This function is only called by comedi drivers during the (*detach) of the
driver. After it is called mite_free() is always called to kfree the allocated
mite_struct pointer.
Rename this function to mite_detach() and merge the kfree() from mite_free()
into it.
Remove the unnecessary clearing of the mite variables since the mite pointer
is going to be kfree()'d anyway.
The mite_detach() function checks if the pointer is valid so remove the
unnecessary checks in the callers.
The check can also be removed in ni_660x since the ni_660x_free_mite_rings()
function does not actually need the 'mite' pointer.
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 inline function is not used. Just remove it.
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 inline function simply returns the 'irq' of the struct pci_dev pointer
that is stored in struct mite_struct. This pci_dev pointer is the same as
the one that is already available in the (*auto_attach) functions where
mite_irq() is used.
Use the pci_dev pointer directly to get the irq and remove the unnecessary
inline function to remove the indirection.
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 defines were probably intended to be used so that ni_mio_common.c
could detect if it was included by a driver that uses ioport or memory
mapped register access. This can actually be detected by checking if
the 'mite' member in the private data has been allocated.
Regardless, the symbols are not used. Just remove them.
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 defines were probably intended to be used so that ni_mio_common.c
could detect if it was included by a driver that uses ioport or memory
mapped register access. This can actually be detected by checking if
the 'mite' member in the private data has been allocated.
Regardless, the symbols are not used. Just remove them.
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 defines were probably intended to be used so that ni_mio_common.c
could detect if it was included by a driver that uses ioport or memory
mapped register access. This can actually be detected by checking if
the 'mite' member in the private data has been allocated.
Regardless, the symbols are not used. Just remove them.
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 define is not used ny the driver. Remove it.
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>
By convention, the first member of the boardinfo is a 'const char *' to
the board name, especially for those comedi drivers that rely on the
comedi core to set dev->board_ptr to the entry matching the board name
when doing a legacy (*attach) to a comedi driver.
Modify the ni_board_struct appropriately for consistency, although the
board name is only used for informational purposes by this driver (only
the driver name is matched by the comedi core).
For aesthetics, also move the members in the boardinfo declarations.
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>
For aesthetics, add some whitespace to the boardinfo declaration.
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 define is not used by the driver. Remove it.
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 define is only used in the comedi_request_region() to specify the
size of the region. Remove the define and just open code the value.
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>
Remove all the boardinfo data the is set to '0'. They will default
to that when the boardinfo is declared.
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>
Pass the irq_polarity to ni_E_init() and remove the defines.
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 defines are only used in the request_irq() calls. Remove them
and just open code the values.
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>
Pass the 'interrupt_pin' to ni_E_init() and remove the macros.
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>
Move some functions to avoid the need for the forward declarations.
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 only member left in this struct is the 'chan' which is the
comedi channel of the encoder.
Remove the struct and refactor the code to pass/use the comedi
channel instead.
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>
Abosorb the s626_read_latch() helper and tidy up this function.
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 array in the encoder private data just makes the driver harder to
follow.
The S626_EVBITS() macro is used to initialize the member in the declaration.
This macro creates a array of four values, based on the encoder channel, that
are used to mask and set the private data 'counter_int_enabs' member. The
values are created using the S626_OVERMASK() and S626_INDXMASK() macros using
the encoder channel number.
Remove the 'my_event_bits' member from the encoder private data as well
as the S626_EVBITS() macro. Refactor the code to use the S626_OVERMASK() and
S626_INDXMASK() macros directly.
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>
There are two functions used for the 'set_mode' callback. One is used for
the channel 0-2 encoders and the other for the channel 3-5 encoders.
Add a helper function, s626_set_mode(), that uses the encoder channel number
to determine which function to call.
Remove the then unnecessary 'set_mode' member and just call s626_set_mode()
directly.
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>
There are two functions used for the 'get_mode' callback. One is used for
the channel 0-2 encoders and the other for the channel 3-5 encoders.
Add a helper function, s626_get_mode(), that uses the encoder channel number
to determine which function to call.
Remove the then unnecessary 'get_mode' member from the encoder private data.
The 'get_mode' callbacks were not being used by the driver in any of the enabled
code. For now block the s626_get_mode() functions with '#ifdef unused' to prevent
a compiler warning.
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>
There are two functions used for the 'reset_cap_flags' callback, One is used for
the channel 0-2 encoders and the other for the channel 3-5 encoders.
