Use the new comedi_subdevice 'readback' member and the core provided
(*insn_read) for the readback of the analog output subdevice channels.
For aesthetics, rename the (*insn_write) function and tidy it up a bit.
Add a comment about the 'readback' member.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The analog output hardware in most comedi drivers does not provide a
way to readback to last values written to the channels. In order to
provide an (*insn_read) for the analog output subdevice, the comedi
drivers save the last values for each channel in the private data.
Add a new member, 'readback', to the comedi_subdevice definition to
provide a common way to save these values.
Introduce a comedi core function, comedi_alloc_subdev_readback(), to
allocate the memory needed to save the values. This memory will be
automatically kfree'd when the driver is detached.
Introduce a comedi core function, comedi_readback_insn_read(), that
the comedi drivers can use for the (*insn_read) of a subdevice to
return the saved values for each channel.
This will allow removing the boilerplate in the comedi drivers to
return the saved values. In some drivers it will also allow removing
the private data completely.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The 'aref' (CR_AREF) values are all AREF_* defines. The SDF_* defines are
subdevice flags. Fix the test in this function.
Reported by: coverity
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The 'aref' (CR_AREF) values are all AREF_* defines. The SDF_* defines are
subdevice flags. Fix the test in this function.
Reported by: coverity
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The previous mask by 0xf ensures that calib_source will not be > 0xf.
Remove the dead code.
Reported by: coverity
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The local variable 'd' is used to read a 32-bit register. One branch of
the code shifts the register value 16 bits to the right to get the high
16-bits. Change the type of the local variable so that this shift works
correctly.
Reported by: coverity
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
These should both be checking the -errno.
Reported by: coverity
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This code wants to allocate two arrays of struct urb pointers not two
arrays of struct urb objects. The size argument for both kcalloc calls
should be sizeof(urb) not sizeof(*urb).
Reported by: coverity
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
These should both be checking the -errno.
Reported by: coverity
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The "ni_labpc" module is a Comedi driver for the National Instruments
Lab-PC series of ISA data acquistion boards, and also provides common
code for the "ni_labpc_pci" and "ni_labpci_cs" modules (for PCI boards
and PCMCIA cards).
Split out the common code into a new module "ni_labpc_common", leaving
the driver for the ISA boards in the existing "ni_labpc" module. This
removes the C preprocessor conditional directives from the resulting
".c" files (although they remain in the included header file
"ni_labpc_isadma.h").
Unneccessary `#include` directives have been removed from the resulting
".c" files.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For consistency, break lines before operators, not after.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix checkpatch issues "CHECK: Logical continuations should be on the
previous line".
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix checkpatch issues "CHECK: Alignment should match open parenthesis".
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix checkpatch issues "CHECK: braces {} should be used on all arms of
this statement".
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Correct checkpatch issue "WARNING: braces {} are not necessary for any
arm of this statement".
Signed-off-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@@
type T;
expression e;
@@
(
- T *
+ T *
)
e
// </smpl>
This semantic patch just removes the cast and adds it back, but when it
does so, it follows the spacing conventions of Linux.
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We initialize 'irqbit' to 0, only to properly set it immediately
afterwards. Just remove the zero-initialization.
Signed-off-by: Chase Southwood <chase.southwood@gmail.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Cc: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
cppcheck was complaining that the variable 'stat' is being reassigned
before the old value is used. Upon inspection, I found that
dt2801_writecmd() cannot fail, always returns 0, and most callers already
do not bother with assigning its return value anyway, so it makes sense to
just change the return type for this function from int to void, and remove
the two assignments to 'stat'.
Signed-off-by: Chase Southwood <chase.southwood@gmail.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Cc: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Coccinelle (using scripts/coccinelle/tests/doublebitand.cocci) found this
assignment because 0 is or'd twice. Upon inspection, I found that the
variable "mode" is only assigned once and used once, and thus can be
cleanly removed. This patch plugs its value straight into writew() and
then removes the variable.
Signed-off-by: Chase Southwood <chase.southwood@gmail.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Cc: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A recent commit introduced a typo in the definition of
APCI1564_DO_CC_INT_DISABLE. The macro is not use yet, so not a huge deal,
but this patch corrects it.
Signed-off-by: Chase Southwood <chase.southwood@gmail.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Cc: H Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The comment (also removed) explains why it was there in the first place, but
that doesn't make much sense.
Signed-off-by: Eli Billauer <eli.billauer@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
These memory barriers were added "just to be safe".
Signed-off-by: Eli Billauer <eli.billauer@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
According to Documentation/memory-barriers.txt, a memory barrier is put
in place by wait_event_*()
Signed-off-by: Eli Billauer <eli.billauer@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
mmiowb() was used to make sure that iowrite32() take place in the correct
order, which is an unnecessary precuation.
Signed-off-by: Eli Billauer <eli.billauer@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The relevant sequences consist of two I/O memory writes. The second write
depends on the first one. mmiowb() was inserted to make sure that no other
thread inserts a "first write" before the current one finished its second.
As a mutex protects this critical sequence, mmiowb() is unnecessary.
Signed-off-by: Eli Billauer <eli.billauer@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
According to Documentation/memory-barriers.txt, a memory barrier is put
in place by wait_event_*()
Signed-off-by: Eli Billauer <eli.billauer@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The Bluetooth maintainer has been complaining about it for a while, and
I shouldn't have merged it over his objections. There also has been no
real work done on it at all to get it out of the staging tree, so just
delete the code for now.
If someone wants to get this fixed up properly, feel free to revert this
commit and send the revert, along with cleanups and we will be glad to
consider it.
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Pali Rohár <pali.rohar@gmail.com>
Cc: Pavel Machek <pavel@ucw.cz>,
Cc: Miguel Oliveira <cmroliv@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
