Hi Kiran,

To use Berkeley DB with Purify (or similar tools such as valgrind), you
will need to rebuild it after configuring with the '--enable-umrw'
flag. This ensures that all bytes in a structure are set, including
padding bytes that make no difference to the operation of Berkeley DB.

Otherwise, tools like Purify will report spurious errors.

If you reconfigure with that flag, rebuild DB and still see this error,
please let us know and we'll look into it further.

Regards,
Michael.

Hi Kiran,

Does your application have problems on AIX when run without Purify?

If so, can you get a stack trace with debug information?

If it runs fine without Purify, can you build with the '--enable-umrw'
flag and check whether that build still crashes when run with Purify?

I don't know of any AIX-specific bugs in Berkeley DB that would explain
this difference in behavior between Linux and AIX, so I'm trying to
work out whether some interaction between Berkeley DB and Purify on AIX
is the problem.

Regards,
Michael.