Partition Manager

Thanks for clarify.
I hope that was my only mistake in this context ;-)

Yes, I was referring to ADFS v4. In particular checking that it formats the correct drive and that formatting 4K drives works.

I certainly wouldn’t use Partition Manager with working drives – unless you don’t mind losing them…not until its had more community testing.

Even if I had every variety of hardware, there’s no way I could thoroughly test Partition Manager as it covers so many File systems, Podules, OS versions, drive types and partition types. I think it’s going to be a case of testing the obvious things and hoping everything else is good!

Format an extFAT 64GB device to FAT32FS via Partition Manager was also working without problems.

Even if I had every variety of hardware, there’s no way I could thoroughly test Partition Manager as it covers so many File systems, Podules, OS versions, drive types and partition types. I think it’s going to be a case of testing the obvious things and hoping everything else is good!

Exactly why I didn’t put support for fixing partitions in DiscKnight. The best I could offer for cases of partition corruption was to search for the Filecore partition on the disc, and copy all the data off to somewhere else.

Partition Manager v0.98 available for testing here

Why does PartMgr (0.98 and others) ignore the flag – Don’t allow windows off screen

0.48 behaved itself.

With RO 5.24 on the Titanium I get:

Internal error: abort on data transfer at &207897E4
*where
Address &207897E4 is at offset &0000FB90 in module 'BASICPM'
*

What is BASICPM ? What does *Help BASICPM say? RO5.24 is rather old now, but I cannot remember a BASICPM.

What is BASICPM

It is a Module provided by PartMgr. It is possibly something to do with the mathematical capabilities of BASIC, or some other reason.

If the setup in !Run is bypassed and !RunImage directly double clicked on then PartMgr lights up nicely on OS5.24 on this Titanium. Disc sizes are correctly reported but then the biggest is only 240GB

Oh yes, I see BASICPM in !PartMgr. However, from my reading of the !Run file, it is only used if BASIC 1.85 or later is not running. In that case, it should have been BASICPM32 that was loaded, not BASICPM. Both modules seem to provide the BASICPM command, and are called BASICPM when loaded.

Maybe Anton could load the standard BASIC 1.85 module?

Here, on this Titanium, if I load BASICPM32 and then try BASICPM -quit <PartMgr$Dir>.!RunImage from the !Run file, I also get the same abort as Anton:

Address &21434524 in Module   BASICPM             @ &21424994 + &FB90
Register dump (at &2007A3F0) is:
pc  = &21434524  Flags=&20000010=nzCv if  Mode=User 32 bit
r0  = &00008F00 : .?.. :      36608
r1  = &004E19BE : ?.N. :    5118398
r2  = &00000010 : .... :         16
r3  = &00000010 : .... :         16
r4  = &0000821A : .?.. :      33306
r5  = &004E9BD8 : ?N. :    5151704
r6  = &000000FF : .... :        255
r7  = &004E9BD6 : ?N. :    5151702
r8  = &00008700 : .?.. :      34560
r9  = &00000002 : .... :          2
r10 = &214442DC : ?BD! :  558121692 => &6975712D : -qui : 1769304365
r11 = &000001FC : ?... :        508
r12 = &00008100 : .?.. :      33024
r13 = &004E9FD8 : ?N. :    5152728
r14 = &21425120 :  QB! :  557994272 => &EAFFFFB0 : ?..? : 3942645680
Code around address is:
214344EC : 049DF004 : .?. : LDREQ   PC,[R13],#4
214344F0 : E3A0400A : .@?? : MOV     R4,#&0A            ; =10
214344F4 : E3A05001 : .P?? : MOV     R5,#1
214344F8 : EB0002B3 : ?..? : BL      &21434FCC
214344FC : E3A00007 : ..?? : MOV     R0,#7
21434500 : EB000364 : d..? : BL      &21435298
21434504 : E49DF004 : .?? : LDR     PC,[R13],#4
21434508 : E5180094 : ?..? : LDR     R0,[R8,#-148]
2143450C : E2112003 : . .? : ANDS    R2,R1,#3
21434510 : 0A000009 : .... : BEQ     &2143453C
21434514 : E1A02182 : ?!?? : MOV     R2,R2,LSL #3
21434518 : E2623020 :  0b? : RSB     R3,R2,#&20         ; =" "
2143451C > E89100C0 > ?.?? : LDMIA   R1,{R6,R7}
21434520 : E1A06236 : 6b?? : MOV     R6,R6,LSR R2
21434524 : E1866317 : .c?? : ORR     R6,R6,R7,LSL R3
21434528 : E4806004 : .`?? : STR     R6,[R0],#4

Caused by r1 not being word aligned?

I’ve just had a try of PartMan v0.88 to see if I could shed any light on what’s going on with the 128MB system memory and…

It correctly has the FAT part within the FileCore part, but why “Reserved”? That’s probably best just omitted given the frankenstein arrangement of how FileCore and FAT dance together on these machines. ;-)

Smells like

LOADFILETOKENCOPY
        LDR     R0,[ARGP,#PAGE]
; Use compile-time architecture flags to decide how to handle the potential unaligned load
...
  [ SupportARMv6
        ; We may be running on ARMv6, or we may be running on an earlier architecture. Support both by emulating pre-ARMv6 unaligned load.
        BIC     R6,R1,#3
        LDMIA   R6,{R6,R7}
  |
        ; We only have to run on pre-ARMv6, where the bottom two address bits are ignored for LDM.
        LDMIA   R1,{R6,R7}
  ]

Oh yes, I see BASICPM in !PartMgr. However, from my reading of the !Run file, it is only used if BASIC 1.85 or later is not running.

