- Use FPGA images with fixed sc12 converter.
- Properly flush channels and restart streaming in the case of an overrun.
Signed-off-by: michael-west <michael.west@ettus.com>
Adds uhd_rc as a link target to static builds of libuhd. This fixes
build errors like this:
```
uhd/lib/cal/database.cpp:12:10: fatal error: cmrc/cmrc.hpp: No such file
or directory
#include <cmrc/cmrc.hpp>
```
This also adds uhd_rc as to $libuhd_libs instead of just listing it
separately, and target objects from uhd_rc to $libuhd_sources.
See the CMake 3.8 documentation on these two variables:
https://cmake.org/cmake/help/v3.8/variable/PROJECT-NAME_SOURCE_DIR.htmlhttps://cmake.org/cmake/help/v3.8/variable/CMAKE_SOURCE_DIR.html
Under normal circumstances, these two are identical. For sub-projects
(i.e., when building UHD as part of something else that is also a CMake
project), only the former is useful. There is no discernible downside of
using UHD_SOURCE_DIR over CMAKE_SOURCE_DIR.
This was changed using sed:
$ sed -i "s/CMAKE_SOURCE_DIR/UHD_SOURCE_DIR/g" \
`ag -l CMAKE_SOURCE_DIR **/{CMakeLists.txt,*.cmake}`
$ sed -i "s/CMAKE_BINARY_DIR/UHD_BINARY_DIR/g" \
`ag -l CMAKE_BINARY_DIR **/{CMakeLists.txt,*.cmake}`
At the same time, we also replace the CMake variable UHD_HOST_ROOT (used
in MPM) with UHD_SOURCE_DIR. There's no reason to have two variables
with the same meaning and different names, but more importantly, this
means that UHD_SOURCE_DIR is defined even in those cases where MPM calls
into CMake files from UHD without any additional patches.
Shoutout to GitHub user marcobergamin for bringing this up.
This gives us type-safety, as well as allowing us to create unit tests for
RFNoC radio_controls without having to create actual RPC servers and clients
in the unit tests.
There are applications (typically for debugging purposes) which need
access to the token and the mb_args. They are thus published via the
property tree.
N320 doesn't have an automatic RX IQ balance correction, so that API is
removed.
The auto-DC offset correction was calling into the manual DC offset
correction code, which means auto-DC offset correction was never enabled
for N320.
This is more of an expressive change than a functional change; Python
seems to add this path to the PYTHONPATH anyway, at least for some
systems. We neverless make this change because:
- It's more explicit/expressive. When tests are run, the PYTHONPATH env
variable is printed, and it now contains this path where it should be,
right at the front. People reading the ctest/python.unittest output
now get told explicitly which path we mean.
- This guarantees that this path is added, even if Python/unittest
should behave differently on other systems or versions.
To clarify: When running unit tests, we want to run the Python code from
build/python, not the installed version. The latter may not yet exist,
and if it does, it's not the version we are editing.
This adds a Verilog definition named `UHD_FPGA_DIR that corresponds to
the location of the UHD "fpga" directory. This allows you to include
files in your out-of-tree modules relative to the FPGA directory. For
example, you could include the library header file rfnoc_chdr_utils.vh
using the following:
`include `"`UHD_FPGA_DIR/usrp3/lib/rfnoc/core/rfnoc_chdr_utils.vh`"
Some simulators may not support `" outside of the context of a `define,
in which case you can do the following:
`define RFNOC_CHDR_UTILS_PATH \
`"`UHD_FPGA_DIR/usrp3/lib/rfnoc/core/rfnoc_chdr_utils.vh`"
`include `RFNOC_CHDR_UTILS_PATH
The test_recv_get_release test should be checking received packets had
the same content as they did on send(), but was instead assigning to the
received buffer.
Shoutouts to GitHub user johnwstanford for pointing out the issue.
inline_io_service::disconnect_receiver() uses the recv_link_if parameter
as a key into the _recv_tbl map. In some error cases, notably when an
underlying link timeout occurs when setting up a transport, that key may
not exist in the table. Attempting to index the table by that
non-existent key causes an std::out_of_range exception to be thrown.
However, in the aforementioned error case, disconnect_receiver() is
called as part of a destructor of an object that is in one of the stack
frames being unwound as the timeout exception is in flight. Throwing an
exception while one is in flight ultimately causes the C++ runtime to
terminate the process. (Generally, throwing an exception in a
destructor, unless caught within the destructor, is considered a bad
practice for this very reason.)
This PR modifies disconnect_receiver() to check for the existence of the
entry in the map and bypass the access if it is not present, thus
preventing the exception from being thrown in this function which is
invoked from another object's destructor.
The previous commit fixed a bug in the DUC, where get_frequency_range()
reported incorrect values. The DDC did not have this bug, but we port
the updates to the unit tests and the documentation from the DUC to the
DDC for consistency's sake.
The tuning range of the DUC depends on the output sample rate (which is
larger), but it was using the input sample rate. This was causing a bug
where for Tx, the DSP tuning range was limited when using multi_usrp API,
and thus would not allow to DSP-tune beyond the current sampling rate.
In this patch, we also re-use the existing calculation of the sampling
rate, and harmonize that code between duc_block_control and
ddc_block_control.
