> Hi there,
> I had some time these days to do some work on the DS2100/3100 port.
> I found 2 scheduling hazards in the macros SAVEALL and RESTOREALL, after
> this fix the memory sizing routine runs well.
> Now I got stuck in kernel_entry, when the bss is cleaned. There are two
> adresses _edata and _end, which are set by the linker as I understand. On
> my machine these are both zero, but they shouldn't be, I think.
> I'm not familiar with the linker internals nor with the ecoff file format,
> so could some, please, make some suggestions to solve the problem!
OK, prompted by this success (!?), I've finally gotten motivated to try
building my own kernel for my DECsystem 5100. As you all might remember,
Paul and I (actually Paul did all the work; I just tested his kernels!)
figured out back in May that this beast is, at least from a PROM standpoint,
pretty much like a 3100, so the calls to write to the 3100 console work on
this thing too. With Paul's last kernel, I get as far as the "Linux version
2.1.14 (paul@slither) ... " banner, and then it crashes.
Is the memory sizing routine the next thing that runs after that banner?
OK, so tonight I built a kernel of my own, using Paul's "linux-184.108.40.206.dec"
source tree. I had to create a couple links (one of which was quite odd,
and I can't figure out where it gets requested!), but after I did that the
"make vmlinux" went fine. I just now remembered the note on Paul's web page
that describes how to put the kernel in the correct format (using objcopy),
and once I did that it loads and gets me to the same point as Paul's kernel.
So where do I go next?
Stu Allen Phone: (716) 231-0073
EDS / GNS Internet: firstname.lastname@example.org