[subexp-daq] How to start a DAQ system after NURDLIB, TRLOII, DRASI, etc. were updated
Håkan T Johansson
f96hajo at chalmers.se
Thu Jan 18 22:39:31 CET 2024
On Thu, 18 Jan 2024, Weber, Guenter Dr. wrote:
>
> Dear Hans,
>
>
> there was some redirect burried in the script which executes scripts which
> execute some more scripts ... and that made me start the DAQ in the old
> folder, not the new one. Sorry for that.
>
>
> I now came across the following line in the script which starts the VULOM:
>
>
> $VULOM4_CTRL --addr=$addr --clear-setup --config=vulom.trlo standalone
> module_trigger
>
>
> Here I need to know the firmware number to define the path VULOM4_CTRL,
> right? Is there a way to do this automatically? Somehow this works already
> in the makefile of NURDLIB.
It looks like this is done with this line in the nurdlib Makefile:
VULOM4_FW:=$(shell cat $(TRLOII_PATH)/fw/vulom4b_trlo/trlo_defs.h 2>&1 | grep MD5SUM_STAMP | sed 's/.*0x//')
such that something like
VULOM4_FW=`cat $(TRLOII_PATH)/fw/vulom4b_trlo/trlo_defs.h 2>&1 | grep MD5SUM_STAMP | sed 's/.*0x//'`
might work in a shell script, if TRLOII_PATH is set, and the fw/ directory
exists and has the current gatewares.
'Funfact': This general mess with the TRLO II version numbers is of my
creation, however a proper fix has so far required too much effort.
Basically, the gateware would have to give info on the offsets and sizes
of various controllable items (fairly easy), and the trloctrl program and
functions would need to use that (much more complex - that it at all works
is thanks to generation of many structures from that trlo_defs.h header,
where some of them also intermix with the .trlo lexer/parser).
> I would like to avoid to hard code this into the environment settings if
> there is a way to figure it out on the fly.
Cheers,
Håkan
>
>
>
> Thanks so much!
>
>
>
> Best greetings
>
> Günter
>
>
>
>
>
>
> ____________________________________________________________________________
> Von: subexp-daq <subexp-daq-bounces at lists.chalmers.se> im Auftrag von Hans
> Toshihide Törnqvist <hans.tornqvist at chalmers.se>
> Gesendet: Donnerstag, 18. Januar 2024 13:28:36
> An: Discuss use of Nurdlib, TRLO II, drasi and UCESB.; Weber, Guenter Dr.
> Betreff: Re: [subexp-daq] How to start a DAQ system after NURDLIB, TRLOII,
> DRASI, etc. were updated
> Dear Günter,
>
> It looks like either the Vetar driver failed, or the fw/driver is old
> and nurdlib expects a file that didn't exist a while back. If it's the
> latter I can add a switch to ignore this file, it's not critical for
> readout and is used to enter a faster readout mode.
>
> Which version of nurdlib is this? The files and line numbers don't match
> with what I see in the source code, and the last time gsi_etherbone.c
> changed was early last year.
>
> The message "Could not open so-and-so" should be for the device, e.g.
> /dev/pcie_wb0. This is used for the readout.
>
> The dactl file should fail with "Failed dactl fopen".
>
> To summarise, I would suggest to make sure that the correct nurdlib and
> r3bfuser are executed by the DAQ scripts. Meanwhile I will prepare a
> switch to ignore the dactl file which could anyhow be useful for old
> configurations, and to clean up the Etherbone error messages.
>
> Best regards,
> Hans
>
> On 2024-01-18 11:29, Weber, Guenter Dr. wrote:
> > Dear friends,
> >
> >
> > I now started the DAQ and it looks like there is a problem with the
> > VETAR2 module. Attached please find the output once the DAQ is started
> > on the RIO4.
> >
> >
> > Here ist the error message:
> >
> >
> > 11: a:1: .Slow module[1]=GSI_VETAR. (crate/crate.c:866)
> > 11: a:1: .Gsi Etherbone init_slow {
> > (module/gsi_etherbone/gsi_etherbone.c:110)
> > 5: a:1: ..Gsi Etherbone Could not open /sys/class/vetar/vetar0/dactl: No
> > such file or directory.
> > (module/gsi_etherbone/gsi_etherbone.c:120)
> > 5: a:1: ..Calling exit(EXIT_FAILURE)...
> > (module/gsi_etherbone/gsi_etherbone.c:120)
> >
> > How to proceed?
> >
> >
> >
> > Many thanks and best greetings
> > Günter
> >
> >
> >
> >
> > ------------------------------------------------------------------------
> > *Von:* subexp-daq <subexp-daq-bounces at lists.chalmers.se> im Auftrag von
> > Håkan T Johansson <f96hajo at chalmers.se>
> > *Gesendet:* Dienstag, 16. Januar 2024 19:20:54
> > *An:* Discuss use of Nurdlib, TRLO II, drasi and UCESB.
> > *Betreff:* Re: [subexp-daq] How to start a DAQ system after NURDLIB,
> > TRLOII, DRASI, etc. were updated
> >
> > On Tue, 16 Jan 2024, Weber, Guenter Dr. wrote:
> >
> >>
> >> Dear friends,
> >>
> >>
> >> I am just trying to figure out how all the various scripts/commands work
> >> together for setting up our DAQ system. Maybe, you can help to clearify
> some
> >> things.
> >>
> >>
> >>
> >> - master.bash
> >
> > This would be intended to run on the readout board (RIO). The other
> > scripts below is for the PC.
