scsi.device 44.2 on A600 - broken?

d0pefish · 03 January 2011, 01:07

Hi!

Has anyone tried using Doobrey's patch of scsi.device, version v44.2 on a real A600?

On my A600, I am using a CF card, and I have been trying scsi 44.2 under various Kickstart (2.05/3.1), filesystem (PFS3/SFS) and CPU (68000/68030) configurations - both as a cooked-in Kickstart modification, and more simply with LoadModule.

What happens is that as soon as the machine reboots with either the modified Kickstart or scsi 44.2 LoadModule'd, the HD light blinks, and then I get "Not a DOS disk in device DH0:". If I try to load HDToolbox, it hangs on scanning the scsi.device.

*Something* is recognised, because the Early Startup menu shows the DH0 partition in the list, and correctly identifies its partition type.

Reboot without v44.2, and the drive works perfectly.

Weirdly enough, scsi 44.2 works perfectly under WinUAE (A600 chipset configuration) on the same drive I am using on the real machine, by way of the CF reader, set as IDE0. I can LoadModule scsi 44.2, the emulated A600 resets, and the CF card boots fine. In a Shell, "version scsi.device" reports back v44.2, and HDToolbox shows my card correctly under scsi.device.

All other versions of scsi.device I have tried have been fine, namely 43.43, 43.45, and obviously the built-in KS2.x and 3.1 versions.

I'm certain that the patch process was alright (otherwise it wouldn't work in UAE, surely?) and I definitely used the A600 .pch file.

It's not the end of the world obviously, but I wondered if anyone else can recreate the problem at their end. If so, what's wrong with the patch?

I've just been trying to compare various device versions and filesystems to find the best speed.

Cheers

FOL · 03 January 2011, 02:42

Quote:

Originally Posted by d0pefish

Hi!

Has anyone tried using Doobrey's patch of scsi.device, version v44.2 on a real A600?

On my A600, I am using a CF card, and I have been trying scsi 44.2 under various Kickstart (2.05/3.1), filesystem (PFS3/SFS) and CPU (68000/68030) configurations - both as a cooked-in Kickstart modification, and more simply with LoadModule.

What happens is that as soon as the machine reboots with either the modified Kickstart or scsi 44.2 LoadModule'd, the HD light blinks, and then I get "Not a DOS disk in device DH0:". If I try to load HDToolbox, it hangs on scanning the scsi.device.

*Something* is recognised, because the Early Startup menu shows the DH0 partition in the list, and correctly identifies its partition type.

Reboot without v44.2, and the drive works perfectly.

Weirdly enough, scsi 44.2 works perfectly under WinUAE (A600 chipset configuration) on the same drive I am using on the real machine, by way of the CF reader, set as IDE0. I can LoadModule scsi 44.2, the emulated A600 resets, and the CF card boots fine. In a Shell, "version scsi.device" reports back v44.2, and HDToolbox shows my card correctly under scsi.device.

All other versions of scsi.device I have tried have been fine, namely 43.43, 43.45, and obviously the built-in KS2.x and 3.1 versions.

I'm certain that the patch process was alright (otherwise it wouldn't work in UAE, surely?) and I definitely used the A600 .pch file.

It's not the end of the world obviously, but I wondered if anyone else can recreate the problem at their end. If so, what's wrong with the patch?

I've just been trying to compare various device versions and filesystems to find the best speed.

Cheers

Download, BB3 or was it BB4.
One of them will patch it for you.

I normally do a fresh install, BB1, then BB2, then BB3, then BB4.
I then back it up, once all working,

.

d0pefish · 03 January 2011, 03:19

Thanks FOL,

Now that you mention it, I see scsi 44.2 is available readily-patched for all Amiga models inside the BB3 archive - I didn't realise this

I did a diff with WinHex of the BB3 archive's A600 version against the scsi.device that I created with spatch from the original .pch file on Doobrey's site, and they're identical - so that's one potential issue eliminated.

One thing I should clarify here: I'm not actually using OS3.9 on my A600, I'm just borrowing components from the various post-OS3.9 updates that are around.

I'm working with a raw Kickstart 3.1 + the scsi.device update; nothing else - so there's no AmigaOS ROM Update file like on a 3.9 + BBx installation. I completely understand how this works, as on my A4000, I run OS3.9, and have all of the BB4 ROM patches cooked-in ROM courtesy of Algor Pro, and SetPatch is told to not apply anything at all from the disk-based AmigaOS ROM Update file. This arrangement works great.

So, given that OS3.9's scsi.device 43.43 and the unofficial 43.45 both work in the way that I'm asking of them - i.e. just applied over vanilla KS3.1 - why wouldn't 44.2?

