SCSI2SD v6 and large drives

[Ferry]

1st partition is 4GB, for WB. 2nd is for programs, 10GB. 3rd is for data, 20GB.

As I said, booting from SCSI2SD without any other device (IDE, disk, etc.), and trying SysInfo, it allows reading from the first partition and also the second one, even if it take a bit to display results, but as soon as I try third, it hangs.

If I boot from IDE, IDE partitions bring no errors, but SCSI2SD ones fail at first try.

Saluditos,

Ferrán.

[Ferry]

BTW, I'm using OS4 Beta for 68k too, from my betatesting days

Saluditos,

Ferrán.

[AC/DC HACKER!]

Originally Posted by Ferry

I have 2 virtual SCSI units in my SD (Samsung 64GB), I'm only using one of them for Amiga, with 5 partitions, and still hanging SysInfo. Are those too many?

If I boot from it (no other device, disk or SD on the system), I can test the speed of the 2 first partitions, as soon as I test the third, SysInfo hangs.

Saluditos,

Ferrán.

Maybe I'm not understanding..? You have 2 virtual SCSI units IN your SD Card? Virtual means not real..usable, but changes are committed with being OK by the user. Apparently I haven't explored the V6 or V5 well enough. I thought SCSI ID is direct relationship to...connection, such as cybppc.device or scsi.device.

[Futaura]

I think Ferrán just means 2 devices set up in scsi2sd-util6, which themselves are of course each tied to a unique SCSI unit/id. To me, I would say these are virtual SCSI devices - there is only one physical bit of hardware, but the firmware is pretending there are more than one.

[Ferry]

Originally Posted by Futaura

I think Ferrán just means 2 devices set up in scsi2sd-util6, which themselves are of course each tied to a unique SCSI unit/id. To me, I would say these are virtual SCSI devices - there is only one physical bit of hardware, but the firmware is pretending there are more than one.

Exactly that, good description.

Originally Posted by AC/DC HACKER!

You have 2 virtual SCSI units IN your SD Card?

As Futaura says, I have 2 "virtual" units on 1 physical device. In a real SCSI chain, each SCSI ID corresponds to a REAL, physical device (hard disk, ZIP drive, scanner, tape drive, etc.) If it were an HD, you could partition it into several partitions: SCSI chain => SCSI ID:HD => partitions (dh0:, dh1:dh2:, etc.)

Since what SCSI2SD does is to allow division of a single PHYSICAL devices into several "SCSI devices", each one with its own SCSI ID, as if it were a real device in a SCSI chain, that's why I call them "virtual": they are not REAL, physical devices, but emulated ones: SCSI chain => 1 x SCSI2SD => several SCSI ID:several "HD" => each "HD": partitions (dh0:, dh1:dh2:, etc.)

Saluditos,

Ferrán.

[AC/DC HACKER!]

I think I finally connected the dots. You're using a HD (Hard Drive (SD Card)) and instead of partitioning the SD Card under SCSI ID 1 (example) you're using the SD Card under ID 1 and ID 2. Which splits the SD Card into multiple drives. Using the Auto for each one to identify which section of the SD Card to be used. But....that...doesn't make it Virtual since you're _using_ the SD card.

Unless...One SCSI ID is using the SD Card (Hard Drive) and the other SCSI ID's aren't. So, "Since what SCSI2SD does is to allow division of a single PHYSICAL devices into several "SCSI devices", each one with its own SCSI ID"...is what I was first thinking. You're just using the Auto part to choose which part of the card to use for each SCSI ID.

It's an odd sense of virtual to me, since virtual means it's there...but..it isn't. Data for the Virtual may be saved to create the Virtual again, but the Virtual is gone once power is gone. I'll point to "Ready, Player One." movie as an example. Or when VR was becoming a big thing, and we'd put on these Screens before our eyes (glasses) and see a computer generated image. Now you see it an Environment that doesn't exist outside of the program/data that creates it. Like the "Idea" of Max Headroom in the 80's. Matt Frewer exists, but Max Headroom is a character.

Ferry, your example helped a bit more. I might give that a shot. But...I'm not sure of the advantages yet vs one SCSI ID for 1 SD Card and multiple partitions.

Before I do that, though, I think solving the data corruption is something that needs to be done.

[Ferry]

Originally Posted by AC/DC HACKER!

I'm not sure of the advantages yet vs one SCSI ID for 1 SD Card and multiple partitions.

Can be useful, for example, for Mac emulation: in ShapeShifter you can use an entire SCSI HD, through its SCSI ID, as a Mac HD. Much faster that using a hardfile.

Before I do that, though, I think solving the data corruption is something that needs to be done.

Looks like latest SCSI2SD firmware, v6.2.14, solves it, I have tested it successfully, so did Futaura.

Saluditos,

Ferrán.

[Futaura]

As I don't need much space on my A1200, I used the multiple device options to spread my partitions across three 4Gb devices, to avoid having to rely on various patches/updates for >4Gb support and to bypass any bugs/problems that entails and/or old applications that don't support it. On the third device I keep my native Linux APUS and Shapeshifter partitions, so just in case something screws up there, my AmigaOS partitions on the other two devices won't be touched.

v6.2.14 seems reliable to me too, which isn't always a foolproof indicator as I found the same with v6.2.5, but seeing as v6.2.14 works for Ferrán too as well as me, I think it has to be the best version yet. I'm using it with with my drives set to Sync 10Mb mode without a problem.

