On Fri, 30 May 2008 02:37:54 -0300 Daniel Ribeiro <[EMAIL PROTECTED]> wrote:
> I did some more research on this and now i am convinced that it is I2S
> MSB with PXA slave.
> The following (untested, its time to sleep ;) ) patches fixes I2S
> emulation on pxa2xx-ssp.c and removes the override hack on ezx.c.
Hi Daniel,
<OffTopic>
Could you please separate paragraphs with a blank line?
It is easier to read on the screen rather then using indentation only.
My lazy brain gets confused by your style.
</OffTopic>
Thanks for the patch. Some in-line questions follow, just to be sure
I'm understanding the code :)
I attach the patch ported to the new code.
SND_SOC_DAIFMT_MSB is for I2S emulation in MSB mode, shouldn't the
format be called SND_SOC_DAIFMT_I2S_MSB?
> [asoc-pxa-ssp-i2s.patch text/x-diff (2,8KB)]
>
> @@ -446,15 +413,17 @@
> return -EINVAL;
> }
>
> + SSPSP_P(port) |= SSPSP_SCMODE(2);
> switch (fmt & SND_SOC_DAIFMT_FORMAT_MASK) {
> case SND_SOC_DAIFMT_DSP_A:
> SSPSP_P(port) |= SSPSP_DMYSTRT(1);
> case SND_SOC_DAIFMT_DSP_B:
> - SSPSP_P(port) |= SSPSP_SCMODE(2);
> break;
> case SND_SOC_DAIFMT_I2S:
> + SSPSP_P(port) |= SSPSP_FSRT;
This is an OR and there's actually no 'break' here, right?
> case SND_SOC_DAIFMT_MSB:
> - /* handled above */
> + SSCR0_P(port) |= SSCR0_EDSS | SCCR0_DataSize(16);
> + SSPSP_P(port) |= SSPSP_SFRMWDTH(16);
> break;
> default:
> return -EINVAL;
> @@ -484,7 +453,13 @@
> [ezx-asoc-i2s.patch text/x-diff (1,5KB)]
>
> @@ -111,8 +111,13 @@
> return ret;
>
> /* set cpu DAI configuration */
> - ret = cpu_dai->dai_ops.set_fmt(cpu_dai, SND_SOC_DAIFMT_DSP_B |
> + if (codec_dai->id == PCAP2_STEREO_DAI)
> + ret = cpu_dai->dai_ops.set_fmt(cpu_dai, SND_SOC_DAIFMT_MSB |
> SND_SOC_DAIFMT_IB_IF | SND_SOC_DAIFMT_CBM_CFM);
^ ^
AFAICS the Ms refer to codec point of view, so we are setting pxa as
slave here, right?
Regards,
Antonio
--
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
Web site: http://www.studenti.unina.it/~ospite
Public key: http://www.studenti.unina.it/~ospite/aopubkey.asc
Fix I2S emulation on pxa2xx-ssp.
