SCSI2SD v6 and large drives

[Ferry]

UnitControl works pretty well, but changes are not permanent. Good thing is that you can use it as a CLI command, so you can put it in the Startup-Sequence to control units.

Saluditos,

Ferrán.

[AC/DC HACKER!]

Right, that's my next use. But MyUnitControl doesn't..at least it didn't go well for my V6.

https://aminet.net/package/util/misc/MyUnitControl

[crazyc]

Originally Posted by Ferry

UnitControl works pretty well, but changes are not permanent. Good thing is that you can use it as a CLI command, so you can put it in the Startup-Sequence to control units.

Saluditos,

Ferrán.

I use unitcontrol this way as well with my ide-scsi bridge via my b1260.

[Ferry]

Originally Posted by Futaura

If you have the SCSI menu set to Auto, Async/Sync is selected by the Sync flag in the RDB of the drive. Not all tools give the option of changing it and it will default to Async. So, are you sure you are not using Async mode? You should get much faster speeds than that, even with 6.2.5, when using Sync mode.

Could you please post the exact settings you are using with the v6?

I downgraded to v6.2.5, set 2 virtual SCSI disk (50 and 9,65 GB respectively) in a 64GB Samsung MicroSD with SD adapter, first one to be used for the Amiga WB and progs in different partitions, second one formatted as a FAT partition, to be mounted in Amiga with a mountlist. In the SCSI2SD menu, SCSI speed limit was set to "No limit (safe)".

I installed and partitioned the 50GB "disk" into several partitions: 4GB for WB, 10GB for Work (apps), etc., I set the Sync RDB flag with RDBFlags, installed PFS3aio in RDB, set all partition to PFS but one (that one was set to FFS for comparison), with default mask and bytes per block (1024), buffers set to 200, and then I copied with DOpus4 my current WB to the first partition, without a single hiccup.

But copying all my apps to the Work partition has been a completely different story: while it was at half the process, it showed an error, forcing to stop the copy and setting the partition as write-protected. Reformatting and copying again with a different program (DiskMaster) made no difference: same error at the same point.

First I thought it could be that I made several tests with different blocksizes and DOSTypes and that could have corrupted somehow the disk, so I put the SD card in my Mac and reformatted it with a "secure" option that writes zeros all over it, so any corruption, RDB, any data on the disk would be definitely erased. Then I repeated all the process above: same results.

Now I'm going to limit SCSI2SD speed to "Sync, 10MB", and see what happens. If this doesn't work, will try "Sync, 5MB"………

Any clue on what could be causing this?

Saluditos,

Ferrán.

[Ferry]

Oh, forgot to add: every time I install PFS3aio in the RDB, its identifier is set to "CFS/0", instead of "PFS/3", so I change it manually. Is that normal, that "CFS/0" identifier?

Saluditos,

Ferrán.

[Jope]

CFS\0 is probably something your partitioning software uses as a placeholder. It is normal that you must manually type in your own dostype.

[Ferry]

Buffff!!! Setting this thing has been a nightmare.

The "Sync, 10MB" mode has been a complete failure, I was not able to even write or read a correct RDB, after a reboot partitions could be there or not; trying to change anything in the partitions setup was impossible, since writting RDB back with modifications resulted in an error message ("Error 7 in drive description")

"Funny" thing is that everytime I installed and partitioned the drive, partitions were there, so I tried to format them and it did, but as soon as I opened HDToolBox again in order to change something minor, like buffers f.ex., saving RDB resulted in error. Or formatting, rebooting to boot from a bootable partition in the SCSI2SD drive, and computer hanging…

Now I have set it to "Sync, 5MB", and everything seems to work well so far… (copying now progs backup, that yesterday failed, so fingers crossed)

Saluditos,

Ferrán.

[AC/DC HACKER!]

Originally Posted by Ferry

Buffff!!! Setting this thing has been a nightmare.

