[Rpcemu] A pleasant blue window...

Ralph Corderoy ralph at inputplus.co.uk
Sat Mar 20 17:28:24 PDT 2010


Hi,

Replying to several emails at once...

Musus Umbra wrote:
> My RO3.70 ROM image didn't pass the checksum; I get:
> 
> $ sha1sum roms/riscos-3.70.rom
> e7cb05c9925dd0d7d9d5ff6f74244066c5c87e34  roms/riscos-3.70.rom

That's not one I recognise out of my collection of duff 3.70s.  :-)

> And the reason it gets that is that the cpu_type is wrong in the
> rpc.cfg file.  And the reason the CPU type is wrong is because the
> bash script never got to change it because the ROM checksum failed and
> aborted the process.

Which is deliberate.  We wanted to know the first point an error
occurred, hence me putting a check for the SHA1 digest of the ROM image.

> > The rom image must be okay as it gets to the desktop in another
> > emulation.
> 
> But interestingly, that same ROM image works perfectly in an older 32-
> bit build of RPCemu I was previously using...  so that's apparently
> not any sort of indicator of correctness :p

It could be there's multiple releases of 3.70 made and some didn't cope
with certain CPUs?

> Try Ralph's script, and then chamging the cpu_type to SA110
> manually...  that worked for me, even with a ROM image that has a
> different hash.

No, don't!  The point of the script was that we'd know what was done it
if worked or failed.  No human intervention involved.  :-)

David wrote:
> > Just to be clear, because you've separated the "got it" from the
> > "ran it", you ensured /tmp/riscos-3.70.rom existed and then entered
> 
> Ah! You mean I should have created a directory called
> "riscos-3.70.rom" inside a directory called "tmp" in my home
> directory?

No, your home directory is nothing to do with this.  Before you go any
further, you need to learn about pathnames on Unix/Linux.  /tmp is
rooted at the top directory, /.  It's a directory that's normally
present for storing temporary things in.  I want you to put your 3.70
ROM image file in that existing directory and call it riscos-3.70.rom.

> No, I think you mean get a rom image, call it "riscos-3.70.rom" and
> put it in a directory called "tmp" in my home directory.

No, that neither.  I never mention your home directory.

> Sorry that wasn't clear in your last post. I searched for it on my
> computer after I'd done the "wget etc." thinking that your bit of
> magic would put it there.

No, my script doesn't supply a ROM image.

> Anyway, I've already done the sha1sum for my rom image for Adny and
> it's a1218ab6bd2f965d317522be68112d9ae3742fa0 which doesn't match
> yours. Is it worth going through the process properly but with the
> wrong sha1sum rom image?

No, because my script would abort when it detected your ROM image's SHA1
digest didn't match what I expected.  It would never even attempt to run
rpcemu.  You'd then tell us and we'd say "A-ha!".

> Didn't understand that bit about "/tmp/riscos-3.70.rom" - I've put a
> rom image from my RPC into the roms directory. did I do wrong again?

Yes, as you now know.  :-)  The idea of the `wget ... | bash' was that
it would download and *run* rpcemu.  You'd then report what happens.
There would be no opportunity for you to do anything to try and fix it,
work around it, etc., that we'd not be privy too.  You simply could not
put your image into the roms directory because you wouldn't have the
chance until the script either failed, or had already run rpcemu.

In short, you were meant to be playing the part of our hands and eyes,
and not thinking for yourself.  :-)

> I get a1218ab6bd2f965d317522be68112d9ae3742fa0 which is different yet
> again.

Yet another I've not come across.

> And it came up with "sha1sum: WARNING: 1 of 1 computed checksum did
> NOT match"

Good.  That bit's as designed, at least.

Musus Umbra wrote:
> The question then becomes "Is the ROM image corrupt?"

If folks want to send me their 3.70 images that don't have a SHA1 digest
of 42c3dd0ae43149849b674ea025cb2206512a5b4a then I'm happy to have a
look and see if I can see the scale of the differences between them.

Cheers,
Ralph.




More information about the Rpcemu mailing list