Does anyone have code/advice about allocating mbufs and
clusters (like in Berkeley's uipc_mbuf.c) out of some fixed chunk of
memory, rather than out of the Unix virtual memory mechanics ?
I am porting the Berkeley 4.3 Internet and sockets code
into a small and simple microprocessor-based box. The box has no
kernel (i.e. I am writing/porting code which talks directly to
hardware, with no routines below me) and has some small amount of RAM.
Running the Berkeley networking requires me to allocate mbufs and
clusters out of this RAM. The Berkeley mbuf allocating code
(uipc_mbuf.c) allocates mbufs out of the rather complicated Unix
virtual memory mechanism. I am a bit confused by this. I modified
this code to allocate mbufs as required by my application, but I
don't trust my code. Did somebody more expert that me do this work
To readers inside DEC
If you followup this message, your
followup will by default be distributed worldwide, not
only inside DEC. This is because I'm sending this both to
a DEC internal and to a worldwide newsgroup.