Like the timer, the support functions for the counter subdevice are broken.

Rewrite the code to follow the comedi API.

The new implementation is based on the (minimal) datasheet I have from
ADDI-DATA.
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 support functions for the timer subdevice are broken.

1) The (*insn_write) assumes that insn->n is always 2 (data[1] is used)
2) The (*insn_read) assumes that insn->n is always 2 (data can be returned in
   data[0] and data[1]).
3) The (*insn_config) does not follow the API. It assumes insn->n is always 4
   (data[1], data[2] and data[3] are used). It also doesn't use data[0] to
   determine what the config "instruction" is.

Rewrite the code to follow the comedi API and add the missing comedi driver
comment block.

The new implementation is based on the (minimal) datasheet I have from
ADDI-DATA.
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 counter subdevice can generate an interrupt. Currently send_sig() is used
to let the task know when the interrupt occurs. Use the dev->read_subdev and
comedi_handle_events() instead.

Remove the, now unused, 'tsk_current' member from the private data and the
unnecessary include of <linux/sched.h>.
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 timer subdevice can generate an interrupt. Currently send_sig() is used
to let the task know when the interrupt occurs. Use the dev->read_subdev
and comedi_handle_events() 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>

This board supports change-of-state interrupts on digital inputs 4 to 19
not 0 to 15.

The current code "works" but it could set inappropriate bits in the mode1
and mode2 registers that setup which channels are enabled. It also doesn't
return the status of the upper 4 channels (19 to 16).

Fix the comment and mask the mode1/mode2 values so that only the interrupt
capable channels can be enabled.

Add the SDF_LSAMPL flag to the subdevice so that 32-bit samples are used
instead of 16-bit ones. This allows returning the upper 4 channels. Use
the remaining bits in the sample to return "event" flags to the user.

The timer and counter subdevices can also generate interrupts and are a bit
hacked. They don't currently follow the comedi API and they use send_sig()
to let the task that know that the interrupt occured. The "event" flags will
be used instead when these subdevices are fixed.
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>

Add `#include` lines to declare stuff used by "plx9080.h" itself.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Document the DMA descriptor `struct plx_dma_desc`, and the DMA abort
function `plx9080_abort_dma()`.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Use the usual block comment style and remove some excess spaces from
single-line comments.  Remove some obvious stuff about register offsets
being zero relative.  Remove some text about the original history of the
file, as it bears hardly any resemblance to it now.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Add macros in "plx9080.h" that define values for the DMATHR register
values.  Use the prefix `PLX_DMATHR_` for the macros.  Make use of the
`BIT(x)` and `GENMASK(h,l)` macros to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
DMACSR0 and DMACSR0 registers.  Use the prefix `PLX_DMACSR_` for the
macros.  Make use of the `BIT(x)` macro to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
DMADPR0 and DMADPR1 registers.  (A little-endian version of the register
value is also placed in the `next` member of `struct plx_dma_desc`.)
Use the prefix `PLX_DMADPR_` for the macros.  Make use of the `BIT(x)`
and `GENMASK(h,l)` macros to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
DMAMODE0 and DMAMODE1 registers.  Use the prefix `PLX_DMAMODE_` for the
macros.  Make use of the `BIT(x)` and `GENMASK(h,l)` macros to define
the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For the PLX PCI 9080, the read-only PCIHIDR register is hard-coded with
the value `0x908010b5`.  Add a macro `PLX_PCIHIDR_9080` that expands to
this value.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
CNTRL register.  Use the prefix `PLX_CNTRL_` for the macros.  Make use
of the `BIT(x)` and `GENMASK(h,l)` macros to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
INTCSR register.  Use the prefix `PLX_INTCSR_` for the macros.  Make use
of the `BIT(x)` and `GENMASK(h,l)` macros to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
DMCFGA register.  Use the prefix `PLX_DMCFGA_` for the macros.  Make use
of the `BIT(x)` and `GENMASK(h,l)` macros to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
DMPBAM register.  Use the prefix `PLX_DMPBAM_` for the macros.  Make use
of the `BIT(x)` and `GENMASK(h,l)` macros to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
LBRD0 and LBRD1 registers.  Use the prefix `PLX_LBRD_` for macros that
apply to both registers, `PLX_LBRD0_` for macros that apply only to the
LBRD0 register, and `PLX_LBRD1_` for macros that apply only to the LBRD1
register.  Make use of the `BIT(x)` and `GENMASK(h,l)` macros to define
the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace `enum bigend_bits` in "plx9080.h" with a bunch of macros
defining values for the BIGEND register.  Use the prefix `PLX_BIGEND_`
for the macro names.  Make use of the `BIT(x)` and `GENMASK(h,l)` macros
to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace `enum marb_bits` in "plx9080.h" with a bunch of macros defining
values for the MARBR and DMAARB registers (which are the same
Mode/Arbitation register at two different offsets).  Use the prefix
`PLX_MARBR_` for the macros.  Make use of the `BIT(x)` and
`GENMASK(h,l)` macros to define the values.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace the existing macros in "plx9080.h" that define values for the
LAS0BA and LAS1BA registers.  Use the prefix `PLX_LASBA_` for the
macros.  Make use of the `BIT(x)` and `GENMASK(h,l)` macros to define
the macros.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Rename the macros for the PLX PCI 9080 LAS0RR and LAS1RR registers in
"plx9080.h", using the prefix `PLX_LASRR_`.  Make use of the `BIT(x)`
and `GENMASK(h,l)` macros to define the values.

