[LispM-Hackers] Loaded ersatz ucode and N928.LOAD on x86

[Dave Richards]

Which begs a question: Why is N928.LOAD so large?  I've read comments in the
code about always being able to count on 2MB.  Is this the normal load size,
or is this a load with all the bells and whistles?  It may not *seem* like a
lot, but remember, it's 1/4 the maximum virtual address space of the
processor.  If this is normal, TI surely didn't give themselves much room
for scalability.

	Dave

> -----Original Message-----
> From: lispm-hackers-admin@lists.unlambda.com
> [mailto:lispm-hackers-admin@lists.unlambda.com]On Behalf Of John
> Morrison
> Sent: Wednesday, March 06, 2002 8:28 PM
> To: lispm-hackers@lists.unlambda.com
> Cc: jm@mak.com
> Subject: [LispM-Hackers] Loaded ersatz ucode and N928.LOAD on x86
>
>
> Hi All;
>
> Just a quick note to let you know I hacked up the NBI file utility
> ("mke3nbi" for "make e3 NBI file") so that it read the asm/C++ binary,
> and tacked on the N928.LOAD load band (all 33MB of it).  I then
> successfully loaded the NBI file over the Ethernet, and they both
> were loaded to different spots in PC memory.
>
> For the terminally curious, the ucode is loaded at 0x11000 (underneath
> the frame buffers in real-mode-accessible RAM so we can jump to it in
> real mode and then enable protected mode.  The load band I just stuck
> starting at 0x100000/1MB.  I can and probably should put it in high
> memory (there is a way to tell Etherboot to stick the bits at an
> offset with respect to the end of physical memory, whatever that
> happens to be at runtime).
>
> \begin{excited-babbling}
>
> Do you realize how convenient we could make installing/running e3, in
> stark contrast to installing Windows, Linux, or anything else?
>
> People would be able to:
>
> (1) go to www.rom-o-matic.com, and burn an Etherboot floppy in about
> 30 seconds,
>
> (2) go to unlambda.com (or wherever) and download a 10-11MB gz file of
> the NBI-format ucode+loadband (that's how small our 33MB NBI file
> compressed to)
>
> (3) simply stick the uncompressed NBI in /tftpboot on a willing
> TFTP/DHCP server
>
> (4) Put in the floppy and hit the reset switch.
>
> No partitioning.  No formatting.  No dual-boot nastiness (I should
> know -- I just ran into all sorts of misery trying to get GRUB to
> understand my 2-disk system).
>
> FYI, it only took about EIGHT SECONDS to load the NBI image over the
> 100Mbit/sec LAN.  It took much longer than that for my ancient,
> wheezing 200MHz machine to pass its BIOS POST checks.
>
> \end{excited-babbling}
>
> Seriously, though, this could be pretty cool.  I have to think there
> is real potential for educational institutions to retask those
> roomfulls of hard-to-manage diskful Windows and Linux PCs to being
> diskless LispMs.  It would be so incredibly easy if we did even a
> halfway-decent job.  (FYI, it is a pet peeve of mine that Lisp is not
> more widely taught in CS courses, and that therefore these graduates
> aren't as clued-in to what the Real Big Programming Issues are.)
>
> -jm (giving up for the night)
>
> --
> ==== John Morrison
> ==== MAK Technologies Inc.
> ==== 185 Alewife Brook Parkway, Cambridge, MA 02138
> ==== http://www.mak.com/
> ==== vox:617-876-8085 x115
> ==== fax:617-876-9208
> ==== jm@mak.com
>
> http://lists.unlambda.com/mailman/listinfo/lispm-hackers