State of the MPC5200 PSC AC97 driver

Hi Sylvain,

Last year you have posted a MPC5200 PSC AC97 driver patch "[PATCH 9/9] sound: Add support for Freescale MPC5200 AC97 interface." with the following comment:

There are various FIXMEs and commented out code here and there. Could you elaborate a bit on the overall state of the driver's functionality,
which areas need improvement and attention?

Seems that you mainly tested BestComm with this driver, what was the overall
BestComm performance, any issues here? Did you use any specific test setup involving some dedicated application, etc.?

Did anyone else tried it and/or has a updated version or can share experience?

Thanks in advance.

Cheers,
m.

Sidenote: last time we tried (2.6.23.1), the AC97 driver didn't work, tested on the phyCORE-MPC5200B-tiny board. No sound, just a little bit noise which can be muted with the mixer. And music finishes about 7 times as fast as it usually should. So we didn't manage to make it work yet.

Robert

 Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de
 Pengutronix - Linux Solutions for Science and Industry
   Handelsregister:  Amtsgericht Hildesheim, HRA 2686
     Hannoversche Str. 2, 31134 Hildesheim, Germany
   Phone: +49-5121-206917-0 |  Fax: +49-5121-206917-9

I have a copy of the driver I hacked to work with the new PSC framework but it doesn't "work" anymore - I have no idea why.

Mail me privately if you want it, I have no time to make patches or look for a good reason for the breakages, but it compiles at least. The Efika hack should be gone now (irrelevant since we released a firmware that obseletes it) and the other fixmes should be by the wayside considering the fscking thing doesn't play sound anymore...

Matt Sealey <[email protected]>
Genesi, Manager, Developer Relations

Hi Matt,

I'd be interested as well; do I read your mail right that you have already managed to hear sound coming out of the MPC5200B PSC AC97 interface? You would be the first one I've heared of, so it would definitely be interesting to see your code.

Robert

 Dipl.-Ing. Robert Schwebel | http://www.pengutronix.de
 Pengutronix - Linux Solutions for Science and Industry
   Handelsregister:  Amtsgericht Hildesheim, HRA 2686
     Hannoversche Str. 2, 31134 Hildesheim, Germany
   Phone: +49-5121-206917-0 |  Fax: +49-5121-206917-9

Indeed, sound was known to work on the efika plateform, altough it is the last patch which not merged upstream, and needs some work.

I leave matt send it to you, as he has been more in touch with this lately.

Friendly,

Sven Luther

Yes ... and it still applies.

It's a minimum boiler plate.
I filled the function at the minimum to get some sound output and being able to hear music coming out of it :)
I also completely skipped the 5200 (not B) support ...

At first there was no DMA at all (full software copy). I added some simple DMA later just to prove it could work and how to use the API.

I just wrote it to get some sound, prove the interface could work under linux and to show how to use DMA. I had hoped someone else would finish it ... (yeah, I know ... big mistake).

My test was listening to Gorrillaz "Feel Good inc" using mplayer ...

At the time several other people tried it and it worked ... unless you did lots of harddrive IO and then it crumbled ...

Sylvain

Currently it fails to play anything with an external Wolfson WM9712 connect= ed=20
to it. It seems the current AC97 reset sequence switches the WM9712 into=20 testmode due to SDATA_OUT and Sync are not held low while the reset is=20 active. Any idea to solve this? Switching both signals to GPIO and force th= em=20
to low level while reset is active? How to cover PSC1 and PSC2 with differe= nt=20
GPIOs for this case?

Juergen
=2D-=20
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
=A0Pengutronix - Linux Solutions for Science and Industry
=A0 Handelsregister: Amtsgericht Hildesheim, HRA 2686
=A0 =A0 =A0 Vertretung Sued/Muenchen, Germany
Phone: +49-8766-939 228 | Fax: +49-5121-206917-9

nd

Cannot reproduce the reset issue now, but in my case the bestcomm was programmed to service the wrong PSC unit. That was the reason I never heard any sound.

Can us give the oftree a better differentiation to program the bestcomm in a correct manner?

Subject: [@num@/@total@] mpc52xx: Change the AC97 driver to be more generic =46rom: Juergen Beisert <[email protected]>

The current AC97 driver for the mpc52xx CPU is fixed to work on PSC2. This patch tries to make it more generic, as it detects the PSC unit for AC97 us= age
to forward this information into the Bestcomm-API.

=2D--

 sound/ppc/mpc52xx_ac97.c |   21 ++++++++++++++++++++-
 1 file changed, 20 insertions(+), 1 deletion(-)

Index: sound/ppc/mpc52xx_ac97.c =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=2D-- sound/ppc/mpc52xx_ac97.c.orig
+++ sound/ppc/mpc52xx_ac97.c
@@ -651,6 +651,7 @@ mpc52xx_ac97_probe(struct of_device *op,
struct mpc52xx_ac97_priv *priv;
struct snd_card *card;
struct resource res;
+ int tx_initiator;
int rv;
=20
dev_dbg(&op->dev, "probing MPC52xx PSC AC97 driver\n");
@@ -699,10 +700,27 @@ mpc52xx_ac97_probe(struct of_device *op,
/* Setup Bestcomm tasks */ spin_lock_init(&priv->dma_lock);
=20
+ /*
+ * PSC1 or PSC2 can be configured for AC97 usage. Select the right
+ * channel, to let the BCOMM unit does its job correctly.
+ */
+ switch (priv->mem_start & 0xFFFF) {
+ case 0x2000: /* PSC1 */
+ tx_initiator =3D 14;
+ break;
+ case 0x2200: /* PSC2 */
+ tx_initiator =3D 12;
+ break;
+ default:
+ dev_dbg(priv->dev, "Unknown PSC unit for AC97 usage!\n");
+ rv - ENODEV;
+ goto err_irq;
+ }
+
priv->tsk_tx =3D bcom_gen_bd_tx_init(AC97_TX_NUM_BD,
priv->mem_start + offsetof(struct mpc52xx_psc,
tfdata),
=2D 12, /* initiator : FIXME */
+ tx_initiator,
2); /* ipr : FIXME */
if (!priv->tsk_tx) {
printk(KERN_ERR DRV_NAME ": bcom_gen_bd_tx_init failed\n");
@@ -759,6 +777,7 @@ err_irqreq:
bcom_gen_bd_tx_release(priv->tsk_tx);
err_bcomm:
mpc52xx_ac97_hwshutdown(priv);
+err_irq:
irq_dispose_mapping(priv->irq);
err_irqmap:
iounmap(priv->psc);

Juergen

=2D-=20
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
=A0Pengutronix - Linux Solutions for Science and Industry
=A0 Handelsregister: Amtsgericht Hildesheim, HRA 2686
=A0 =A0 =A0 Vertretung Sued/Muenchen, Germany
Phone: +49-8766-939 228 | Fax: +49-5121-206917-9

Hi,

if someone is interested: Here the full patch to get sound support for MPC5200b and a current 2.6.25 kernel.

