Re: Problem with dbCaLink task in 3.14.4

[Marty Kraimer <mrk@aps.anl.gov>]

Mark Rivers wrote:
> Folks,
>
> I am getting the following error when processing an EPICS "seq" record with CA links:
>
> dbCaLink: A call to "assert ((epicsMutexLock((pca->lock))==epicsMutexLockOK))" failed in ..../dbCa.c at 757
> Current time THU OCT 30 2003 18:48:42.968775352.
> EPICS Release EPICS R3.14.4 $R3-14-4$ $2003/09/23 21:35:11$.
> Please E-mail this message and the output from "tt (0x1558050)"
> to the author or to tech-talk@aps.anl.gov
> filename="../../../src/libCom/taskwd/taskwd.c" line number=170
> task 0x1558050 suspended
>
> Here is the output of "tt":
> ioc13lab> tt (0x1558050)
> 19e530 vxTaskEntry    +60 : 1a70bcc ()
> 1a70c1c epicsThreadPrivateGet+cc : dbCaTask ()
> 19ff9e0 dbCaTask       +2d4: epicsAssert ()
> 1a6f9dc epicsAssert    +f0 : epicsThreadSuspendSelf ()
> 1a70644 epicsThreadSuspendSelf+14 : taskSuspend ()
> value = 0 = 0x0
>
> The problem happens when the following record is processed:
> grecord(seq,"$(P)AbortScansSEQ") {
>         field(LNK1,"$(P)scanH.EXSC  CA MS")
>         field(LNK2,"$(P)scan1.EXSC  CA MS")
>         field(LNK3,"$(P)scan2.EXSC  CA MS")
>         field(LNK4,"$(P)scan3.EXSC  CA MS")
>         field(LNK5,"$(P)scan4.EXSC  CA MS")
> }
>
> This record is just writing 0 to the "EXSC" field of 5 "sscan" records using CA links.  The problem does not appear to be in the sscan record itself, because I can write to the .EXSC fields using CA from a client, or using "dbpf" from the vxWorks prompt, and it works fine.
>
> Changing this record to:
> grecord(seq,"$(P)AbortScansSEQ") {
>         field(LNK1,"$(P)scanH.EXSC  PP MS")
>         field(LNK2,"$(P)scan1.EXSC  PP MS")
>         field(LNK3,"$(P)scan2.EXSC  PP MS")
>         field(LNK4,"$(P)scan3.EXSC  PP MS")
>         field(LNK5,"$(P)scan4.EXSC  PP MS")
> }
> i.e. using database links rather than CA links "fixes" the problem, and there are no errors.
>
> I've attempted to reproduce the problem in a simple example application and have not been able to do so yet.
>
> Any ideas what the problem could be?
>
> Thanks,
> Mark

Nothing obvious. The failure occurs on the statement

epicsMutexMustLock(pca->lock);

Other than memory corruption it is hard to explain why this fails.

Does this happen the very first time $(P)AbortScansSEQ is processed?

Before it is processed issue the instruction

dbcar "$(P)AbortScansSEQ",3

Does it show everything connected?

Marty