Skip to contentSkip to conversation

alsa/soc: add locking to mpc5200-psc-ac97 driver

PATCH In review Grant Likely 5 participants 15 messages Jun 29, 2009 – Jul 2, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 alsa/soc: add locking to mpc5200-psc-ac97 driver

From: Grant Likely <[email protected]>

AC97 bus register read/write hooks need to provide locking, but the mpc5200-psc-ac97 driver does not. This patch adds a mutex around the register access routines.

---

 sound/soc/fsl/mpc5200_dma.c      |    1 +
 sound/soc/fsl/mpc5200_dma.h      |    1 +
 sound/soc/fsl/mpc5200_psc_ac97.c |   17 ++++++++++++++++-
 3 files changed, 18 insertions(+), 1 deletions(-)
sound/soc/fsl/mpc5200_dma.c +1 -0
index efec33a..f0a2d40 100644
@@ -456,6 +456,7 @@ int mpc5200_audio_dma_create(struct of_device *op)
 		return -ENODEV;
 
 	spin_lock_init(&psc_dma->lock);
+	mutex_init(&psc_dma->mutex);
 	psc_dma->id = be32_to_cpu(*prop);
 	psc_dma->irq = irq;
 	psc_dma->psc_regs = regs;
sound/soc/fsl/mpc5200_dma.h +1 -0
index 2000803..8d396bb 100644
@@ -55,6 +55,7 @@ struct psc_dma {
 	unsigned int irq;
 	struct device *dev;
 	spinlock_t lock;
+	struct mutex mutex;
 	u32 sicr;
 	uint sysclk;
 	int imr;
sound/soc/fsl/mpc5200_psc_ac97.c +16 -1
index 794a247..7eb5499 100644
@@ -34,13 +34,20 @@ static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg)
 	int status;
 	unsigned int val;
 
+	mutex_lock(&psc_dma->mutex);
+
 	/* Wait for command send status zero = ready */
 	status = spin_event_timeout(!(in_be16(&psc_dma->psc_regs->sr_csr.status) &
 				MPC52xx_PSC_SR_CMDSEND), 100, 0);
 	if (status == 0) {
 		pr_err("timeout on ac97 bus (rdy)\n");
+		mutex_unlock(&psc_dma->mutex);
 		return -ENODEV;
 	}
+
+	/* Force clear the data valid bit */
+	in_be32(&psc_dma->psc_regs->ac97_data);
+
 	/* Send the read */
 	out_be32(&psc_dma->psc_regs->ac97_cmd, (1<<31) | ((reg & 0x7f) << 24));
 
@@ -50,16 +57,19 @@ static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg)
 	if (status == 0) {
 		pr_err("timeout on ac97 read (val) %x\n",
 				in_be16(&psc_dma->psc_regs->sr_csr.status));
+		mutex_unlock(&psc_dma->mutex);
 		return -ENODEV;
 	}
 	/* Get the data */
 	val = in_be32(&psc_dma->psc_regs->ac97_data);
 	if (((val >> 24) & 0x7f) != reg) {
 		pr_err("reg echo error on ac97 read\n");
+		mutex_unlock(&psc_dma->mutex);
 		return -ENODEV;
 	}
 	val = (val >> 8) & 0xffff;
 
+	mutex_unlock(&psc_dma->mutex);
 	return (unsigned short) val;
 }
 
@@ -68,16 +78,21 @@ static void psc_ac97_write(struct snd_ac97 *ac97,
 {
 	int status;
 
+	mutex_lock(&psc_dma->mutex);
+
 	/* Wait for command status zero = ready */
 	status = spin_event_timeout(!(in_be16(&psc_dma->psc_regs->sr_csr.status) &
 				MPC52xx_PSC_SR_CMDSEND), 100, 0);
 	if (status == 0) {
 		pr_err("timeout on ac97 bus (write)\n");
-		return;
+		goto out;
 	}
 	/* Write data */
 	out_be32(&psc_dma->psc_regs->ac97_cmd,
 			((reg & 0x7f) << 24) | (val << 8));
+
+ out:
+	mutex_unlock(&psc_dma->mutex);
 }
 
 static void psc_ac97_warm_reset(struct snd_ac97 *ac97)
#1 lore raw
Quoted 5 lines · Grant Likely

From: Grant Likely <[email protected]>

AC97 bus register read/write hooks need to provide locking, but the mpc5200-psc-ac97 driver does not.  This patch adds a mutex around the register access routines.

