* Moving a size and data type information from daq/channels.h to
include/daq_data_types.h.
* Extending standalone_edc to use DBR_TIME* to look for late data.
* Updating the stream generators to handle more data types.

Turned src/drv/param.c src/drv/crc.c into a helper library.
Modified daqd and standalone_edc to consume this as a library.

Removed more now unused code from standalone_edc.

Tests:
 * daqd
 * nds

The test is to:

* start a simulated IFO (via fe_multi_stream_test)
* start daqd
* read live data from nds
* verify the data

This updates the NDS2 client detection, reimplements fe_stream_check_nds,
and adds some test harness logic.

Missing return values and return types keep daq from building on debian 10.

"2to3 -w .", and changes to shebang lines

Updated the FindEPICSBase.cmake to make full targets for ca, cas, Com, gdd.

These files appear to be generated during the cmake configure process,
and therefore shouldn't be in the repo.

This brings in:
* Updates to the cmake build to support FrameCPP 2.6+
* Various startup scripts for Debian8 & Gentoo 3.0
* Daqd files updated to match branch 3.5
* Moved over the ix/zmq/mx stream work from 3.5
* The split of daq_core.h into daq_core.h & daq_core_defs.h to
centralize declarations.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4827 6dcd42c9-f523-4c6d-aada-af552506706e

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4767 6dcd42c9-f523-4c6d-aada-af552506706e

Instead of not taking a minute trend size which breaks compatibility
with old daqdrc files, allow the minute trend buffer size to be
specified (but still defaults to the correct value if not specified) and
abort if it is not 60.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4766 6dcd42c9-f523-4c6d-aada-af552506706e

Previously it had started at the beginning of the buffer and move
dcu by dcu.  Now, use data from the channel scatter gather maps to
set where each dcu starts.

Working this way should isolate the effect of arithmetic errors in
writing into the producer move buffer.

This is only used in the DAQD_SHMEM, using this in more of the producers
would require more extensive changes in the initialize_vmpic code +
accounting for where TP data starts.

This fixes data layout issues that showed up in daqd_shmem if an expected
dcu was marked as bad.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4764 6dcd42c9-f523-4c6d-aada-af552506706e

daqd_shmem used to be protected by a 'DEV_BUILD' option, build it all
the time, if you can.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4763 6dcd42c9-f523-4c6d-aada-af552506706e

Added monitor_thread_scheduling.py, this is a pcaspy script which takes
two arguments, a PID and a prefix.  It then exports a set of PVs,
currently one per child thread of the given PID that contains the last
cpu the thread was scheduled on.  It refreshes the data at a 4hz rate.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4760 6dcd42c9-f523-4c6d-aada-af552506706e

set parameter "GDS_BROADCAST" = "1"; for a gds broadcaster.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4759 6dcd42c9-f523-4c6d-aada-af552506706e

Changed a local buffer from being stack allocated to heap allocated.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4758 6dcd42c9-f523-4c6d-aada-af552506706e

The configuration on the running system is:

start trend 600 60;

The last value (60) can be any multiple of 60 and defines how many minutes are in a
minute trend.

This option is being changed to:

start trend optionalint;

So in the production system it would be:

start trend 600;

Do not allow multi-minute trends.  It is too easy to misconfigure and end up with your minute trends averaging data longer than you thought.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4748 6dcd42c9-f523-4c6d-aada-af552506706e

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4729 6dcd42c9-f523-4c6d-aada-af552506706e

Compress the zmq stream by removing unneeded dcu headers and TP table
entries.

Fix some issues with data/TP layout in misc programs.

