BlueSCSI Issues - Bus Error

[rjkucia]

This is a continuation of an old thread: https://tinkerdifferent.com/threads/macintosh-se-wont-boot-system-7.792/

tl;dr - Booting System 7 off of my BlueSCSI didn't work. I got a RaSCSI instead, and that worked.

Since then I put the BlueSCSI away and didn't think about it too much. I've upgraded to an SE/30 board in my SE, and have been pretty happy with the RaSCSI. I wanted to try out the BlueSCSI again and see if I was still having the same issues. Using an image containing 7.5.3 that is confirmed to work when booting from my RaSCSI, I got a "bus error" bomb during boot. I've tried multiple SD cards with similar results.

Since the common denominator at this point is the BlueSCSI itself, what would you all recommend for starting to troubleshoot?

 

[jon_pi]

I had similar issues with an external BlueSCSI I recently tried. No matter what it would just crash the scsi bus. I tried every help tip online and different sd cards, and multiple different “confirmed working” images, and the logs always showed good. I ended up just returning mine for a refund because it was too much hassle. The supplier told me they tested it after and it worked so it must be something wrong with my computer. I got a scsi2sd 5.5 instead and it worked flawlessly first try. Seems to me the BlueSCSI is not the most reliable, or is extremely finicky to make work, either way I wouldn’t try one again.

 

[Chopsticks]

firstly don't use lido, either use the apple scsi hd format utility or perhaps scsi director. I mention this because I did benchmarks with a large number of vintage Mac formatting utilities and I found with my se/30 that the LIDO hd driver along with most others actually performed a lot worse in transfers, reads/writes etc then the generic apple driver. I had success with scsi director 4 though, and it has some handle testing and benchmarking utilities in it.

secondly, as cheap and cost effective that the BlueSCSI may be its not electrically designed very well to be honest. I think the guy behind that project is probably more a software than a hardware guy. BlueSCSI is based on the ArdSCSino-stm32 by Ztto in case you're wondering on the linage of the code. but basically the STM micro hardware is being used out of spec in the BlueSCSI because the guy who made the PCB failed to read the Datasheet and implement correct IO pin currect protection among other things. I do tend to be hearing and reading or more people having issues with the BlueSCSI these days and while I haven't had any trouble with the ones I have, I don't use them outside of scsi hd and benchmarking testing ever. My understanding is the several people have approached the project creator about some of the issues/concerns and that the he wasn't very receptive to the feedback...

finally while expensive and probably a bit to expensive for some people, I don't think you could go wrong the the scs2sd 5.5, these are reliable devices and have long history of good support, firmware update etc


btw have you recapped you se/30, I read the forum post you linked and its not really clear about that...

 

[rjkucia]

firstly don't use lido

Is this a BlueSCSI-specific issue? The images work fine on my BlueSCSI. I'm not sure how they were formatted, I believe I got the images pre-formatted online.

I don't think you could go wrong the the scs2sd 5.5

I might check that out if I can't get this figured out, thanks.

btw have you recapped you se/30,

Haha well in that forum post it was still an SE - I swapped the board out for an SE/30 board, which I did buy recapped. So no worries there!

 

[Phipli]

I might check that out if I can't get this figured out, thanks.

I think I've heard that SCSI2SD boards are unobtainable at the moment due to the component shortage, the ZuluSCSI boards are the closest you'll find at the moment.

 

[Chopsticks]

Is this a BlueSCSI-specific issue? The images work fine on my BlueSCSI. I'm not sure how they were formatted, I believe I got the images pre-formatted online.


I might check that out if I can't get this figured out, thanks.


Haha well in that forum post it was still an SE - I swapped the board out for an SE/30 board, which I did buy recapped. So no worries there!

not BlueSCSI specific, I was reading an old Macformat magazine that had some benchmarks between a bunch of scsi format utilities and it was a showing that most of the 3rd party utilise drivers weren't as fast as the apple driver.. long story short I did a bunch of tests myself and come to the same conclusion they'd back in the day.

there's nothing wring with the RaSCSI either just that its probably not the best option for a boot drive due to needing to turn it on and let the rascsi software boot up first

I think I've heard that SCSI2SD boards are unobtainable at the moment due to the component shortage, the ZuluSCSI boards are the closest you'll find at the moment.

yes that's very true, I wrote that post quite late at night and I actually meant to write ZuluSCSI

 

[tecneeq]

Since the common denominator at this point is the BlueSCSI itself, what would you all recommend for starting to troubleshoot?