Does your touchscreen driver use this mutex? Or was this mutex needed just for the AC97 driver?

Signature · 2 lines
Jon Smirl
[email protected]
#2 lore raw
Quoted 2 lines · Jon Smirl

Does your touchscreen driver use this mutex? Or was this mutex needed just for the AC97 driver?

It's in the register accesses so everything accessing the chip registers will use it.

#3 lore raw
Quoted 7 lines · Mark Brown
Quoted 2 lines · Jon Smirl

Does your touchscreen driver use this mutex? Or was this mutex needed just for the AC97 driver?

It's in the register accesses so everything accessing the chip registers will use it.

The mutexes are needed. The ac97 bus doesn't have any knowledge about what codec(s) will be wired up to it so it must protect against multiple access. In my case both the wm9712 audio codec driver and the wm9712 touchscreen driver perform register accesses.

g.

Signature · 2 lines
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.
#4 lore raw
In reply to Grant Likely

Hi Grant,

Quoted 51 lines · Grant Likely

From: Grant Likely <[email protected]>

AC97 bus register read/write hooks need to provide locking, but the mpc5200-psc-ac97 driver does not. This patch adds a mutex around the register access routines.

---

 sound/soc/fsl/mpc5200_dma.c      |    1 +
 sound/soc/fsl/mpc5200_dma.h      |    1 +
 sound/soc/fsl/mpc5200_psc_ac97.c |   17 ++++++++++++++++-
 3 files changed, 18 insertions(+), 1 deletions(-)
sound/soc/fsl/mpc5200_dma.c +1 -0
index efec33a..f0a2d40 100644
@@ -456,6 +456,7 @@ int mpc5200_audio_dma_create(struct of_device *op)
 		return -ENODEV;
 
 	spin_lock_init(&psc_dma->lock);
+	mutex_init(&psc_dma->mutex);
 	psc_dma->id = be32_to_cpu(*prop);
 	psc_dma->irq = irq;
 	psc_dma->psc_regs = regs;
sound/soc/fsl/mpc5200_dma.h +1 -0
index 2000803..8d396bb 100644
@@ -55,6 +55,7 @@ struct psc_dma {
 	unsigned int irq;
 	struct device *dev;
 	spinlock_t lock;
+	struct mutex mutex;
 	u32 sicr;
 	uint sysclk;
 	int imr;
sound/soc/fsl/mpc5200_psc_ac97.c +2 -0
index 794a247..7eb5499 100644
@@ -34,13 +34,20 @@ static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg)
 	int status;
 	unsigned int val;
 
+	mutex_lock(&psc_dma->mutex);
+
 	/* Wait for command send status zero = ready */
 	status = spin_event_timeout(!(in_be16(&psc_dma->psc_regs->sr_csr.status) &

sound/soc/fsl/mpc5200_psc_ac97.c static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg) · line 50

Quoted 10 lines
sound/soc/fsl/mpc5200_psc_ac97.c +5 -0
 				MPC52xx_PSC_SR_CMDSEND), 100, 0);
 	if (status == 0) {
 		pr_err("timeout on ac97 bus (rdy)\n");
+		mutex_unlock(&psc_dma->mutex);
 		return -ENODEV;
 	}
+
+	/* Force clear the data valid bit */
+	in_be32(&psc_dma->psc_regs->ac97_data);
+

No mutex involved here. I think this is either a seperate patch or it needs at least to be mentioned in the patch description.

