Experimental Physics and
Industrial Control System

Dirk Zimoch <[email protected]> · Wed, 14 Feb 2018 15:36:39 +0100

Hi everyone,

I got a crash in a 3.14.12.4 CA client after it exhausts all file

descriptors.

The last thing it prints is:

CAC: exception during virtual circuit creation "CAC: TCP circuit

creation failure because "Too many open files""

Then it crashes.

The client is an overview panel for all our IOCs. Thus the excessive use

of file descriptors is understandable. Also the error message makes sense.

The real problem is the crash afterwards. Here is the stack trace:

#0 0x00007f5e9943e0e6 in assertIdenticalMutex (this=0x0, guard=...,

chan=..., sidIn=4294967295, typeIn=65535, countIn=0)

    at /usr/local/epics/base-3.14.12/include/epicsGuard.h:84

#1 tcpiiu::installChannel (this=0x0, guard=..., chan=...,

sidIn=4294967295, typeIn=65535, countIn=0) at ../tcpiiu.cpp:1936

#2 0x00007f5e9942c3fb in cac::transferChanToVirtCircuit (this=<value

optimized out>, cid=<value optimized out>, sid=4294967295,

 typeCode=65535, count=0, minorVersionNumber=13, addr=...,

currentTime=...) at ../cac.cpp:631

#3 0x00007f5e9943add8 in udpiiu::searchRespAction (this=<value

optimized out>, msg=<value optimized out>, addr=<value optimized out>,

    currentTime=<value optimized out>) at ../udpiiu.cpp:690

#4 0x00007f5e9943aeea in udpiiu::postMsg (this=0x1d120d0, net_addr=...,

pInBuf=<value optimized out>, blockSize=24, currentTime=...)

    at ../udpiiu.cpp:857

#5 0x00007f5e9943ce69 in udpRecvThread::run (this=0x1d224f8) at

../udpiiu.cpp:394

#6 0x00007f5ea2066d21 in epicsThreadCallEntryPoint (pPvt=0x1d22518) at

../../../src/libCom/osi/epicsThread.cpp:85

#7 0x00007f5ea206c672 in start_routine (arg=0x1d014e0) at

../../../src/libCom/osi/os/Linux/osdThread.c:395

#8  0x0000003465c07aa1 in start_thread () from /lib64/libpthread.so.0
#9  0x00000034650e8bcd in clone () from /lib64/libc.so.6

As you can see the crash is fully in EPICS base. No user code involved.

I find the "this=0x0" arguments highly suspicious.

Is this a known problem? Has it already been fixed in later EPICS base

versions? (If not I may try to hunt down the problem but I do not want

to waste my time...)

Dirk

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System