Consider the following Python REPL code:
>>> import uhd
>>> U = uhd.usrp.MultiUSRP('type=x300')
>>> U.set_rx_rate(10e6)
>>> U.set_tx_rate(10e6)
>>> # Creating a streaming is required, or the input rate will not be
>>> # set:
>>> S = U.get_tx_stream(uhd.usrp.StreamArgs("fc32", "sc16"))
>>> treq = uhd.types.TuneRequest(1e9)
>>> treq.rf_freq = 1e9
>>> treq.dsp_freq = 50e6
>>> treq.dsp_freq_policy = uhd.types.TuneRequestPolicy.manual
>>> treq.rf_freq_policy = uhd.types.TuneRequestPolicy.manual
>>> tres = U.set_rx_freq(treq, 0)
>>> print(str(tres))
Tune Result:
Target RF Freq: 1000.000000 (MHz)
Actual RF Freq: 1000.000000 (MHz)
Target DSP Freq: 50.000000 (MHz)
Actual DSP Freq: 5.000000 (MHz)
>>> # Note the last two lines: The *target* DSP freq was already clipped
>>> # to 5 MHz. These lines show 50.0 MHz when this patch is applied.
This bugfix is accompanied some related changes:
- The unit test is amended to verify the behaviour
- The API documentation is amended with details on its behaviour
This sets the ARCH and PART_ID environment variables so that the
selected part family is used for simulations by default. This can be
overridden by changing them in the Makefile for the testbench if a
testbench requires a specific part family. Prior to this change, the
default was always ARCH=kintex7, PART_ID=xc7k410t/ffg900/-2, which
required support for that part to be installed.
Some archs require linking against libatomic, others don't. We add some
CMake code that checks for libatomic.so requirement if:
- We are not on MSVC, AND
- Compiling std::atomics code would cause a linker error.
We then check for the existence of libatomic.so, and fail if we can't
find it.
The script was previously only Python 2 compatible.
Python 2 support is now removed, so we don't maintain backwards
compatibility with Python 2.
This also fixes all linter warnings.
A recv() of zero samples on an RX stream should return immediately
(i.e., without respect to the timeout and regardless of the availability
of samples), surfacing any stream error conditions via metadata. This
convention was broken in a2f10ee9, causing a recv() of zero samples to
wait for the entire timeout period and then return ERROR_CODE_TIMEOUT if
no samples are available. This commit restores the desired semantics.
In x300_eth_mgr.cpp the variable init is created but not initialized.
Only some of the variables within the struct are then set before init
is assigned to a different variable. Initialize the variable to
prevent unexpected values.
- Reduce PLL1 DLD lock count to 4,000 (0xFA0), or 100ms
- Change loop to check for lock every 10ms
Signed-off-by: michael-west <michael.west@ettus.com>
Modified to send 2 packets back to back at each packet size to test
output during sequential input packets. Also fixed whitespace.
Signed-off-by: michael-west <michael.west@ettus.com>
Added delay to error packets so overrun error is back in-band. Avoids
dropping good data packets in the case of an overrun.
Signed-off-by: michael-west <michael.west@ettus.com>
Re-wrote converter to remove clock cycle delay on i_tready when handling
residual output and fixed improper handling of tlast during residual
data processing. Resolves some USB overflow issues when using sc12 data
type on B200 devices.
Signed-off-by: michael-west <michael.west@ettus.com>
Fix implicit typecasts that could potentially lose data. Doing this to
show that these typecasts are done on purpose (and to resolve warnings
from VS).
Fix the "Enum.3: Prefer class enums over "plain" enums" warning for the
node_type enum and update the calls to the enumerators as proposed by
the C++ Core Guidelines.
There were some rare corner cases where the EOB could get lost in the
DUC due to the dds_timed logic not always passing it through as it
should. This resulted in an underflow error message at the end of
transmission.
This commit also fixes an issue where part of the last packet
used a frequency shift of 0 instead of the requested frequency
shift, and an issue where the first few samples of a burst used the
wrong frequency shift value.
Part of the fix includes adding a TUSER port to dds_sin_cos_lut_only.
The TUSER port is built into the IP but was disabled. It is now
enabled and set to 1 bit wide. This has a very small effect on
resource usage and can be left unconnected when not needed.
The dds_freq_tune block was shared by the DUC and DDC. To avoid
affecting the DDC, a new version, dds_freq_tune_duc, is being
added for the DUC to use that has the necessary fixes.
The new dds_wrapper.v is a wrapper for the dds_sin_cos_lut_only IP.
This IP has the undesirable behavior that new inputs must be provided
to push previous outputs through the IP. This wrapper hides that
complexity by adding some logic to ensure all data gets pushed through
automatically. This logic uses the TUSER port on the IP.
Finally, a testbench for dds_timed was added.
PkgComplex adds functions for doing complex arithmetic in SystemVerilog
simulation.
PkgMath provides mathematical operations and constants that aren't
built into SystemVerilog, such as a constant for pi and the function
round().
PkgRandom adds randomization functions beyond what standard Verilog
supports but that don't require any special licenses or simulators.
cmake has newer variables to identify specific Visual Studio version.
This change updates to use those. This also changes how images
files are globbed to be relative paths, which is useful if build files
are moved from machine to machine. This case is possible in Azure
Pipeline's CI.
Signed-off-by: Steven Koo <steven.koo@ni.com>
Fixed issue where variable tmp_file going out of scope could have
leaked the storage it points to. Replaced the declaration of tmp_file
by using a unique pointer with custom deleter that takes care of
closing the tmpfile when it is not needed anymore.