Signed-off-by: Daniel Ribeiro <[EMAIL PROTECTED]>
Index: linux-2.6.25.3/sound/soc/pxa/pxa2xx-ssp.c
===================================================================
--- linux-2.6.25.3.orig/sound/soc/pxa/pxa2xx-ssp.c
+++ linux-2.6.25.3/sound/soc/pxa/pxa2xx-ssp.c
@@ -392,42 +392,13 @@
unsigned int fmt)
{
void __iomem *mmio_base = ssp[cpu_dai->id].ssp->mmio_base;
- int val;
+ int val, tmp;
/* reset port settings */
__raw_writel(0, mmio_base + SSCR0);
__raw_writel(0, mmio_base + SSCR1);
__raw_writel(0, mmio_base + SSPSP);
- /* NOTE: I2S emulation is still very much work in progress here */
-
- /* FIXME: this is what wince uses for msb */
- if ((fmt & SND_SOC_DAIFMT_FORMAT_MASK) == SND_SOC_DAIFMT_MSB) {
- __raw_writel(SSCR0_EDSS | SSCR0_TISSP | SSCR0_DataSize(16),
- mmio_base + SSCR0);
- goto master;
- }
-
- /* check for I2S emulation mode - handle it separately */
- if ((fmt & SND_SOC_DAIFMT_FORMAT_MASK) == SND_SOC_DAIFMT_I2S) {
- /* 8.4.11 */
-
- /* Only SSCR0[NCS] or SSCR0[ECS] bit fields settings are optional */
- __raw_writel(SSCR0_EDSS | SSCR0_PSP | SSCR0_DataSize(16),
- mmio_base + SSCR0);
-
- /* set FIFO thresholds */
- __raw_writel(SSCR1_RxTresh(14) | SSCR1_TxTresh(1),
- mmio_base + SSCR1);
-
- /* normal: */
- /* all bit fields must be cleared except: FSRT = 1 and
- * SFRMWDTH = 16, DMYSTART=0,1) */
- __raw_writel(SSPSP_FSRT | SSPSP_SFRMWDTH(16) | SSPSP_DMYSTRT(0),
- mmio_base + SSPSP);
- goto master;
- }
-
val = __raw_readl(mmio_base + SSCR0) | SSCR0_PSP;
__raw_writel(val, mmio_base + SSCR0);
__raw_writel(SSCR1_RxTresh(14) | SSCR1_TxTresh(1) |
@@ -465,15 +436,20 @@
}
val = __raw_readl(mmio_base + SSPSP);
+ val |= SSPSP_SCMODE(2);
switch (fmt & SND_SOC_DAIFMT_FORMAT_MASK) {
case SND_SOC_DAIFMT_DSP_A:
val |= SSPSP_DMYSTRT(1);
case SND_SOC_DAIFMT_DSP_B:
- val |= SSPSP_SCMODE(2);
break;
case SND_SOC_DAIFMT_I2S:
+ val |= SSPSP_FSRT;
case SND_SOC_DAIFMT_MSB:
- /* handled above */
+ val |= SSPSP_SFRMWDTH(16);
+
+ tmp = __raw_readl(mmio_base + SSCR0) | SSCR0_EDSS |
+ SSCR0_DataSize(16);
+ __raw_writel(tmp, mmio_base + SSCR0);
break;
default:
return -EINVAL;
@@ -505,12 +481,20 @@
pr_debug("pxa2xx_ssp_hw_params: dma %d", dma);
+ sscr0 = __raw_readl(mmio_base + SSCR0);
+
/* we can only change the settings if the port is not in use */
- if (__raw_readl(mmio_base + SSCR0) & SSCR0_SSE)
- return 0;
+ if (sscr0 & SSCR0_SSE)
+ goto ret;
+
+ /* check if we are running on I2S mode */
+ /* FIXME: Is there a better way to check this? */
+ if ((sscr0 & (SSCR0_EDSS | SSCR0_DataSize(16) | SSCR0_PSP)) ==
+ (SSCR0_EDSS | SSCR0_DataSize(16) | SSCR0_PSP))
+ goto ret;
/* clear selected SSP bits */
- sscr0 = __raw_readl(mmio_base + SSCR0) & ~(SSCR0_DSS | SSCR0_EDSS);
+ sscr0 &= ~(SSCR0_DSS | SSCR0_EDSS);
__raw_writel(sscr0, mmio_base + SSCR0);
/* bit size */
@@ -530,6 +514,7 @@
}
__raw_writel(sscr0, mmio_base + SSCR0);
+ret:
pr_debug("SSCR0 0x%08x SSCR1 0x%08x SSTO 0x%08x SSPSP 0x%08x SSSR 0x%08x SSACD 0x%08x",
__raw_readl(mmio_base + SSCR0), __raw_readl(mmio_base + SSCR1),
__raw_readl(mmio_base + SSTO), __raw_readl(mmio_base + SSPSP),
pgpRoENXrIKiC.pgp
Description: PGP signature
