-------- Original-Nachricht -------- > Datum: Tue, 12 Feb 2008 05:09:45 -0800 > Von: Guy Shaw <Guy.Shaw at Sun.COM> > An: powerpc-discuss at opensolaris.org > Betreff: [powerpc-discuss] sspsh is alive
> > A mutant version of Rich Teer's mini-sample shell > is up and running. > > The terminal is not setup correctly. > I have to enter, for example, > > /sbin/hello-write^J > ^^ > > and the characters do not echo. > > But, it works. It runs the given program > and comes back and stays alive. > > It is exercising a good bit of libc. > I ran into another unimplemented libc function, > memchr(). The other processors have highly > optimized assembly language versions of memchr(). > So, long ago, memchr.s for PowerPC was just left > as a placeholder, which called libc_unimplemented(). > I didn't want to switch to memchr.c, and fiddle > with the Makefile, so I just implemented memchr() > in assembly language. It is very straight-forward, > and leaves performance on the table, but I would > not care about that for quite some time. > > I had to make up my own init() function. > Init does the following: > calls libc_init() > opens fds 0, 1, and 2 to the console > sets 'environ' > umask(022) > > Some of that properly belongs in the real /sbin/init, > not at the start of every program. > So, I might make up a mini-init to do this > sort of setup, then launch a shell. > > The rest belongs in one of those C run-time (crt) > files. > > -- Guy Shaw Guy, you rock!! Thanks for the frequent and detailed updates. This allows others to reproduce your scenarios (even after a substantial amount of time of doing nothing related to Polaris). "Merci" on behalf of the community :-) -- Martin
