* linux-next: manual merge of the sound-asoc tree with the slave-dma tree
@ 2012-03-05 2:59 Stephen Rothwell
0 siblings, 0 replies; 5+ messages in thread
From: Stephen Rothwell @ 2012-03-05 2:59 UTC (permalink / raw)
To: Mark Brown, Liam Girdwood
Cc: linux-next, linux-kernel, Vinod Koul, Sascha Hauer, Lars-Peter Clausen
[-- Attachment #1: Type: text/plain, Size: 514 bytes --]
Hi all,
Today's linux-next merge of the sound-asoc tree got a conflict in
sound/soc/imx/imx-pcm-dma-mx2.c between commit c0fa6c8af7ab ("ASoC: i.MX:
Add missing dma_async_issue_pending") from the slave-dma tree and commit
c307e8e32e11 ("ASoC: imx-pcm-dma: Use dmaengine PCM helper functions")
from the sound-asoc tree.
The latter removed the code modified by the former, so hopefully the
former change is subsumed into the latter.
--
Cheers,
Stephen Rothwell sfr@canb.auug.org.au
[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]
^ permalink raw reply [flat|nested] 5+ messages in thread
* linux-next: manual merge of the sound-asoc tree with the slave-dma tree
@ 2013-12-03 0:38 Stephen Rothwell
2013-12-03 16:48 ` Mark Brown
0 siblings, 1 reply; 5+ messages in thread
From: Stephen Rothwell @ 2013-12-03 0:38 UTC (permalink / raw)
To: Mark Brown, Liam Girdwood, Vinod Koul
Cc: linux-next, linux-kernel, Nicolin Chen
[-- Attachment #1: Type: text/plain, Size: 2128 bytes --]
Hi all,
Today's linux-next merge of the sound-asoc tree got a conflict in
sound/soc/fsl/fsl_ssi.c between commit bf02c7cb4314 ("ASoC: fsl_ssi: Add
dual fifo mode support") from the slave-dma tree and commit 2924a9981006
("ASoC: fsl_ssi: Add monaural audio support for non-ac97 interface") from
the sound-asoc tree.
I fixed it up (see below) and can carry the fix as necessary (no action
is required).
--
Cheers,
Stephen Rothwell sfr@canb.auug.org.au
diff --cc sound/soc/fsl/fsl_ssi.c
index f43be6d4c549,90ff1071e29c..000000000000
--- a/sound/soc/fsl/fsl_ssi.c
+++ b/sound/soc/fsl/fsl_ssi.c
@@@ -143,7 -143,7 +143,8 @@@ struct fsl_ssi_private
bool ssi_on_imx;
bool imx_ac97;
bool use_dma;
+ bool use_dual_fifo;
+ u8 i2s_mode;
struct clk *clk;
struct snd_dmaengine_dai_dma_data dma_params_tx;
struct snd_dmaengine_dai_dma_data dma_params_rx;
@@@ -388,38 -417,9 +418,15 @@@ static int fsl_ssi_setup(struct fsl_ssi
* because it is also running without an active substream. Normally SSI
* is only enabled when there is a substream.