The "Sync, 10MB" mode has been a complete failure, I was not able to even write or read a correct RDB, after a reboot partitions could be there or not; trying to change anything in the partitions setup was impossible, since writting RDB back with modifications resulted in an error message ("Error 7 in drive description")

"Funny" thing is that everytime I installed and partitioned the drive, partitions were there, so I tried to format them and it did, but as soon as I opened HDToolBox again in order to change samething minor, like buffers f.ex., saving RDB resulted in error. Or formatting, rebooting to boot from a bootable partition in the SCSI2SD drive, and computer hanging…

Now I have set it to "Sync, 5MB", and everything seems to work well so far… (copying now progs backup, that yesterday failed, so fingers crossed)

Saluditos,

Ferrán.

(Not into luck) Sync 5MB is doing very well with my CSPCC's SCSI. Sync 10MB screws any SD Card I'm using once Partitions are set using PFS3, SFS, doesn't matter. There's a bug FOR Amiga..somewhere. Well, okay, not FOR Amigas..but..there's a bug somewhere.

I'm thinking it doesn't matter if it's CSPPC or not since Futaura has reported close to same findings. But I haven't contacted Michael yet, as Futaura has.

I'm currently testing an OS 3.1.4.1 install on a small SD..when I'm able. Maybe there will be a report by Futaura soon?

[Futaura]

You still have problems when using 6.2.5? I'm using 6.2.5 in Sync 10mb mode now and yesterday I reformatted and recopied all my partitions from my HDDs (SCSI and IDE). All data tested ok with DirDiff (took a while!) and partitions checked ok with PFSDoctor. I've not heard any more from Michael yet, but presumably he is still worked through my last feedback. 6.2.9 was definitely a problem for my system. This is on a BPPC / 060 card.

In case it matters, I'm using a 64GB SanDisk Extreme PRO SDXC UHS-I card, not that I need all that space - it was just cheap on Black Friday . I currently have it split into 3 X 4GB (as denoted in scsi2sd-util), although in hindsight I probably should have set them a little bit under. No matter, I left a fair amount of free space at the end of each. I also have one 20GB device setup which I'm not using. All checkboxes are on in scsi2sd-util, with the exception of termination and LUN mapping which are both off. Speed limit set to Sync 10MB, just in case the safe default goes faster (it probably doesn't). As mentioned earlier, it is the second device on my SCSI chain with a Yamaha CDRW on the end of the cable with termination enabled. I have since moved my SCSI2SD into my removable caddy/rack, which used to house my HDD, which I can use to easily remove and hookup to a PC.

Partitions were set up on my Amiga using the OS3.9 HDToolbox and RDBFlags to switch synchronous on (so SCSI settings in the boot menu are set to default/auto). I now even boot from SCSI2SD, with a backup boot partition on IDE CF. Most partitions are PFS3 (using the regular, not direct SCSI, 18.5 last commercial version) and I even managed and remembered how to transfer my Linux APUS and Shapeshifter partitions across from my HDD!

@Ferry, bytes per block should be set to 512 with PFS3 (was the case with the commercial version - not sure about aio). I know my HDToolbox defaults to 1024 (better for FFS). I stick to using a mask of 0x7ffffffc and maxtransfer of 0x00ffffff on my BPPC. If you are having problems with RDB corruption then those settings are not relevant though. Have you been able to tell if the RDB corruption happens during writing, reading or both? For example, if you write the RDB in async mode, can HDToolbox still load it if you switch to Sync 10MB mode? What other devices do you have on your SCSI chain, if any? I see the 1260 SCSI Kit manual refers to using passive termination - BPPC and CSPPC need active termination. SCSI2SD has active rather than passive termination. Do you still have a SCSI HDD, and if so, does that work ok in Sync 10MB mode (unitcontrol will show which mode a drive actually uses)? You should never need to format the SD card, since it's structure and integrity should not be affected by SCSI2SD - SCSI2SD essentially corrupts the data in its buffer before writing it to the SD card.

[Ferry]

@Futaura My SCSI2SD card is directly connected to the Blizzard's rear connector in the A1200, using an adaptor IDC50-to-DB25 and a female-to-female DB25 SCSI cable from a Zip SCSI drive, no other drives are connected. This is a Blizz1260+SCSI card.

As for another SCSI drives, I've been using for years (couple of decades, maybe? ) a ATX tower case with three SCSI disks (2 x DEC plus 1 x Quantum Fireball, all with 2GB capacity), two optical drives (Yamaha CDRW and Plextor slot-in CDR) and an external Zip SCSI drive, which provided termination. Ocassionaly, I connect an SCSI Epson GT-7000 to the Zip unit passthrough.

