[Csnd] UDP server crash

The Csound Mailinglist
1 
I get a very consistent crash when sending code via UDP with the following .csd. 
If I change the instr number from 8800 to 8799 or lower, there is no crash.
Any clue about the reason?

# udp.csd
<CsoundSynthesizer>
<CsOptions>
-odac
--port=9100

</CsOptions>

<CsInstruments>
sr     = 44100
ksmps  = 64
nchnls = 2
0dbfs  = 1

instr 8800
    turnoff
endin

</CsInstruments>

<CsScore>
</CsScore>
</CsoundSynthesizer>

$ csound udp.csd

# baz.inc
instr baz
    println "baz"
    turnoff
endin

$ cat baz.inc | nc -u localhost 9100

Csound mailing list Csound@listserv.heanet.ie https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to https://github.com/csound/csound/issues Discussions of bugs and features can be posted here

2

Nothing comes to mind immediately, but have you tried using a number instrument instrument instead of “baz” and see if it makes any difference?

Prof. Victor Lazzarini
Maynooth University
Ireland

3

Yes, with numbered instruments there is no crash

Csound mailing list Send bugs reports to Discussions of bugs and features can be posted here

4

I tested here on Mac (10.14.6, Intel) and did not get any crash. I
had just pulled the latest from develop and rebuilt csound prior to
testing.

What version of Csound are you using?

5

I tested on linux, --Csound version 6.15 (double samples) Mar 27 2021

6

Have you run under valgrind? Might give a clue as where to look.

Get TypeApp for Android

7

On Ubuntu Studio, I also see this problem with:

–Csound version 6.14 (double samples) Apr 25 2020

[commit: fb2e61cf2bea6caf5117c30c120430b0fa4ee56c]

Csound mailing list Send bugs reports to Discussions of bugs and features can be posted here

8

I rebuilt on Linux Mint with latest from develop and can reproduce the
issue. I rebuilt a debug build and with gdb has this:

writing 512 sample blks of 64-bit floats to dac
SECTION 1:
    instr baz uses instrument number 1

Thread 2 "csound" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7ffff599d700 (LWP 9311)]
0x00007ffff7f132a8 in named_instr_assign_numbers (csound=0x55555555b2a0,
    engineState=0x7ffff0031820)
    at /home/steven/work/csound/Engine/csound_orc_compile.c:1190
1190 while ((!engineState->instrtxtp[num] ||
(gdb) bt
#0 0x00007ffff7f132a8 in named_instr_assign_numbers
    (csound=0x55555555b2a0, engineState=0x7ffff0031820)
    at /home/steven/work/csound/Engine/csound_orc_compile.c:1190
#1 0x00007ffff7f149ed in csoundCompileTreeInternal
    (csound=0x55555555b2a0, root=0x7ffff00317c0, async=1)
    at /home/steven/work/csound/Engine/csound_orc_compile.c:1693
#2 0x00007ffff7f155fd in csoundCompileOrcInternal
    (csound=0x55555555b2a0, str=0x7ffff509c020 "instr baz\n println
\"baz\"\n turnoff\nendin\n \n", async=1)
    at /home/steven/work/csound/Engine/csound_orc_compile.c:1934
#3 0x00007ffff7e0f234 in csoundCompileOrcAsync
    (csound=0x55555555b2a0, str=0x7ffff509c020 "instr baz\n println
\"baz\"\n turnoff\nendin\n \n") at
/home/steven/work/csound/Top/threadsafe.c:489
#4 0x00007ffff7e103dc in udp_recv (pdata=0x555555797180)
    at /home/steven/work/csound/Top/server.c:213
#5 0x00007ffff7bc9609 in start_thread (arg=<optimized out>)
    at pthread_create.c:477
#6 0x00007ffff7af0293 in clone ()
    at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95

Not sure why that would be a problem on Linux but didn't happen on
Mac. I did double check on Mac and realized that my first test was
off (nc required 127.0.0.1 instead of localhost) but the results was
the same with no crash.

9

This does not seem related to UDPr. I can reproduce a similar crash using ctcsound, so it might be some kind of run condition when assigning numbers to named instruments.

import ctcsound
orc = """
sr = 44100
ksmps = 64
0dbfs = 1

instr 8700
    turnoff
endin

"""

csound = ctcsound.Csound()
options = ["-odac"]
for opt in options:
    csound.setOption(opt)
csound.compileOrc(orc)
csound.start()
pt = ctcsound.CsoundPerformanceThread(csound.csound())
pt.play()

for i in range(10):
    csound.compileOrc(fr'''
    instr foo{i}
        turnoff
    endin
    ''')

Csound mailing list Send bugs reports to Discussions of bugs and features can be posted here

10

That’s what I suspected.