`comedi_write()` is the handler for the "write" file operation for
COMEDI devices.  It mostly runs without using the main mutex of the
COMEDI device, but uses the `attach_lock` rw_semaphore to protect
against the COMEDI device becoming "detached".  A file object can write
data for a COMEDI asynchonous command if it initiated the command.  The
COMEDI subdevice is marked as busy when the command is started.  At some
point, the "write" handler detects that the command has terminated and
so marks the subdevice as non-busy.

In order to mark the subdevice as non-busy, the "write" handler needs to
release the `attach_lock` rw_semaphore and `acquire the main `mutex`.
There is a vulnerable point between the two, so it checks that the
device is still attached after acquiring the mutex.  However, it does
not currently check that the conditions for becoming non-busy still
hold.  Add some more checks that the subdevice is still busy with a
command initiated by the same file object, and that the command is in
the correct direction (in case the subdevice supports both "read" and
"write").
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Reviewed-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Rearrange the code in `comedi_write()` to reduce the amount of
indentation.  The code never reiterates the `while` loop once `count`
has become non-zero, so we can check that in the `while` condition to
save an indentation level.  (Note that `nbytes` has been checked to be
non-zero before entering the loop, so we can remove that check.)  Move
the code that makes the subdevice "become non-busy" outside the `while`
loop, using a new flag variable `become_nonbusy` to decide whether it
needs to be done.  This simplifies the wait queue handling so there is a
single place where the task is removed from the wait queue, and we can
remove the `on_wait_queue` flag variable.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Reviewed-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Change the MODULE_DESCRIPTION to something more useful than the
generic "Comedi low-level driver".
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The Description is a bit long winded and the same information is in
the Devices. Shorten the Description and tidy up the Devices.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This function is called as part of the pci_driver (*probe) before
doing the (*auto_attach) of the comedi driver. For aesthetics, move
the function to a more logical place in the driver.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, move these defines after the register defines and rename
them to have namespace associated with the driver.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, move this enum after the register defines and rename
it to have namespace associated with the driver.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This member of the boardinfo is identical to the offset of the boardinfo
in the boardtypes array. It's also passed as the 'context' to the driver
(*auto_attach).

The 'cardtype' is only needed by the (*auto_attach) to determine which
PCI BAR to use and in pci_dio_reset() to handle the board specific code.

Remove the 'cardtype' member and use the 'context' value instead.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, use a common define for the interrupt control and status
registers.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, rename this define and fix the alignment.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, replace these defines with a macro.

Refactor the switch in pci_dio_reset() to use common code.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, replace these defines with some macros.

Refactor the switch in pci_dio_reset() to not require the fallthrough
comment.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

These boards use the same offsets for the interrupt control registers.
For aesthetics, remove the current defines and use common ones.

Fix the switch() in pci_dio_reset() to use common code.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

These boards use the same offsets for the interrupt control registers.
For aesthetics, remove the current defines and use common ones.

Fix the switch() in pci_dio_reset() to use common code.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, use a common comment for the switch() that disables and clears
interrupts.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, move the disable of the channel freeze for the PCI-1752
and PCI-1756 boards out of the switch used to disable and clear interrupts.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, use a default case in the switch (board->cardtype) used
to reset the various boards.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

These defines are only used to initialize the diosubd_data 'addr' members
in the boardinfo. For aesthetics, just open-code the values and remove the
defines.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

These defines are only used to initialize the diosubd_data 'addr' members
in the boardinfo. For aesthetics, just open-code the values and remove the
defines.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Currently the board reset function also resets the digital output channels
to 0. This works but it makes the reset function a bit messy and each
board type has to be handled special.

Move the digital output reset into the subdevice init where it can be
handle based on the subdevice setup.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The board reset function disables and clears all interrupts. It also
resets all the digital output channels to 0.

Interrupts are not currently used by this driver. For asthetics, do
the board reset early in the (*auto_attach) to make sure the interrupts
are disabled in case this feature is added.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The board reset function disables and clears all interrupts. It also
resets all the digital output channels to 0.

Interrupts are not used by this driver so the disable/clear during the
(*detach) is not necessary.

Reseting all the digital outputs to 0 might not be desired depending
on what the outputs are connected to.

