This provides a non-blocking variant of bladerf_submit_stream_buffer(),
which provides API users with more control over how they might want
to handle asynchronous buffer submission.
This API functionality has been exposed to allow users to implement
schemes similar to that used to address Issue #438.
This will be returned by API functions when an operation would need
to block to complete, but the caller has requested non-blocking
operation. In this situation, the caller may need to wait and
resubmit the request at a later time.
This "Queue full" event may occur when using the scheduled
retunes. Since this is an event that the user can respond to
and act upon, an appropriate status/error code has been added.
As of FPGA v0.2.0, the frequency tuning algorithm is run in the FPGA to
eliminate time wasted by communicating LMS6002D registers over USB. This
is now the default if the FPGA version is >= v0.2.0.
This default can be overridden via a BLADERF_DEFAULT_TUNE_MODE
environment variable, with options being "host" or "fpga".
The main landing page in the Doxygen is now filled out with
some general information.
Existing examples have been simplified and moved to
dedicated pages that are linked to from the main page and
relevant sections of the documentation elsewhere.
Explicitly noted user buffer size required by bladerf_get_serial().
Made note of extraneous byte used by bladerf_image structure to
avoid confusion. This extra byte will be kept to avoid breaking
compatibility.
This changeset is directly based upon a patch and suggestions kindly
provided by Marian and Diana from Null Team (www.null.ro) and YateBTS
(www.yatebts.com). The original patch may be found on the issue tracker
for issue #95.
Per issue #95, it has been found that on many Linux-based
USB 3.0 setups, it is not possible to re-connect to a bladeRF after
a previous application using the bladeRF failed to close its device
handle. (For example, as the result of an application crashing or being
killed.)
To recover from this situation, we now perform a USB port reset when
opening the device, prior to attempting to actually access it.
Should this behavior cause any new issues, it can be disabled...
... at compile-time by providing
-DENABLE_LIBUSB_DEV_RESET_ON_OPEN=False to CMake
... at run-time via the bladerf_set_usb_reset_on_open(false),
when the aforementioned compile-time flag is True.
A huge "thank you" goes out Marian, Diana, and everyone at Null Team and
YateBTS!
API functionality to probe for FX3 bootloader devices and download
firmware to detected devices has been introduced.
This is based upon the boot image format/procedure documented in
Cypress's AN76405: "EZ-USB (R) FX3 (TM) Boot Options"
http://www.cypress.com/?docID=49862
The existing bladerf_sync_rx() snippet in the libbladeRF API docs has
been replaced by a more complete snippet in libbladeRF/doc/examples.
From now on, major snippets and examples to be included in
Doxygen-generated documentation shall now be placed in the
libbladeRF/doc/examples directory, and must be buildable and testable.
The Doxygen "snippet" command to shall be used to include the relevant
code from the example program.
libbladeRF: [WIP] Use a simple delay to wait for TX samples in example
API support for timestamp metadata has been introduced. This is made
available when using the newly introduced BLADERF_FORMAT_SC16_Q11_META
format value.
When using the synchronous interface, metadata is exchanged between the
caller and the API via the updated bladerf_metadata structure. Users of
the asynchronous interface must currently manually manage the metadata
headers as desired, as documented in libbladeRF.h.
When using the BLADERF_FORMAT_SC16_Q11_META format in conjunction with
the synchronous interface, it is possible to retrieve the timestamp for
received samples and an indication of overrun. An RX operation may also
be scheduled for at a specific timestamp value the future. Similarly,
bursts of samples can be scheduled to be transmitted at a specific
timestamp value, or at the first available timestamp ("now").