Index: sound/ppc/Kconfig =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=2D-- sound/ppc/Kconfig.orig 2008-04-17 15:33:27.000000000 +0200
+++ sound/ppc/Kconfig 2008-04-17 15:38:03.000000000 +0200
@@ -53,3 +53,19 @@ config SND_PS3_DEFAULT_START_DELAY
depends on SND_PS3
default "2000"
endmenu
+
+
+# ALSA ppc drivers
+
+menu "ALSA PPC devices"
+ depends on SND!=3Dn && PPC
+
+config SND_PPC_MPC52xx_AC97
+ tristate "Freescale MPC52xx AC97 interface support"
+ depends on SND && PPC_MPC52xx
+ select SND_AC97_CODEC
+ help
+ Say Y or M if you want to support any AC97 codec attached to
+ the Freescqle MPC52xx AC97 interface.
+
+endmenu
Index: sound/ppc/Makefile =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=2D-- sound/ppc/Makefile.orig 2008-04-17 15:33:27.000000000 +0200
+++ sound/ppc/Makefile 2008-04-17 15:38:03.000000000 +0200
@@ -4,7 +4,9 @@
#
=20
snd-powermac-objs :=3D powermac.o pmac.o awacs.o burgundy.o daca.o tumbler=
=2Eo keywest.o beep.o
+snd-mpc52xx-ac97-objs :=3D mpc52xx_ac97.o
=20
# Toplevel Module Dependency
obj-$(CONFIG_SND_POWERMAC) +=3D snd-powermac.o
obj-$(CONFIG_SND_PS3) +=3D snd_ps3.o
+obj-$(CONFIG_SND_PPC_MPC52xx_AC97) +=3D snd-mpc52xx-ac97.o
Index: sound/ppc/mpc52xx_ac97.c =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=2D-- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ sound/ppc/mpc52xx_ac97.c 2008-04-17 16:13:46.000000000 +0200
@@ -0,0 +1,807 @@
+/*
+ * Driver for the PSC of the Freescale MPC52xx configured as AC97 interface
+ *
+ *
+ * Copyright (C) 2006 Sylvain Munaut <[email protected]>
+ *
+ * This file is licensed under the terms of the GNU General Public License
+ * version 2. This program is licensed "as is" without any warranty of any
+ * kind, whether express or implied.
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/interrupt.h>
+#include <linux/spinlock.h>
+
+#include <sound/core.h>
+#include <sound/initval.h>
+#include <sound/pcm.h>
+#include <sound/pcm_params.h>
+#include <sound/ac97_codec.h>
+
+#include <asm/of_platform.h>
+#include <linux/dma-mapping.h>
+#include <asm/mpc52xx_psc.h>
+
+#include <sysdev/bestcomm/bestcomm.h>
+#include <sysdev/bestcomm/gen_bd.h>
+
+
+#define DRV_NAME "mpc52xx-psc-ac97"
+
+
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */
+/* Structs / Defines =
*/
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */ +
+/* Private structure */
+struct mpc52xx_ac97_priv {
+ struct device *dev;
+ resource_size_t mem_start;
+ resource_size_t mem_len;
+ int irq;
+ struct mpc52xx_psc __iomem *psc;
+ struct mpc52xx_psc_fifo __iomem *fifo;
+
+ struct bcom_task *tsk_tx;
+ spinlock_t dma_lock;
+
+ struct snd_card *card;
+ struct snd_pcm *pcm;
+ struct snd_ac97 *ac97;
+
+ struct snd_pcm_substream *substream_playback;
+
+ int period_byte_size;
+ u32 period_start, period_end, period_next_p;
+};
+
+/* Register bit definition (AC97 mode specific) */
+#define PSC_AC97_SLOT_BIT(n) (1<<(12-n))
+#define PSC_AC97_SLOTS_XMIT_SHIFT 16
+#define PSC_AC97_SLOTS_RECV_SHIFT 0
+
+/* Bestcomm options */
+#define AC97_TX_NUM_BD 32
+#define AC97_RX_NUM_BD 32
+
+static int mpc52xx_ac97_tx_fill(struct mpc52xx_ac97_priv *priv) +{
+ struct snd_pcm_runtime *rt;
+
+ u32 dma_data_ptr;
+
+ rt =3D priv->substream_playback->runtime;
+
+ dma_data_ptr =3D virt_to_phys(rt->dma_area);
+
+ priv->period_byte_size =3D frames_to_bytes(rt, rt->period_size);
+ priv->period_start =3D dma_data_ptr;
+ priv->period_end =3D dma_data_ptr + priv->period_byte_size * rt->periods;
+ priv->period_next_p =3D dma_data_ptr;
+
+ spin_lock(&priv->dma_lock);
+ while (!bcom_queue_full(priv->tsk_tx)) {
+ struct bcom_gen_bd *bd;
+
+ /* Submit a new one */
+ bd =3D (struct bcom_gen_bd *) bcom_prepare_next_buffer(priv->tsk_tx);
+ bd->status =3D priv->period_byte_size;
+ bd->buf_pa =3D priv->period_next_p;
+ bcom_submit_next_buffer(priv->tsk_tx, NULL);
+
+ /* Next pointer */
+ priv->period_next_p +=3D priv->period_byte_size;
+ if (priv->period_next_p >=3D priv->period_end)
+ priv->period_next_p =3D priv->period_start;
+ }
+ spin_unlock(&priv->dma_lock);
+
+ return 0;
+}
+
+
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */
+/* ISR routine =
*/
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */ +
+static irqreturn_t mpc52xx_ac97_tx_irq(int irq, void *dev_id) +{
+ struct mpc52xx_ac97_priv *priv =3D dev_id;
+ struct snd_pcm_runtime *rt;
+ struct bcom_gen_bd *bd;
+
+ rt =3D priv->substream_playback->runtime;
+
+ if (!bcom_buffer_done(priv->tsk_tx)) {
+ dev_dbg(priv->dev, "tx mismatch? Check correct output PSC\n");
+ bcom_disable(priv->tsk_tx);
+ }
+
+ spin_lock(&priv->dma_lock);
+ while (bcom_buffer_done(priv->tsk_tx)) {
