Hi Martin,
Thanks for looking into it. Meanwhile I collected more information and it
does get weirder.
I'm attaching a basic C++ program that does nothing useful, it just RX
streams until CTRL+C kills it. Compiling instructions for Linux are in the
first lines of the code.
On Ubuntu 18.04, everything works even with UHD 3.11, meaning that I can
kill the binary and rerun it right away as many times as I want.
On Ubuntu 16.04, UHD 3.11 seems to have the "state" that I mentioned
earlier in this thread, meaning that CTRL+C kills the board until reboot
("/usr/lib/uhd/utils/b2xx_fx3_utils -D" may work too, but I haven't tried
it as a soft reboot).
I wrote this test code in a hurry, and hopefully there aren't distracting
bugs. The same does happen in production code, which I can't attach for
obvious reasons.
Thanks,
Dario
On Thu, Apr 19, 2018 at 3:34 PM, Martin Braun <[email protected]>
wrote:
> On 03/21/2018 10:12 PM, Dario Fertonani wrote:
> > No comments on this? It's a major bug introduced in the new release, in
> > my opinion.
> >
> > Anybody can replicate it by running the official test file linked below,
> > when compiled with UHD 3.11 from Ubuntu PPA.
> >
> > https://github.com/manuts/uhd-examples/blob/master/rx_multi_
> samples.cpp
>
> Just for the record, this is not the source location for official examples.
>
> > Just run it long enough (with proper "nsamps" option), kill it via
> > CTRL+C while still streaming, then relaunch. It dies instantly with
> > errors like the following.
> >
> > [ERROR] [STREAMER] recv packet demuxer unexpected sid 0x1fffc
> > [ERROR] [STREAMER] recv packet demuxer unexpected sid 0x10006
> > DError: Receiver error ERROR_CODE_OVERFLOW (Out of sequence error)
>
> Dario,
>
> sorry for the delayed response. We'll need to triage this and see what's
> going on.
>
> Thanks for bringing this up,
>
> Martin
>
>
>
> >
> > In my case I ran the following command, but the nsamps value doesn't
> > really matter as long as it's long enough for the user to kill the first
> > run via CTRL+C.
> >
> > $ ./rx_multi_samples --nsamps 10000000
> >
> >
> > Thanks,
> > Dario
> >
> >
> >
> > On Thu, Mar 8, 2018 at 5:30 PM, Dario Fertonani
> > <[email protected] <mailto:[email protected]>> wrote:
> >
> > Thank you Martin, logging is fixed now.
> > However, something must have changes in the rx_streamer behavior.
> > Basic receive code that has compiled/worked since the 3.9.x days no
> > longer works. This is for B210 under Ubuntu 16.04, g++ 5.4, UHD fro
> > 3.11 from PPA.
> >
> > The receiver works the first time after B210 boot (and associated
> > FPGA image load). Then, after CTRL+C or equivalent interruptions,
> > the same binary fails right away with these errors (the SID change
> > at every run)
> >
> > [ERROR] [STREAMER] recv packet demuxer unexpected sid 0xfffffff6
> > [ERROR] [STREAMER] recv packet demuxer unexpected sid 0xfffcfff5
> >
> > as soon as the recv function is called. This suggests that 1)
> > there's some stale state from the previous run, and 2) a B210 reboot
> > clears that state.
> > Is there a way to clean that state from C++ API? Or, more in
> > general, a way to get the rx_stream to work without rebooting the
> board?
> >
> > Thanks,
> > Dario
> >
> > On Thu, Mar 8, 2018 at 12:49 PM, Martin Braun
> > <[email protected] <mailto:[email protected]>> wrote:
> >
> > On 03/08/2018 09:56 AM, Dario Fertonani wrote:
> > > Hi Martin,
> > >
> > > Our code doesn't compile on Ubuntu 16.04 with UHD 3.11 from
> PPA (UHD
> > > 3.10.x works well). This is on "amd64" ark, with g++ that dies
> out with
> > > a missing header (copied below)
> > >
> > > fatal error: uhd/utils/msg.hpp: No such file or directory
> > >
> > >
> > > I can probably fix the headers with some trial and error, but
> there's a
> > > deeper question: Did anything significant change in the "msg"
> API?
