[Top] [All Lists]

RE: linux-2.14.2.dec is obsolete!

To: "''" <>
Subject: RE: linux-2.14.2.dec is obsolete!
From: Rob Mohr <>
Date: Tue, 10 Mar 1998 14:21:23 -0800

I have a DECstation 3100 with 24MB RAM, color framebuffer, 19 inch color
monitor, and a TK50 tape unit.  The system is networked to my Pentium
133 system running Windows95.  I am willing to help out with this
porting effort however I can.  My system is currently set up to boot
either Ultrix 4.4 or NetBSD 1.3.  I have a small (170MB) harddisk that I
can use to test Linux kernals and miniroots on.  I do not have any
experience hacking kernals, but with a little direction I can modify
source and recompile tests as needed.  Let me know what I might be able
to do to help out.

Rob Mohr (work) (home)

(P.S.  Please send replies to both addresses listed above.)

> -----Original Message-----
> From:
> []
> Sent: Tuesday, March 10, 1998 11:03 AM
> To:
> Subject:      linux-2.14.2.dec is obsolete!
> Fellow DECstation Hackers,
> I've got good and bad news for you. First the good one:
> The DECstation specific stuff is reintegrated into the latest
> Linux/MIPS source
> tree which is 2.1.73! This kernel is tested on a kmin and a 3max+ and
> may
> work on maxines (I hope).
> Now the bad one [Hmmm... where's that asbesto suit...]:
> This Kernel will no longer work on pmax-alike DECstations, which are:
> DS2100, DS3100, DS5100 and DS5000/200! But relax, a few minor changes
> and
> you'll be back :-). I haven't had a chance to make these myself just
> because I
> don't have the hardware.
> Improvements over
> Well, to be honest, not very much. 2.1.73 dies exactly where the
> latest 2.1.14
> has died before: "Kernel panic: VFS: Unable to mount root fs on
> 00:00", but
> what would you do without device drivers?. On the other hand, it dies
> in a more
> cleaningly manner :-).
> Changes since
> o Building process: The kernel itself and the "bootloader" code are
> clearly
> seperated from each other. The bootloader does no longer use any
> kernel
> functionality and the kernel does no longer use any prom_functions
> (bye bye,
> prom_printf). The kernel is compiled into ELF format and then
> converted into an
>  a.out image just like for the other MIPS machines. This kernel image
> is linked
> together with the bootloader into an ECOFF image, which can be booted
> via the
> DECstation boot PROM. I have added the option "make netboot" to the
> Makefile to
> do this. There is no need to remove sections any longer.
> o Startup Procedure: This netbootable Image
> (arch/mips/dec/boot/nbImage) will
> be loaded at address 0x80200000, which leaves enough room to work with
> compressed kernel images. This might be necessary in the future for
> certain
> PROM versions. The kernel itself is then copied down to 0x80000000 and
> started.
> This, together with the now no longer needed bootloader, saves nearly
> 200 KB
> memory. PROM functions, on the other hand, will no longer work.
> o Primitive serial console for DECstations with scc chip.
> o Proper kernel initialization: No more "skipping console-init" or
> "skipping
> device_init".
> o Memory sizing routine for pmax-alike Machines (untested!).
> o Code cleanup and better integration into the main source tree.
> Development environment notes:
> This kernel will not compile anymore with the 2.7.*-binutils. You _do_
> need
> 2.8.1-binutils with support for the a.out-mips-little target. Due to
> an
> unresolved linker bug you have to remove the -N option from the final
> kernel
> link (in arch/mips/Makefile).
> Do _not_ use "make menuconfig", it insists on enabling CONFIG_CDROM
> which results in
> an unresolved symbol cdrom_init(). Use "make config" instead.
> You would not see anything unless you enable CONFIG_SERIAL and
> Todo:
> I cannot do this myself, any feedback is appreciated, patches are
> welcome:
> o Test if the startup procedure works on pmax machines.
> o Test and debug the pmax_get_memory routine.
> o Serial console support for DECstations with the DC7085 chip, or
> raster
> console if you like.
> o Test if serial console works on maxines. The serial console is
> hardcoded for
> maxines, regardless of the settings for "console" and "osconsole". If
> you do not see
> anything try playing around with tinycon.c
> o Test if this kernel works so far on R4000 based DECstations.
> Things I am working on:
> o Passing command line parameters to the kernel.
> o Improving serial console for scc chips.
> o Ramdisk support.
> o Better interrupt handling code integration.
> o Turbochannel support.
> For us all:
> o Writing device drivers.
> o Writing device drivers.
> o And, of course, writing device drivers.
> I will be uploading my source tree this evening and it should soon
> appear at:
> Stay tuned,
> Harald

<Prev in Thread] Current Thread [Next in Thread>