+ /* Get the buffer back */
+ bcom_retrieve_buffer(priv->tsk_tx, NULL, NULL);
+
+ /* Submit a new one */
+ bd =3D (struct bcom_gen_bd *) bcom_prepare_next_buffer(priv->tsk_tx);
+ bd->status =3D priv->period_byte_size;
+ bd->buf_pa =3D priv->period_next_p;
+ bcom_submit_next_buffer(priv->tsk_tx, NULL);
+ bcom_enable(priv->tsk_tx);
+
+ /* Next pointer */
+ priv->period_next_p +=3D priv->period_byte_size;
+ if (priv->period_next_p >=3D priv->period_end)
+ priv->period_next_p =3D priv->period_start;
+ }
+ spin_unlock(&priv->dma_lock);
+
+ snd_pcm_period_elapsed(priv->substream_playback);
+
+ return IRQ_HANDLED;
+}
+
+
+static irqreturn_t mpc52xx_ac97_irq(int irq, void *dev_id)
+{
+ struct mpc52xx_ac97_priv *priv =3D dev_id;
+
+ static int icnt =3D 0;
+
+#if 1
+ /* Anti Crash during dev ;) */
+ if ((icnt++) > 5000)
+ out_be16(&priv->psc->mpc52xx_psc_imr, 0);
+#endif
+
+ /* Print statuts */
+ dev_dbg(priv->dev, "isr: %04x", in_be16(&priv->psc->mpc52xx_psc_imr));
+ out_8(&priv->psc->command,MPC52xx_PSC_RST_ERR_STAT);
+
+ return IRQ_HANDLED;
+}
+
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */
+/* PCM interface =
*/
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */ +
+/* HW desc */
+
+static struct snd_pcm_hardware mpc52xx_ac97_hw =3D {
+ .info =3D SNDRV_PCM_INFO_INTERLEAVED |
+ SNDRV_PCM_INFO_MMAP |
+ SNDRV_PCM_INFO_MMAP_VALID,
+ .formats =3D SNDRV_PCM_FMTBIT_S32_BE,
+ .rates =3D SNDRV_PCM_RATE_8000_48000,
+ .rate_min =3D 8000,
+ .rate_max =3D 48000,
+ .channels_min =3D 1,
+ .channels_max =3D 2, /* Support for more ? */
+ .buffer_bytes_max =3D 1024*1024,
+ .period_bytes_min =3D 512,
+ .period_bytes_max =3D 16*1024,
+ .periods_min =3D 8,
+ .periods_max =3D 1024,
+ .fifo_size =3D 512,
+};
+
+
+/* Playback */
+
+static int mpc52xx_ac97_playback_open(struct snd_pcm_substream *substream) +{
+ struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data;
+
+ dev_dbg(priv->dev, "mpc52xx_ac97_playback_open(%p)\n", substream);
+
+ substream->runtime->hw =3D mpc52xx_ac97_hw;
+
+ priv->substream_playback =3D substream;
+
+ return 0; /* FIXME */
+}
+
+static int mpc52xx_ac97_playback_close(struct snd_pcm_substream *substream) +{
+ struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data;
+ dev_dbg(priv->dev, "mpc52xx_ac97_playback_close(%p)\n", substream);
+ priv->substream_playback =3D NULL;
+ return 0; /* FIXME */
+}
+
+static int mpc52xx_ac97_playback_prepare(struct snd_pcm_substream *substre= am)
+{
+ struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data;
+
+ dev_dbg(priv->dev, "mpc52xx_ac97_playback_prepare(%p)\n", substream);
+
+ /* FIXME, need a spinlock to protect access */
+ if (substream->runtime->channels =3D=3D 1)
+ out_be32(&priv->psc->ac97_slots, 0x01000000);
+ else
+ out_be32(&priv->psc->ac97_slots, 0x03000000);
+
+ snd_ac97_set_rate(priv->ac97, AC97_PCM_FRONT_DAC_RATE,
+ substream->runtime->rate);
+
+ return 0; /* FIXME */
+}
+
+
+/* Capture */
+
+static int mpc52xx_ac97_capture_open(struct snd_pcm_substream *substream) +{
+/* struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data; */
+ return 0; /* FIXME */
+}
+
+static int mpc52xx_ac97_capture_close(struct snd_pcm_substream *substream) +{
+/* struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data; */
+ return 0; /* FIXME */
+}
+
+static int
+mpc52xx_ac97_capture_prepare(struct snd_pcm_substream *substream) +{
+/* struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data; */
+ return 0; /* FIXME */
+}
+
+
+/* Common */
+
+static int mpc52xx_ac97_hw_params(struct snd_pcm_substream *substream,
+ struct snd_pcm_hw_params *params)
+{
+ struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data;
+ int rv;
+
+ dev_dbg(priv->dev, "mpc52xx_ac97_hw_params(%p)\n", substream);
+
+ rv =3D snd_pcm_lib_malloc_pages(substream,
+ params_buffer_bytes(params));
+ if (rv < 0) {
+ printk(KERN_ERR "hw params failes\n"); /* FIXME */
+ return rv;
+ }
+
+ dev_dbg(priv->dev, "%d %d %d\n", params_buffer_bytes(params),
+ params_period_bytes(params), params_periods(params));
+
+ return 0;
+}
+
+static int mpc52xx_ac97_hw_free(struct snd_pcm_substream *substream) +{
+ struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data;
+
+ dev_dbg(priv->dev, "mpc52xx_ac97_hw_free(%p)\n", substream);
+
+ return snd_pcm_lib_free_pages(substream);
+}
+
+static int mpc52xx_ac97_trigger(struct snd_pcm_substream *substream, int c= md)
+{
+ struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data;
+ int rv =3D 0;
+
+ dev_dbg(priv->dev, "mpc52xx_ac97_trigger(%p,%d)\n", substream, cmd);
+
+ switch (cmd) {
+ case SNDRV_PCM_TRIGGER_START:
+ /* Enable TX taks */
+ bcom_gen_bd_tx_reset(priv->tsk_tx);
