Hi,
> Ralf's fixes to the dynamic linker fixed the Xt applications. Now,
> only the PAM remains to be fixed.
The PAM problem is also a dynamic linker problem. It's (I hope) one the
last major showstopper class bugs that prevent just building a
Linux/MIPS distribution from 95% unchanged source packages - and
fullfilling all the promises to those loading hardware to me ...
Btw, I stopped the mega load test (Running many compilers and tools in
parallel) I posted recently after some more hours because the machine ran
out of swap, some processes died therefore and load < 100 isn't that
interesting ...
System call latencies were pretty lousy bad for MIPS kernel because
we were using a syscall handler written in C. I rewrote the entire
thing in assembler and now we're as fast as a MMX Pentium accounting
for the different clock rates. Pretty nice, because we actually have
to save/restore many more registers and deal with a more complex
system call convention. Seven nops left to eleminate ...
> So, where do we stand now? So, the only bits missing now are:
>
> 1. me finishing the support for the X server. just a hack
> here and there.
>
> 1.b. Getting the other important bits of the RRM code in the
> kernel.
>
> 2. me fixing the mouse.
>
> 3. Would it be possible to negotiate with SGI management
> the posibility of shipping the IRIX runtime libraries and
> the X server as found on IRIX with a Linux disrtibution?
Bad point in time, it has been pointed out to me that SGI's stock
was doing pretty bad ;-)
> 4. A nice, easy-to use install program. Taking the existing
> Red Hat/Mustand install program and port it should be
> pretty easy.
Ralf
|