Quoted 51 lines
sound/soc/fsl/mpc5200_psc_ac97.c +9 -1
 	/* Send the read */
 	out_be32(&psc_dma->psc_regs->ac97_cmd, (1<<31) | ((reg & 0x7f) << 24));
 
@@ -50,16 +57,19 @@ static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg)
 	if (status == 0) {
 		pr_err("timeout on ac97 read (val) %x\n",
 				in_be16(&psc_dma->psc_regs->sr_csr.status));
+		mutex_unlock(&psc_dma->mutex);
 		return -ENODEV;
 	}
 	/* Get the data */
 	val = in_be32(&psc_dma->psc_regs->ac97_data);
 	if (((val >> 24) & 0x7f) != reg) {
 		pr_err("reg echo error on ac97 read\n");
+		mutex_unlock(&psc_dma->mutex);
 		return -ENODEV;
 	}
 	val = (val >> 8) & 0xffff;
 
+	mutex_unlock(&psc_dma->mutex);
 	return (unsigned short) val;
 }
 
@@ -68,16 +78,21 @@ static void psc_ac97_write(struct snd_ac97 *ac97,
 {
 	int status;
 
+	mutex_lock(&psc_dma->mutex);
+
 	/* Wait for command status zero = ready */
 	status = spin_event_timeout(!(in_be16(&psc_dma->psc_regs->sr_csr.status) &
 				MPC52xx_PSC_SR_CMDSEND), 100, 0);
 	if (status == 0) {
 		pr_err("timeout on ac97 bus (write)\n");
-		return;
+		goto out;
 	}
 	/* Write data */
 	out_be32(&psc_dma->psc_regs->ac97_cmd,
 			((reg & 0x7f) << 24) | (val << 8));
+
+ out:
+	mutex_unlock(&psc_dma->mutex);
 }
 
 static void psc_ac97_warm_reset(struct snd_ac97 *ac97)

_______________________________________________
Linuxppc-dev mailing list
[email protected]
https://lists.ozlabs.org/listinfo/linuxppc-dev

Signature · 2 lines
Pengutronix e.K.                           | Wolfram Sang                |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |

_______________________________________________
Linuxppc-dev mailing list
[email protected]
https://lists.ozlabs.org/listinfo/linuxppc-dev

#5 lore raw

On Tue, Jun 30, 2009 at 2:18 AM, Wolfram Sang<[email protected]> wrote:

Wolfram, do you have any way to test the on the Phytec hardware? The WM9712 on the baseboard supports a touchscreen, is it brought out to a header?

Quoted 133 lines

Hi Grant,

Quoted 57 lines · Grant Likely

From: Grant Likely <[email protected]>

AC97 bus register read/write hooks need to provide locking, but the mpc5200-psc-ac97 driver does not.  This patch adds a mutex around the register access routines.

---

 sound/soc/fsl/mpc5200_dma.c      |    1 + sound/soc/fsl/mpc5200_dma.h      |    1 + sound/soc/fsl/mpc5200_psc_ac97.c |   17 ++++++++++++++++-

 3 files changed, 18 insertions(+), 1 deletions(-)
sound/soc/fsl/mpc5200_dma.c +1 -0
index efec33a..f0a2d40 100644
@@ -456,6 +456,7 @@ int mpc5200_audio_dma_create(struct of_device *op)
               return -ENODEV;
 
       spin_lock_init(&psc_dma->lock);
+     mutex_init(&psc_dma->mutex);
       psc_dma->id = be32_to_cpu(*prop);
       psc_dma->irq = irq;
       psc_dma->psc_regs = regs;
sound/soc/fsl/mpc5200_dma.h +1 -0
index 2000803..8d396bb 100644
@@ -55,6 +55,7 @@ struct psc_dma {
       unsigned int irq;
       struct device *dev;
       spinlock_t lock;
+     struct mutex mutex;
       u32 sicr;
       uint sysclk;
       int imr;
sound/soc/fsl/mpc5200_psc_ac97.c +2 -0
index 794a247..7eb5499 100644
@@ -34,13 +34,20 @@ static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg)
       int status;
       unsigned int val;
 