+ mpc52xx_ac97_tx_fill(priv);
+ bcom_enable(priv->tsk_tx);
+/*
+ out_be16(&priv->psc->mpc52xx_psc_imr, 0x0800); // 0x0100
+ out_be16(&priv->psc->mpc52xx_psc_imr, 0x0100); // 0x0100
+*/
+ /* FIXME: Shouldn't we check for overrun too ? */
+ /* also, shouldn't we just activate TX here ? */
+
+ break;
+
+ case SNDRV_PCM_TRIGGER_STOP:
+ /* Disable TX task */
+ bcom_disable(priv->tsk_tx);
+ out_be16(&priv->psc->mpc52xx_psc_imr, 0x0000); // 0x0100
+
+ break;
+
+ default:
+ rv =3D -EINVAL;
+ }
+
+ /* FIXME */
+ return rv;
+}
+
+static snd_pcm_uframes_t mpc52xx_ac97_pointer(struct snd_pcm_substream *su= bstream)
+{
+ struct snd_pcm_runtime *runtime =3D substream->runtime;
+ struct mpc52xx_ac97_priv *priv =3D substream->pcm->private_data;
+ u32 count;
+
+ count =3D priv->tsk_tx->bd[priv->tsk_tx->outdex].data[0] - priv->period_s=
tart;
+
+ return bytes_to_frames(runtime, count);
+}
+
+
+/* Ops */
+
+static struct snd_pcm_ops mpc52xx_ac97_playback_ops =3D {
+ .open =3D mpc52xx_ac97_playback_open,
+ .close =3D mpc52xx_ac97_playback_close,
+ .ioctl =3D snd_pcm_lib_ioctl,
+ .hw_params =3D mpc52xx_ac97_hw_params,
+ .hw_free =3D mpc52xx_ac97_hw_free,
+ .prepare =3D mpc52xx_ac97_playback_prepare,
+ .trigger =3D mpc52xx_ac97_trigger,
+ .pointer =3D mpc52xx_ac97_pointer,
+};
+
+static struct snd_pcm_ops mpc52xx_ac97_capture_ops =3D {
+ .open =3D mpc52xx_ac97_capture_open,
+ .close =3D mpc52xx_ac97_capture_close,
+ .ioctl =3D snd_pcm_lib_ioctl,
+ .hw_params =3D mpc52xx_ac97_hw_params,
+ .hw_free =3D mpc52xx_ac97_hw_free,
+ .prepare =3D mpc52xx_ac97_capture_prepare,
+ .trigger =3D mpc52xx_ac97_trigger,
+ .pointer =3D mpc52xx_ac97_pointer,
+};
+
+
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */
+/* AC97 Bus interface =
*/
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */ +
+static unsigned short mpc52xx_ac97_bus_read(struct snd_ac97 *ac97,
+ unsigned short reg)
+{
+ struct mpc52xx_ac97_priv *priv =3D ac97->private_data;
+ int timeout;
+ unsigned int val;
+
+ /* Wait for it to be ready */
+ timeout =3D 1000;
+ while ((--timeout) && (in_be16(&priv->psc->mpc52xx_psc_status) &
+ MPC52xx_PSC_SR_CMDSEND) )
+ udelay(10);
+
+ if (!timeout) {
+ printk(KERN_ERR DRV_NAME ": timeout on ac97 bus (rdy)\n");
+ return 0xffff;
+ }
+
+ /* Do the read */
+ out_be32(&priv->psc->ac97_cmd, (1<<31) | ((reg & 0x7f) << 24));
+
+ /* Wait for the answer */
+ timeout =3D 1000;
+ while ((--timeout) && !(in_be16(&priv->psc->mpc52xx_psc_status) &
+ MPC52xx_PSC_SR_DATA_VAL) )
+ udelay(10);
+
+ if (!timeout) {
+ printk(KERN_ERR DRV_NAME ": timeout on ac97 read (val)\n");
+ return 0xffff;
+ }
+
+ /* Get the data */
+ val =3D in_be32(&priv->psc->ac97_data);
+ if ( ((val>>24) & 0x7f) !=3D reg ) {
+ printk(KERN_ERR DRV_NAME ": reg echo error on ac97 read\n");
+ return 0xffff;
+ }
+ val =3D (val >> 8) & 0xffff;
+
+ return (unsigned short) val;
+}
+
+static void mpc52xx_ac97_bus_write(struct snd_ac97 *ac97,
+ unsigned short reg, unsigned short val)
+{
+ struct mpc52xx_ac97_priv *priv =3D ac97->private_data;
+ int timeout;
+
+ /* Wait for it to be ready */
+ timeout =3D 1000;
+ while ((--timeout) && (in_be16(&priv->psc->mpc52xx_psc_status) &
+ MPC52xx_PSC_SR_CMDSEND) )
+ udelay(10);
+
+ if (!timeout) {
+ printk(KERN_ERR DRV_NAME ": timeout on ac97 write\n");
+ return;
+ }
+
+ /* Write data */
+ out_be32(&priv->psc->ac97_cmd, ((reg & 0x7f) << 24) | (val << 8));
+}
+
+static void mpc52xx_ac97_bus_reset(struct snd_ac97 *ac97)
+{
+ struct mpc52xx_ac97_priv *priv =3D ac97->private_data;
+
+ dev_dbg(priv->dev, "ac97 codec reset\n");
+
+ /* Do a cold reset */
+ /*
+ * Note: This could interfere with some external AC97 mixers, as it
+ * could switch them into test mode, when SYNC or SDATA_OUT are not
+ * low while RES is low!
+ */
+ out_8(&priv->psc->op1, 0x02);
+ udelay(10);
+ out_8(&priv->psc->op0, 0x02);
+ udelay(50);
+
+ /* PSC recover from cold reset (cfr user manual, not sure if useful) */
+ out_be32(&priv->psc->sicr, in_be32(&priv->psc->sicr));
+}
+
+
+static struct snd_ac97_bus_ops mpc52xx_ac97_bus_ops =3D {
+ .read =3D mpc52xx_ac97_bus_read,
+ .write =3D mpc52xx_ac97_bus_write,
+ .reset =3D mpc52xx_ac97_bus_reset,
+};
+
+
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */
+/* Sound driver setup =
*/
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */ +
+static int mpc52xx_ac97_setup_pcm(struct mpc52xx_ac97_priv *priv) +{
+ int rv;
+
+ rv =3D snd_pcm_new(priv->card, DRV_NAME "-pcm", 0, 1, 1, &priv->pcm);
+ if (rv) {
+ dev_dbg(priv->dev, "%s: snd_pcm_new failed\n", DRV_NAME);
+ return rv;
+ }
+
+ rv =3D snd_pcm_lib_preallocate_pages_for_all(priv->pcm,
+ SNDRV_DMA_TYPE_CONTINUOUS, snd_dma_continuous_data(GFP_KERNEL),