The tower is attached to the Blizzard's rear DB25 connector with a long DB25-to-Centronics50, internally it goes thru all units (3 HD and 2 optical drives), goes outside again with another Centronics50 and the Zip is connected with a Centronics50-to-DB25 cable. And when I connect the scanner, I do with with a DB25-to-HD68 cable. As you can see, it's a loooong SCSI chain, and it has been working flawlessly for many, many years, but, in fact, that's nothing for a SCSI-2 chain, that allows up to 3 meters of cable.

All HDs are set to Sync transfer, providing almost 8MB/s without a hiccup.

But all tests with the SCSI2SD card have been made with it as the only SCSI device on the chain, boot has been done from a SD2IDE44 adapter.

As for the setting, my card has terminator set to ON, the rest of settings are the same. Mask is also the same, but Maxtransfer is set to 0x0007FE00 (the default one). Can it affect somehow to transfer speed? As for the bytes per block, PSF3, both original and AIO versions do support 1024 bytes AFAIK, so it should not be a problem.

Regarding RDB corruption, I can't really know if it is produced at writing or at reading, but I'd say is at writing, because several reads of the same RDB with RDB-Informer (http://aminet.net/disk/misc/RDB-Informer.lha) show the very same info, so I guess it is correctly reading bad written info: if the problem was at reading, it would show different results, not always the same. But I can't really know…

As for termination, nowhere in the manual of the kit is recommended one over the other, active or passive, in fact none of these words are mentioned. Anyway, since it's the only device in the chain, termination in the card is set to ON. Anyway, I plan to install it in a way that I can attach my hard drives together with the card when needed, but it will usually be alone, so, after a lot of reading, I have already bought an active termination, looks like they are far better that passive ones.

Saluditos,

Ferrán.

[AC/DC HACKER!]

Originally Posted by Futaura

You still have problems when using 6.2.5? I'm using 6.2.5 in Sync 10mb mode now and yesterday I reformatted and recopied all my partitions from my HDDs (SCSI and IDE). All data tested ok with DirDiff (took a while!) and partitions checked ok with PFSDoctor. I've not heard any more from Michael yet, but presumably he is still worked through my last feedback. 6.2.9 was definitely a problem for my system. This is on a BPPC / 060 card.

I used that Mask, as well as others. Mostly 0xFFFFFF. If I use Sync 10/MB it will eventually screw my RDB/FileSystem. PFS3-AIO, normal. If I stay with Sync 5/MB,,all is good and seems to be remains stable.

In case it matters, I'm using a 64GB SanDisk Extreme PRO SDXC UHS-I card, not that I need all that space - it was just cheap on Black Friday . I currently have it split into 3 X 4GB (as denoted in scsi2sd-util), although in hindsight I probably should have set them a little bit under. No matter, I left a fair amount of free space at the end of each. I also have one 20GB device setup which I'm not using. All checkboxes are on in scsi2sd-util, with the exception of termination and LUN mapping which are both off. Speed limit set to Sync 10MB, just in case the safe default goes faster (it probably doesn't). As mentioned earlier, it is the second device on my SCSI chain with a Yamaha CDRW on the end of the cable with termination enabled. I have since moved my SCSI2SD into my removable caddy/rack, which used to house my HDD, which I can use to easily remove and hookup to a PC.

Since Yuletide just started I might not have the opportunity to check this further for a day or 2. I appreciate the info - that's helpful. I haven't added any space between partitions, but I do have space at the end of SD Cards and CompactFlashes. Again, not having any problems with Sync 5. So I'm excited to see what Michael develops. I'll write him soon, also.

Partitions were set up on my Amiga using the OS3.9 HDToolbox and RDBFlags to switch synchronous on (so SCSI settings in the boot menu are set to default/auto). I now even boot from SCSI2SD, with a backup boot partition on IDE CF. Most partitions are PFS3 (using the regular, not direct SCSI, 18.5 last commercial version) and I even managed and remembered how to transfer my Linux APUS and Shapeshifter partitions across from my HDD!

