On Tuesday 30 January 2001  4:38 pm, Reinoud Zandijk wrote:
> Hi hi All,
>
> Good news from the new RiscPC/RC7500/NC/Imago (hopefully) bootloader front
> ..... Its loading debugged kernels now !! ... All was needed was a one line
> patch *snif*.

Excellent, you mention imago, does this mean you're talking to millipede 
about it?  (I was about to attempt to start talking to them about netbsd 
support...)

> Oh well... I also managed to get the amount of Kernel VM from 48Mb to 80Mb
> bij moving the video memory up from 0xf4000000 to 0xf7000000 where nothing
> seemed to have been mapped. There is more memory to gain but i'd rather do
> that when i have a nice overview of the addresses instead of the current
> list.

We've pushed a lot more upwards to get more space.  I'm working on tidying a 
diff up for this.

> The distro is (as was before) obtainable for testing at the /pub directory
> of ftp://130.89.65.13 . This time it's the new updated bootloader, the new
> updated rpc_machdep.c and a patch against ../sys/arch/arm32 that modifies
> the amount of VM and moves the video memory up.

I'll give it a whirl.

> A slight warning : it might be nessisary to build from scratch; i.e.
> delete the compile directory and rerun config .... I had some strange
> problems a few times with compiling after I changed the VM setting...
> maybe the dependencies aren't allways that well :( ... maybe related to
> the .s vs .S ?

Yes the .s v .S this is a pain, there's a pr about this, with a patch (from 
Richard Earnshaw) to fix this, PR 12051 if you want to grab the diff.  It 
does make it easier.

> The only 2 small things next to a command line prompt for parameters like
> the i386 f.e. is the fact that the text on the screen is still red until
> you switch virt consoles and a very wierd one ONLY with kernels with the
> in kernel debugger build in : the text displayed before the kernel
> switches to its pagetables is drawn VERY SLOW... this is nothing to worry
> about but it puzzles me still ! After the switch everything is OK.

How about not displaying anything till the kernel does the page tablesw itch 
(I need to look over your changes more closely to see how they update 
rpc_machdep)

> I'll focus now on making the interface a bit user friendlier and on the
> two small issues mentioned above....
>
> For testers :  I would really appreciate feedback on operating on
> different memory configurations... esp. the Kinetic machines.

Ok will try ... nope it doesn't boot, I get:

I got 629 memory pages to mess with
Getting memory configuration

Found DRAM (0) at 0x10000000 for	16384k
Found DRAM (1) at 0x20400000 for	28672k
Found DRAM (2) at 0x30000000 for	32768k
Found VRAM (0) at 0x02000000 for	1024k

Address not recognised at line 6800

Does it require a patched kernel to actually boot at this point?   I'm using 
a normal kernel (not had time to recompile, and your patches conflict with my 
current working tree)

Cheers,
Chris