*/
- if (ssi_private->imx_ac97) {
- /*
- * Setup the clock control register
- */
- write_ssi(CCSR_SSI_SxCCR_WL(17) | CCSR_SSI_SxCCR_DC(13),
- &ssi->stccr);
- write_ssi(CCSR_SSI_SxCCR_WL(17) | CCSR_SSI_SxCCR_DC(13),
- &ssi->srccr);
-
- /*
- * Enable AC97 mode and startup the SSI
- */
- write_ssi(CCSR_SSI_SACNT_AC97EN | CCSR_SSI_SACNT_FV,
- &ssi->sacnt);
- write_ssi(0xff, &ssi->saccdis);
- write_ssi(0x300, &ssi->saccen);
-
- /*
- * Enable SSI, Transmit and Receive
- */
- write_ssi_mask(&ssi->scr, 0, CCSR_SSI_SCR_SSIEN |
- CCSR_SSI_SCR_TE | CCSR_SSI_SCR_RE);
-
- write_ssi(CCSR_SSI_SOR_WAIT(3), &ssi->sor);
- }
+ if (ssi_private->imx_ac97)
+ fsl_ssi_setup_ac97(ssi_private);
+ if (ssi_private->use_dual_fifo) {
+ write_ssi_mask(&ssi->srcr, 0, CCSR_SSI_SRCR_RFEN1);
+ write_ssi_mask(&ssi->stcr, 0, CCSR_SSI_STCR_TFEN1);
+ write_ssi_mask(&ssi->scr, 0, CCSR_SSI_SCR_TCH_EN);
+ }
+
return 0;
}
[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: linux-next: manual merge of the sound-asoc tree with the slave-dma tree
2013-12-03 0:38 Stephen Rothwell
@ 2013-12-03 16:48 ` Mark Brown
0 siblings, 0 replies; 5+ messages in thread
From: Mark Brown @ 2013-12-03 16:48 UTC (permalink / raw)
To: Stephen Rothwell
Cc: Liam Girdwood, Vinod Koul, linux-next, linux-kernel, Nicolin Chen
[-- Attachment #1: Type: text/plain, Size: 529 bytes --]
On Tue, Dec 03, 2013 at 11:38:52AM +1100, Stephen Rothwell wrote:
> Hi all,
>
> Today's linux-next merge of the sound-asoc tree got a conflict in
> sound/soc/fsl/fsl_ssi.c between commit bf02c7cb4314 ("ASoC: fsl_ssi: Add
> dual fifo mode support") from the slave-dma tree and commit 2924a9981006
> ("ASoC: fsl_ssi: Add monaural audio support for non-ac97 interface") from
> the sound-asoc tree.
>
> I fixed it up (see below) and can carry the fix as necessary (no action
> is required).
Looks good to me, thanks.
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]
^ permalink raw reply [flat|nested] 5+ messages in thread
* linux-next: manual merge of the sound-asoc tree with the slave-dma tree
@ 2013-12-20 2:14 Stephen Rothwell
0 siblings, 0 replies; 5+ messages in thread
From: Stephen Rothwell @ 2013-12-20 2:14 UTC (permalink / raw)
To: Mark Brown, Liam Girdwood, Vinod Koul
Cc: linux-next, linux-kernel, Nicolin Chen
[-- Attachment #1: Type: text/plain, Size: 5686 bytes --]
Hi all,
Today's linux-next merge of the sound-asoc tree got a conflict in
sound/soc/fsl/fsl_ssi.c between commit 0da9e55e71bc ("ASoC: fsl_ssi: Add
dual fifo mode support") from the slave-dma tree and commits 2924a9981006
("ASoC: fsl_ssi: Add monaural audio support for non-ac97 interface") and
aafa85e71a75 ("ASoC: fsl_ssi: Add DAI master mode support for SSI on i.MX
series") from the sound-asoc tree.
I fixed it up (see below) and can carry the fix as necessary (no action
is required).
--
Cheers,
Stephen Rothwell sfr@canb.auug.org.au
diff --cc sound/soc/fsl/fsl_ssi.c
index f43be6d4c549,b2ebaf811599..000000000000
--- a/sound/soc/fsl/fsl_ssi.c
+++ b/sound/soc/fsl/fsl_ssi.c
@@@ -143,7 -140,10 +140,11 @@@ struct fsl_ssi_private
bool ssi_on_imx;
bool imx_ac97;
bool use_dma;
+ bool use_dual_fifo;
+ bool baudclk_locked;
+ u8 i2s_mode;
+ spinlock_t baudclk_lock;
+ struct clk *baudclk;
struct clk *clk;
struct snd_dmaengine_dai_dma_data dma_params_tx;
struct snd_dmaengine_dai_dma_data dma_params_rx;
@@@ -388,38 -417,9 +418,15 @@@ static int fsl_ssi_setup(struct fsl_ssi
* because it is also running without an active substream. Normally SSI
* is only enabled when there is a substream.
