2011-06-09 11:37:59

by Prashant Shah

Subject: [PATCH] Staging: comedi: dyna_pci10xx: replace semaphore with mutex

- Replace all semaphores with mutex
- Check NULL value for devpriv during detach callback

Signed-off-by: Prashant P. Shah <[email protected]>
---
drivers/staging/comedi/drivers/dyna_pci10xx.c | 24 +++++++++++-------------
1 files changed, 11 insertions(+), 13 deletions(-)

diff --git a/drivers/staging/comedi/drivers/dyna_pci10xx.c b/drivers/staging/comedi/drivers/dyna_pci10xx.c
index 7acb2ed..50abb6e 100644
--- a/drivers/staging/comedi/drivers/dyna_pci10xx.c
+++ b/drivers/staging/comedi/drivers/dyna_pci10xx.c
@@ -39,7 +39,7 @@

#include "../comedidev.h"
#include "comedi_pci.h"
-#include <linux/semaphore.h>
+#include <linux/mutex.h>

#define PCI_VENDOR_ID_DYNALOG 0x10b5
#define DRV_NAME "dyna_pci10xx"
@@ -126,7 +126,7 @@ static struct comedi_driver driver_dyna_pci10xx = {
struct dyna_pci10xx_private {
struct pci_dev *pci_dev; /* ptr to PCI device */
char valid; /* card is usable */
- struct semaphore sem;
+ struct mutex mutex;

/* device base address registers */
unsigned long BADR0, BADR1, BADR2, BADR3, BADR4, BADR5;
@@ -152,7 +152,7 @@ static int dyna_pci10xx_insn_read_ai(struct comedi_device *dev,
chan = CR_CHAN(insn->chanspec);
range = thisboard->range_codes_ai[CR_RANGE((insn->chanspec))];

- down(&devpriv->sem);
+ mutex_lock(&devpriv->mutex);
/* convert n samples */
for (n = 0; n < insn->n; n++) {
/* trigger conversion */
@@ -176,7 +176,7 @@ conv_finish:
d &= 0x0FFF;
data[n] = d;
}
- up(&devpriv->sem);
+ mutex_unlock(&devpriv->mutex);

/* return the number of samples read/written */
return n;
@@ -193,13 +193,13 @@ static int dyna_pci10xx_insn_write_ao(struct comedi_device *dev,
chan = CR_CHAN(insn->chanspec);
range = thisboard->range_codes_ai[CR_RANGE((insn->chanspec))];

- down(&devpriv->sem);
+ mutex_lock(&devpriv->mutex);
for (n = 0; n < insn->n; n++) {
/* trigger conversion and write data */
outw_p(data[n], devpriv->BADR2);
smp_mb(); udelay(10);
}
- up(&devpriv->sem);
+ mutex_unlock(&devpriv->mutex);
return n;
}

@@ -213,11 +213,11 @@ static int dyna_pci10xx_di_insn_bits(struct comedi_device *dev,
if (insn->n != 2)
return -EINVAL;

- down(&devpriv->sem);
+ mutex_lock(&devpriv->mutex);
smp_mb();
d = inw_p(devpriv->BADR3);
udelay(10);
- up(&devpriv->sem);
+ mutex_unlock(&devpriv->mutex);

/* on return the data[0] contains output and data[1] contains input */
data[1] = d;
@@ -238,15 +238,13 @@ static int dyna_pci10xx_do_insn_bits(struct comedi_device *dev,
* in data[1], each channel cooresponding to a bit.
* s->state contains the previous write data
*/
-
+ mutex_lock(&devpriv->mutex);
if (data[0]) {
- down(&devpriv->sem);
s->state &= ~data[0];
s->state |= (data[0] & data[1]);
smp_mb();
outw_p(s->state, devpriv->BADR3);
udelay(10);
- up(&devpriv->sem);
}

/*
@@ -255,7 +253,7 @@ static int dyna_pci10xx_do_insn_bits(struct comedi_device *dev,
* output values if it was a purely digital output subdevice.
*/
data[1] = s->state;
-
+ mutex_unlock(&devpriv->mutex);
return 2;
}

@@ -339,7 +337,7 @@ found:
return -EIO;
}

- sema_init(&devpriv->sem, 1);
+ mutex_init(&devpriv->mutex);
dev->board_ptr = &boardtypes[board_index];
devpriv->pci_dev = pcidev;

--
1.7.1

2011-06-09 12:00:15

by Ian Abbott

Subject: Re: [PATCH] Staging: comedi: dyna_pci10xx: replace semaphore with mutex

On 09/06/11 12:52, Prashant P. Shah wrote:
> - sema_init(&devpriv->sem, 1);
> + mutex_init(&devpriv->mutex);
> dev->board_ptr = &boardtypes[board_index];
> devpriv->pci_dev = pcidev;

You could call mutex_destroy() in your _detach routine as well. It's
used for debugging mutexes:

static int dyna_pci10xx_detach(struct comedi_device *dev)
{
if (devpriv && devpriv->pci_dev) {
comedi_pci_disable(devpriv->pci_dev);
mutex_destroy(&devpriv->mutex);
}
return 0;
}

--
-=( Ian Abbott @ MEV Ltd. E-mail: <[email protected]> )=-
-=( Tel: +44 (0)161 477 1898 FAX: +44 (0)161 718 3587 )=-

2011-06-13 07:17:55

by Prashant Shah

Subject: Re: [PATCH] Staging: comedi: dyna_pci10xx: replace semaphore with mutex

Hi,

I am resending the patches again. Cleaned up the code a bit and
removed unwanted delays.

Regards.