+ 128*1024, 128*1024);
+ if (rv) {
+ dev_dbg(priv->dev,
+ "%s: snd_pcm_lib_preallocate_pages_for_all failed\n",
+ DRV_NAME);
+ return rv;
+ }
+
+ snd_pcm_set_ops(priv->pcm, SNDRV_PCM_STREAM_PLAYBACK,
+ &mpc52xx_ac97_playback_ops);
+ snd_pcm_set_ops(priv->pcm, SNDRV_PCM_STREAM_CAPTURE,
+ &mpc52xx_ac97_capture_ops);
+
+ priv->pcm->private_data =3D priv;
+ priv->pcm->info_flags =3D 0;
+
+ strcpy(priv->pcm->name, "Freescale MPC52xx PSC-AC97 PCM");
+
+ return 0;
+}
+
+static int mpc52xx_ac97_setup_mixer(struct mpc52xx_ac97_priv *priv) +{
+ struct snd_ac97_bus *ac97_bus;
+ struct snd_ac97_template ac97_template;
+ int rv;
+
+ rv =3D snd_ac97_bus(priv->card, 0, &mpc52xx_ac97_bus_ops, NULL, &ac97_bus=
);
+ if (rv) {
+ printk(KERN_ERR DRV_NAME ": snd_ac97_bus failed\n");
+ return rv;
+ }
+
+ memset(&ac97_template, 0, sizeof(struct snd_ac97_template));
+ ac97_template.private_data =3D priv;
+
+ rv =3D snd_ac97_mixer(ac97_bus, &ac97_template, &priv->ac97);
+ if (rv) {
+ printk(KERN_ERR DRV_NAME ": snd_ac97_mixer failed\n");
+ return rv;
+ }
+
+ return 0;
+}
+
+static int mpc52xx_ac97_hwinit(struct mpc52xx_ac97_priv *priv) +{
+ /* Reset everything first by safety */
+ out_8(&priv->psc->command,MPC52xx_PSC_RST_RX);
+ out_8(&priv->psc->command,MPC52xx_PSC_RST_TX);
+ out_8(&priv->psc->command,MPC52xx_PSC_RST_ERR_STAT);
+
+ /* Do a cold reset of codec */
+ /*
+ * Note: This could interfere with some external AC97 mixers, as it
+ * could switch them into test mode, when SYNC or SDATA_OUT are not
+ * low while RES is low!
+ */
+ out_8(&priv->psc->op1, 0x02);
+ udelay(10);
+ out_8(&priv->psc->op0, 0x02);
+ udelay(50);
+
+ /* Configure AC97 enhanced mode */
+ out_be32(&priv->psc->sicr, 0x03010000);
+
+ /* No slots active */
+ out_be32(&priv->psc->ac97_slots, 0x00000000);
+
+ /* No IRQ */
+ out_be16(&priv->psc->mpc52xx_psc_imr, 0x0000);
+
+ /* FIFO levels */
+ out_8(&priv->fifo->rfcntl, 0x07);
+ out_8(&priv->fifo->tfcntl, 0x07);
+ out_be16(&priv->fifo->rfalarm, 0x80);
+ out_be16(&priv->fifo->tfalarm, 0x80);
+
+ /* Go */
+ out_8(&priv->psc->command,MPC52xx_PSC_TX_ENABLE);
+ out_8(&priv->psc->command,MPC52xx_PSC_RX_ENABLE);
+
+ return 0;
+}
+
+static int mpc52xx_ac97_hwshutdown(struct mpc52xx_ac97_priv *priv) +{
+ /* No IRQ */
+ out_be16(&priv->psc->mpc52xx_psc_imr, 0x0000);
+
+ /* Disable TB & RX */
+ out_8(&priv->psc->command,MPC52xx_PSC_RST_RX);
+ out_8(&priv->psc->command,MPC52xx_PSC_RST_TX);
+
+ /* FIXME : Reset or put codec in low power ? */
+
+ return 0;
+}
+
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */
+/* OF Platform Driver =
*/
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */ +
+static int __devinit
+mpc52xx_ac97_probe(struct of_device *op, const struct of_device_id *match) +{
+ struct device_node *dn =3D op->node;
+ struct mpc52xx_ac97_priv *priv;
+ struct snd_card *card;
+ struct resource res;
+ int tx_initiator;
+ int rv;
+ const unsigned int *devno;
+
+ dev_dbg(&op->dev, "probing MPC52xx PSC AC97 driver\n");
+
+ /* Get card structure */
+ rv =3D -ENOMEM;
+ card =3D snd_card_new(SNDRV_DEFAULT_IDX1, SNDRV_DEFAULT_STR1,
+ THIS_MODULE, sizeof(struct mpc52xx_ac97_priv));
+ if (!card)
+ goto err_early;
+
+ priv =3D card->private_data;
+
+ /* Init our private structure */
+ priv->card =3D card;
+ priv->dev =3D &op->dev;
+
+ /* Get resources (mem,irq,...) */
+ rv =3D of_address_to_resource(dn, 0, &res);
+ if (rv)
+ goto err_early;
+
+ priv->mem_start =3D res.start;
+ priv->mem_len =3D res.end - res.start + 1;
+
+ if (!request_mem_region(priv->mem_start, priv->mem_len, DRV_NAME)) {
+ dev_err(&op->dev, "%s: request_mem_region failed\n", DRV_NAME);
+ rv =3D -EBUSY;
+ goto err_early;
+ }
+
+ priv->psc =3D ioremap(priv->mem_start, priv->mem_len);
+ if (!priv->psc) {
+ dev_err(&op->dev, "%s: ioremap failed\n", DRV_NAME);
+ rv =3D -ENOMEM;
+ goto err_iomap;
+ }
+ /* the fifo starts right after psc ends */
+ priv->fifo =3D (struct mpc52xx_psc_fifo*)&priv->psc[1]; /* FIXME */
+
+ priv->irq =3D irq_of_parse_and_map(dn, 0);
+ if (priv->irq =3D=3D NO_IRQ) {
+ dev_err(&op->dev, "%s: irq_of_parse_and_map failed\n",
+ DRV_NAME);
+ rv =3D -EBUSY;
+ goto err_irqmap;
+ }
+
+ /* Setup Bestcomm tasks */
+ spin_lock_init(&priv->dma_lock);
+
+ /*
+ * PSC1 or PSC2 can be configured for AC97 usage. Select the right
+ * channel, to let the BCOMM unit does its job correctly.
+ */
+ devno =3D of_get_property(dn, "cell-index", NULL);
+ switch (*devno) {
+ case 0: /* PSC1 */
+ tx_initiator =3D 14;
+ break;
+ case 1: /* PSC2 */
+ tx_initiator =3D 12;
+ break;
+ default:
+ dev_dbg(priv->dev, "Unknown PSC unit for AC97 usage!\n");
+ rv =3D -ENODEV;
+ goto err_irq;
+ }