> >
> > Dario,
> >
> > everything related to the logging API (msg.hpp falls into this
> > category)
> > was completely re-done between 3.10 and 3.11. I'll refer to the
> > manual
> > page: http://files.ettus.com/manual/page_logging.html
> > <http://files.ettus.com/manual/page_logging.html>
> >
> > ...and this email:
> > http://lists.ettus.com/pipermail/usrp-users_lists.
> ettus.com/2017-April/024653.html
> > <http://lists.ettus.com/pipermail/usrp-users_lists.
> ettus.com/2017-April/024653.html>
> >
> > -- M
> >
> > >
> > > Thanks,
> > > Dario
> > >
> > >
> > > On Mon, Mar 5, 2018 at 1:51 PM, Martin Braun via USRP-users
> > > <[email protected]
> > <mailto:[email protected]>
> > <mailto:[email protected]
> > <mailto:[email protected]>>> wrote:
> > >
> > > Two more comments:
> > >
> > > - Binary installers are currently building and will be
> > made available
> > > shortly on the usual spots: Ubuntu PPA is here:
> > > https://launchpad.net/~ettusresearch/+related-packages
> > <https://launchpad.net/~ettusresearch/+related-packages>
> > > <https://launchpad.net/~ettusresearch/+related-packages
> > <https://launchpad.net/~ettusresearch/+related-packages>>
> > >
> > > ...and Windows+Fedora binaries will be published here:
> > >
> > http://files.ettus.com/binaries/uhd/uhd_003.011.000.
> 000-release/ <http://files.ettus.com/binaries/uhd/uhd_003.011.000.
> 000-release/>
> > >
> > <http://files.ettus.com/binaries/uhd/uhd_003.011.000.
> 000-release/
> > <http://files.ettus.com/binaries/uhd/uhd_003.011.000.
> 000-release/>>
> > >
> > > (but aren't there yet).
> > >
> > > - Because we changed out dev model, you will need to
> > manually reset
> > > maint branch if you're tracking that on git. The following
> > commands will
> > > work:
> > >
> > > $ git checkout maint
> > > $ git remote update
> > > $ git reset --hard @{u}
> > >
> > > Cheers,
> > > Martin
> > >
> > >
> > > On 03/05/2018 12:20 PM, Martin Braun wrote:
> > > > Hi all,
> > > >
> > > > today we've tagged the release for UHD 3.11.0.0. We
> > found a bunch of
> > > > issues during RC testing, but not enough to warrant
> > another release
> > > > candidate.
> > > > Some of the issues we fixed were actually
> > usability-related, but there
> > > > was an actual bug with the images downloader, which
> > might not properly
> > > > update the correct images after doing git pull, and we
> > found and
> > > fixed a
> > > > bug in the X300 firmware which manifests as the GPSDO
> > not being found
> > > > (if you were running master branch with the X300, we
> > highly recommend
> > > > upgrading).
> > > >
> > > > There is one known N310 issue which is still open as of
> > this release:
> > > > When running UHD natively on the device in a
> > multi-threading scenario,
> > > > it is possible to encounter race conditions which lead
> > to UHD dropping
> > > > command packets. We will be releasing a bugfix release
> > within a few
> > > > weeks to address this issue, along with anything else we
> > might
> > > have missed.
> > > >
> > > > Please find the tag on github here:
> > > >
> > >
> > https://github.com/EttusResearch/uhd/releases/
> tag/release_003_011_000_000
> > <https://github.com/EttusResearch/uhd/releases/
> tag/release_003_011_000_000>
> > >
> > <https://github.com/EttusResearch/uhd/releases/
> tag/release_003_011_000_000
> > <https://github.com/EttusResearch/uhd/releases/
> tag/release_003_011_000_000>>
> > > >
> > > > Cheers!
