Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Federico Vaga <federico.vaga@gmail.com>

The biggest change (as for number of lines) is renaming he ztt_instance
structure pointers to ztt. Also, remove a trailing semicolon in a #define,
fix a kfree call and remove an old comment that doesn't apply any more.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Federico Vaga <federico.vaga@gmail.com>

This adds 4 bytes to the control (there are atill 16 free) to fit
a whole struct addr_zio (same as sockaddr_zio, to be defined when we
merge PF_ZIO). The structure was almost ready for that,
but we missed the sa_family 16-bit field.

This also changes the minor number for the ZIO control structure, and
all the few users of the address-related fields.
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

A new zio_attribute flag is declared: ZATTR_CONTROL. If this flag
is setted the zio_attribute appear in the control, otherwise
the attribute is visibile in the sysfs but not in the control
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

The ZIO core use the dev_id parameter but this patch is compatible only
with kernel 3.2 or greater.
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Federico Vaga <federico.vaga@gmail.com>

The helper zio_trigger_abort() is introduced; it can be used to abort a
data transfer.

The bit flag ZTI_COMPLETING is introduced: if high ZIO is completing the
transfert operation by invokng the data_done() function; if low the
transfer is running or the trigger is waiting for event. If abort() is
called when this glag is high, the functions does nothing.

The helper zio_trigger_data_done() is introduced; insted of call the
data_done() function with cset->ti->t_op->data_done() and manually set
the ZTI_COMPLETING flag, this hellper must be used. If the trigger
operation data_done is not set, then use the __zio_internal_data_done
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

If default_(zbuf|trig) is not empty then ZIO use this as first choice
for the buffer|trigger; otherwise it tries to use the
preferred_(buffer|trigger) defined at zio_device and valid for any
cset. The last choice is the ZIO default.
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

The core should ensure its users that create calls are not reentrant.
We use the locks that already exist in the buffer type and the trigger
type.

Then, we init both ti->lock and bi->lock, so our users can rely on
them (see next patch).

Related to locks: fix a comment in zio.h
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Acked-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

This patch introduces zio_driver and zio_device_id structure,
which are used to define a zio's driver. ZIO uses these
structures to handle the match between devices and drivers
connected on the zio bus.

To make ZIO work on the Linux bus mechanism, ZIO creates two
zio_device structures. The first zio_device is allocated and it is
registered by the module which wants to declare the existence
of a new ZIO device. This first zio_device is only used by
ZIO to match the device with a driver. When the first zio
device matchs with a driver, ZIO creates a new zio device child
of the first one. The new device is filled with all the device
information and sysfs attributes. This second zio device
represents the real device to use; ZIO passes this zio device
as argument to the driver probe function, so the driver can
use the real zio device for its operations.
From the user space point of view, both ZIO devices are
available at /sys/bus/zio/devices/; the first zio device has
the prefix "hwdev-", the second zio device (the real one) has
the prefix "zio-"
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

cset and bi use a different set of sysfs default attributes, so now they
use their own device_type with their default attributes
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

remove the folder where zio lists all the available triggers, buffers
and devices. Now you can find devices at /sys/bus/zio/devices/, and
retrive the list of the available triggers and buffers from:
/sys/bus/zio/available_buffers
/sys/bus/zio/available_triggers
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Now all the zio objects are handled with the device structure
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

ad7887 and ad7888 are now handled by the same driver.
The driver requires data only for enabled channels
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Every device must has the zio attribute "resolution-bits" on any level of
the device hierarchy. Internally, the preferred name for fields
and variables is "nbits".
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

The current control used to be per-trigger (i.e., per-cset).
However, some parameters are per-channel (like the gain),
so each channel now has its own current control structure.

Also, zio-dump now shows device and trigger attributes, both standard
and
extended ones.
Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Federico Vaga <federico.vaga@gmail.com>
Acked-by: Alessandro Rubini <rubini@gnudd.com>