I have a few theories:

There is a bug in 44.2 that shows up on real hardware, but not in WinUAE
44.2 just doesn't like my CF card whereas all other versions of scsi.device do (?!) - I need to get back home and test a proper hard disk on it
I am missing something really, really silly.

FOL · 03 January 2011, 03:24

Quote:

Originally Posted by d0pefish

Thanks FOL,

Now that you mention it, I see scsi 44.2 is available readily-patched for all Amiga models inside the BB3 archive - I didn't realise this

I did a diff with WinHex of the BB3 archive's A600 version against the scsi.device that I created with spatch from the original .pch file on Doobrey's site, and they're identical - so that's one potential issue eliminated.

One thing I should clarify here: I'm not actually using OS3.9 on my A600, I'm just borrowing components from the various post-OS3.9 updates that are around.

I'm working with a raw Kickstart 3.1 + the scsi.device update; nothing else - so there's no AmigaOS ROM Update file like on a 3.9 + BBx installation. I completely understand how this works, as on my A4000, I run OS3.9, and have all of the BB4 ROM patches cooked-in ROM courtesy of Algor Pro, and SetPatch is told to not apply anything at all from the disk-based AmigaOS ROM Update file. This arrangement works great.

So, given that OS3.9's scsi.device 43.43 and the unofficial 43.45 both work in the way that I'm asking of them - i.e. just applied over vanilla KS3.1 - why wouldn't 44.2?

I have a few theories:

There is a bug in 44.2 that shows up on real hardware, but not in WinUAE
44.2 just doesn't like my CF card whereas all other versions of scsi.device do (?!) - I need to get back home and test a proper hard disk on it
I am missing something really, really silly.

To be honest, whenever I "just borrowed" 44.22 it never worked with vanilla KS3.1, not even on my A1200.

Only ever had it working, when installing BB3 correctly.
But once it was in rom, I didnt need it anymore.

Out of curiousity, what Filesystem are you using?

d0pefish · 03 January 2011, 03:45

Currently, I've got a 4GB SFS 1.84 partition. Later versions of SFS don't work on KS 2.05 - my real ROM - so the drive wouldn't available to softkick from

- I learned this one the hard way.

My previous attempt was with PFS3 - and the result was the same.

One thing I did notice is that once 44.2 was loaded, SysInfo (loaded from floppy) showed a load of random characters where you'd normally see the description of the filesystem. So it was as if 44.2 was causing the filesystem to be mis-read off the RDB or something.

Perhaps I'll get some screenshots tomorrow to clarify what I mean

There are plenty ways I can carry on, I mean, I don't really *need* 44.2 as there are other means to get the 16GB card fully accessible on the machine (the end goal), I'm just curious more than anything

And if it means there's possibly a bug in the patched driver, perhaps someone can help create a fix for it.

Zetr0 · 03 January 2011, 04:46

@d0pefish / FOL

doesn't scsi.device 44.2 require a patched FFS to work?

mfilos · 03 January 2011, 10:46

d0pefish my friend...
I tried the same yesterday while teasing my beloved A600.
Actually I was planning of making a custom 3.9 kickstart as I did in my A1200D PPC. For some reasons I haven't managed to boot from my custom Kickstart (despite without HD I got the purple screen with Kickstart 3.9 (45.57).
I didn't have extra time to search it... but I used the 1MB kickstart I made, via ACATune and works like a charm!!! Pretty fast and slick and BBlank module that i putted as extra is cool also lol

Check your PM btw...

FOL · 03 January 2011, 12:51

Quote:

Originally Posted by mfilos

but I used the 1MB kickstart I made, via ACATune and works like a charm!!!

Exactly what I done.

@Zetr0, I have tried using FFS 45.15, no joy. Never tried 45.16 though.

@d0pefish, If its showing random characters, then it needs Filesystems deleted and reloaded. This is how I got round that, when I came across that problem.
Obviously doing so, will erase data on the HDD.

d0pefish · 03 January 2011, 14:46

Cheers guys,

Perhaps the freezing is caused by the machine stumbling on the random data when it's looking for the filesystem. I'll zero out the first few megs of the flash card and attempt HDtoolbox again

Minuous · 03 January 2011, 16:23

What CPU does your A600 have?

d0pefish · 03 January 2011, 17:08

Currently a full 68030, ACA630/30

d0pefish · 04 January 2011, 00:07

Ok, zeroing out the drive and prepping the drive with scsi.device 44.2 loaded and resident worked.

