Skip to contentSkip to conversation

comedi: Change error return code for if statement in the function,cb_pcimdas_ai_rinsn

PATCH Hartley Sweeten 2 participants 2 messages Feb 25, 2015 – Feb 26, 2015 2 messages not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw
Quoted 15 lines · Nicholas Krause

This changes us using the incorrect error,-ETIMEOUT when checking if the channel we are allocating to on the device structure pointer passed to this function is greater then the maximum available channels for this device to the correct error for a channel being out of range,-ECHRNG.

---
 drivers/staging/comedi/drivers/cb_pcimdas.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/comedi/drivers/cb_pcimdas.c +0 -0
index 70dd2c9..d91a6f3 100644

drivers/staging/comedi/drivers/cb_pcimdas.c static int cb_pcimdas_ai_rinsn(struct comedi_device *dev, · line 127

Quoted 8 lines
drivers/staging/comedi/drivers/cb_pcimdas.c +1 -1
@@ -121,7 +121,7 @@ static int cb_pcimdas_ai_rinsn(struct comedi_device *dev,
 		maxchans = s->n_chan;
 
 	if (chan > (maxchans - 1))
-		return -ETIMEDOUT;	/* *** Wrong error code. Fixme. */
+		return -ECHRNG;
 
 	/* configure for sw initiated read */
 	d = inb(devpriv->BADR3 + 5);

Hmm... This isn't quite right...

The 16 single-ended / 8 differential analog input channels on this board is set with a switch on the PCB. The state of the switch should be read when the driver is attached and the subdevice initialized with the correct number of channels. The core will then validate the "chan" number before calling the (*insn_read) operation.

Regards,
Hartley

A message that is not in the archive (sent off-list or not yet received): [email protected]

#3 lore raw
Quoted 44 lines · Nicholas Krause
Quoted 41 lines · Hartley Sweeten
Quoted 2 lines · Nicholas Krause

This changes us using the incorrect error,-ETIMEOUT when checking if the channel we are allocating to on the device structure pointer

passed

Quoted 1 line

to this function is greater then the maximum available channels for

this

Review

Quoted 8 lines

device to the correct error for a channel being out of range,-ECHRNG.

---
  drivers/staging/comedi/drivers/cb_pcimdas.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
(quoted diff) +0 -0

b/drivers/staging/comedi/drivers/cb_pcimdas.c

drivers/staging/comedi/drivers/cb_pcimdas.c static int cb_pcimdas_ai_rinsn(struct

Quoted 4 lines

index 70dd2c9..d91a6f3 100644

drivers/staging/comedi/drivers/cb_pcimdas.c +0 -0
@@ -121,7 +121,7 @@ static int cb_pcimdas_ai_rinsn(struct

comedi_device *dev,

drivers/staging/comedi/drivers/cb_pcimdas.c static int cb_pcimdas_ai_rinsn(struct · line 127

Quoted 8 lines
drivers/staging/comedi/drivers/cb_pcimdas.c +1 -1
  		maxchans = s->n_chan;
 
  	if (chan > (maxchans - 1))
-		return -ETIMEDOUT;	/* *** Wrong error code. Fixme. */
+		return -ECHRNG;
 
  	/* configure for sw initiated read */
  	d = inb(devpriv->BADR3 + 5);

Hmm... This isn't quite right...

The 16 single-ended / 8 differential analog input channels on this board is
set with a switch on the PCB. The state of the switch should be read when
the driver is attached and the subdevice initialized with the correct number
of channels. The core will then validate the "chan" number before calling
the (*insn_read) operation.

Quoted 6 lines
Quoted 2 lines

Regards,
Hartley

Hartley,
If that is the case then why is the check for max channels here. I can send in a v2 removing this check as it seems unneeded based on my understanding and your response to this patch. Nick

The patch is fine, but now superceded by Hartley's series of patches which has more extensive changes to the driver:

http://driverdev.linuxdriverproject.org/pipermail/driverdev-devel/2015-February/065754.html

That set of patches assumes the switches remain in the same state through the lifetime of the device instance, which seems reasonable for a PCI card.

Signature · 2 lines
-=( Ian Abbott @ MEV Ltd.    E-mail: <[email protected]> )=-
-=(                          Web: http://www.mev.co.uk/  )=-