> > > >
> > > > Martin
> > > >
> > > >
> > > > Changelog:
> > > > ## 003.011.000.000
> > > > * N310: Added driver
> > > > * UBX: Add support for CAL antenna for Rev E
> > > > * Added the module peripheral manager (MPM) with all
> > subcomponents
> > > (N310
> > > > drivers, mpm_shell, RPC server, BIST, etc.)
> > > > * UHD: Added rpclib as a internally tracked dependency
> > > > * UHD: Reduced the usage of Boost
> > > > * UHD: Updated uhd_images_downloader: Now uses more
> > elaborate manifest
> > > > to optimize downloads
> > > > * UHD: Introduced uhdlib internal include paths
> > > > * UHD: Add support for configuration files. USRP
> > settings can now
> > > be set
> > > > using a uhd.conf file in addition to device args
> > > > * UHD: Add narrow and narrow_cast
> > > > * gr-usrptest: Various bugfixes
> > > > * Updated required Vivado version to 2017.4
> > > > * Updated all license headers to use SPDX identifiers
> > and correctly
> > > > identify Ettus Research as part of National Instruments
> > > > * This release includes all bugfixes and features from
> > previous
> > > > releases, in particular, the 3.10.* release cycle
> > > >
> > >
> > >
> > > _______________________________________________
> > > USRP-users mailing list
> > > [email protected]
> > <mailto:[email protected]>
> > <mailto:[email protected]
> > <mailto:[email protected]>>
> > >
> > http://lists.ettus.com/mailman/listinfo/usrp-users_
> lists.ettus.com
> > <http://lists.ettus.com/mailman/listinfo/usrp-users_
> lists.ettus.com>
> > >
> > <http://lists.ettus.com/mailman/listinfo/usrp-users_
> lists.ettus.com
> > <http://lists.ettus.com/mailman/listinfo/usrp-users_
> lists.ettus.com>>
> > >
> > >
> >
> >
> >
>
>
// Compile:
// g++ rxTest.cpp -o rxTest -std=c++11 -O3 -luhd -lboost_system
// Run:
// ./rxTest
#include <iostream>
#include <uhd/usrp/multi_usrp.hpp>
//
int main( )
{
//constants
const double SamplingRate_Hz = 8e6;
const double CarrierFreq_Hz = 100e6;
const unsigned ChunkSamples = 1024;
//init board and set frequencies
auto rfBoardPtr = uhd::usrp::multi_usrp::make( uhd::device_addr_t( "" ) );
rfBoardPtr->set_clock_source( "internal" );
rfBoardPtr->set_time_unknown_pps( 0.0 );
rfBoardPtr->set_rx_rate( SamplingRate_Hz );
rfBoardPtr->set_rx_freq( uhd::tune_request_t( CarrierFreq_Hz ) );
//stream
std::vector< std::complex< float > > rxSampleVec( ChunkSamples );
uhd::stream_args_t streamParams( "fc32" , "sc16" );
streamParams.channels = { 0 };
auto rxStreamPtr = rfBoardPtr->get_rx_stream( streamParams );
uhd::stream_cmd_t rxStreamCmd( uhd::stream_cmd_t::STREAM_MODE_START_CONTINUOUS );
rxStreamCmd.stream_now = true;
rxStreamPtr->issue_stream_cmd( rxStreamCmd );
std::cout << "Receiving samples..." << std::endl;
while ( true )
{
uhd::rx_metadata_t rxMetaData;
if ( rxStreamPtr->recv( rxSampleVec.data( ) , ChunkSamples , rxMetaData ) < ChunkSamples )
{
std::cerr << "Error in UHD recv." << std::endl;
std::cerr << rxMetaData.to_pp_string( false ) << std::endl;
return -1;
}
}
//return
return 0;
}
_______________________________________________
USRP-users mailing list
[email protected]
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