*/
- if (ssi_private->imx_ac97) {
- /*
- * Setup the clock control register
- */
- write_ssi(CCSR_SSI_SxCCR_WL(17) | CCSR_SSI_SxCCR_DC(13),
- &ssi->stccr);
- write_ssi(CCSR_SSI_SxCCR_WL(17) | CCSR_SSI_SxCCR_DC(13),
- &ssi->srccr);
-
- /*
- * Enable AC97 mode and startup the SSI
- */
- write_ssi(CCSR_SSI_SACNT_AC97EN | CCSR_SSI_SACNT_FV,
- &ssi->sacnt);
- write_ssi(0xff, &ssi->saccdis);
- write_ssi(0x300, &ssi->saccen);
-
- /*
- * Enable SSI, Transmit and Receive
- */
- write_ssi_mask(&ssi->scr, 0, CCSR_SSI_SCR_SSIEN |
- CCSR_SSI_SCR_TE | CCSR_SSI_SCR_RE);
-
- write_ssi(CCSR_SSI_SOR_WAIT(3), &ssi->sor);
- }
+ if (ssi_private->imx_ac97)
+ fsl_ssi_setup_ac97(ssi_private);
+ if (ssi_private->use_dual_fifo) {
+ write_ssi_mask(&ssi->srcr, 0, CCSR_SSI_SRCR_RFEN1);
+ write_ssi_mask(&ssi->stcr, 0, CCSR_SSI_STCR_TFEN1);
+ write_ssi_mask(&ssi->scr, 0, CCSR_SSI_SCR_TCH_EN);
+ }
+
return 0;
}
@@@ -438,64 -438,19 +445,28 @@@ static int fsl_ssi_startup(struct snd_p
struct snd_soc_pcm_runtime *rtd = substream->private_data;
struct fsl_ssi_private *ssi_private =
snd_soc_dai_get_drvdata(rtd->cpu_dai);
- int synchronous = ssi_private->cpu_dai_drv.symmetric_rates;
+ unsigned long flags;
- /*
- * If this is the first stream opened, then request the IRQ
- * and initialize the SSI registers.
+ /* First, we only do fsl_ssi_setup() when SSI is going to be active.
+ * Second, fsl_ssi_setup was already called by ac97_init earlier if
+ * the driver is in ac97 mode.
*/
- if (!ssi_private->first_stream) {
- ssi_private->first_stream = substream;
-
- /*
- * fsl_ssi_setup was already called by ac97_init earlier if
- * the driver is in ac97 mode.
- */
- if (!ssi_private->imx_ac97)
- fsl_ssi_setup(ssi_private);
- } else {
- if (synchronous) {
- struct snd_pcm_runtime *first_runtime =
- ssi_private->first_stream->runtime;
- /*
- * This is the second stream open, and we're in
- * synchronous mode, so we need to impose sample
- * sample size constraints. This is because STCCR is
- * used for playback and capture in synchronous mode,
- * so there's no way to specify different word
- * lengths.
- *
- * Note that this can cause a race condition if the
- * second stream is opened before the first stream is
- * fully initialized. We provide some protection by
- * checking to make sure the first stream is
- * initialized, but it's not perfect. ALSA sometimes
- * re-initializes the driver with a different sample
- * rate or size. If the second stream is opened
- * before the first stream has received its final
- * parameters, then the second stream may be
- * constrained to the wrong sample rate or size.