Bus error could be software, from defective desktop file to wrong enablers.
I would wipe the image with zeros, or create a blank one, activate termination, use ID0, and have it alone on the internal SCSI bus.
Then i would zap the PRAM clean and boot a floppy with the patched apple scsi hd format utility, create partitions and filesystems, copy the system from the floppy to the target filesystem, reboot once again into the floppy, then into the filesystem on the BlueSCSI.

This is as clean and as minimalist as possible.

 

[jeremywork]

firstly don't use lido, either use the apple scsi hd format utility or perhaps scsi director. I mention this because I did benchmarks with a large number of vintage Mac formatting utilities and I found with my se/30 that the LIDO hd driver along with most others actually performed a lot worse in transfers, reads/writes etc then the generic apple driver. I had success with scsi director 4 though, and it has some handle testing and benchmarking utilities in it.

not BlueSCSI specific, I was reading an old Macformat magazine that had some benchmarks between a bunch of scsi format utilities and it was a showing that most of the 3rd party utilise drivers weren't as fast as the apple driver.. long story short I did a bunch of tests myself and come to the same conclusion they'd back in the day.

This is something I began testing when I first acquired a Stratos CF AztecMonster and experienced instability.

I still need to perform more tests to feel confident in understanding what's going on, but I also did see slower throughput with Lido than other formatters; however in this particular case it was one of few that produced stable results. In general I've been happy with the results from Anubis. Also stable, but quite a bit faster.

Spoiler: Tests recorded so far:

Test notes from two examples of the Stratos CF AztecMonster

-(#2) with 4GB SanDisk 30MB/s on Quadra 950 (#3) internal bus-

• Formatting with Drive Setup 1.7.3 produced seemingly usable partitions indicating up to 4.5MB/s in ATTO Performance Utility. Over a gigabyte of mixed files can be copied onto the drive without error, but copying back eventually hangs the system. Drive Setup's block scan completes successfully, but upon restart the OS prompts initialization of two 0KB ProDOS volumes which cannot complete. At this point no other partitions are mountable or recoverable in DiskWarrior.
• Formatting with Lido 7 produced usable volumes reading about 1.7MB/s in ATTO Performance Utility. Over a gigabyte of mixed files can be copied onto the drive without error, but copying back produces warnings that some files could not be copied due to disk error. The disk can be used as a boot drive despite this issue.
• Formatting with Transoft® SCSI Director™ 4.0 produced seemingly usable partitions reading about 4.5MB/s in ATTO Performance Utility. Copying over a gigabyte of mixed files to one partition completed but counted down to -2 files remaining, then presented a message about the partition being damaged and unusable. Copying about half a gigabyte to the other partition completed with the same negative file count bug, and copying back resulted in one unreadable file. Rebuilding the first partition in DiskWarrior completed, but the program warns a disk malfunction occurred when writing the replacement directory (2351, -36.) The same warning is given for the second partition, but it can re-mount after the failure.
• Formatting with synchronous mode in Micr0Net Utility passed testing and yielded a partition reading about 4.0MB/s in ATTO Performance Utility. Copying over a gigabyte of mixed files worked, but upon completion the system indicated the disk cannot be used because it cannot be found. Rebuilding the partition in DiskWarrior completed, but the program warns a disk malfunction occurred when writing the replacement directory (2351, -36.) Trying to boot from it caused hangs and bus errors.
• Formatting with FWB Hard Disk Toolkit 2.0 resulted in functional partitions reading about 1.7MB/s in ATTO Performance Utility.
• Silverlining Pro passes all parts of the setup phase but finds a regular number of bad blocks (seemingly) evenly dispersed, which prevents the format from completing.
• MacinStor™ recognizes the device on the built-in SCSI bus but does not support initializing it.
• Disk Manager Mac® twice hung the system at the same point in block scanning two different 4GB SanDisk CF cards (30MB/s)

-(#2) with 4GB SanDisk 30MB/s on Quadra 950 (#3) Sixty Eight Thousand RAID Warrior-

• Formatting with Anubis™ Utility V2.54h produced usable volumes reading over 8.0MB/s in ATTO Performance Utility. Copying data would usually work, but occasionally the SCSI bus would lock for 10 seconds ore more before resuming. Infrequently, the system would not recover from these lockups. Trying to boot from a viable System Folder results in nearly constant lockups, taking over half an hour to boot to desktop when successful.
• MacinStor™ recognizes the RAID Warrior's bus but cannot detect any devices attached to it, even if they have mounted volumes.
• Transoft® SCSI Director™ 4.0 recognizes the RAID Warrior's bus but cannot detect any devices attached to it, even if they have mounted volumes.