Remove the board reset and just use comedi_pci_detach() directly for
the driver (*detach).
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

These defines are only used to initialize the diosubd_data 'addr' members
in the boardinfo. For aesthetics, just open-code the values and remove the
defines.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

These defines are only used to initialize the 'timer_regbase' boardinfo.
For aesthetics, just open-code the values and remove the defines.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The "board id" register is always 4-bits (4 di channels) and the register
used to read the bits is always > 0.

Simplify the 'boardid' boardinfo by replacing it with a 'id_reg' member
and open-coding the subdevice n_chan.

For aesthetics, remove all the *_BOARDID defines and just open-code the
register values in the boardinfo.

Add the missing boardinfo for the pci1739 board id register and increase
the nsubdevs to handle it.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Currently the di/do subdevices store a pointer to the diosubd_data in s->private.
The (*insn_bits) functions then use that to get to the 'addr' needed to access
the registers.

The only member of diosubd_data that is needed by the (*insn_bits) functions is
the 'addr'. For aesthetics, just store the 'addr' in s->private.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Currently the (*insn_bits) functions used the 'regs' member to determine how
many registers need to be read or written to update the subdevice. We can use
the subdevice 'n_chan' to determine this and make the code a bit clearer.

The (*auto_attach) also uses this member to determine how many 8255 devices
need to be initialized. These subdevices do not use the 'chans' member of
diosubd_data. Move the 'regs' value to the 'chans' to allow removing the
'regs' member completely.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The boards supported by this driver either use 8-bit or 16-bit I/O. The
'io_access' member of the boardinfo is used by the (*auto_attach) to
determine which (*insn_bits) function to use.

Simplify the boardinfo a bit by refactoring the 'io_access' member into
a bit-field flag 'is_16bit'. Use the new flag and remove the switch ()
code in the (*auto_attach).
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This function initializes a digitial input subdevices. For aesthetics,
absorb it into the (*auto_attach).

Remove the improper initialization of the SDF_LSAMPL subdev_flag and
len_chanlist. These are only used by subdevices that support async
commands.

Also, remove the unnecessary 'specflags' from the diosubd_data. Only
the boardid subdevice uses it.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This function initializes a digitial output subdevice. For aesthetics,
absorb it into the (*auto_attach).

Remove the improper initialization of the SDF_LSAMPL subdev_flag and
len_chanlist. These are only used by subdevices that support async
commands.

Also remove the unnecessary initilaization of the subdevice 'state'.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, use a const pointer to access the diosubd_data in the
boardinfo when doing the (*auto_attach)..
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetics, post-increment the 'subdev' index when used to get a
comedi_subdevice pointer instead of incrementing it after the subdevice
is initialized.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

All the boards use PCI BAR2 for the dev->iobase except for the pci1736
which uses PCI BAR0. Just use the board->cardtype to determine which
PCI BAR to use.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Reformat the bolck comment in the kernel CodingStyle.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

These are not necessary and just add cruft. Remove them.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

BIT macro is used for defining BIT location instead of
shifting operator - coding style issue
Signed-off-by: NRanjith Thangavel <ranjithece24@gmail.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

ni_release_ai_mite_channel(), ni_release_ao_mite_channel(),
ni_release_gpct_mite_channel() and ni_release_cdo_mite_channel()
call functions which interpret -1 as a special value meaning "no channel".
This patch adds explicit "no_channel" versions instead.

On the other hand, after "no_channel" versions are used,
ni_set_ai_dma_channel(), ni_set_ao_dma_channel(),
ni_set_gpct_dma_channel(), ni_set_cdo_dma_channel() are called with actual
"channel" parameter being always unsigned, so their signatures are changed
accordingly.

A side benefit of the changes is suppressesing 4 sparse warnings:
"warning: shift too big (4294967295) for type int".
Signed-off-by: NAndrzej Pietrasiewicz <andrzejtp2010@gmail.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Fixed multiple comment blocks that didn't comply with the
kernels coding style, and fixed a few spacing issues as well.
Signed-off-by: NDaniel H. Hemmingsen <dhh.kernel@gmail.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

BIT macro is used for defining BIT location instead of
shifting operator - coding style issue
Signed-off-by: NRanjith Thangavel <ranjithece24@gmail.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

BIT macro is used for defining BIT location instead of
shifting operator - coding style issue
Signed-off-by: NRanjith Thangavel <ranjithece24@gmail.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>