+     mutex_lock(&psc_dma->mutex);
+
       /* Wait for command send status zero = ready */

      status = spin_event_timeout(!(in_be16(&psc_dma->psc_regs->sr_csr.status) & MPC52xx_PSC_SR_CMDSEND), 100, 0);
      if (status == 0) {
              pr_err("timeout on ac97 bus (rdy)\n");
+             mutex_unlock(&psc_dma->mutex);
              return -ENODEV;
      }

sound/soc/fsl/mpc5200_psc_ac97.c static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg) · line 50

Quoted 4 lines
sound/soc/fsl/mpc5200_psc_ac97.c +4 -0
+
+     /* Force clear the data valid bit */
+     in_be32(&psc_dma->psc_regs->ac97_data);
+

No mutex involved here. I think this is either a seperate patch or it needs at least to be mentioned in the patch description.

Quoted 45 lines

      /* Send the read */
      out_be32(&psc_dma->psc_regs->ac97_cmd, (1<<31) | ((reg & 0x7f) << 24));

sound/soc/fsl/mpc5200_psc_ac97.c +5 -0
@@ -50,16 +57,19 @@ static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg)
       if (status == 0) {
               pr_err("timeout on ac97 read (val) %x\n",
                               in_be16(&psc_dma->psc_regs->sr_csr.status));
+             mutex_unlock(&psc_dma->mutex);
               return -ENODEV;
       }
       /* Get the data */
       val = in_be32(&psc_dma->psc_regs->ac97_data);
       if (((val >> 24) & 0x7f) != reg) {
               pr_err("reg echo error on ac97 read\n");
+             mutex_unlock(&psc_dma->mutex);
               return -ENODEV;
       }
       val = (val >> 8) & 0xffff;
 
+     mutex_unlock(&psc_dma->mutex);
       return (unsigned short) val;
  }
 
@@ -68,16 +78,21 @@ static void psc_ac97_write(struct snd_ac97 *ac97,
  {
       int status;
 
+     mutex_lock(&psc_dma->mutex);
+
       /* Wait for command status zero = ready */

      status = spin_event_timeout(!(in_be16(&psc_dma->psc_regs->sr_csr.status) & MPC52xx_PSC_SR_CMDSEND), 100, 0);
      if (status == 0) {
              pr_err("timeout on ac97 bus (write)\n");
-             return;
+             goto out;
      }
      /* Write data */
      out_be32(&psc_dma->psc_regs->ac97_cmd,
                      ((reg & 0x7f) << 24) | (val << 8));
+
+ out:
+     mutex_unlock(&psc_dma->mutex);
 }

Quoted 6 lines

 static void psc_ac97_warm_reset(struct snd_ac97 *ac97)

_______________________________________________
Linuxppc-dev mailing list
[email protected]
https://lists.ozlabs.org/listinfo/linuxppc-dev

--
Pengutronix e.K.                           | Wolfram Sang                | Industrial Linux Solutions                 | http://www.pengutronix.de/  |

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkpJrkcACgkQD27XaX1/VRsY/QCgyx8IMANjokbNnrOQlXINRLeW lT4AnAy3ES9r3wriGkRN7ivnLA3zyRHb
=RUPr
-----END PGP SIGNATURE-----

Signature · 2 lines
Jon Smirl
[email protected]
#6 lore raw
Quoted 3 lines · Jon Smirl

Wolfram, do you have any way to test the on the Phytec hardware? The WM9712 on the baseboard supports a touchscreen, is it brought out to a header?

You can probably test adequately by writing a dummy AC97 driver that sits in a thread and does reads (which should be all that's required to test the interaction).