When I prep the drive (create RDB, partition and add filesystems) on a machine with <44.2, the drive doesn't work under 44.2.

Now that the drive is prepared under 44.2, it's not accessible under <44.2; the partitions just appear as NDOS.

In order to access my drive, I have to load a kickstart with 44.2 applied from floppy first

Will have to find an EPROM!

FOL · 04 January 2011, 00:09

Quote:

Originally Posted by d0pefish

Ok, zeroing out the drive and prepping the drive with scsi.device 44.2 loaded and resident worked.

When I prep the drive (create RDB, partition and add filesystems) on a machine with <44.2, the drive doesn't work under 44.2.

Now that the drive is prepared under 44.2, it's not accessible under <44.2; the partitions just appear as NDOS.

In order to access my drive, I have to load a kickstart with 44.2 applied from floppy first

Will have to find an EPROM!

What Filesystem?
What Partition size?

d0pefish · 04 January 2011, 00:46

PFS3, and 3.8GB this time.

Thanks

Photon · 04 January 2011, 00:52

I feel like I'm not on my home street on this, but I think it's always a good idea to partition disks on the ((lowest OS version)) system that's going to use them. I had to set up my CF cards on my A500 with kick 2.0/WB 2.1 to gift it with the proper FFS version to work on my 3 CF-adapted Amigas.

Normally, a boot-floppy works to boot "anything", but I have limited experience with PFS.

Asking someone to make (or burn yourself) a ROM with patches would solve it, so...

Me, I don't need those strange patches, but I never go beyond kick/wb 3.1. I think it's fast enough as it is, but then I rarely do much work in WB, someone who sits and waits for icons to load may feel differently. There should be a "real" snapshot patch for that instead and it would be fast from kick/wb 1.2 and up

d0pefish · 04 January 2011, 17:59

Well, I've no idea how, but I've fixed it

Perhaps half the issue was involving WinUAE during some of the drive prep stages. This time I just got all filesystem/formatting tools, HDToolbox, and scsi 44.2 and loadmodule onto a floppy.

After prepping and formatting the drive all on the 600, I checked whether it DH0: was mounted and listable both before and after loading the 44.2 scsi module. This time it did.

So I used WinUAE to copy my ClassicWB setup across, and prepared a custom KS3.1 rom with scsi 44.2, exec v44 by Piru and various other patches from OS3.9.

Now ACATune can load it, and I don't need the LoadModule line.

It was worth it. I now get 3.2MB/sec transfer rate on the A600's CF card. Before the ACA630, it was about 0.6MB. After the ACA but before patches, it was 2.1MB/sec.

Now I'm about to clone the hell out of my CF card so I don't have to endure all this again!

Thanks very much to the various people who have PMed me with some helpful files!

Cosmos · 04 January 2011, 18:30

You can add too this TLSF patch http://uk.aminet.net/util/boot/TLSFMem_19.readme

Romable since the begining

alexh · 04 January 2011, 18:37

How stable is that patch? I saw a big improvement when running the original but there was a round of bug-fixes due at the time that Chris quit the project.

http://eab.abime.net/showthread.php?p=394066

I think they were due to reporting the wrong value with MEMF_LARGEST?

FOL · 04 January 2011, 19:15

Quote:

Originally Posted by Cosmos

You can add too this TLSF patch http://uk.aminet.net/util/boot/TLSFMem_19.readme

Romable since the begining

I thought this was added to the graphics.library 40.25 Beta 8?

Cosmos · 04 January 2011, 19:56

@FOL

Oh... TLSFMem patch the exec.library !!

@alexh

Thanks for the URL = I'll see tomorrow... So tired, I'm going into my bed right now...

03 January 2011, 01:07	#1
d0pefish Registered User Join Date: May 2006 Location: Newcastle-upon-Tyne, England UK Age: 35 Posts: 103	scsi.device 44.2 on A600 - broken? Hi! Has anyone tried using Doobrey's patch of scsi.device, version v44.2 on a real A600? On my A600, I am using a CF card, and I have been trying scsi 44.2 under various Kickstart (2.05/3.1), filesystem (PFS3/SFS) and CPU (68000/68030) configurations - both as a cooked-in Kickstart modification, and more simply with LoadModule. What happens is that as soon as the machine reboots with either the modified Kickstart or scsi 44.2 LoadModule'd, the HD light blinks, and then I get "Not a DOS disk in device DH0:". If I try to load HDToolbox, it hangs on scanning the scsi.device. Something is recognised, because the Early Startup menu shows the DH0 partition in the list, and correctly identifies its partition type. Reboot without v44.2, and the drive works perfectly. Weirdly enough, scsi 44.2 works perfectly under WinUAE (A600 chipset configuration) on the same drive I am using on the real machine, by way of the CF reader, set as IDE0. I can LoadModule scsi 44.2, the emulated A600 resets, and the CF card boots fine. In a Shell, "version scsi.device" reports back v44.2, and HDToolbox shows my card correctly under scsi.device. All other versions of scsi.device I have tried have been fine, namely 43.43, 43.45, and obviously the built-in KS2.x and 3.1 versions. I'm certain that the patch process was alright (otherwise it wouldn't work in UAE, surely?) and I definitely used the A600 .pch file. It's not the end of the world obviously, but I wondered if anyone else can recreate the problem at their end. If so, what's wrong with the patch? I've just been trying to compare various device versions and filesystems to find the best speed. Cheers