-(#1) with 128GB SanDisk 120MB/s on Quadra 950 (#3) Sixty Eight Thousand RAID Warrior-

• Formatting with Anubis™ Utility V2.54h produced usable volumes reading over 8.0MB/s in ATTO Performance Utility. Copying data would usually work, but occasionally the SCSI bus would lock for 10 seconds ore more before resuming. Infrequently, the system would not recover from these lockups. Trying to boot from a viable System Folder results in nearly constant lockups, taking over half an hour to boot to desktop when successful.
• MacinStor™ recognizes the RAID Warrior's bus but cannot detect any devices attached to it, even if they have mounted volumes.
• Transoft® SCSI Director™ 4.0 recognizes the RAID Warrior's bus but cannot detect any devices attached to it, even if they have mounted volumes.

-(#1) with 128GB SanDisk 120MB/s on Quadra 950 (#3) ATTO SiliconExpress IV-

• Formatting with Anubis™ Utility V2.54h produced seemingly usable partitions reading over 8.0MB/s in ATTO Performance Utility, but were not stable and quickly became corrupt.

-(#1) with 128GB SanDisk 120MB/s on Quadra 950 (#4) internal bus-

• Formatting with Anubis™ Utility V2.54h produced usable volumes reading about 3.5MB/s (40MHz) in ATTO Performance Utility. No issues with use or bootability detected.

-(#2) with 4GB SanDisk 30MB/s on Quadra 950 (#4) internal bus-

• Formatting with Drive Setup 1.7.3 produced usable volumes reading about 3.7MB/s (45MHz); 3.5MB/s (40MHz); 2.8MB/s (33MHz) in ATTO Performance Utility. Over a gigabyte of data could be copied onto the volume without issue, but retreival failed for a very small number of files. DiskWarrior could complete a rebuild, but would indicate a disk failure while replacing the directory.
• Formatting with Lido 7 produced usable volumes reading about 1.7MB/s (45MHz); 0.7MB/s (40MHz); 0.6MB/s (33MHz); 1.0MB/s (75MHz) in ATTO Performance Utility. Over a gigabyte of data could be copied onto the volume and retrieved from it without issue. DiskWarrior can complete a rebuild and successfully replace the directory. No issues with use or bootability detected.

-(#1) with 128GB SanDisk 120MB/s on Quadra 950 (#4) ATTO SiliconExpress IV-

• Formatting with Anubis™ Utility V2.54h produced usable volumes reading about 8.3MB/s (45MHz); 8.3MB/s (40MHz); 8.1MB/s (33MHz); 7.5MB/s (75MHz) in ATTO Performance Utility. No issues with use or bootability detected. DiskWarrior can complete a rebuild and successfully replace the directory.

-(#2) with 4GB SanDisk 30MB/s on Quadra 800 internal/external bus-

• Formatting with Lido 7 produced usable volumes reading about 1.5MB/s in ATTO Performance Utility. Over a gigabyte of data could be copied onto the volume and retrieved from it without issue. DiskWarrior can complete a rebuild and successfully replace the directory. No issues with use or bootability detected.
• Formatting with Anubis™ Utility V2.54h produced usable volumes reading about 4.0MB/s in ATTO Performance Utility. Over a gigabyte of data could be copied onto the volume and retrieved from it without corruption, though occasionally the SCSI bus would lock up and hang the system. DiskWarrior can complete a rebuild and successfully replace the directory. No issues with use or bootability detected.

-(#2) with 4GB SanDisk 30MB/s on Quadra 800 Sixty Eight Thousand RAID Warrior-

• Formatting with Anubis™ Utility V2.54h produced usable volumes reading about 8.0MB/s in ATTO Performance Utility. Copying data would usually work, but occasionally the SCSI bus would lock for 10 seconds ore more before resuming. Infrequently, the system would not recover from these lockups. Trying to boot from a viable System Folder results in nearly constant lockups, taking over half an hour to boot to desktop when successful.
• MacinStor™ recognizes the RAID Warrior's bus but cannot detect any devices attached to it, even if they have mounted volumes.
• Transoft® SCSI Director™ 4.0 recognizes the RAID Warrior's bus but cannot detect any devices attached to it, even if they have mounted volumes.