#7 lore raw
In reply to Jon Smirl

Hi Jon,

Quoted 3 lines

Wolfram, do you have any way to test the on the Phytec hardware? The WM9712 on the baseboard supports a touchscreen, is it brought out to a header?

Sorry, I didn't get it: Shall I test something specific?

The touchscreen connector is X19 BTW (assuming a PCM-973 baseboard).

Regards,

Wolfram

Signature · 2 lines
Pengutronix e.K.                           | Wolfram Sang                |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |

_______________________________________________
Linuxppc-dev mailing list
[email protected]
https://lists.ozlabs.org/listinfo/linuxppc-dev

#8 lore raw
Quoted 7 lines · Wolfram Sang

Hi Jon,

Quoted 3 lines

Wolfram, do you have any way to test the on the Phytec hardware? The WM9712 on the baseboard supports a touchscreen, is it brought out to a header?

Sorry, I didn't get it: Shall I test something specific?

I don't own a touchscreen.

AFAIK no one has ever plugged a touchscreen into the PCM-973 to see if works since there hasn't been a driver previously. Just do a general test so that you can tell customers that it works.

Quoted 20 lines

The touchscreen connector is X19 BTW (assuming a PCM-973 baseboard).

Regards,

  Wolfram

--
Pengutronix e.K.                           | Wolfram Sang                | Industrial Linux Solutions                 | http://www.pengutronix.de/  |

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkpLJMEACgkQD27XaX1/VRs/cwCgn7p7ffxr8WGoW7MwALkBMKtD VtYAoIa5/viinyvFF5cQP3BvH9WP3a5I
=Zrq3
-----END PGP SIGNATURE-----

Signature · 2 lines
Jon Smirl
[email protected]
#9 lore raw
Quoted 8 lines
Quoted 2 lines

Sorry, I didn't get it: Shall I test something specific?

I don't own a touchscreen.

AFAIK no one has ever plugged a touchscreen into the PCM-973 to see if works since there hasn't been a driver previously. Just do a general test so that you can tell customers that it works.

I am not sure, we have a suitable touchscreen. Also, I am afraid this is a bit too much to do besides my regular day work right now. Sorry.

Regards,

Wolfram

Signature · 2 lines
Pengutronix e.K.                           | Wolfram Sang                |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |

_______________________________________________
Linuxppc-dev mailing list
[email protected]
https://lists.ozlabs.org/listinfo/linuxppc-dev

#10 lore raw
Quoted 5 lines · Wolfram Sang
Quoted 5 lines
Quoted 2 lines

Sorry, I didn't get it: Shall I test something specific?

I don't own a touchscreen.

Quoted 7 lines
Quoted 4 lines

AFAIK no one has ever plugged a touchscreen into the PCM-973 to see if works since there hasn't been a driver previously.
Just do a general
test so that you can tell customers that it works.

I am not sure, we have a suitable touchscreen. Also, I am afraid this is a bit too much to do besides my regular day work right now. Sorry.

I've now tested this on a client's board which uses the pcm030 with a custom base board (based on the development board) which uses the same codec. It works there. I don't have a discrete touchscreen to wire up to the PCM-973 though.

g.

Signature · 2 lines
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.
#11 lore raw

-----Original Message-----
From: = linuxppc-dev-bounces+emillbrandt=3Ddekaresearch.com@lists.ozlabs.org [mailto:[email protected].= o
rg] On Behalf Of Grant Likely
Sent: Wednesday, July 01, 2009 10:45
To: Wolfram Sang
Cc: [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH] alsa/soc: add locking to mpc5200-psc-ac97 driver

Quoted 6 lines · Wolfram Sang
Quoted 6 lines
Quoted 2 lines

Sorry, I didn't get it: Shall I test something specific?

I don't own a touchscreen.

AFAIK no one has ever plugged a touchscreen into the PCM-973 to see

if

Quoted 5 lines
Quoted 3 lines

works since there hasn't been a driver previously. Just do a general
test so that you can tell customers that it works.