I've used RDBFlags to set Sync, and that goes well with Sync 5. I don't care for OS 3.9 since 3.1.4. But...for exploring purposes, I'll check into that also but I don't see that logically making any difference. Yup, I'm not using Direct SCSI either.

[Futaura]

My PFS3 docs say to set the block size to 512, and only use 1024 if that's all the device supports, but it probably doesn't matter as PFS3 probably overrides that itself. Not that this is related to RDB corruption at all as that setting is irrelevant. Maxtransfer 0x0007FE00 should be ok - the manual suggests following the values in the docs for the P5 SCSI software.

I probably need to do some more testing, as my system was my old development machine, it is still running a bunch of debug tools like mungwall and cyberguard. These slow the system down a fair bit. I already disabled mungwall yesterday, which speeded things up, but maybe I will now get problems with Sync 10MB mode too.

[Futaura]

I'd recommend that everybody with data corruption issues contact Michael asap - email address at http://www.codesrc.com/contact.shtml. The more systems that his fixes can be tested on, the better. I've received another new version this morning, but I think maybe my heavily patched OS3.9/4 system may be a factor also. I haven't tested yet, but will do after I've tested 6.2.5 a little more.

[Futaura]

I've retried my tests with 6.2.5, after disabling Cyberguard, Mungwall and all other debug tools. Still completely stable for me with no corruption in Sync 10MB mode. It is probably hard to make direct comparisons because we are all using different SCSI controllers - despite them all being developed by Phase 5, there are of course 3 different controller chips and 3 different drivers.

I've repeated my tests with the latest 6.2.10 test version and that seems ok to me too. However, it needs testing on more systems than just mine to draw any real conclusions.

[AC/DC HACKER!]

Originally Posted by Futaura

I'd recommend that everybody with data corruption issues contact Michael asap - email address at http://www.codesrc.com/contact.shtml. The more systems that his fixes can be tested on, the better. I've received another new version this morning, but I think maybe my heavily patched OS3.9/4 system may be a factor also. I haven't tested yet, but will do after I've tested 6.2.5 a little more.

My OS 3.1.4.1 is updated from OS 3.5 which is "heavily" patched as well. Or what we used to call a "Power User". But not in a clueless sense...check what you use, check for consistency. On a Main SD Card. I also have a 16 GB SD Card that is open to any test I want to do (including reformat completely). I'm getting the same results between the them - though limited to 64GB..I don't want to rebuild but will. The 16GB ...I don't care if I rebuild. Sync 5/MB remains stable without anything else checked thought SCS2SD V6 Utility (MacOS---could use Windows version, but...didn't start there). SFS/00, PFS/00, PFS/03..all okay. Once past 10/MBs...problems from whatever FS soon follow.

I'll Email soon. There is/are some kind of difference(s) between BPPC and CSPPC. You just think you're is better, huh?? Yah, c'mon, admit it! (I Am joking!) Honestly, though (laughing..humor is Good!), I've changed between the 3 Firmware for CSPPC checking differences also...and none make a difference except for CyberGrahX V4 with CVPPC. (Also testing newer P96).

Originally Posted by Futaura

I've repeated my tests with the latest 6.2.10 test version and that seems ok to me too. However, it needs testing on more systems than just mine to draw any real conclusions.

I looked recently and didn't see that version. *Mock tears* Haha.

[Ferry]

I already sent a message to Michael with all the data concerning to my data corruption case.

@futaura, thanks for letting us know there's a new version of the firmware, I hadn't seen it.

Saluditos,

Ferrán.

Edit: Hmmm, looks like the version you are using is not publicly available.

[Futaura]

Yes, it was a test version - hopefully Michael will send it to you to test, unless he already plans an imminent release.

It always worries me when something works on one system, but not on others - partly because that probably means it could still cause trouble on the system it appears to work fine on . Even if the software we are running is having an impact on the issue, ultimately it is irrelevant as the firmware needs to work in all instances, but trying to figure out what software could be making a difference could sometimes help with figuring out what needs to be fixed in the firmware. I guess we are all running pretty different patches, etc, on our systems. Anybody else use Executive like me?