- */
- if (first_runtime->sample_bits) {
- snd_pcm_hw_constraint_minmax(substream->runtime,
- SNDRV_PCM_HW_PARAM_SAMPLE_BITS,
- first_runtime->sample_bits,
- first_runtime->sample_bits);
- }
- }
-
- ssi_private->second_stream = substream;
+ if (!dai->active && !ssi_private->imx_ac97) {
+ fsl_ssi_setup(ssi_private);
+ spin_lock_irqsave(&ssi_private->baudclk_lock, flags);
+ ssi_private->baudclk_locked = false;
+ spin_unlock_irqrestore(&ssi_private->baudclk_lock, flags);
}
+ /* When using dual fifo mode, it is safer to ensure an even period
+ * size. If appearing to an odd number while DMA always starts its
+ * task from fifo0, fifo1 would be neglected at the end of each
+ * period. But SSI would still access fifo1 with an invalid data.
+ */
+ if (ssi_private->use_dual_fifo)
+ snd_pcm_hw_constraint_step(substream->runtime, 0,
+ SNDRV_PCM_HW_PARAM_PERIOD_SIZE, 2);
+
return 0;
}
@@@ -962,8 -1151,11 +1167,11 @@@ static int fsl_ssi_probe(struct platfor
/* Older 8610 DTs didn't have the fifo-depth property */
ssi_private->fifo_depth = 8;
+ ssi_private->baudclk_locked = false;
+ spin_lock_init(&ssi_private->baudclk_lock);
+
if (of_device_is_compatible(pdev->dev.of_node, "fsl,imx21-ssi")) {
- u32 dma_events[2];
+ u32 dma_events[2], dmas[4];
ssi_private->ssi_on_imx = true;
ssi_private->clk = devm_clk_get(&pdev->dev, NULL);
[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]
^ permalink raw reply [flat|nested] 5+ messages in thread
* linux-next: manual merge of the sound-asoc tree with the slave-dma tree
@ 2014-01-13 1:43 Stephen Rothwell
0 siblings, 0 replies; 5+ messages in thread
From: Stephen Rothwell @ 2014-01-13 1:43 UTC (permalink / raw)
To: Mark Brown, Liam Girdwood, Vinod Koul
Cc: linux-next, linux-kernel, Nicolin Chen, Nicolin Chen
[-- Attachment #1: Type: text/plain, Size: 4171 bytes --]
Hi all,
Today's linux-next merge of the sound-asoc tree got a conflict in
sound/soc/fsl/fsl_ssi.c between commit 0da9e55e71bc ("ASoC: fsl_ssi: Add
dual fifo mode support") from the slave-dma tree and commit 2b56b5f02029
("ASoC: fsl_ssi: Set default slot number for common cases") from the
sound-asoc tree.
I fixed it up (see below) and can carry the fix as necessary (no action
is required).
--
Cheers,
Stephen Rothwell sfr@canb.auug.org.au
diff --cc sound/soc/fsl/fsl_ssi.c
index 2101fc5c5739,7864ec5cf5f9..000000000000
--- a/sound/soc/fsl/fsl_ssi.c
+++ b/sound/soc/fsl/fsl_ssi.c
@@@ -140,8 -162,9 +162,10 @@@ struct fsl_ssi_private
bool ssi_on_imx;
bool imx_ac97;
bool use_dma;
+ bool use_dual_fifo;
bool baudclk_locked;
+ bool irq_stats;
+ bool offline_config;
u8 i2s_mode;
spinlock_t baudclk_lock;
struct clk *baudclk;
@@@ -421,12 -711,17 +712,23 @@@ static int fsl_ssi_setup(struct fsl_ssi
if (ssi_private->imx_ac97)
fsl_ssi_setup_ac97(ssi_private);
+ if (ssi_private->use_dual_fifo) {
+ write_ssi_mask(&ssi->srcr, 0, CCSR_SSI_SRCR_RFEN1);
+ write_ssi_mask(&ssi->stcr, 0, CCSR_SSI_STCR_TFEN1);
+ write_ssi_mask(&ssi->scr, 0, CCSR_SSI_SCR_TCH_EN);
+ }
+
+ /*
+ * Set a default slot number so that there is no need for those common
+ * cases like I2S mode to call the extra set_tdm_slot() any more.
