On 2013-07-25 17:47, H Hartley Sweeten wrote:
On Thursday, July 25, 2013 6:39 AM, Ian Abbott wrote:
On 2013-07-24 19:48, H Hartley Sweeten wrote:
When updating the output channels make sure the channels configured as
inputs stay inputs.
Signed-off-by: H Hartley Sweeten <hswee...@visionengravers.com>
Cc: Ian Abbott <abbo...@mev.co.uk>
Cc: Greg Kroah-Hartman <gre...@linuxfoundation.org>
---
drivers/staging/comedi/drivers/pcmuio.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/drivers/staging/comedi/drivers/pcmuio.c
b/drivers/staging/comedi/drivers/pcmuio.c
index 076a08a..5b0ade2 100644
--- a/drivers/staging/comedi/drivers/pcmuio.c
+++ b/drivers/staging/comedi/drivers/pcmuio.c
@@ -215,8 +215,14 @@ static int pcmuio_dio_insn_bits(struct comedi_device *dev,
s->state &= ~mask;
s->state |= (mask & bits);
- /* invert the state and update the channels */
+ /*
+ * Invert the state and update the channels.
+ *
+ * The s->io_bits mask makes sure inputs are '0' so
+ * that the output pins stay in a high-z state.
+ */
val = s->state ^ ((0x1 << s->n_chan) - 1);
+ val &= s->io_bits;
pcmuio_write(dev, val, asic, 0, port);
}
That's fine although the original function has other problems, for
example the value returned in data[1] should reflect the modified data,
not the original data and probably shouldn't be modifying s->state based
on the data read from the hardware. I think the function should be
something like this:
static int pcmuio_dio_insn_bits(struct comedi_device *dev,
struct comedi_subdevice *s,
struct comedi_insn *insn,
unsigned int *data)
{
unsigned int chanmask = ((1 << s->n_chan) - 1;
unsigned int mask = data[0] & chanmask;
unsigned int bits = data[1];
int asic = s->index / 2;
int port = (s->index % 2) ? 3 : 0;
unsigned int val;
if (mask) {
s->state &= ~mask;
s->state |= (mask & bits);
/*
* Invert the state and update the channels.
*
* The s->io_bits mask makes sure inputs are '0' so
* that the output pins stay in a high-z state.
*/
val = ~s->state & s->io_bits & chanmask;
pcmuio_write(dev, val, asic, 0, port);
}
/* get inverted state of the channels from the port */
val = pcmuio_read(dev, asic, 0, port);
/* get the true state of the channels */
data[1] = ~val & chanmask;
return insn->n;
}
This routine made my head hurt...
I'm not sure about the variable name 'chanmask', 'allchans' or
'allchanmask' might be better.
End result I "think" both solutions end up doing the same thing. But, I like
your proposal better.
Even with the s->io_bits taken into account when writing data to the
port, I think the old version still goes wrong if the value read differs
from the value last written for the bits that aren't being modified.
Also, I think the value written by pcmuio_dio_insn_config() should be
like the above for both INSN_CONFIG_DIO_OUTPUT and INSN_CONFIG_DIO_INPUT
- after modifying s->io_bits, the value written to the port should be
something like `~s->state & s->io_bits & ((1 << s->n_chans) - 1)`.
I'll try to get this series rebased today. Please let me know if you want me
to add the 'irq' member to the private data for the 2nd interrupt.
I think that would be cleaner.
--
-=( Ian Abbott @ MEV Ltd. E-mail: <abbo...@mev.co.uk> )=-
-=( Tel: +44 (0)161 477 1898 FAX: +44 (0)161 718 3587 )=-
_______________________________________________
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