Refactor the two callbacks into a single s626_reset_cap_flags() function and use
the encoder channel number to handle the differences.
Remove the then unnecessary 'reset_cap_flags' member and just call s626_reset_cap_flags()
directly.
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>
There are two functions used for the 'pulse_index' callback, One is used for
the channel 0-2 encoders and the other for the channel 3-5 encoders.
Refactor the two callbacks into a single s626_pulse_index() function and use
the encoder channel number to handle the differences.
Remove the then unnecessary 'pulse_index' member and just call s626_pulse_index()
drectly.
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>
There are two functions used for the 'get_int_src' callback. One is used for
the channel 0-2 encoders and the other for the channel 3-5 encoders.
Refactor the two callbacks into a single s626_get_int_srsc() function and use
the encoder channel number to handle the differences.
Remove the then unnecessary 'get_int_src' member from the encoder private data.
The 'get_int_src' callbacks were not being used by the driver. For now block
the s626_get_int_src() function with '#ifdef unused' to prevent a compiler
warning.
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>
There are two functions used for the 'set_int_src' callback, One is used for
the channel 0-2 encoders and the other for the channel 3-5 encoders.
Refactor the two callbacks into a single s626_set_int_src() function and use
the encoder channel number to handle the differences.
Remove the then unnecessary 'set_int_src' member and just call s626_set_int_src()
drectly.
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>
There are two functions used for the 'get_load_trig' callback. One is used for
the channel 0-2 encoders and the other for the channel 3-5 encoders.
Refactor the two callbacks into a single s626_get_load_trig() function and use
the encoder channel number to handle the differences.
Remove the then unnecessary 'get_enable' member from the encoder private data.
The 'get_load_trig' callbacks were not being used by the driver. For now block
the s626_get_load_trig() function with '#ifdef unused' to prevent a compiler
warning.
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>
There are two functions used for the 'set_load_trig' callback, One is used for
the channel 0-2 encoders and the other for the channel 3-5 encoders.
Refactor the two callbacks into a single s626_set_load_trig() function and use
the encoder channel number to handle the differences.
Remove the then unnecessary 'set_load_trig' member and just call s626_set_load_trig()
drectly.
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>
There are two functions used for the 'get_enable' callback, s626_get_enable_a()
function is used for the channel 0-2 encoders and s626_get_enable_b() is used
for the channel 3-5 encoders.
Refactor the two callbacks into a single s626_get_enable() function and use the
encoder channel number to handle the differenced.
Remove the then unnecessary 'get_enable' member and just call s626_set_enable()
directly.
The 'get_enable' callbacks were not being used by the driver. For now block the
s626_get_enable() function with '#ifdef unused' to prevent a compiler warning.
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>
There are two functions used for the 'set_enable' callback, s626_set_enable_a()
function is used for the channel 0-2 encoders and s626_set_enable_b() is used
for the channel 3-5 encoders.
Refactor the two callbacks into a single s626_set_enable() function and use the
encoder channel number to handle the differenced.
Remove the then unnecessary 'set_enable' member and just call s626_set_enable()
directly.
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 member of the encoder private data can be calculated based on
the comedi channel stored in the 'chan' member of the private data.
Do that instead and remove the unnecessary 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>
Redefining the Counter Preload/Latch registers as a macro that calculates
the register offset based on the comedi channel number.
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 members of the encoder private data can be calculated based on
the comedi channel stored in the 'chan' member of the private data.
Do that instead and remove the unnecessary members.
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>
Add the comedi "chan" number to the encoder private data so we can then
remove some of the members of the private data.
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>
Start cleaning up the 'struct s626_enc_info' by redefining the my_cra
and my_crb members based on the comedi channel number.
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>
On moving the function into the driver proper, also check the device is
asserting the shared interrupt line.
This patch also fixes the interrupt handling for the digital input
change-of-state interrupts.
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>
This board supports an interrupt that can be generated by an AND/OR
combination of 16 of the input channels.
Create a separate subdevice to handle this interrupt.
The apci1564_di_config() function is used to configure which
inputs are used to generate the interrupt. Currently this function
is broken since it does not follow the comedi API for insn_config
functions. Fix this function by implementing the config instruction
INSN_CONFIG_DIGITAL_TRIG.
Add the remaining subdevice operations necessary for the interrupt
subdevice to support async commands.