+ */
+ if (!ssi_private->imx_ac97) {
+ write_ssi_mask(&ssi->stccr, CCSR_SSI_SxCCR_DC_MASK,
+ CCSR_SSI_SxCCR_DC(2));
+ write_ssi_mask(&ssi->srccr, CCSR_SSI_SxCCR_DC_MASK,
+ CCSR_SSI_SxCCR_DC(2));
+ }
+
return 0;
}
@@@ -1170,8 -1343,35 +1359,35 @@@ static int fsl_ssi_probe(struct platfor
ssi_private->baudclk_locked = false;
spin_lock_init(&ssi_private->baudclk_lock);
- if (of_device_is_compatible(pdev->dev.of_node, "fsl,imx21-ssi")) {
+ /*
+ * imx51 and later SoCs have a slightly different IP that allows the
+ * SSI configuration while the SSI unit is running.
+ *
+ * More important, it is necessary on those SoCs to configure the
+ * sperate TX/RX DMA bits just before starting the stream
+ * (fsl_ssi_trigger). The SDMA unit has to be configured before fsl_ssi
+ * sends any DMA requests to the SDMA unit, otherwise it is not defined
+ * how the SDMA unit handles the DMA request.
+ *
+ * SDMA units are present on devices starting at imx35 but the imx35
+ * reference manual states that the DMA bits should not be changed
+ * while the SSI unit is running (SSIEN). So we support the necessary
+ * online configuration of fsl-ssi starting at imx51.
+ */
+ switch (hw_type) {
+ case FSL_SSI_MCP8610:
+ case FSL_SSI_MX21:
+ case FSL_SSI_MX35:
+ ssi_private->offline_config = true;
+ break;
+ case FSL_SSI_MX51:
+ ssi_private->offline_config = false;
+ break;
+ }
+
+ if (hw_type == FSL_SSI_MX21 || hw_type == FSL_SSI_MX51 ||
+ hw_type == FSL_SSI_MX35) {
- u32 dma_events[2];
+ u32 dma_events[2], dmas[4];
ssi_private->ssi_on_imx = true;
ssi_private->clk = devm_clk_get(&pdev->dev, NULL);
@@@ -1234,16 -1435,13 +1451,22 @@@
dma_events[0], shared ? IMX_DMATYPE_SSI_SP : IMX_DMATYPE_SSI);
imx_pcm_dma_params_init_data(&ssi_private->filter_data_rx,
dma_events[1], shared ? IMX_DMATYPE_SSI_SP : IMX_DMATYPE_SSI);
+ if (!of_property_read_u32_array(pdev->dev.of_node, "dmas", dmas, 4)
+ && dmas[2] == IMX_DMATYPE_SSI_DUAL) {
+ ssi_private->use_dual_fifo = true;
+ /* When using dual fifo mode, we need to keep watermark
+ * as even numbers due to dma script limitation.
+ */
+ ssi_private->dma_params_tx.maxburst &= ~0x1;
+ ssi_private->dma_params_rx.maxburst &= ~0x1;
+ }
- } else if (ssi_private->use_dma) {
+ }
+
+ /*
+ * Enable interrupts only for MCP8610 and MX51. The other MXs have
+ * different writeable interrupt status registers.
+ */
+ if (ssi_private->use_dma) {
/* The 'name' should not have any slashes in it. */
ret = devm_request_irq(&pdev->dev, ssi_private->irq,
fsl_ssi_isr, 0, ssi_private->name,
[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2014-01-13 1:44 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-03-05 2:59 linux-next: manual merge of the sound-asoc tree with the slave-dma tree Stephen Rothwell
2013-12-03 0:38 Stephen Rothwell
2013-12-03 16:48 ` Mark Brown
2013-12-20 2:14 Stephen Rothwell
2014-01-13 1:43 Stephen Rothwell
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).