A user here was reading in ~400MB month-long data files
into sac 101.3b, and ran out of swap space.
I looked at the behavior on both rhel4 linux and solaris8,
and it is similar. "Read" opens the input file, reads the header,
closes the file, opens it again and re-reads the header,
mallocs space for the trace, and then reads that. I don't
see any place where it free's the malloc'ed memory, and each
malloc result address is different. On subsequent "read file"'s
it never seems to re-use the buffer it allocated, and keeps
on grabbing more virtual memory, eventually filling swap.
I've tried to ask for sac source, but I'm not sure my request
is actually being submitted to IRIS.
Anybody seen this behavior and have a fix/workaround?