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 addi_private struct defined in addi-data/addi_common.h is very bloated
and contains many fields which addi_apci_1564 does not require. In the
interest of eventually removing this driver's dependency on
addi_common.h, we can create a private data struct specifically for
addi_apci_1564 containing only the fields it will actually use.
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 global variables ui_InterruptStatus_1564 and ui_InterruptData are both
set but never used. Just remove them from the driver.
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>
This member of the private data struct is only set at one location in the
entire driver, and then never even used for anything. Let's just remove
its use.
Signed-off-by: Chase Southwood <chase.southwood@gmail.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Cc: H Hartley Sweeeten <hsweeten@visionengravers.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The DT2821 series board have jumpers that set the output range for
the two Analog Output channels. The range_table for the Analog Output
subdevice provides all possible ranges to the user. When a unipolar
range is selected the board expects the data to be in a straight
binary format. When a bipolar range is select the data should be in
two's complement format.
Currently, the user passes some configuration options when attaching
to the driver to select the data format for each channel. If the
user does not pass the config options, the data format is assumed to
be straight binary.
The Analog Output subdevice now has a range_table that provides the
user will all possible ranges. Use the range information to determine
if the data values need to be munged into two's complememnt values
and deprecate the config options.
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 direction of the 16 DIO channels is controlled with two bits in the
DACSR register. The HBOE bit controls the direction of channels 15-8 and
the LBOE bit channels 7-0.
The DACSR register is also used to control the Analog Output async command
support. Currently, starting or canceling an Analog Output command would
also reset the DIO directions so that all channels were inputs.
Fix the driver so that the DIO direction is not changed when starting or
canceling an Analog Output async command.
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>
For aesthetics, rename the register map defines so that the bit defines
are associated with the registers. Also, convert all the bit defines into
bit-shifts to make them clear.
Remove all the verbose comments. All the registers and bits are named
according to the User Manual for the DT2821 Series, Document UM-05073-G.
That manual can be checked if necessary for the verbose information.
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>
A common timer is used for analog input and output async commands. The
(*do_cmdtest) for both subdevices calculates the divisor as part of
Step 4 when validating the trigger arguments. The divisor is calculated
again in the (*do_cmd) for both subdevices in order to set the timer.
The comedi core only calls the (*do_cmd) is called after a successful
(*do_cmdtest). Save the divisor from the (*do_cmdtest) in the private
data and use that value in the (*do_cmd).
The extra check of the cmd->convert_arg in dt282x_ai_cmd() is not necessary.
The convert_arg was already checked in the (*do_cmdtest), and it's not used
in the (*do_cmd).
Tidy up dt282x_ns_to_timer(), the parameters are all unsigned int's and the
mask of the 'flags' can be moved here to simplify the callers.
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>
Change the MODULE_DESCRIPTION to something more useful than the
generic "Comedi low-level driver".
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>
Tidy up the multi-line comments at the beginning of the file to follow
the CodingStyle.
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 DT2821 series boards do not have programmable analog output ranges.
The range is set, per-channel, with jumpers on the board.
Currently the user has to pass some configuration options when attaching
to the driver in order to set the range for each channel. If they do not
pass the range options they will default to 0. If they pass an invalid
range option they will also default to 0. The 0 range option is bipolar
10V. If the user passes a range option that does not match the hardware
setting it doesn't cause any problems in the driver it just causes the
user space comedilib library to incorrectly convert the hardware values
to physical values.
Make the attach a bit simpler by deprecating these options and providing
a range table with all the valid ranges for the analog outputs. The user
can then select the correct range when they use the comedilib functions.
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>
Most of the boards supported by this driver can do either single-ended or
differential Analog Input but there are a couple that can only do one or
the other. On the boards that can do both the type of input is set with
jumpers on the board.
The 'it->options[4]' passed by the user during the board (*attach) is used
to determine how the Analog Inputs are configured. This is used to set
the subdevice flags and number of channels correctly.
Fix the check of this option so that it only applies when the board can
actually do differential input. Also, default to differential inputs if
that is the only mode the board supports.
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>
For aesthetics, add some whitespace to the Analog Input subdevice init.
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>
For aesthetics, add some whitespace to the Analog Output subdevice init.
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 enum is used to index the 'options' values that are passed
by the user attaches to the driver. These options are identified
in the comedi comment block.
For aesthetics, remove the enum and just open code the index values.
Refactor the Analog Input subdevice init so that the it->options[4],
indicating differential inputs, does not have to be checked multiple
times.
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>