Renamed the daq_multi_cycle_... dataBlockSize to make it different
than the dcu_message_header_t dataBlockSize field.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4725 6dcd42c9-f523-4c6d-aada-af552506706e

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4703 6dcd42c9-f523-4c6d-aada-af552506706e

When run on gcc 4.4 the zmq_recv process would segfault on the
construction of the zmq receiving object.  This is a rewrite of that
code back to C in hopes that that cleared up an issue specific to gcc 4.4.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4685 6dcd42c9-f523-4c6d-aada-af552506706e

zmq_recv was not passing the data crc through, fix that.
producer_shmem needed to stop using an old rmIpc structure that no longer
applies.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4679 6dcd42c9-f523-4c6d-aada-af552506706e

Added debug output to:
daqd_shmem
fe_stream_test
zmq_multi_stream
zmq_dc_recv/zmq_recv

to show config & data crc values on the output streams.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4678 6dcd42c9-f523-4c6d-aada-af552506706e

The daqd_shmem had been overwriting TP data with regular data from the next dcu.
Advance the destination pointer the proper length to not overwrite test points.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4660 6dcd42c9-f523-4c6d-aada-af552506706e

Copying test point data out from the shmem buffer into the move buffer.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4659 6dcd42c9-f523-4c6d-aada-af552506706e

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4657 6dcd42c9-f523-4c6d-aada-af552506706e

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4656 6dcd42c9-f523-4c6d-aada-af552506706e

A number of structures have common portions, extract the common header fields.

Fixed a spurious copy/paste error in comm_impl.cc.

Moved the zmq_recv from using a max number of dcus to using a max transmission size.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4653 6dcd42c9-f523-4c6d-aada-af552506706e

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4652 6dcd42c9-f523-4c6d-aada-af552506706e

Combined the data/gds structures into one.  Updated the logic to allow
moving data from the fe simulation through to the daqd.  Partial work
on the daqd_shmem system (not complete).

Pulled some logic out of comm.y and placed it in its own files
comm_impl.cc/hh, to aid in debugging/tool introspection.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4651 6dcd42c9-f523-4c6d-aada-af552506706e

daqd_shmem is similar to a standiop.  However it reads data from one combined shared memory segment, does not require a timer or adc card.

Updates to the fe_stream_test to put data in the proper places.

Additional test scripts and commands to help manipulate and introspect mbuf memory buffers.
src/drv/mbuf/mbuf_probe
src/zmq_stream/scripts/dump_zmq_core_structs.py

Added a zmq_recv command to receive data from a zmq stream and publish it to a local mbuf.

Adding a new c userspace interface to the mbuf shared memory.

git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4650 6dcd42c9-f523-4c6d-aada-af552506706e

If built without zmq present the src/zmq_stream will not be built and the daqd builds will use a run_number_client that always returns a run number of 0. This is in response to a request to build the daqd variants w/o having a hard zmq dependency.

The change has been done to both the CMake and autotools build systems.


git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4625 6dcd42c9-f523-4c6d-aada-af552506706e

Also renamed /dev/symmetricom /dev/gpstime


git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4619 6dcd42c9-f523-4c6d-aada-af552506706e

Remove the throw() from full_frame to allow it to crash with the default handler and show better debug output


git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4618 6dcd42c9-f523-4c6d-aada-af552506706e

Also adjusts the build system to use the vendored copy of the zmq
bindings if the zmq.hpp file cannot be found.


git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4616 6dcd42c9-f523-4c6d-aada-af552506706e

The rcv build of daqd had code to read cycle count information from an
array of rmIpcStr structures.

1. This was unneeded in this build of daqd
2. The array was never initialized and resulted in invalid pointers
3. It only ran in production due to optimization recognizing a unused variable and removing the bad read.

This code breaks the producer.cc out to producer_rcv.cc and removes the unneeded rmIpcStr structures and accesses.


git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4593 6dcd42c9-f523-4c6d-aada-af552506706e

As requested by Keith:
- The run_number database should have date/time (or GPS) stamps for each record. (added)
- There should be an EPICS variable that displays the current "run number" (FW1_CONFIGURATION_NUMBER)
- Ability to save the database to arbitrary location (so it can be backed up) (Added the -f flag to run_number_server)
- Example 'systemd' service file for run_number_server installation


git-svn-id: https://redoubt.ligo-wa.caltech.edu/svn/advLigoRTS/trunk@4577 6dcd42c9-f523-4c6d-aada-af552506706e