- 04 Aug, 2012 40 commits
-
-
Antti Palosaari authored
Signed-off-by:
Antti Palosaari <crope@iki.fi> Signed-off-by:
Mauro Carvalho Chehab <mchehab@redhat.com>
-
Antti Palosaari authored
I think it is better to allow driver finish pending operations and disconnect device. Killing all ongoing USB transfers could have very bad effect, even break hardware in worst case. Signed-off-by:
-
Antti Palosaari authored
Almost all hacks are now removed and much less code. Reducing af9015 driver code from 2084 to 1609, it is almost 500 LOC less! Signed-off-by:
-
Antti Palosaari authored
Compilation was disabled as it was failing with allyesconfig. Rename duplicate functions and enable building. Signed-off-by:
-
Antti Palosaari authored
It was possible to free resources by unloading module while initialization was still ongoing on delayed work. Use PID to make decision on .disconnect() if caller is our work or some other as work also calls .disconnect() in error case. There could be better solution still... Signed-off-by:
-
Antti Palosaari authored
Debug prints write operation data instead of read data was intention. Signed-off-by:
-
Antti Palosaari authored
Signed-off-by:
-
Antti Palosaari authored
There is no USB match flag for used USB interface. It is rather common there is multiple interfaces offering different services. For example one for television and one for remote controller. Signed-off-by:
-
Antti Palosaari authored
Signed-off-by:
-
Antti Palosaari authored
No worth to keep it on own file as it is only 30 LOC. Signed-off-by:
-
Antti Palosaari authored
Use Kernel logging insteads of own macros. Get rid of old debugs and use dynamic debug. Signed-off-by:
-
Antti Palosaari authored
Unregister all if there is any error meet during frontend initialization. Signed-off-by:
-
Antti Palosaari authored
Calling usb_clear_halt() during device init is not correct. 2 of 7 AF9015 devices I have timeouts next USB control message after usb_clear_halt(). It was originally performed between tuner_attach() and rc_query() and likely not causing problems since rc_query() is repeated continously. None could see it when first rc_query() failed... Secondly it was not called for .generic_bulk_ctrl_endpoint_response. Signed-off-by:
-
Antti Palosaari authored
Move work to the struct dvb_usb_device that we can access it inside .disconnect(). Also many other minor changes. Signed-off-by:
-
Antti Palosaari authored
It could be null in case of driver does not set it properly, like missing dynamic ID handling. Signed-off-by:
-
Antti Palosaari authored
Some error checking changes and refactoring. Signed-off-by:
-
Antti Palosaari authored
Adapter priv is not really needed, use device priv instead. There is only driver or two using that. Device caps are not needed. There was only "has I2C adapter" capability defined. It is useless as we can see same just checking existence of i2c_algo callback. And also remove some totally not used at all variables. Signed-off-by:
-
Antti Palosaari authored
Signed-off-by:
-
Antti Palosaari authored
Signed-off-by:
-
Antti Palosaari authored
Remove Cypress USB-interface firmware downloading routines. That is common module and having single vendor chip routines in common module is wrong. Just move those elsewhere. Move single function out from the dvb_usb_firmware.c and remove that file. Signed-off-by:
-
Antti Palosaari authored
Delay firmware download and whole driver initialization using workqueue. udev causes problems when blocking firmware download was done during module init. This will likely resolve all DVB USB firmware downloading issues we have had during recent years. Fixes bug in case of DVB USB driver: https://bugzilla.redhat.com/show_bug.cgi?id=827538Signed-off-by:
-
Antti Palosaari authored
It is no longer needed because all frontends are attached as a one go. Signed-off-by:
-
Antti Palosaari authored
Get rid of (struct dvb_usb_adapter_fe_properties) as we no longer need it. Frontends are now defined as a array of pointers inside adapter struct. Signed-off-by:
-
Antti Palosaari authored
Move .frontend_attach(), .tuner_attach() and .streaming_ctrl() from the frontend to adapter. Signed-off-by:
-
Antti Palosaari authored
Filtering given PIDs from the transport stream is done by the DVB USB bridge. It is highly possible there is limitations what kind of stream DVB USB bridge can PID filter, but it still does not make sense to define filters for each frontend as frontend could offer different stream types for different standards. Likely new way is to enable / disable PID filters are needed to make decision at runtime (callback). PID filters are quite legacy stuff as those are aimed cut stream smaller to fit for the USB1.1... Signed-off-by:
-
Antti Palosaari authored
Move USB stream properties from frontend to adapter. It is property of adapter. Signed-off-by:
-
Antti Palosaari authored
Add new callback in order to resolve used TS (MPEG transport stream) configuration. Normal TS is 188 byte payload only but there is currently 204 byte TS and raw TS supported too. Traditionally TS type was mapped as a frontend property but it does not work no longer after we changed from MFE (multi-frontend) to SFE (single-frontend). So it is not possible to map TS for the given FE as there could be only one FE instead of multiple. Signed-off-by:
-
Antti Palosaari authored
Move stream from the frontend to adapter. There could be only one stream per adapter. One adapter can has multiple frontends but only one can stream at the time. Signed-off-by:
-
Antti Palosaari authored
New callback to resolve current USB stream configuration. Signed-off-by:
-
Antti Palosaari authored
Signed-off-by:
-
Antti Palosaari authored
Change URB count, buffer size and type [BULK/ISOC] dynamically when needed if existing URB buffers are big enough. Signed-off-by:
-
Antti Palosaari authored
Add .get_rc_config() callback and remove old static configs. Refactor remote controller routines. Signed-off-by:
-
Antti Palosaari authored
Those must be initialized earlier as we now pass (struct dvb_usb_device *) to the firmware download callbacks too. Signed-off-by:
-
Antti Palosaari authored
Most commonly only one firmware is used by the driver and it is not needed to selected run time. So restore old functionality but allow .get_firmware_name() callback to override it. Signed-off-by:
-
Antti Palosaari authored
* remove old legacy code totally * move default RC keymap definition the the (struct dvb_usb_driver_info) Signed-off-by:
-
Antti Palosaari authored
That callback is called only once when device is connected. Call is done after the possible firmware is downloaded to the device, just after the .power_ctrl() and before adapters are created. Signed-off-by:
-
Antti Palosaari authored
Callback to resolve adapter count of current device. Old static .num_adapters field can be still used but the new .get_adapter_count() has priority if both offered by the driver. Signed-off-by:
-
Antti Palosaari authored
Pass all the needed data to the DVB USB core using (struct usb_device_id) .driver_info. That simplifies old code a lot and saves memory as all device IDs and names are not defined inside (struct dvb_usb_device_properties) as earlier. Signed-off-by:
-
Antti Palosaari authored
Signed-off-by:
-
Antti Palosaari authored
Use callback to return firmware name instead of static firmware name. There is some chips that needs to select firmware name according to chip revision. Signed-off-by:
-
