In IPF 4.x, what is the performance gain in compiling your rulset into an LKM?
From Darren: “Very significant. I did do some benchmarking of this, originally but I forget what the performance measurements were, now. “The improvements in performance come from two areas. The first is that the packet matching is now all in C, rather than using intermediate structures. The second is that rather than compare each field, one at a time, in each rule, it sorts the fields to be matched for each rule as an optimisation and only does comparisons when the matching is different.” • In IPF 4.x, I’m trying to build with gcc3 but it’s not working! If you get an error like this: gcc -I. -g -I../.. -D_BSD_SOURCE -DSOLARIS2=9 -c ../../lib/debug.c -o debug.o ../../lib/debug.c: In function `debug’: ../../lib/debug.c:30: error: `__builtin_va_alist’ undeclared (first use in this function) ../../lib/debug.c:30: error: (Each undeclared identifier is reported only once ../../lib/debug.c:30: error: for each function it appears in.) *** Error code 1 make: Fatal error: Command failed for tar