+
+ priv->tsk_tx =3D bcom_gen_bd_tx_init(AC97_TX_NUM_BD,
+ priv->mem_start + sizeof(struct mpc52xx_psc) +
+ offsetof(struct mpc52xx_psc_fifo, tfdata),
+ tx_initiator,
+ 2); /* ipr : FIXME */
+ if (!priv->tsk_tx) {
+ dev_err(&op->dev, "%s: bcom_gen_bd_tx_init failed\n",
+ DRV_NAME);
+ rv =3D -ENOMEM;
+ goto err_bcomm;
+ }
+
+ /* Low level HW Init */
+ mpc52xx_ac97_hwinit(priv);
+
+ /* Request IRQ now that we're 'stable' */
+ rv =3D request_irq(priv->irq, mpc52xx_ac97_irq, 0, DRV_NAME, priv);
+ if (rv < 0) {
+ dev_err(&op->dev, "%s: request_irq failed\n", DRV_NAME);
+ goto err_irqreq;
+ }
+
+ rv =3D request_irq(bcom_get_task_irq(priv->tsk_tx),
+ mpc52xx_ac97_tx_irq, 0, DRV_NAME "_tx", priv);
+ if (rv < 0) {
+ dev_err(&op->dev, "%s: request_irq failed\n", DRV_NAME);
+ goto err_txirqreq;
+ }
+
+ /* Prepare sound stuff */
+ rv =3D mpc52xx_ac97_setup_mixer(priv);
+ if (rv)
+ goto err_late;
+
+ rv =3D mpc52xx_ac97_setup_pcm(priv);
+ if (rv)
+ goto err_late;
+
+ /* Finally register the card */
+ snprintf(card->shortname, sizeof(card->shortname), DRV_NAME);
+ snprintf(card->longname, sizeof(card->longname),
+ "Freescale MPC52xx PSC-AC97 (%s)", card->mixername);
+
+ rv =3D snd_card_register(card);
+ if (rv) {
+ dev_err(&op->dev, "%s: snd_card_register failed\n", DRV_NAME);
+ goto err_late;
+ }
+
+ dev_set_drvdata(&op->dev, priv);
+
+ return 0;
+
+err_late:
+ free_irq(bcom_get_task_irq(priv->tsk_tx), priv);
+err_txirqreq:
+ free_irq(priv->irq, priv);
+err_irqreq:
+ bcom_gen_bd_tx_release(priv->tsk_tx);
+err_bcomm:
+ mpc52xx_ac97_hwshutdown(priv);
+err_irq:
+ irq_dispose_mapping(priv->irq);
+err_irqmap:
+ iounmap(priv->psc);
+err_iomap:
+ release_mem_region(priv->mem_start, priv->mem_len);
+err_early:
+ if (card)
+ snd_card_free(card);
+ return rv;
+}
+
+static int mpc52xx_ac97_remove(struct of_device *op)
+{
+ struct mpc52xx_ac97_priv *priv;
+
+ dev_dbg(&op->dev, "removing MPC52xx PSC AC97 driver\n");
+
+ priv =3D dev_get_drvdata(&op->dev);
+ if (priv) {
+ /* Sound subsys shutdown */
+ snd_card_free(priv->card);
+
+ /* Low level HW shutdown */
+ mpc52xx_ac97_hwshutdown(priv);
+
+ /* Release bestcomm tasks */
+ free_irq(bcom_get_task_irq(priv->tsk_tx), priv);
+ bcom_gen_bd_tx_release(priv->tsk_tx);
+
+ /* Release resources */
+ iounmap(priv->psc);
+ free_irq(priv->irq, priv);
+ irq_dispose_mapping(priv->irq);
+ release_mem_region(priv->mem_start, priv->mem_len);
+ }
+
+ dev_set_drvdata(&op->dev, NULL);
+
+ return 0;
+}
+
+
+static struct of_device_id mpc52xx_ac97_of_match[] =3D {
+ {
+ .type =3D "sound",
+ .compatible =3D "mpc5200b-psc-ac97", /* B only for now */
+ },
+};
+
+static struct of_platform_driver mpc52xx_ac97_of_driver =3D {
+ .owner =3D THIS_MODULE,
+ .name =3D DRV_NAME,
+ .match_table =3D mpc52xx_ac97_of_match,
+ .probe =3D mpc52xx_ac97_probe,
+ .remove =3D mpc52xx_ac97_remove,
+ .driver =3D {
+ .name =3D DRV_NAME,
+ },
+};
+
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */
+/* Module =
*/
+/* =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D */ +
+static int __init mpc52xx_ac97_init(void)
+{
+ int rv;
+
+ printk(KERN_INFO "Sound: MPC52xx PSC AC97 driver\n");
+
+ rv =3D of_register_platform_driver(&mpc52xx_ac97_of_driver);
+ if (rv) {
+ printk(KERN_ERR DRV_NAME ": "
+ "of_register_platform_driver failed (%i)\n", rv);
+ return rv;
+ }
+
+ return 0;
+}
+
+static void __exit mpc52xx_ac97_exit(void)
+{
+ of_unregister_platform_driver(&mpc52xx_ac97_of_driver);
+}
+
+module_init(mpc52xx_ac97_init); +module_exit(mpc52xx_ac97_exit);
+
+MODULE_AUTHOR("Sylvain Munaut <[email protected]>"); +MODULE_DESCRIPTION(DRV_NAME ": Freescale MPC52xx PSC AC97 driver"); +MODULE_LICENSE("GPL");
+
Index: include/asm-powerpc/mpc52xx_psc.h =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=2D-- include/asm-powerpc/mpc52xx_psc.h.orig 2008-04-17 16:12:09.000000000 =
+0200
+++ include/asm-powerpc/mpc52xx_psc.h 2008-04-17 16:13:38.000000000 +0200
@@ -28,6 +28,10 @@
#define MPC52xx_PSC_MAXNUM 6
=20
/* Programmable Serial Controller (PSC) status register bits */
+#define MPC52xx_PSC_SR_UNEX_RX 0x0001
+#define MPC52xx_PSC_SR_DATA_VAL 0x0002
+#define MPC52xx_PSC_SR_DATA_OVR 0x0004
+#define MPC52xx_PSC_SR_CMDSEND 0x0008
#define MPC52xx_PSC_SR_CDE 0x0080
#define MPC52xx_PSC_SR_RXRDY 0x0100
#define MPC52xx_PSC_SR_RXFULL 0x0200
@@ -132,8 +136,10 @@ struct mpc52xx_psc {
u8 reserved5[3];
u8 ctlr; /* PSC + 0x1c */
u8 reserved6[3];
=2D u16 ccr; /* PSC + 0x20 */
=2D u8 reserved7[14];
+ u32 ccr; /* PSC + 0x20 */
+ u32 ac97_slots; /* PSC + 0x24 */
+ u32 ac97_cmd; /* PSC + 0x28 */
+ u32 ac97_data; /* PSC + 0x2c */
u8 ivr; /* PSC + 0x30 */
u8 reserved8[3];
u8 ip; /* PSC + 0x34 */