-(#2) with 4GB SanDisk 30MB/s on IIfx Sixty Eight Thousand SCSI Bolt-

• Formatting with Anubis™ Utility V2.54h produced usable volumes reading about 9.2MB/s (50MHz) in ATTO Performance Utility. No issues with use or bootability detected. DiskWarrior can complete a rebuild and successfully replace the directory.

-(#2) with 32GB SanDisk 120MB/s on IIfx Sixty Eight Thousand SCSI Bolt-

• Formatting with Anubis™ Utility V2.54h produced usable volumes reading about 9.5MB/s (50MHz) in ATTO Performance Utility. No issues with use or bootability detected. DiskWarrior can complete a rebuild and successfully replace the directory.

( http://www.jeremywork.com/loose_cf_aztecmonster.html )

 

[Chopsticks]

This is something I began testing when I first acquired a Stratos CF AztecMonster and experienced instability.

I still need to perform more tests to feel confident in understanding what's going on, but I also did see slower throughput with Lido than other formatters; however in this particular case it was one of few that produced stable results. In general I've been happy with the results from Anubis. Also stable, but quite a bit faster.

Spoiler: Tests recorded so far:

( http://www.jeremywork.com/loose_cf_aztecmonster.html )

interesting, I actually don't think I've used Anubis before, I mostly use a scuznet for my sd card to scsi solution so I'll have to download a copy and see what performance I get as looking at you linked results it appears that I might get even fast results

 

[volvo242gt]

Seems similar to what @AndyO is experiencing with his Classic...

 

[Zitruskeks]

I had strange issues with the BlueSCSI on my newly restored SE/30 as well. Booted with a premade Image and seemed to work, but had errors now and then, Bus Errors and other crashing. Didn't really thought much as I currently have a crude mix of RAM in it, all 1Meg Modules, Bank A 9chip parity, and Bank B mixed with 8chip no parity and 3 chip parity modules. But I checked the system with Tech Tools Pro 2.5 and Snooper 2.0 and everything run fine.


While trying to setup my own installation of 7.5.3 the installer always (tried like 20 times) crashed on Disk3 while "setting um defaults". Under SheepShaver using the same drive images all ran fine. Also tested the install on my PiSCSI and it worked well with that, too.

I was using my non fake SanDisk 32GB SD that was working in every other device I had thrown it in, from tablets, the PiSCSI ...

I had found this here and similar threads and began looking for alternatives of the BlueSCSI because I was thinking that while the STM32F103 is 5V tolerant on his IO, it can only drive the signals to 3.3V.
But then i also read somewhere (unfortunately didn't save the link) that one guy found that a cheep chinese non-brand SD card worked on his BlueSCSI after he also had problems, and that he thought the SD card socket doesn't give good contacts and the thicker chinese card was better. I searched my stuff for a non brand card and indeed, it was like 1.5 times as thick as the SanDisk. So for a giggle I tried setting up 7.5.3 again and lo and behold - it is working fine! Even upgraded to 7.5.5 without a problem.

So maybe trying this could be a thing. Or try adding tape or a shim to the backside of the SD card to give it more thickness.

 

[rjkucia]

Sorry for the lack of an update - I was able to resolve my initial issue by updating the firmware (with some advice in the TD thread https://tinkerdifferent.com/threads/bluescsi-issues-bus-error.2107/page-2#post-20021)

Right now I'm having similar-but-slightly-different bus errors pop up, but I think that might be a bad SD card.

 

[Chopsticks]

sd card support can be a bit hit or miss at times with Arduino based project or other similar devices, generally a library such as SdFat ts used to provide support for sdcard read/write. sometimes when these libraries get updated it can cause issues with some sd cards. its not so common these days as the few most commonly used libraries are pretty mature. however this is why its often said to try another sdcard, I usually use genuine sands cards or Samsung cards, if one won't work usually the other brand will.

the other thing to tip is getting some 99% IPA and soaking the sdcard socket on the bluebell with it and insert and remove a sdcard a dozen times, its possible there is flux or oxidisation on the pins causing intermittent contact issues etc