[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: <sys/socket.h> mess.
From: |
Roland McGrath |
Subject: |
Re: <sys/socket.h> mess. |
Date: |
Sun, 10 Jun 2001 17:00:57 -0400 (EDT) |
Btw, I just gave up and added your wins.uva.nl address to the posters list.
> It basically matches 4.4 BSD. The only real difference is the
> definition of SOMAXCONN (which was defined to 5 on 4.4 BSD). There
> are some difference in the names of the types of structure members
> (u_int vs. socklen_t, caddr_t vs. void *) and we have some additional
> constants (AF_INET6, PF_INET6) and CMSG_* access macros (these should
> be compatible with the 4.4 BSD `struct cmsghdr').
>
> `struct msghdr' and `struct cmsghdr' are identical to what's used on
> Linux except that cmsg_len is of type socklen_t instead of size_t
> (should be socklen_t according to the latest POSIX draft).
>
> So if it wasn't for SOMAXCONN, we could put it in
> sysdeps/unix/bsd/bsd4.4/bits.
I call that more than close enough that we should stick it there.
FreeBSD's <sys/socket.h> defines SOMAXCONN to 128. If there comes a time
when the Hurd socket bits/layouts diverge from BSD, then I can't see a
reason not to punt the BSD bits entirely, move the linux file into sysdeps/gnu,
and use that.