Regards,
Juergen

=2D-=20
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
=A0Pengutronix - Linux Solutions for Science and Industry
=A0 Handelsregister: Amtsgericht Hildesheim, HRA 2686
=A0 =A0 =A0 Vertretung Sued/Muenchen, Germany
Phone: +49-8766-939 228 | Fax: +49-5121-206917-9

Hooray! :)

Does it work, though, with your board?

Matt Sealey <[email protected]>
Genesi, Manager, Developer Relations

Hmm, my board runs an MPC5200B. How can we solve this u16 versus u32 issue = for=20
both CPUs?

In the oftree one need somthing like that:

		// PSC1 is ac97
		ac97@2000 {
			device_type =3D "sound";
			compatible =3D "mpc5200b-psc-ac97","fsl,mpc5200b-psc-ac97";
			cell-index =3D <0>; // or <1> if it is PSC2! Very important!
			reg =3D <2000 100>;
			interrupts =3D <2 2 0>;
			interrupt-parent =3D <&mpc5200_pic>;
		};

Don't know if it uses the current correct content.

Juergen

=2D-=20
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
=C2=A0Pengutronix - Linux Solutions for Science and Industry
=C2=A0 Handelsregister: Amtsgericht Hildesheim, HRA 2686
=C2=A0 =C2=A0 =C2=A0 Vertretung Sued/Muenchen, Germany
Phone: +49-8766-939 228 | Fax: +49-5121-206917-9

Use a union like the rest of the bits of the PSC/FIFO structures which differ per-chip or per-functionality, I'd say.

You can't solve this with the device tree.. you literally have to write to the right place, which is dictated to you by the chip you run (which is determined from the device tree) but actually writing to that place can't be done by writing a 16-bit value into the ccr location (since it will write into the low-order 16-bits of the 32-bit ccr, which is not actually the CCR register on 5200).

If running a 5200 or 5200B has been determined, then the driver can then mangle it's CCR setting code to match. The PSC UART driver already has a small workaround for this although I am not sure it is the cleanest method in the world..

Matt Sealey <[email protected]>
Genesi, Manager, Developer Relations

yes. ;-)

=2D-=20
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
=C2=A0Pengutronix - Linux Solutions for Science and Industry
=C2=A0 Handelsregister: Amtsgericht Hildesheim, HRA 2686
=C2=A0 =C2=A0 =C2=A0 Vertretung Sued/Muenchen, Germany
Phone: +49-8766-939 228 | Fax: +49-5121-206917-9

"priv->psc" is of type "struct mpc52xx_ac97_priv*". If I add 0x58 to it,=20 wouldn't I add 0x58 times the size of "struct mpc52xx_ac97_priv"?

Juergen

=2D-=20
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
=C2=A0Pengutronix - Linux Solutions for Science and Industry
=C2=A0 Handelsregister: Amtsgericht Hildesheim, HRA 2686
=C2=A0 =C2=A0 =C2=A0 Vertretung Sued/Muenchen, Germany
Phone: +49-8766-939 228 | Fax: +49-5121-206917-9

I always got a result of MBAR+PSC_OFFSET(n)+0x58 out of it as I expected.

priv->psc is of type struct mpc52xx_psc * which means it's just pointer arithmetic. If you add a value to it (not increment or so as if it's an array) then it just adds the value, no?

Matt Sealey <[email protected]>
Genesi, Manager, Developer Relations

Of course no -- that'll be pointer arithmetic unless you cast the pointer
to an integer type.

