[PATCH 1/3] ASoC: amd: use DMA addr rather than CPU pa for audio buffer

[PATCH 1/3] ASoC: amd: use DMA addr rather than CPU pa for audio buffer

[Ravulapati Vishnu vardhan rao]

We shouldn't assume CPU physical address we get from page_to_phys()
is same as DMA address we get from dma_alloc_coherent(). On x86_64,
we won't run into any problem with the assumption when dma_ops is
nommu_dma_ops. However, DMA address is IOVA when IOMMU is enabled.
And it's most likely different from CPU physical address when AMD
IOMMU is not in passthrough mode.

Signed-off-by: Vijendar Mukunda <vijendar.mukunda@amd.com>
Tested-by: RAVULAPATI VISHNU VARDHAN RAO <Vishnuvardhanrao.Ravulapati@amd.com>
---
 sound/soc/amd/raven/acp3x-pcm-dma.c | 23 +++++++++++------------
 1 file changed, 11 insertions(+), 12 deletions(-)

diff --git a/sound/soc/amd/raven/acp3x-pcm-dma.c b/sound/soc/amd/raven/acp3x-pcm-dma.c
index a4ade6b..49c4872 100644
--- a/sound/soc/amd/raven/acp3x-pcm-dma.c
+++ b/sound/soc/amd/raven/acp3x-pcm-dma.c
@@ -31,7 +31,7 @@ struct i2s_stream_instance {
 	u16 num_pages;
 	u16 channels;
 	u32 xfer_resolution;
-	struct page *pg;
+	dma_addr_t dma_addr;
 	u64 bytescount;
 	void __iomem *acp3x_base;
 };
@@ -211,11 +211,13 @@ static irqreturn_t i2s_irq_handler(int irq, void *dev_id)
 static void config_acp3x_dma(struct i2s_stream_instance *rtd, int direction)
 {
 	u16 page_idx;
-	u64 addr;
+	dma_addr_t addr;
 	u32 low, high, val, acp_fifo_addr;
-	struct page *pg = rtd->pg;
 
-	/* 8 scratch registers used to map one 64 bit address */
+	addr = rtd->dma_addr;
+	/* 8 scratch registers used to map one 64 bit address.
+	 * For 2 pages (8192 * 2 bytes), it will be 16 registers.
+	 */
 	if (direction == SNDRV_PCM_STREAM_PLAYBACK)
 		val = 0;
 	else
@@ -229,7 +231,6 @@ static void config_acp3x_dma(struct i2s_stream_instance *rtd, int direction)
 
 	for (page_idx = 0; page_idx < rtd->num_pages; page_idx++) {
 		/* Load the low address of page int ACP SRAM through SRBM */
-		addr = page_to_phys(pg);
 		low = lower_32_bits(addr);
 		high = upper_32_bits(addr);
 
@@ -239,7 +240,7 @@ static void config_acp3x_dma(struct i2s_stream_instance *rtd, int direction)
 				+ 4);
 		/* Move to next physically contiguos page */
 		val += 8;
-		pg++;
+		addr += PAGE_SIZE;
 	}
 
 	if (direction == SNDRV_PCM_STREAM_PLAYBACK) {
@@ -340,10 +341,8 @@ static int acp3x_dma_hw_params(struct snd_pcm_substream *substream,
 			       struct snd_pcm_hw_params *params)
 {
 	int status;
-	u64 size;
-	struct page *pg;
-	struct snd_pcm_runtime *runtime = substream->runtime;
-	struct i2s_stream_instance *rtd = runtime->private_data;
+	uint64_t size;
+	struct i2s_stream_instance *rtd = substream->runtime->private_data;
 
 	if (!rtd)
 		return -EINVAL;
@@ -355,8 +354,8 @@ static int acp3x_dma_hw_params(struct snd_pcm_substream *substream,
 
 	memset(substream->runtime->dma_area, 0, params_buffer_bytes(params));
 	pg = virt_to_page(substream->dma_buffer.area);
-	if (pg) {
-		rtd->pg = pg;
+	if (substream->dma_buffer.area) {
+		rtd->dma_addr = substream->dma_buffer.addr;
 		rtd->num_pages = (PAGE_ALIGN(size) >> PAGE_SHIFT);
 		config_acp3x_dma(rtd, substream->stream);
 		status = 0;
-- 
2.7.4

[PATCH 2/3] ASoC: amd: use dma_ops of parent device for acp3x dma driver

[Ravulapati Vishnu vardhan rao]

AMD platform device acp_audio_dma can only be created by parent PCI
device driver. Pass struct device of the parent
to snd_pcm_lib_preallocate_pages() so dma_alloc_coherent()
can use correct dma_ops. Otherwise, it will
use default dma_ops which is nommu_dma_ops on x86_64 even when
IOMMU is enabled and set to non passthrough mode.

Though platform device inherits some dma related fields during its
creation, we can't simply pass its struct device
to snd_pcm_lib_preallocate_pages() because dma_ops is not among the
inherited fields. Even it were, drivers/iommu/amd_iommu.c would
ignore it because get_device_id() doesn't handle platform device.

This change shouldn't give us any trouble even struct device of the
parent becomes null or represents some non PCI device in the future,
because get_dma_ops() correctly handles null struct device or uses
the default dma_ops if struct device doesn't have it set.