[Futaura]

Ok, so I've changed my mind. I can still trigger data corruption in v6.2.14 .

Specifically, copying large files using DOpus Magellan. I'm currently trying to improve AmiSSL v4 for m68k, and it was when copying over the new library that I stumbled upon the corruption. Now going to file another report with Michael. I can also trigger it using the Copy command with a 1Mb or larger buffer. DOpus Magellan ramps up its buffer size to at least 2Mb for large files, hence the problem.

[AC/DC HACKER!]

I've been tracing a power problem which was messing with my CyberStormPPC + CyberVisionPPC. I was thinking at first that the firmware of SCSI2SD V6 was causing a problem..or something changed in my motherboard. Argo, I haven't done anything with SCS2SD for a bit.

Futaura, great catch..I'm glad someone else has noticed it happening with Directory Opus v5.xx. Wondering if my CSPPC got broken somehow, I've renewed my appreciation for WarpDTs also.

I haven't gotten anything more from Michael to test...and I figured he's checking on stuff as I checked his Wiki also. Appreciate your update.

[aperez]

Originally Posted by Futaura

Ok, so I've changed my mind. I can still trigger data corruption in v6.2.14 .

Specifically, copying large files using DOpus Magellan. I'm currently trying to improve AmiSSL v4 for m68k, and it was when copying over the new library that I stumbled upon the corruption. Now going to file another report with Michael. I can also trigger it using the Copy command with a 1Mb or larger buffer. DOpus Magellan ramps up its buffer size to at least 2Mb for large files, hence the problem.

For the time being, you should revert to 6.2.5, then. There is a fix in the works for the data corruption in synchronous mode, which will be released as 6.2.15, in the coming weeks.

[Ferry]

Originally Posted by Futaura

Ok, so I've changed my mind. I can still trigger data corruption in v6.2.14 .

Specifically, copying large files using DOpus Magellan. I'm currently trying to improve AmiSSL v4 for m68k, and it was when copying over the new library that I stumbled upon the corruption. Now going to file another report with Michael. I can also trigger it using the Copy command with a 1Mb or larger buffer. DOpus Magellan ramps up its buffer size to at least 2Mb for large files, hence the problem.

Ouch! Really? Hmmm, bad news… I’ll check that asap and report back.


Saluditos,


Ferrán.

[brett71]

Are the corruption issues being reported here just to do with Directory Opus 5+, or does it also affect the last of the Opus 4 releases? (4.12 I think?)

[AC/DC HACKER!]

I haven't noticed any problems with Opus 4.12 so far.

I got an Email from Michael containing next test. I made my usual quick tests and it passed. These were with DOpus 5.91. But speed testers like SysInfo 4.x and SpeedTest are showing 2MB/s. But normal use is showing it's a lot faster. I'll check more later.

[Ferry]

Originally Posted by brett71

Are the corruption issues being reported here just to do with Directory Opus 5+, or does it also affect the last of the Opus 4 releases? (4.12 I think?)

Latest DOpus 4 unofficial release is v4.17pre21:

http://dopus.free.fr/

The latest release in Aminet is v4.16JR:

http://aminet.net/package/util/dopus/DOpus416JRbin

Saluditos,

Ferrán.

[AMike]

I use Dopus 4.12, but only for small datasize because it's slower than workbench copy. Is there a specific testszenario for Dopus 4.x? I use 6.2.14 since December and it seems stable.

[Ferry]

Originally Posted by AC/DC HACKER!

I haven't noticed any problems with Opus 4.12 so far.

I've copied an entire partition -a just-intalled WB v3.1.4.1-, checked it and reported not a single difference, but I havn't checked if there are any +2MB sized files.

I got an Email from Michael containing next test. I made my usual quick tests and it passed. These were with DOpus 5.91. But speed testers like SysInfo 4.x and SpeedTest are showing 2MB/s. But normal use is showing it's a lot faster. I'll check more later.

Yes, there's a new v6.2.15 firmware in his page:

http://www.codesrc.com/files/scsi2sd-v6/v6.2.15/

Is that the one you received?

Saluditos,

Ferrán.

[Futaura]

Originally Posted by Ferry

Yes, there's a new v6.2.15 firmware in his page:

http://www.codesrc.com/files/scsi2sd-v6/v6.2.15/

Is that the one you received?

I received this too, but for some reason my Amiga simply refuses to boot with 6.2.15 - it hangs with a black screen and I can't even get to the boot menu. I had to revert to 6.2.14 to get it booting. This is with my SCSI drives having the synchronous flag set in the RDB.

[aperez]

Have you informed Michael yet? Also, it would be helpful to know which SCSI controller you're using. Necessary, even.

[Futaura]

Originally Posted by aperez

Have you informed Michael yet? Also, it would be helpful to know which SCSI controller you're using. Necessary, even.

Of course. Michael sent it to me to test before release, but due to illness I was not able to test it until 1st Feb. I note the release on the website is the same build (binary compare on firmware file) as I had tested. I did try downloading, flashing and testing twice, but when I get a chance I'll try again. Just wanted to mention it here in case somebody else had the same issue.