In fact, I'm now thinking I need to retest the 6.2.10 build with all my debugging tools enabled again, just to check if the corruption comes back.

[AC/DC HACKER!]

@Futaura

I checked with v6.2.9 and it SEEMS to be doing well with copying if I do not use any version of Directory Opus. Workbench and ADOS Copying are showing solid copies while DOus 4 or 5.(81/91) are stopping shortly after copying a bunch of files. WHDLoad confirmed this when I attempted to run my generic Uropa 2 WHD install and WHDLoad reported a bad A1200 kickstart image to my 16GB Test SD card. Copied Kickstarts again from Shell and...ran, zero problems.

Before all that I used SysSpeed to check Sync 10MB/s and the SCSI2SD V6 light froze..I couldn't get it to reset, even with a 10, 15, or 20 multiple seconds Amiga+Amiga+CTRL presses. Which is also what happens when I'm using Dopus.

I'll send an Email now.

[AC/DC HACKER!]

Email sent, reply and attachment received. Been testing it.

Copying with Directory Opus (File Manager) isn't causing a solid light from the SCS2SD V6 now, copies are going all the way through. However, data corruption is still happening with what has been copied. WHDLoad reports that there is corruption, and I can see it within Intro animation from a game I'm loading that way. WHDLoad also reported that Kickstart image it's using is corrupt. I also sometimes get a prompt from the SCSI that there's a Parity checksum problem.

I've enabled that...hasn't helped. Not yet...from several tests.

Reset the Amiga then delete the files. Then copy from Workbench (Drag'N'Drop) or Shell from original source which is on my IDE ...and thus not connected to the same SCSI chain as SCSI2SD, and it's totally functioning fine...from a CF adapter. When I do that, everything "looks" great, WHDLoad isn't prompting errors, and the CSPPC SCSI isn't reporting any errors. All looks good.

Seems there's a conflict or bug still getting in the way of the File Managers...? However, when I copy from Workbench or Shell, the speeds also fantastic. I'm noticing around 7 to 8MB/s. SysSpeed is what I use to also check speeds.. It's not locking with Sync 10 MB..and shows ~8MB/s. Once the "final bug" is located, this thing will be rockin'!

So far, as long as I don't use any of Directory Opus versions, it looks great. But that shouldn't have to be. I am mostly using PFS3-AIO as the File System.

"I've repeated my tests with the latest 6.2.10 test version and that seems ok to me too. However, it needs testing on more systems than just mine to draw any real conclusions."

Yup, the different chips are playing a role I bet. OS 3.1.4.1's FFS is also noticing the same corruption here. I think it's not File System related.

I didn't do this yet.. Click'N'Drag my Uropa 2 WHDLoad install from my CF to SD..since that copies well, and then use directory open to copy that to another part of the SD...see if it shows errors. If so, it's still in the SCSI2SD firmware.

Update: That got corruption.

[Futaura]

FYI, my standard boot is DOpus Magellan in WBR mode. However, I've not tried copying with it yet - only Copy command from shell or BackItAllUp. The Copy command that I'm using defaults to a 64K buffer - I think I mentioned earlier that raising this to 128K or 256K caused major SCSI bus fatal errors (errors messages from blizzppc.device required a power off/on). This was consistently reproducible with 6.2.9, but no such errors with 6.2.5 or 6.2.10. As it happens SCSI2SD uses a 64K write buffer internally.

I'll check later, but does DOpus have a copy buffer setting? If so, I'd bet this is making a difference. I'll check with devmon later also, which will show the write sizes being sent.

Parity checking should usually be enabled on all devices. The only time I've ever seen parity errors was years ago when my Epson scanner was hooked up. I know I keep banging on about termination, but it is even more important on the CSPPC, seeing as it is 68pin and the controller itself doesn't act as a terminator (unlike the Blizzards). Ideally, you need to place an active terminator at both ends of the cable and the SCSI2SD in the middle with termination disabled. I have been told that using SCSI2SD's built-in termination is probably a bad idea on a 68pin bus: "Just looking at this scsi2sd HW it seems to auto terminate simple 50pin scsi but with a 68pin bus that probably causes a real mess." - from RS himself (I have a contact on the MorphOS team).