Signed-off-by: Vijendar Mukunda <vijendar.mukunda@amd.com>
Tested-by: Ravulapati, Vishnu vardhan rao <Vishnuvardhanrao.Ravulapati@amd.com>
---
 sound/soc/amd/raven/acp3x-pcm-dma.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/sound/soc/amd/raven/acp3x-pcm-dma.c b/sound/soc/amd/raven/acp3x-pcm-dma.c
index 49c4872..c74b094b 100644
--- a/sound/soc/amd/raven/acp3x-pcm-dma.c
+++ b/sound/soc/amd/raven/acp3x-pcm-dma.c
@@ -384,9 +384,14 @@ static snd_pcm_uframes_t acp3x_dma_pointer(struct snd_pcm_substream *substream)
 
 static int acp3x_dma_new(struct snd_soc_pcm_runtime *rtd)
 {
-	snd_pcm_lib_preallocate_pages_for_all(rtd->pcm, SNDRV_DMA_TYPE_DEV,
-					      rtd->pcm->card->dev,
-					      MIN_BUFFER, MAX_BUFFER);
+	struct snd_soc_component *component = snd_soc_rtdcom_lookup(rtd,
+								    DRV_NAME);
+	struct device *parent = component->dev->parent;
+
+	snd_pcm_lib_preallocate_pages_for_all(rtd->pcm,
+							SNDRV_DMA_TYPE_DEV,
+							parent, MIN_BUFFER,
+							MAX_BUFFER);
 	return 0;
 }
 
-- 
2.7.4

[PATCH 3/3] ASoC : amd: reduced period size

[Ravulapati Vishnu vardhan rao]

Reduced period size and offsets.

Signed-off-by:Ravulapati, Vishnu vardhan rao <Vishnuvardhanrao.Ravulapati@amd.com>
---
 sound/soc/amd/raven/acp3x.h | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/sound/soc/amd/raven/acp3x.h b/sound/soc/amd/raven/acp3x.h
index 4f2cadd..033e2a9 100644
--- a/sound/soc/amd/raven/acp3x.h
+++ b/sound/soc/amd/raven/acp3x.h
@@ -23,17 +23,17 @@
 #define ACP_SRAM_PTE_OFFSET	0x02050000
 #define PAGE_SIZE_4K_ENABLE 0x2
 #define MEM_WINDOW_START	0x4000000
-#define PLAYBACK_FIFO_ADDR_OFFSET 0x400
-#define CAPTURE_FIFO_ADDR_OFFSET  0x500
+#define PLAYBACK_FIFO_ADDR_OFFSET 0x00
+#define CAPTURE_FIFO_ADDR_OFFSET  0x04
 
 #define PLAYBACK_MIN_NUM_PERIODS    2
 #define PLAYBACK_MAX_NUM_PERIODS    8
-#define PLAYBACK_MAX_PERIOD_SIZE    16384
-#define PLAYBACK_MIN_PERIOD_SIZE    4096
+#define PLAYBACK_MAX_PERIOD_SIZE    8192
+#define PLAYBACK_MIN_PERIOD_SIZE    1024
 #define CAPTURE_MIN_NUM_PERIODS     2
 #define CAPTURE_MAX_NUM_PERIODS     8
-#define CAPTURE_MAX_PERIOD_SIZE     16384
-#define CAPTURE_MIN_PERIOD_SIZE     4096
+#define CAPTURE_MAX_PERIOD_SIZE     8192
+#define CAPTURE_MIN_PERIOD_SIZE     1024
 
 #define MAX_BUFFER (PLAYBACK_MAX_PERIOD_SIZE * PLAYBACK_MAX_NUM_PERIODS)
 #define MIN_BUFFER MAX_BUFFER
-- 
2.7.4

Re: [PATCH 3/3] ASoC : amd: reduced period size

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 421 bytes --]

On Mon, Jul 29, 2019 at 05:38:31PM +0530, Ravulapati Vishnu vardhan rao wrote:
> Reduced period size and offsets.

Why?

>  #define PLAYBACK_MAX_NUM_PERIODS    8
> -#define PLAYBACK_MAX_PERIOD_SIZE    16384
> -#define PLAYBACK_MIN_PERIOD_SIZE    4096
> +#define PLAYBACK_MAX_PERIOD_SIZE    8192
> +#define PLAYBACK_MIN_PERIOD_SIZE    1024

Is there a need to also reduce the maximum, the hardware culd support it
before?

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH 1/3] ASoC: amd: use DMA addr rather than CPU pa for audio buffer

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 891 bytes --]

On Mon, Jul 29, 2019 at 05:38:29PM +0530, Ravulapati Vishnu vardhan rao wrote:
> We shouldn't assume CPU physical address we get from page_to_phys()
> is same as DMA address we get from dma_alloc_coherent(). On x86_64,
> we won't run into any problem with the assumption when dma_ops is
> nommu_dma_ops. However, DMA address is IOVA when IOMMU is enabled.
> And it's most likely different from CPU physical address when AMD
> IOMMU is not in passthrough mode.

This doesn't build for me:

sound/soc/amd/raven/acp3x-pcm-dma.c: In function ‘acp3x_dma_hw_params’:
sound/soc/amd/raven/acp3x-pcm-dma.c:356:2: error: ‘pg’ undeclared (first use in this function)
  pg = virt_to_page(substream->dma_buffer.area);
  ^~
sound/soc/amd/raven/acp3x-pcm-dma.c:356:2: note: each undeclared identifier is reported only once for each function it appears in

Please check and resend.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]