> >
> >>
> >> source ../env/env.sh
> >> ./trloii_setup.sh
> >> # gdb --args \
> >> ../r3bfuser/build_cc_ppc-linux_4.2.2_debug_drasi/m_read_meb.drasi \
> >> --label=MCAL1 \
> >> --triva=master,fctime=10,ctime=50 \
> >> --log-no-rate-limit \
> >> --server=drasi,dest=lyserv \
> >> --buf=size=400Mi \
> >> --max-ev-size=0x1000000 \
> >> --subev=crate=1,type=20,subtype=2,control=9,procid=1 \
> >> --eb=lyserv \
> >> "$@"
> >>
> >> In the first line some environment variables are set. The second line
> tells
> >> the VULOM4 how to operate (i. e. selection from modes of operation
> defined
> >> in vulom.trlo). Then DRASI is started with a bunch of parameters. Is
> there
> >> somewhere an explanation of the purpose of all these settings?
> >
> > The drasi command line options should be described here:
> >
> > http://fy.chalmers.se/~f96hajo/drasi/doc/drasi_cmdline.html
> > <http://fy.chalmers.se/~f96hajo/drasi/doc/drasi_cmdline.html>
> >
> > (Note: 'lyserv' is the name of your PC, at least as seen from the RIO.)
> >
> > The TRLO II setup file (vulom.trlo) is worse in terms of documentation.
> > The format as such is hopefully rather straightforward. A description is
> > here: http://fy.chalmers.se/~f96hajo/trloii/man5/trloconf.5.html
> > <http://fy.chalmers.se/~f96hajo/trloii/man5/trloconf.5.html> .
> > The actual meaning is more complex, the TRLO II description can be found
> > here:
> >
> > http://fy.chalmers.se/~f96hajo/trloii/vulom4_trlo/descr_defs_frame.html
> > <http://fy.chalmers.se/~f96hajo/trloii/vulom4_trlo/descr_defs_frame.html>
> >
> > (left pane), which just is a colourised version of this:
> > http://fy.chalmers.se/~f96hajo/trloii/description.txt
> > <http://fy.chalmers.se/~f96hajo/trloii/description.txt>
> >
> > It does not describe the .trlo file as such, but what the gateware tries
> > to achieve.
> >
> >> And what is in the last line? I never came accross "$@" before.
> >
> > That would be all the (non-shifted, i.e. not removed) options given to the
> > script itself, i.e. master.bash.
> >
> >> - eb.bash
> >>
> >> ../drasi/bin/lwrocmerge \
> >>
> >> --label=MCAL_EB \
> >> --merge-mode=event \
> >> --server=trans,flush=1 \
> >> --server=stream,flush=1 \
> >> --buf=size=1600Mi \
> >> --max-ev-size=20Mi \
> >> --eb-master=rio4-mcal-1 \
> >> --file-writer \
> >> --drasi=rio4-mcal-1
> >>
> >> What is the purpose of this command?
> >
> > This runs an 'event builder' on the PC. Not strictly needed, but this
> > way, the RIO only ever needs to send data once, to this event-builder
> > (EB). File writing, and streaming of on-line data is then handled by this
> > process. It is cheaper to have better network cards and so on in a plain
> > PC.
> >
> >> Why are values for buffer size and max
> >> event size different then in the command before?
> >
> > The nomenclaturure is unfortunately a bit convoluted below with ucesb
> > regards 'buffer size'. The --buf in the two commands above give the size
> > of the memory used to buffer data inside the process, and depends on how
> > much memory each machine has. The --max-ev-size tells how large each .lmd
> > event can be. The --max-ev-size configured in the event builder needs to
> > be as large as the sum of all event sources it has. I.e. should be able
> > to be the same as in your single source above.
> >
> > In the readout itself, any event that is read out cannot be larger than
> > this. If it is, then the DAQ will report a fatal failure.
> >
> >> - fanout.bash
> >>
> >> #!/bin/bash
> >> while :
> >> do
> >> ~/mbsrun/rio4/mcalstruck/ucesb/empty/empty \
> >> stream://localhost \
> >> --server=stream:8001,bufsize=10Mi,flush=1,dataport:7001
> >> sleep 5
> >> done
> >>
> >> This looks like setting up a connection point to the outside world,
> right?
> >
> > Yes. It accepts connections and will serve them data for online analysis.
> >
> > stream://localhost tells it where to read data (here the PC itself where
> > it is running). Using the 'stream' protocol, which esentially is just raw
> > LMD events. (As is the 'transport protocol'.) That will be served on the
> > default port used for the stream server set up by the EB above.
> >
> > --server is then the server that accepts connections, and serves data on
> > another port (8001).
> >
> >> And we find the third (random?) value for a buffer size.
> >
> > That bufsize is actually the LMD buffer size used by the stream server.
> > The corresponding value in the readout and EB is set implicitly from the
> > maximum event size...
> >
> >> - rate.bash
> >>
> >> ../drasi/bin/lwrocmon --rate rio4-mcal-1
> >>
> >> Ok, this I understand. We can look the DAQ over the shoulder and see what
> >> the thing is doing in terms of number of trigger events, transported
> data,
> >> etc.
> >
> >> - log.bash
> >>
> >> ../drasi/bin/lwrocmon --log rio4-mcal-1 localhost
> >>
> >> This is also easy to understand.
> >
> > :-)
> >
> > Then also try to do (on the PC)
> >
> > drasi/bin/lwrocmon --tree rio4-mcal-1 localhost
> >
> > which would show a 'tree-view' monitoring of the running system.
> >
> >
> > Cheers,
> > Håkan
> >
> --
> subexp-daq mailing list
> subexp-daq at lists.chalmers.se
> https://lists.chalmers.se/mailman/listinfo/subexp-daq
>
>
More information about the subexp-daq
mailing list