Define a macro `PLX_LASRR_PREFETCH` for the "prefetchable memory" bit in
this register, and define a macro `PLX_LASRR_MLOC_MASK` to mask the PCI
memory location control bits.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Rename the macros in "plx9080.h" that define the offsets of registers,
following the pattern `PLX_REG_<NAME>`, where `<NAME>` is the register
name from the PLX PCI 9080 Data Book.

Add defines for the "Mailbox" registers, and add parameterized macros
for the mailbox registers and the DMA control registers.  Make use of
the parameterized versions of the macros where it seems appropriate.

The registers for supporting the I2O (Intelligent Input/Output) feature
are largely left undefined, just defining enough to allow the I2O
feature to be disabled.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The defines related to the Power-Up Test Suite (PUTS) are just cruft that
has nothing to do with the PLX PCI-9080 chip.  They seem to have been
inherited from "drivers/net/plx9060.h" in the kernel 2.2.16 sources for
use by the "wanxl" driver.  Remove them.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For the PLX local address space range registers, LAS0RR and LAS1RR, bit
0 indicates whether the local address space will be mapped to memory
space or I/O space.  If mapped to I/O space, bit 1 must be set to 0, and
bits 31 to 2 form the address decoding mask, which should be -2^N mod
2^32 for a range of length 2^N.

The `LRNG_IO_MASK` macro is supposed to specify the address decoding
bits for I/O space.  It currently has the value `0xfffffffa`, but should
be `0xfffffffc`, or possibly `0xfffffffe` (it doesn't really matter,
since bit 1 is required to be set to 0).  Change it to `0xfffffffc`.

Similarly, for the PLX local address space local base address (remap)
registers, LAS0BA and LAS1BA, bits 31 to 2, masked with the
corresponding "range" register form the local base address for the local
address space.  The `LMAP_IO_MASK` macro is supposed to mask the valid
bits for I/O space.  Change its value from `0xfffffffa` to `0xfffffffc`
to match `LRNG_IO_MASK`.
Signed-off-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The checkpatch.pl warns about two `udelay(x)` calls, one of 100
microseconds, and one of 10 microseconds.  The 100 microseconds one is
used when waiting for FPGA to become ready to accept firmware, and is
not that critical, so replace it with a call to `usleep_range(100,
1000)`.  The 10 microseconds one is called as each 16-bit word of
firmware data is written.  Replace it with a fairly tight
`usleep_range(10, 20)` to avoid slowing down firmware loading too much.
The firmware is fairly short, so this would only slow it down firmware
loading by about 20 milliseconds or so.
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>

Rename functions to avoid CamelCase warnings from checkpatch, and to use
namespace associated with the driver.
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>

Rename the macros that define values for the reference DACs register to
avoid CamelCase, and to make it clearer which register they are
associated with.  Add a macro `DB2K_REG_DACS_SET` for the value `0x0080`
that triggers setting one of the references.
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>

Rename the macros that define values for the trigger control register to
avoid CamelCase, and to make it clearer which register they are
associated with.
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>

Rename the macros defining values for the DAC status register to avoid
CamelCase, and to make it clear which register they are associated with.
Refactor the macros defining the regular DAC channel "busy" bits into a
single macro that takes the DAC channel number as a parameter.

Add a macro to define the offset of the read-only DAC status register.
It is the same offset as the DAC control register, which is write-only.

The code in `daqboard2000_ao_eoc()` that checks the status for
completion of the DAC conversion looks wrong.  The register has a "busy"
bit for each channel, but the existing code only works for channels 0
and 1.  The driver only supports two DAC channels at the moment, so the
bug is currently harmless, but fix it so we can support four DAC
channels on some board models.
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>

Rename the macros used to define values for the DAC control register to
avoid CamelCase and to make it clearer which register they are
associated with.  Refactor the macros used to define values to enable or
disable DAC channels to use the channel number as a parameter.  None of
these macros are currently used by the driver.
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>

Rename the macros associated with the acquisition status register to
avoid CamelCase and to make it clear which register they are associated
with.

Add a macro to define the offset of the read-only acquisition status
register.  It's the same offset as the acquisition control register,
which is write-only.
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>

Rename the macros defining values for the acquisition control register
to avoid CamelCase, and to make it clearer which register they are
associated with.
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>

Rename the macros defining register offsets to avoid CamelCase, and to
use namespace associated with the driver.
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>

Rename the macros defining values for the Serial EEPROM Control Register
to avoid CamelCase.
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>

Fix checkpatch issue: "CHECK: Please use a blank line after
function/struct/union/enum declarations".
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>

Fix checkpatch issues of the form "CHECK: spaces preferred around that
'*' (ctx:VxV)".
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>

Reformat one of the block comments to conform to the usual style (it's
the only one that doesn't).
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>

Remove some commented out code.  Some of it uses constructs that don't
exist in the driver, and probably come from the source code for the MS
Windows driver.
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>

This patch Replace all occurences of (1<<x) by BIT(x) in the file pcmmio.c to
get rid of checkpatch.pl "CHECK" output "Prefer using the BIT macro"
Signed-off-by: NRavishankar Karkala Mallikarjunayya <ravishankarkm32@gmail.com>
Reviewed-by: NIan Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>