- 03 Jul, 2012 4 commits
-
-
Federico Vaga authored
Can't get private data during device creation Signed-off-by:Federico Vaga <federico.vaga@gmail.com>
-
Federico Vaga authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by: -
Alessandro Rubini authored
-
- 19 Jun, 2012 9 commits
-
-
Federico Vaga authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by:
Alessandro Rubini <rubini@gnudd.com> Acked-by:
-
Federico Vaga authored
-
Federico Vaga authored
Signed-off-by:
-
- 16 Apr, 2012 3 commits
-
-
Alessandro Rubini authored
-
Alessandro Rubini authored
-
Alessandro Rubini authored
Signed-off-by:
-
- 09 Apr, 2012 3 commits
-
-
Alessandro Rubini authored
Signed-off-by:
-
Alessandro Rubini authored
Signed-off-by:
-
Alessandro Rubini authored
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:
-
- 06 Apr, 2012 9 commits
-
-
Alessandro Rubini authored
Signed-off-by:
-
Federico Vaga authored
Signed-off-by:
-
Federico Vaga authored
Signed-off-by:
-
Federico Vaga authored
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 authored
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 authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by:
-
Federico Vaga authored
Signed-off-by: -
Federico Vaga authored
Signed-off-by:
-
- 15 Mar, 2012 8 commits
-
-
Federico Vaga authored
Signed-off-by:
-
Federico Vaga authored
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 authored
Signed-off-by:
-
Federico Vaga authored
Now all the zio objects are handled with the device structure Signed-off-by:
-
Federico Vaga authored
Signed-off-by:
-
Federico Vaga authored
ad7887 and ad7888 are now handled by the same driver. The driver requires data only for enabled channels Signed-off-by:
-
Federico Vaga authored
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 authored
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:
-
- 14 Mar, 2012 4 commits
-
-
Federico Vaga authored
Signed-off-by:
-
Federico Vaga authored
If a zio object declares a set of read only attributes and it doesn't need s_op->info_get(), it can avoid the s_op assignment, so zattr_show now checks if s_op is defined before calling it. Signed-off-by:
-
Federico Vaga authored
When disabling a trigger, ZIO invokes t_op->abort() to stop acquisition. If the abort function invokes data_done() to return partial blocks, the ZTI_BUSY flag is correctly cleared. But, if the abort function frees all active blocks the flag remains set. With this patch, always clear the ZTI_BUSY flag when the trigger is disabled so it is correct if/when the trigger is re-enabled. Signed-off-by:
-
Federico Vaga authored
The bug was evident in zio_change_current_buffer. When you change the buffer, zio registers a new bi, but the old __bi_register() overwrites the current bi in use by the channel; so, when zio tries to unregister the old bi, there is not any reference to it. Now you must assign a bi to a channel after its registration. For simmetry and to prevent future bugs, also ti assignment must be done after __ti_registration(). Signed-off-by:
-
