversation

Add Loongson HD Audio support

PATCH RESEND In review Yanteng Si 3 participants 7 messages Jun 7, 2023

Subsystems (3)
lore
Share & tools
lore

Yanteng Si

author
#0 lore raw

v3 cover Add Loongson HD Audio support

v3:
* Adding some comments.
* Seting polling_mode flag.

v2:
* Handling with two new flags added to struct hdac_bus.

v1:
* The Loongson HDA controller is compatible with High Definition Audio Specification Revision 1.0a.
See <https://loongson.github.io/LoongArch-Documentation/Loongson-
7A1000-usermanual-EN.html#hda-controller>

Yanteng Si (4):
ALSA: hda: Add Loongson LS7A HD-Audio support
ALSA: hda: Using polling mode for loongson controller by default
ALSA: hda: Workaround for SDnCTL register on loongson
ALSA: hda/intel: Workaround for WALLCLK register for loongson
controller

 include/linux/pci_ids.h     |  3 +++
 include/sound/hdaudio.h     |  2 ++
 sound/hda/hdac_controller.c |  5 ++++-
 sound/hda/hdac_device.c     |  1 +
 sound/hda/hdac_stream.c     |  6 +++++-
 sound/pci/hda/hda_intel.c   | 20 ++++++++++++++++++++
 sound/pci/hda/patch_hdmi.c  |  1 +
 7 files changed, 36 insertions(+), 2 deletions(-)
Signature · 1 line
2.31.4
author
#1 lore raw

PATCH RESEND v3 1/4 ALSA: hda: Add Loongson LS7A HD-Audio support

Add the new PCI ID 0x0014 0x7a07 and the new PCI ID 0x0014 0x7a37 Loongson HDA controller.

---
 include/linux/pci_ids.h    | 3 +++
 sound/hda/hdac_device.c    | 1 +
 sound/pci/hda/hda_intel.c  | 7 +++++++
 sound/pci/hda/patch_hdmi.c | 1 +
 4 files changed, 12 insertions(+)
include/linux/pci_ids.h +3 -0
index 22f4be95b3f1..543998b44e4c 100644
@@ -158,6 +158,9 @@
 
 #define PCI_VENDOR_ID_LOONGSON		0x0014
 
+#define PCI_DEVICE_ID_LOONGSON_HDA      0x7a07
+#define PCI_DEVICE_ID_LOONGSON_HDMI     0x7a37
+
 #define PCI_VENDOR_ID_SOLIDIGM		0x025e
 
 #define PCI_VENDOR_ID_TTTECH		0x0357
sound/hda/hdac_device.c +1 -0
index 6c043fbd606f..bbf7bcdb449a 100644
@@ -645,6 +645,7 @@ struct hda_vendor_id {
 };
 
 static const struct hda_vendor_id hda_vendor_ids[] = {
+	{ 0x0014, "Loongson" },
 	{ 0x1002, "ATI" },
 	{ 0x1013, "Cirrus Logic" },
 	{ 0x1057, "Motorola" },
sound/pci/hda/hda_intel.c +7 -0
index 3226691ac923..9c353dc7740c 100644
@@ -237,6 +237,7 @@ enum {
 	AZX_DRIVER_CTHDA,
 	AZX_DRIVER_CMEDIA,
 	AZX_DRIVER_ZHAOXIN,
+	AZX_DRIVER_LOONGSON,
 	AZX_DRIVER_GENERIC,
 	AZX_NUM_DRIVERS, /* keep this as last entry */
 };
@@ -360,6 +361,7 @@ static const char * const driver_short_names[] = {
 	[AZX_DRIVER_CTHDA] = "HDA Creative",
 	[AZX_DRIVER_CMEDIA] = "HDA C-Media",
 	[AZX_DRIVER_ZHAOXIN] = "HDA Zhaoxin",
+	[AZX_DRIVER_LOONGSON] = "HDA Loongson",
 	[AZX_DRIVER_GENERIC] = "HD-Audio Generic",
 };
 
