Alexander Bluhm <[email protected]> wrote: > On Thu, Jan 16, 2020 at 02:35:25PM +0100, Mark Kettenis wrote: > > As for your suggestion. I think it is a bit of a mess. I wasn't > > entirely sure about the consequences of such a change for ports. > > However, "Stevens" has this as "char ipopt_list[]" which means that it > > was signed on some platforms and unsigned on others. Interestingly > > enough it has been int8_t since 1.1 in out tree... > > The only non-kernel user in our tree is lib/libc/rpc/svc_tcp.c. > There the value is always casted to u_char. The defines for this > field in netinet/ip.h are numbers also above 127. So char looks > like a historic mistake to we. We should fix it.
This signed-type is not unique to openbsd, this is a very old definition. Have other systems changed it to an unsigned type, and avoided any unknown fallout? If we find one system which changed it, I think we're good. > Who expects fallout with IP options? We block them in pf per > default. don't make this about whether ip options are received, the argument is spurious since it's not like we can delete this structure ...
