Timeout ill-behaved terminals to avoid tasks hanging CICS

CEGN hangs forever after RACF terminal timeout

For the last 8 1/2 years, customer has been running with a USERMOD (V4.1 version ANCEGN4) to handle a situation where badly-behaved customer terminals (over which customer has no contol) cause a task to hang forever, taking up a task slot and eventually causing MAXTASKs and CICS recycle. Details are in the referenced PMR and the original one (40246.7TD.000). Current frequency of the problem for customer is about 50 tasks a day, causing a recycle every 3 days or so (MAXTASKS set to 150).

The hang is caused by the following processing:
1. A terminal times out (RACF TIMEOUT set to 15 minutes) and CEGN is started to handle this
2. CEGN does a RECEIVE to save the TERMINAL BUFFER
3. The RECEIVE times out and an ABEND AZCT is issued
4. During abend processing a RESETSR is issued to cancel the outstanding RECEIVE
5. The TERMINAL is still in SEND mode, and CICS needs to SEND an EB. It therefore SENDs a SIGNAL command which is an expedited flow requesting the TERMINAL to send CD so that CICS can send the EB
6. This SEND is never posted complete by VTAM and so CEGN hangs

Instead of having a USERMOD ported to each release, this should be fixed in CICS. There are several ways to implement this, but the 2 main ways seem to be:
1. Have a TIMEOUT on the SEND SIGNAL that causes it to be cancelled and allows CEGN to end (would assume if no response to this then the other end is completely dead)
2. Have a new extra parameter on the GNTRAN SIT (e.g. ,NONE) and use this to control whether the USERMOD code (which would now be put into CICS proper - the USERMOD just skips the CEGN RECEIVE in step 2 above by pretending it isn't a 3270, this already has access to the SIT) is used, or whether the existing code runs

Method 1 would solve several customer issues, not just this one.