03 January 2011, 03:19	#3
d0pefish Registered User Join Date: May 2006 Location: Newcastle-upon-Tyne, England UK Age: 35 Posts: 103	Thanks FOL, Now that you mention it, I see scsi 44.2 is available readily-patched for all Amiga models inside the BB3 archive - I didn't realise this I did a diff with WinHex of the BB3 archive's A600 version against the scsi.device that I created with spatch from the original .pch file on Doobrey's site, and they're identical - so that's one potential issue eliminated. One thing I should clarify here: I'm not actually using OS3.9 on my A600, I'm just borrowing components from the various post-OS3.9 updates that are around. I'm working with a raw Kickstart 3.1 + the scsi.device update; nothing else - so there's no AmigaOS ROM Update file like on a 3.9 + BBx installation. I completely understand how this works, as on my A4000, I run OS3.9, and have all of the BB4 ROM patches cooked-in ROM courtesy of Algor Pro, and SetPatch is told to not apply anything at all from the disk-based AmigaOS ROM Update file. This arrangement works great. So, given that OS3.9's scsi.device 43.43 and the unofficial 43.45 both work in the way that I'm asking of them - i.e. just applied over vanilla KS3.1 - why wouldn't 44.2? I have a few theories: There is a bug in 44.2 that shows up on real hardware, but not in WinUAE 44.2 just doesn't like my CF card whereas all other versions of scsi.device do (?!) - I need to get back home and test a proper hard disk on it I am missing something really, really silly.

04 January 2011, 00:52	#15
Photon Moderator Join Date: Nov 2004 Location: Eksjö / Sweden Posts: 5,449	I feel like I'm not on my home street on this, but I think it's always a good idea to partition disks on the ((lowest OS version)) system that's going to use them. I had to set up my CF cards on my A500 with kick 2.0/WB 2.1 to gift it with the proper FFS version to work on my 3 CF-adapted Amigas. Normally, a boot-floppy works to boot "anything", but I have limited experience with PFS. Asking someone to make (or burn yourself) a ROM with patches would solve it, so... Me, I don't need those strange patches, but I never go beyond kick/wb 3.1. I think it's fast enough as it is, but then I rarely do much work in WB, someone who sits and waits for icons to load may feel differently. There should be a "real" snapshot patch for that instead and it would be fast from kick/wb 1.2 and up Last edited by Photon; 04 January 2011 at 04:35. Reason: Unclosed ((TCB Braket Alert...))

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Broken Squirrell SCSI adaptor.	ElectroBlaster	support.Hardware	0	26 April 2012 12:22
A1200 sfs scsi.device 44.2 broken?	elvio.lusiani	support.Apps	0	01 February 2011 00:05
broken scsi pin	lost_loven	support.Hardware	2	02 January 2011 14:23
Broken A600 hd	mtb	support.Hardware	13	14 August 2008 09:00
BPPC scsi cable broken...	Bamiga2002	support.Hardware	2	02 December 2004 14:01

03 January 2011, 03:45	#5
d0pefish Registered User Join Date: May 2006 Location: Newcastle-upon-Tyne, England UK Age: 35 Posts: 103	Currently, I've got a 4GB SFS 1.84 partition. Later versions of SFS don't work on KS 2.05 - my real ROM - so the drive wouldn't available to softkick from - I learned this one the hard way. My previous attempt was with PFS3 - and the result was the same. One thing I did notice is that once 44.2 was loaded, SysInfo (loaded from floppy) showed a load of random characters where you'd normally see the description of the filesystem. So it was as if 44.2 was causing the filesystem to be mis-read off the RDB or something. Perhaps I'll get some screenshots tomorrow to clarify what I mean There are plenty ways I can carry on, I mean, I don't really need 44.2 as there are other means to get the 16GB card fully accessible on the machine (the end goal), I'm just curious more than anything And if it means there's possibly a bug in the patched driver, perhaps someone can help create a fix for it.