I use 5.28 and BASIC is 1.81. I get the same error message like Anton.

Maybe Anton could load the standard BASIC 1.85 module?

If I change RMEnsure BASIC 1.85 in !Run to 1.81 all I try is working.
Is there anything in PartMgr that needed 1.85?

Thanks for the reports. I added a bespoke build of 32-bit BASIC 1.85 to get it running on the A9home, but as you’ve all noticed its the ARMv4 version.

I’ll correct in due course. Commenting out the lines that check the BASIC version in !Run should switch it back to ROM BASIC.

Why does PartMgr (0.98 and others) ignore the flag – Don’t allow windows off screen

It allows the window to move off-screen, which is primarily for legacy machines with low resolutions. If it’s causing an issue I can make it resolution dependent, so it’s only allowed off-screen on low-res desktops.

No issues. It just seemed odd that an App can bypass system settings.

needs to be built with the SupportARMv6 option to create an aligned address for the LDM

Should be fixed in the build I’ve just uploaded

It works now on my Titanium too.

Here also.
Tested on my a9home. IDEFS is working, USB devices are not detected.

Tested on my a9home. IDEFS is working, USB devices are not detected.

What filesystem do the USB devices appear as? If it’s SCSIFS, could you post a debug log so I can see what it did.

why “Reserved”? That’s probably best just omitted given the frankenstein arrangement of how FileCore and FAT dance together

It’s in the partition table as a Reserved partition, so it gets shown.

A9home supports MassFS, DOSFS and WIN95FS… PartMgr see nothing.
Can create a debug log tomorrow morning.

It will be MassFS I suspect, which PM doesn’t currently know about . If you could eMail me a copy of the relevant FS Module and it’s associated Filer Module, I can look at adding it.

I’m close to releasing the Partition Manager build that will create a usable Pi drive, although I have an issue with AcornHTTP I’ve yet to figure out.

AcornHTTP has a dependency on !Boot which is preventing hot-swapping over the SD card so a new one can be formatted. When the Module loads, it tries to access several files in !Boot, but the worst issue is URL_GetURL/URL_ReadData fail if !Boot isn’t accessible.

Short of creating a bespoke AcornHTTP that works when !Boot doesn’t exist, I’m not sure how to proceed. PM needs to connect to sites to pull down HardDisc4, the Pi firmware and various packages that need pre-installing for the new OS drive to be usable when booted.

And on the subject of HardDisc4. As it’s a generic HardDisc4, there are several issues with it that make it unsuitable for a Pi out of the box, for example:

1. The network config defaults to Access, not DHCP/IP
2. NetTime is not configured to sync from a Time server
3. !PackMan is missing

To end up with a usable SD card, I’m having to alter various things such as adding the missing TimeSetup into RO520Hook and RO530Hook (it has to go here because Choices.Boot is copied from these on first boot), reconfiguring !Internet for DHCP and pulling down and installing packages such as !PackMan etc.

For the minute I’ve hardcoded all of this, but I really need to create a full package manager so you can select which packages you want pre-installed onto the SD card. For the minute I’ve just added the bare minimum.

All of this is working…I just need to find a workaround for AcornHTTP’s dependence on !Boot for it to be usable on a Pi.

Short of creating a bespoke AcornHTTP that works when !Boot doesn’t exist, I’m not sure how to proceed.

Create a RAMdisc of a useful size, then hijack the Choices variables to point to the RAMdisc, so the fetcher can do whatever without messing with the SD card.

Although initialising a Pi Boot drive will fail at the download stage, due to the AcornHTTP !Boot dependency I’ve mentioned above, I’ve put Partition Manager v0.99 up for testing.

I’ve detailed the caveats and known issues in the download link.

I’ll continue to look for a workaround to the !Boot dependency before a final release. In the meantime, you can perform a full Pi Boot drive initialise with a drive attached via USB.

Hi Jon.

I have just just used 0.99 to create a Pi Boot drive on a Pi 4 and I am now using Netsurf 3.10 to post this. So, good progress.

However, I have the following comments on the contents of the Boot Loader directory (specifically for a Pi 4)

 - I did not find a Cmos file, so I copied one from elsewhere
 - Bootcode.bin is not needed on a Pi 4, so I deleted it.
 - Start.elf is not needed on a Pi 4, so I deleted it.
 - Fixup.dat is not needed on a Pi 4, so I deleted it.
 - I replaced the following files from a known good Pi 4 boot SDcard
   config.txt, fixup4.dat, start4.elf 

Without the above changes, I was getting nowhere towards a workable system. The screen was fixed at an workable low resolution.

With the changes, this RISCOSPi2 SDcard seems good. I had no trouble with hot swapping the SDcard for the download.