@@ -2809,6 +2811,11 @@ static const struct pci_device_id azx_ids[] = {
 	  .driver_data = AZX_DRIVER_GENERIC | AZX_DCAPS_PRESET_ATI_HDMI },
 	/* Zhaoxin */
 	{ PCI_DEVICE(0x1d17, 0x3288), .driver_data = AZX_DRIVER_ZHAOXIN },
+	/* Loongson HDAudio*/
+	{PCI_DEVICE(PCI_VENDOR_ID_LOONGSON, PCI_DEVICE_ID_LOONGSON_HDA),
+	  .driver_data = AZX_DRIVER_LOONGSON },
+	{PCI_DEVICE(PCI_VENDOR_ID_LOONGSON, PCI_DEVICE_ID_LOONGSON_HDMI),
+	  .driver_data = AZX_DRIVER_LOONGSON },
 	{ 0, }
 };
 MODULE_DEVICE_TABLE(pci, azx_ids);
sound/pci/hda/patch_hdmi.c +1 -0
index 5c0b1a09fd57..260d3e64f658 100644
@@ -4505,6 +4505,7 @@ static int patch_gf_hdmi(struct hda_codec *codec)
  * patch entries
  */
 static const struct hda_device_id snd_hda_id_hdmi[] = {
+HDA_CODEC_ENTRY(0x00147a47, "Loongson HDMI",	patch_generic_hdmi),
 HDA_CODEC_ENTRY(0x1002793c, "RS600 HDMI",	patch_atihdmi),
 HDA_CODEC_ENTRY(0x10027919, "RS600 HDMI",	patch_atihdmi),
 HDA_CODEC_ENTRY(0x1002791a, "RS690/780 HDMI",	patch_atihdmi),
Signature · 1 line
2.31.4
author
#2 lore raw

PATCH RESEND v3 2/4 ALSA: hda: Using polling mode for loongson controller by default

In reply to Yanteng Si

On loongson controller, RIRBSTS.RINTFL cannot be cleared, azx_interrupt() is called all the time. We disable RIRB interrupt, and use polling mode by default.

---
 include/sound/hdaudio.h     | 1 +
 sound/hda/hdac_controller.c | 5 ++++-
 sound/pci/hda/hda_intel.c   | 5 +++++
 3 files changed, 10 insertions(+), 1 deletion(-)
include/sound/hdaudio.h +1 -0
index 97f09acae302..a0bb40a4b721 100644
@@ -347,6 +347,7 @@ struct hdac_bus {
 	bool corbrp_self_clear:1;	/* CORBRP clears itself after reset */
 	bool polling_mode:1;
 	bool needs_damn_long_delay:1;
+	bool not_use_interrupts:1;	/* prohibiting the RIRB IRQ */
 
 	int poll_count;
 
sound/hda/hdac_controller.c +4 -1
index 3c7af6558249..7f3a000fab0c 100644
@@ -79,7 +79,10 @@ void snd_hdac_bus_init_cmd_io(struct hdac_bus *bus)
 	/* set N=1, get RIRB response interrupt for new entry */
 	snd_hdac_chip_writew(bus, RINTCNT, 1);
 	/* enable rirb dma and response irq */
-	snd_hdac_chip_writeb(bus, RIRBCTL, AZX_RBCTL_DMA_EN | AZX_RBCTL_IRQ_EN);
+	if (bus->not_use_interrupts)
+		snd_hdac_chip_writeb(bus, RIRBCTL, AZX_RBCTL_DMA_EN);
+	else
+		snd_hdac_chip_writeb(bus, RIRBCTL, AZX_RBCTL_DMA_EN | AZX_RBCTL_IRQ_EN);
 	/* Accept unsolicited responses */
 	snd_hdac_chip_updatel(bus, GCTL, AZX_GCTL_UNSOL, AZX_GCTL_UNSOL);
 	spin_unlock_irq(&bus->reg_lock);
sound/pci/hda/hda_intel.c +5 -0
index 9c353dc7740c..b7a7a92d03ef 100644
@@ -1875,6 +1875,11 @@ static int azx_first_init(struct azx *chip)
 	if (chip->driver_type == AZX_DRIVER_GFHDMI)
 		bus->polling_mode = 1;
 
+	if (chip->driver_type == AZX_DRIVER_LOONGSON) {
+		bus->polling_mode = 1;
+		bus->not_use_interrupts = 1;
+	}
+
 	err = pcim_iomap_regions(pci, 1 << 0, "ICH HD audio");
 	if (err < 0)
 		return err;
Signature · 1 line
2.31.4
author
#3 lore raw

PATCH RESEND v3 3/4 ALSA: hda: Workaround for SDnCTL register on loongson

In reply to Yanteng Si

On loongson controller, after calling snd_hdac_stream_updateb() to enable DMA engine, the SDnCTL.STRM will become to zero. We need to access SDnCTL in dword to keep SDnCTL.STRM is not changed.