03 January 2011, 04:46	#6
Zetr0 Ya' like it Retr0? Join Date: Jul 2005 Location: United Kingdom Age: 48 Posts: 9,768	@d0pefish / FOL doesn't scsi.device 44.2 require a patched FFS to work?

03 January 2011, 10:46	#7
mfilos Paranoid Amigoid Join Date: Mar 2008 Location: Athens/Greece Age: 44 Posts: 1,979	d0pefish my friend... I tried the same yesterday while teasing my beloved A600. Actually I was planning of making a custom 3.9 kickstart as I did in my A1200D PPC. For some reasons I haven't managed to boot from my custom Kickstart (despite without HD I got the purple screen with Kickstart 3.9 (45.57). I didn't have extra time to search it... but I used the 1MB kickstart I made, via ACATune and works like a charm!!! Pretty fast and slick and BBlank module that i putted as extra is cool also lol Check your PM btw...

03 January 2011, 14:46	#9
d0pefish Registered User Join Date: May 2006 Location: Newcastle-upon-Tyne, England UK Age: 35 Posts: 103	Cheers guys, Perhaps the freezing is caused by the machine stumbling on the random data when it's looking for the filesystem. I'll zero out the first few megs of the flash card and attempt HDtoolbox again

03 January 2011, 16:23	#10
Minuous Coder/webmaster/gamer Join Date: Oct 2001 Location: Canberra/Australia Posts: 2,516	What CPU does your A600 have?

03 January 2011, 17:08	#11
d0pefish Registered User Join Date: May 2006 Location: Newcastle-upon-Tyne, England UK Age: 35 Posts: 103	Currently a full 68030, ACA630/30

04 January 2011, 00:07	#12
d0pefish Registered User Join Date: May 2006 Location: Newcastle-upon-Tyne, England UK Age: 35 Posts: 103	Ok, zeroing out the drive and prepping the drive with scsi.device 44.2 loaded and resident worked. When I prep the drive (create RDB, partition and add filesystems) on a machine with <44.2, the drive doesn't work under 44.2. Now that the drive is prepared under 44.2, it's not accessible under <44.2; the partitions just appear as NDOS. In order to access my drive, I have to load a kickstart with 44.2 applied from floppy first Will have to find an EPROM!

04 January 2011, 00:46	#14
d0pefish Registered User Join Date: May 2006 Location: Newcastle-upon-Tyne, England UK Age: 35 Posts: 103	PFS3, and 3.8GB this time. Thanks

04 January 2011, 17:59	#16
d0pefish Registered User Join Date: May 2006 Location: Newcastle-upon-Tyne, England UK Age: 35 Posts: 103	Well, I've no idea how, but I've fixed it Perhaps half the issue was involving WinUAE during some of the drive prep stages. This time I just got all filesystem/formatting tools, HDToolbox, and scsi 44.2 and loadmodule onto a floppy. After prepping and formatting the drive all on the 600, I checked whether it DH0: was mounted and listable both before and after loading the 44.2 scsi module. This time it did. So I used WinUAE to copy my ClassicWB setup across, and prepared a custom KS3.1 rom with scsi 44.2, exec v44 by Piru and various other patches from OS3.9. Now ACATune can load it, and I don't need the LoadModule line. It was worth it. I now get 3.2MB/sec transfer rate on the A600's CF card. Before the ACA630, it was about 0.6MB. After the ACA but before patches, it was 2.1MB/sec. Now I'm about to clone the hell out of my CF card so I don't have to endure all this again! Thanks very much to the various people who have PMed me with some helpful files!

04 January 2011, 18:30	#17
Cosmos Banned Join Date: Jan 2007 Location: France Posts: 655	You can add too this TLSF patch http://uk.aminet.net/util/boot/TLSFMem_19.readme Romable since the begining

04 January 2011, 18:37	#18
alexh Thalion Webshrine Join Date: Jan 2004 Location: Oxford Posts: 13,986	How stable is that patch? I saw a big improvement when running the original but there was a round of bug-fixes due at the time that Chris quit the project. http://eab.abime.net/showthread.php?p=394066 I think they were due to reporting the wrong value with MEMF_LARGEST?

04 January 2011, 19:56	#20
Cosmos Banned Join Date: Jan 2007 Location: France Posts: 655	@FOL Oh... TLSFMem patch the exec.library !! @alexh Thanks for the URL = I'll see tomorrow... So tired, I'm going into my bed right now...

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)