On May 14, 2013, at 8:21 AM, Gisle Vanem <gva...@broadpark.no> wrote:
> I really don't understand the motivation behind the Win32-code > for pcap_compile(). I assume the motivation is that pcap_compile() is not thread-safe. It's no more thread-safe on UN*X, so if the goal is to allow programs to use it without having to lock out other threads using it, we should perhaps use a pthread mutex if they're available... ...except if you're using Grand Central Dispatch rather than pthreads, in which case we should do what's appropriate for GCD. Then again, sufficiently recent versions of Flex can generate thread-safe lexical analyzers; if we can also generate a thread-safe parser (e.g., with Bison or some more modern flavor of YACC), we could probably incorporate the remaining global variables into a structure and make it thread-safe *if* the right tools are used to generate the parser and lexer. > Why doesn't other libpcap functions needs this critical-section protection > too? Because they're thread-safe. _______________________________________________ tcpdump-workers mailing list tcpdump-workers@lists.tcpdump.org https://lists.sandelman.ca/mailman/listinfo/tcpdump-workers
