Date : Sun, 21 Aug 2011 17:52:24
From : rs423@... (Mick)
Subject: [BeebSoc] Re: Amstrad emailer / DataCentre
Where was I?? Oh yes.....
J.G.Harston wrote:
>
> All versions of HADFS from version 1 onwards can access extra
> devices using calls to external support, such as the IDE and GoMMC
> drivers in the Support ROM at mdfs.net/hadfs. HADFS version 5.30
> onwards have IDE and GoMMC drivers built in. I strongly recommend
> not using HADFS versions earlier than 5.50 as they only store 5-bit
> datestamps, so run out of dates after 2012 (1981+2^5).
>
The Olympic bug strikes <eek!>. The version of HADFS I got from you
years ago didn't do hard discs at all which is the reason I didn't use
it. Now it does, and it is the fastest, I shall be using 5.5 as my
number 1 filing system. Reading is fast, and that counts for a lot when
searching through 1700 music tracks from the past.
>
>> version I had / have didn't. Oh, and is the maximum size of a HADFS
>> drive 16MB? It was the default size using HDinit (512MB for ADFS) so I
>> presume (perhaps wrongly) that it is. This is more than adequate
>>
>
> HADFS version 5 uses 16-bit disk sector numbers (ie &0000 to &FFFF)
> so can only access 16M of disk space (&000000 to &FFFFFF). So,
> HADFS 5 accesses four 16M drives on each IDE device. HDInit allows
> you to reserve the first 64M of a 512M ADFS drive image to hold
> these four HADFS drives so you can have HADFS and ADFS partitions
> coexisting on the same device.
>
Now you've got me confused. According to my free space map for drives 0
and 1, adding free and used spare up equals 536870912 bytes. If I'm not
mistaken (mathematically I frequently am) 1024 bytes equals 1k yes?
1024 x 1024 = 1 Mb (1048576) x 512 =536870912. So I don't understand. Is
the image of one of the drives actually larger than what it reports? If
so, is it all added to one of the virtual drives or both?? The reason I
ask is should I format drive 0 again using HD init, would it also
destroy the HADFS partition???
> HADFS version 6 uses 24-bit disk sector numbers (ie &000000 to
> &FFFFFF) and so can access 4G of disk space (&00000000 to &FFFFFFFF)
>
Oooooo <greedy smile>. That's great!
> but I haven't finished carefully integrating the code into HADFS.
> Having an extra 8 bits of disk address means changing the layout of
> the Free Space Map, Directory entries and open file buffers.
>
A complete re-build by the sound of it.
> I will update HDInit so that with HADFS 6 you will have the
> opposite option of reserving a 512M ADFS partition at the start of
> a 4G HADFS partition.
>
I like this idea. To have the choice of HADFS as host rather than
parasite has /got/ to be an option.
> (Even though the ADFS filesystem structure uses 24-bit sector
> numbers itself, the disk access API places the drive number in the
> top 3 bits, so ADFS is limited to 21 bits to specify the sector
> number on a particular drive, resulting in the 512M drive size
> limit.)
>
Oh. I did wonder at the size limit seeing as 256 byte sectors times 24
bit (16,677,216) would make 4 Gig wouldn't it? I'd say "Why not use
32-bit" but Rick may well then have a point about everything ever ever
written for a beeb ;-)
>
>> though. If page remains at &1900 or below, then that would be superb.
>>
>
> HADFS claims a single page of workspace, so with only DFS and HADFS
> claiming memory PAGE is as &1A00. If you want to lose floppy disk
> access you could remove DFS and PAGE will drop to &1800.
>
1 page is modest compared with ADFS. Why &1800? Do you mean &0F00?
Before I added DFS, page was at &0E00.With DFS, it rose to &1900.
Cheers,
Mick.