Experimental Physics and
Industrial Control System

<[email protected]> · Fri, 7 Nov 2014 15:57:31 +0000

Hi,

I have come across a difference in behaviour with asyn and the EPICS stringin record (asyn 4-22 / base 3.14.12.2 / win64) that occurs when the value received by the driver happens to be larger than the stringin maximum size. A record was changed from "periodic scan" to "I/O Intr" resulting in a crash due to a strcpy() (in monitor() from stringinRecord.c of base) being unable to find a NULL terminator and thus overwriting memory. Looking at devAsynOctet.c  it seems that stringin records that are periodically scanned are always guaranteed to have a NULL terminated value, whereas ones processed via interruptCallbackSi() are NULL terminated only if the data received is less than 40 characters long (I noticed that in base 3.14.12.4 the stringinRecord.c monitor() routine now uses strncpy() and so does not require NULL termination). For my case I've just added NULL termination for "I/O Intr" too, but from looking at the stringin record reference manual page I wasn't sure if the ori!
 ginal I/O Instr stringin behaviour was actually more correct? 

Regards, 

Freddie

-- 
Scanned by iCritical.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System