Yingkun Meng <[email protected]>

---
 include/sound/hdaudio.h   | 1 +
 sound/hda/hdac_stream.c   | 6 +++++-
 sound/pci/hda/hda_intel.c | 1 +
 3 files changed, 7 insertions(+), 1 deletion(-)
include/sound/hdaudio.h +1 -0
index a0bb40a4b721..94efbbeaa531 100644
@@ -348,6 +348,7 @@ struct hdac_bus {
 	bool polling_mode:1;
 	bool needs_damn_long_delay:1;
 	bool not_use_interrupts:1;	/* prohibiting the RIRB IRQ */
+	bool access_sdnctl_in_dword:1;	/* accessing the sdnctl register by dword */
 
 	int poll_count;
 
sound/hda/hdac_stream.c +5 -1
index 1f56fd33b9af..2633a4bb1d85 100644
@@ -150,7 +150,11 @@ void snd_hdac_stream_start(struct hdac_stream *azx_dev)
 					stripe_ctl);
 	}
 	/* set DMA start and interrupt mask */
-	snd_hdac_stream_updateb(azx_dev, SD_CTL,
+	if (bus->access_sdnctl_in_dword)
+		snd_hdac_stream_updatel(azx_dev, SD_CTL,
+				0, SD_CTL_DMA_START | SD_INT_MASK);
+	else
+		snd_hdac_stream_updateb(azx_dev, SD_CTL,
 				0, SD_CTL_DMA_START | SD_INT_MASK);
 	azx_dev->running = true;
 }
sound/pci/hda/hda_intel.c +1 -0
index b7a7a92d03ef..fc4787c7782a 100644
@@ -1878,6 +1878,7 @@ static int azx_first_init(struct azx *chip)
 	if (chip->driver_type == AZX_DRIVER_LOONGSON) {
 		bus->polling_mode = 1;
 		bus->not_use_interrupts = 1;
+		bus->access_sdnctl_in_dword = 1;
 	}
 
 	err = pcim_iomap_regions(pci, 1 << 0, "ICH HD audio");
Signature · 1 line
2.31.4
author
#4 lore raw

PATCH RESEND v3 4/4 ALSA: hda/intel: Workaround for WALLCLK register for loongson controller

In reply to Yanteng Si

On loongson controller, the value of WALLCLK register is always 0, which is meaningless, so we return directly.

---
 sound/pci/hda/hda_intel.c | 7 +++++++
 1 file changed, 7 insertions(+)
sound/pci/hda/hda_intel.c +7 -0
index fc4787c7782a..ef831770ca7d 100644
@@ -655,6 +655,13 @@ static int azx_position_ok(struct azx *chip, struct azx_dev *azx_dev)
 	unsigned int pos;
 	snd_pcm_uframes_t hwptr, target;
 
+	/*
+	 * The value of the WALLCLK register is always 0
+	 * on the Loongson controller, so we return directly.
+	 */
+	if (chip->driver_type == AZX_DRIVER_LOONGSON)
+		return 1;
+
 	wallclk = azx_readl(chip, WALLCLK) - azx_dev->core.start_wallclk;
 	if (wallclk < (azx_dev->core.period_wallclk * 2) / 3)
 		return -1;	/* bogus (too early) interrupt */
Signature · 1 line
2.31.4
maintainer Acked-by on the v3 series
#5 lore raw
In reply to Yanteng Si
Quoted 34 lines · Yanteng Si

v3:
* Adding some comments.
* Seting polling_mode flag.

v2:
* Handling with two new flags added to struct hdac_bus.

v1:
* The Loongson HDA controller is compatible with High Definition Audio Specification Revision 1.0a.
See <https://loongson.github.io/LoongArch-Documentation/Loongson-
7A1000-usermanual-EN.html#hda-controller>

Yanteng Si (4):
ALSA: hda: Add Loongson LS7A HD-Audio support
ALSA: hda: Using polling mode for loongson controller by default
ALSA: hda: Workaround for SDnCTL register on loongson
ALSA: hda/intel: Workaround for WALLCLK register for loongson
controller

 include/linux/pci_ids.h     |  3 +++
 include/sound/hdaudio.h     |  2 ++
 sound/hda/hdac_controller.c |  5 ++++-
 sound/hda/hdac_device.c     |  1 +
 sound/hda/hdac_stream.c     |  6 +++++-
 sound/pci/hda/hda_intel.c   | 20 ++++++++++++++++++++
 sound/pci/hda/patch_hdmi.c  |  1 +
 7 files changed, 36 insertions(+), 2 deletions(-)

--
2.31.4

#6 lore raw
In reply to Yanteng Si
Quoted 14 lines · Yanteng Si

v3:
* Adding some comments.
* Seting polling_mode flag.

v2:
* Handling with two new flags added to struct hdac_bus.

v1:
* The Loongson HDA controller is compatible with High Definition Audio Specification Revision 1.0a.
See <https://loongson.github.io/LoongArch-Documentation/Loongson-
7A1000-usermanual-EN.html#hda-controller>

Quoted 6 lines

Yanteng Si (4):
ALSA: hda: Add Loongson LS7A HD-Audio support
ALSA: hda: Using polling mode for loongson controller by default
ALSA: hda: Workaround for SDnCTL register on loongson
ALSA: hda/intel: Workaround for WALLCLK register for loongson
controller

Applied all four patches now.

The patch 3 had a typo (missing Signed-off-by: for Yingkun, so I corrected it manually.

thanks,

Takashi