server does not properly cleanup when failing to initialize data structures for new circuit because of a low memory situation

[Jeff Hill]

It has been discovered that the CA server does not cleanup properly if there is insufficent memory during creation of data structures for a new CA circuit.

Additional information:
Many thanks to Pam Gurd for her comments below helping to debug the situation.

I think the bottom line is that there’s too much stuff trying to run on this IOC. Still shouldn’t suspend, but I think we (I?) have to figure out how to lighten the load.

    Pam.

dtl-hprf-ioc3> 0x1208670 (CAS-TCP): memPartAlloc: block too big - 29376 in partition 0x1e390c.

VME Bus Error accessing A24: 0x006406

data access

Exception current instruction address: 0x00191324

Machine Status Register: 0x0000b030

Data Access Register: 0xeeeeeef2

Condition Register: 0x22000042

Data storage interrupt Register: 0x0000b030

Task: 0x1208670 "CAS-TCP"

CAS: unable to init the event facility

filename="../../../src/libCom/taskwd/taskwd.c" line number=170

task 0x1208670 suspended

CAS-TCP 19cfec4 1208670 181 SUSPEND 191324 1208420 110001 0

dtl-hprf-ioc3> tt 0x1208670

19e528 vxTaskEntry +60 : 19cfec4 ()

19cff14 epicsThreadPrivateGet+cc : 1a6badc ()

1a6bde4 rsrvGetUInt32 +3b8: create_tcp_client ()

1a6d14c create_tcp_client+174: destroy_tcp_client ()

1a6c9e8 destroy_tcp_client+60 : db_add_extra_labor_event ()

1a4e2f0 db_add_extra_labor_event+2c : epicsMutexLock ()

19c976c epicsMutexLock +18 : semTake ()

value = 0 = 0x0

Original Mantis Bug: mantis-100
    http://www.aps.anl.gov/epics/mantis/view_bug_page.php?f_id=100

[Jeff Hill (johill-lanl)]

fixed in R3.14.7

[Andrew Johnson (anj)]

R3.14.7 Released