I am not sure, we have a suitable touchscreen. Also, I am afraid this

is a bit

Quoted 1 line

too much to do besides my regular day work right now. Sorry.

I've now tested this on a client's board which uses the pcm030 with a custom base board (based on the development board) which uses the same codec. It works there. I don't have a discrete touchscreen to wire up to the PCM-973 though.

g.

--=20

There is a header on the phytec PCM-973 base board (X20) that brings out aux1, aux2, aux3, aux4 (wiper) ADCs from the wm9712. The ADCs use chip registers the same way the touchscreen does. There is even support for doing this through the wm97xx-core driver (wm97xx_read_aux_adc). If you wanted to test it out I would put a pot or some fixed voltage divider on one of those inputs and give it a try. I don't know what the internal ADC ref is, I think its 1.65V so whatever signal you put in should be less than that.

Eric

_________________________________________________________________________=
________________

This e-mail and the information, including any attachments, it contains = are intended to be a confidential communication only to the person or = entity to whom it is addressed and may contain information that is = privileged. If the reader of this message is not the intended recipient, = you are hereby notified that any dissemination, distribution or copying = of this communication is strictly prohibited. If you have received this = communication in error, please immediately notify the sender and destroy = the original message.

Thank you.

Please consider the environment before printing this email.

#12 lore raw
In reply to Wolfram Sang
Quoted 7 lines · Wolfram Sang

sound/soc/fsl/mpc5200_psc_ac97.c static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg) · line 50

Quoted 4 lines
sound/soc/fsl/mpc5200_psc_ac97.c +4 -0
+
+     /* Force clear the data valid bit */
+     in_be32(&psc_dma->psc_regs->ac97_data);
+

No mutex involved here. I think this is either a seperate patch or it needs at least to be mentioned in the patch description.

Oops, that was sloppy. Yes, I'll put this into a separate patch. Thanks.

g.

#13 lore raw
Quoted 10 lines · Grant Likely
Quoted 7 lines · Wolfram Sang

sound/soc/fsl/mpc5200_psc_ac97.c static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg) · line 50

Quoted 4 lines
sound/soc/fsl/mpc5200_psc_ac97.c +4 -0
+
+     /* Force clear the data valid bit */
+     in_be32(&psc_dma->psc_regs->ac97_data);
+

No mutex involved here. I think this is either a separate patch or it needs at least to be mentioned in the patch description.

Oops, that was sloppy.  Yes, I'll put this into a separate patch.  Thanks.

Now that you have added the mutexes, do you ever need to force clear the valid bit?
Maybe log an error if this happens so that we can track down why.

Signature · 2 lines
Jon Smirl
[email protected]
#14 lore raw
Quoted 10 lines · Jon Smirl
Quoted 9 lines · Grant Likely
Quoted 7 lines · Wolfram Sang

sound/soc/fsl/mpc5200_psc_ac97.c static unsigned short psc_ac97_read(struct snd_ac97 *ac97, unsigned short reg) · line 50

Quoted 4 lines
sound/soc/fsl/mpc5200_psc_ac97.c +4 -0
+
+     /* Force clear the data valid bit */
+     in_be32(&psc_dma->psc_regs->ac97_data);
+

No mutex involved here. I think this is either a separate patch or it needs at least to be mentioned in the patch description.

Quoted 5 lines
Quoted 1 line

Oops, that was sloppy.  Yes, I'll put this into a separate patch.  Thanks.

Now that you have added the mutexes, do you ever need to force clear the valid bit?
Maybe log an error if this happens so that we can track down why.

I know exactly why it happened. spin_event_timeout() had a bug that would cause it to report a false positive timeout. The bug is fixed and queued for merging.

However, I still think this explicit clear should be applied. It is just a cheap register read and it prevents the driver from getting wedged after a timeout does event, for whatever reason.

g.

Signature · 2 lines
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.