WBR, Sergei

Hello,

Juergen Beisert írta:

First of all, thanks for fixing the driver. I just tested it on the EFIKA, it (almost) works. In its current state loading the kernel module results in a segfault and oops. The fix is to add a missing line to the device-tree. The forth version of the EFIKA fix works fine (taken from http://www.powerdeveloper.org/platforms/efika/devicetree ):

efika:~ # cat /mnt/sound.forth
\ FORTH

s" /builtin/sound" find-device
\ Audio is on PSC2, just for informational purposes 1 encode-int s" cell-index" property
device-end
efika:~ #

But trying to fix it up from prom_init.c does not seem to work, the missing 'cell-index' property does not show up. Please let me know, how to add it properly. Here is what I tried:

factory:/usr/src/linux-2.6.25 # diff -u arch/powerpc/kernel/prom_init.c.orig arch/powerpc/kernel/prom_init.c

@@ -2212,6 +2212,7 @@
 

static void __init fixup_device_tree_efika(void)
{

+       int sound_ci = 1;
       int sound_irq[3] = { 2, 2, 0 };
       int bcomm_irq[3*16] = { 3,0,0, 3,1,0, 3,2,0, 3,3,0,
                               3,4,0, 3,5,0, 3,6,0, 3,7,0,
@@ -2257,6 +2258,8 @@
               rv = prom_getprop(node, "interrupts", prop, sizeof(prop));
               if (rv == PROM_ERROR) {
                       prom_printf("Adding sound interrupts property\n");
+                        prom_setprop(node, "/builtin/sound", "cell-index",
+                                     &sound_ci, sizeof(int));
                       prom_setprop(node, "/builtin/sound", "interrupts",
                                    sound_irq, sizeof(sound_irq));
               }

factory:/usr/src/linux-2.6.25 #

Bye,
CzP

00

I'm not sure if "cell-index" is a correct property. I copied it from the ua= rt=20
driver, because this driver needs something to distinguish between PSC1 and= =20
PSC2. Maybe there is a better and correct oftree solution? Any oftree exper= t=20
here?

Juergen

=2D-=20
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
=A0Pengutronix - Linux Solutions for Science and Industry
=A0 Handelsregister: Amtsgericht Hildesheim, HRA 2686
=A0 =A0 =A0 Vertretung Sued/Muenchen, Germany
Phone: +49-8766-939 228 | Fax: +49-5121-206917-9

cell-index is correct by legacy, but I think it has been said by that you could just as well pick which PSC you are running on by it's address.

I would prefer the address approach, but everyone's going to keep cell-index around anyway so why not leave it there and why not keep using it?

Matt Sealey <[email protected]>
Genesi, Manager, Developer Relations

I was the one who started using cell-index in the first place; but I've had a change of opinion and I think doing it by address would be better. That being said, it doesn't hurt to have the cell-index property in there. If/when we rework the drivers to use address, then it will just be a harmless and superfluous property.

BTW, is anyone trying to shepherd this driver into the ALSA tree? Its out of my area of expertise and responsibility, so I haven't been pursuing it.

Cheers,
g.

Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.

We (well, I) were going to try but we got some oopses; from one thing or another, but with Juergen's fixes and cell-index in place, now it plays audio.

Now it can be pushed to ALSA on that basis, but I'm still trying to get my head around where we need to throw patches and if it goes into ALSA Drivers first or the kernel or both? It's kind of convoluted compared to the ease of just CC'ing netdev or linux-ide..

If you want to push it, please do, but also I guess it'll need to be bundled with the cell-index patch for the fixups (we couldn't get it to make the property for some odd reason) or a change to using addresses in the driver for PSC selection (Juergen's original patch did this anyway)

Matt Sealey <[email protected]>
Genesi, Manager, Developer Relations

I can pick up the cell-index patch, but I won't pick up the ALSA stuff; that's not my tree. Post the driver patch to the alsa list (but still cc linuxppc and me). The driver needs to go into ALSA drivers first. The kernel is updated from there.

Cheers,
g.

Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.

Matt,

te:

rs

Please don't forget what Sylvian said about this driver: "I also completely= =20
skipped the 5200 (not B) support ..." and your own "I think it should be le= ft=20
noted here that the CCR size changed from 16 bits to 32 bits from 5200 to=20 5200B in order to reduce confusion.". Its not solved yet. So any user of an= =20
old MPC5200 will be surprised...

Juergen

=2D-=20
Dipl.-Ing. Juergen Beisert | http://www.pengutronix.de
=C2=A0Pengutronix - Linux Solutions for Science and Industry
=C2=A0 Handelsregister: Amtsgericht Hildesheim, HRA 2686
=C2=A0 =C2=A0 =C2=A0 Vertretung Sued/Muenchen, Germany
Phone: +49-8766-939 228 | Fax: +49-5121-206917-9

+static struct of_device_id mpc52xx_ac97_of_match[] = {
+ {
+ .type = "sound",
+ .compatible = "mpc5200b-psc-ac97", /* B only for now */
+ },
+};

It shouldn't even load, the matchlist is 5200b only..? No alarms and no surprises.. :)

Matt Sealey <[email protected]>
Genesi, Manager, Developer Relations

This probably isn't the ideal solution; but I've got no objection to bringing it in as a work around until someone can make the driver cleaner.

However, this change needs to be more robust. It must first check if the cell-index property already exists before trying to create it (just like how adding the interrupts property works).

Cheers,
g.

Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.

It misses a 'MODULE_DEVICE_TABLE(of, mpc52xx_ac97_of_match);' and a nul termination of struct mpc52xx_ac97_of_match.
This will allow autoload of the drivers, as it is done in most other kernel drivers.

Olaf

All,

Thanks for the input and comments.

I am also having a look at the I2S driver. There were attempts some time ago and I found few pieces of code floating on net but couldn't find anything cleaned up. Did anyone tried to put the results of those attempts together?

Cheers,
m.

I'm actually working on the I2S driver now and I'll probably have something to post about a month from now. This is funded development, so I'll actually have lots of time to spend on it when I get back from ELC.

Cheers,
g.

Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.

Nice, I'll most probably work on it quite soon as well. It would be good to coordinate efforts, will ping you when I am about to start.

Cheers,
m.