Partition Manager feedback

[JonAbbott]

Well it runs now. Only the one partition shows

I have a vague idea of the partition structure from reverse engineering the Simtec Module, but can't see where its storing it on the disc. It does however have similarities with HCCS, so they might be derivatives.

Could you dump the beginning of the disc please:

Code: Select all

SYS "IDEFS_DiscOp",,1,4<<29,&10000,&1000
*SAVE SimtecPT 10000+1000

[IanJeffray]

Could you dump the beginning of the disc please

Attached.

[IanJeffray]

Could you dump the beginning of the disc please

Attached.

It was. I'm sure. "Funny things" seem to happen on this forum though - are my posts being silently edited?

[JonAbbott]

Could you dump the beginning of the disc please

Attached.

It was. I'm sure. "Funny things" seem to happen on this forum though - are my posts being silently edited?

That would be me, stripping off attachments and inserting the text from the debug logs etc. so I can read them in a browser.

That dump wasn't the physical drive, so I removed it. I had a brain freeze and forgot FileCore can only see the logical drives. I'm currently figuring out how to physically address the drive, which is proving somewhat complicated without any documentation and the SWI I need being a NOP.

Do you have a spare drive we can test with? Ideally I'd want a drive with two partitions and an unpartitioned area, which we can then throw SWI at to see what responses we get.

I've worked out what I think will read the partition table, but it might be writing it - it's not obvious from the code as IDEFS_SetDriveConfig, which you'd think would be writing to the disk, appears to be reading going by the ATA command it's using.

Long story short, the only way to address the physical drive with SimTec appears to be ATA commands via IDEFS_Command, but it's entry register values aren't obvious. It appears to take the ATA command in a register, when I'd normally expect a pointer to an ATA command block and two of the registers make no sense as they both appear to be a transfer size. I spend 5 hours staring at the Module code yesterday to no avail!

[IanJeffray]

Do you have a spare drive we can test with?

Sure. I can put the Simtec card in another machine with another FS to load things from. I've some old spinning rust and some CF cards that I can hang from the Simtec for testing.

[JonAbbott]

Sure. I can put the Simtec card in another machine with another FS to load things from. I've some old spinning rust and some CF cards that I can hang from the Simtec for testing.

Great. I'll work out some tests and get back to you.

[JonAbbott]

Sure. I can put the Simtec card in another machine with another FS to load things from. I've some old spinning rust and some CF cards that I can hang from the Simtec for testing.

Could you attach a drive we can test with please and upload the output from the following:

1. create two identical 50MB partitions
2. format the 1st for FileCore
3. *SPOOL the output from the following commands

Code: Select all

SYS "IDEFS_Enumerate",-1 TO A%
FOR Z%=&10000 TO &12000 STEP 4:!Z%=-1:NEXT
SYS "IDEFS_XferData",A%,&41,0,&10000,&1000,1<<9 TO ,,R2%,R3%,R4%
PRINT "0: ";~R2%,~R3%,~R4%
*SAVE data1 10000+1100

FOR Z%=&10000 TO &12000 STEP 4:!Z%=-1:NEXT
SYS "IDEFS_XferData",A%,&41,0,&10000,&1000,0 TO ,,R2%,R3%,R4%
PRINT "1: ";~R2%,~R3%,~R4%
*SAVE data2 10000+1100

FOR Z%=&10000 TO &12000 STEP 4:!Z%=-1:NEXT
SYS "IDEFS_XferData",A%,&41,1,&10200,&C00,1<<9
*SAVE data3 10000+1100

I'm not certain if addressing is in LBA - data1/data3 should confirm that. I'm also not certain if R5 is required on entry or what it is (I think its the LBA size), which data2 should confirm. Finally, a comparison of R2,R3,R4 on exit should confirm if it updates the input parameters and if the R5 entry parameter affects them.

From looking at the scandisc function in IDETool, I'm guessing the partition table is within the first 6 LBA and verified by "andy" being the 1st or 2nd word in the sector.

Once I'm certain on those parameters, we'll try a write.

[IanJeffray]

format the 1st for FileCore

I'm not entirely sure what that means. !IDETool formats everything or nothing after partitioning. So I chose format.
I picked a 1024 byte LFAU, first partition is called "One", second is called "Two".
I used a brand new 2GB CF card, so there should be no other 'Acorn' data or anything on the disc.

Here's the output files and the output from the spool.

[IanJeffray]

I thought I'd try PartMgr on this new setup. No joy with the Simtec of course.

But what on earth is going on with my ADFS drive. It's a basic 16GB CF card, formatted to its maximum as normal with HForm, yet PartMgr seems to be showing ... well I honestly don't know what it's showing ... it READS TO ME like it's saying it thinks the whole disc is 2.91GB, and within that, there's a 2.91GB "Reserved" section and a 14.91GB FileCore section. Ummm??

addfuss.png (25.31 KiB) Viewed 6510 times

[JonAbbott]

!IDETool formats everything or nothing after partitioning. So I chose format.
I picked a 1024 byte LFAU, first partition is called "One", second is called "Two".
I used a brand new 2GB CF card, so there should be no other 'Acorn' data or anything on the disc.

Here's the output files and the output from the spool.

Thanks, that confirms everything I needed to read from the disk.

Given your comment about !IDETool formatting everything and what I see in the data it appears Simtec does not have a partition table. It looks like it trawls the disc for partitions with each partition pointing to the next in a linked list across the disc.

Could you please confirm the exact size of "One" so I can figure out which words hold the partition size and offset to the next partition.

what on earth is going on with my ADFS drive. It's a basic 16GB CF card, formatted to its maximum as normal with HForm, yet PartMgr seems to be showing ... well I honestly don't know what it's showing ... it READS TO ME like it's saying it thinks the whole disc is 2.91GB, and within that, there's a 2.91GB "Reserved" section and a 14.91GB FileCore section. Ummm??

addfuss.png

A debug log will probably explain what you're seeing, but if I was to guess:

• the drive was initially formatted with an HForm that writes a MBR partition table
• the MBR partition table has a 2.91GB partition entry that isn't the "official" Acorn type
• the drive was later reformatted as "16GB" with a formatter that doesn't clear or update the MBR

So I suspect you have an MBR that's mismatched with the contents of the drive.