BlueSCSI Pico1 stalls in MESSAGE OUT with Adaptec AHA-1542B and AHA-2940UW, but works on Macintosh

[mika-sc]

Summary

My BlueSCSI Pico1 is selected successfully as SCSI ID 0 by two different Adaptec PC SCSI host adapters, but communication then stalls in the MESSAGE_OUT phase and ends in a watchdog timeout.

The same BlueSCSI unit, SD card, and image can be detected on a Macintosh.

The problem is reproducible with both:

• 2026.04.27-rel
• 2024.04.02-release

Therefore, it does not appear to be a regression limited to the latest firmware.

Hardware

• BlueSCSI platform: Pico1 / RP2040

• BlueSCSI PCB revision: 2023.09a

• Host adapters tested:

  • Adaptec AHA-1542B
  • Adaptec AHA-2940UW

• AHA-1542B BIOS version: 3.20

• AHA-2940UW BIOS version: 2.20

• Connection: internal narrow SCSI

• SD card: 4 GB FAT32, Class 4

• SCSI target ID: 0

• Target LUN: 0

The same BlueSCSI unit works when connected to a Macintosh.

Images tested

Hard disk image

/HD0.img
ID: 0
LUN: 0
Block size: 512
Type: Fixed
Size: 516096 kB
Quirks: None

CD-ROM image

/CD0.iso
ID: 0
LUN: 0
Block size: 2048
Type: Optical
Size: 156262 kB
Quirks: None

Both device types fail at the same SCSI bus phase, before any disk data is read.

Configuration

[SCSI]
System=Generic
Debug=1

The log confirms that the Generic preset is active and that the device has Quirks: None.

Reproduction steps

1. Connect the BlueSCSI Pico1 to the PC SCSI host adapter.
2. Configure an image as SCSI ID 0, LUN 0.
3. Power on the BlueSCSI and the PC.
4. Allow the Adaptec BIOS to scan the SCSI bus.
5. The adapter selects ID 0, but the transfer stalls in MESSAGE_OUT.
6. BlueSCSI resets the bus after approximately 15 seconds.

Main failure log

This is reproducible with both firmware versions:

DBG BUS RESET
DBG ---- SELECTION: 0
DBG ---- MESSAGE_OUT

--------------

WATCHDOG TIMEOUT, attempting bus reset

scsiDev.cdb: 0x00 0x00 0x00 0x00 0x00 0x00
             0x00 0x00 0x00 0x00 0x00 0x00
scsiDev.phase: 2
SCSI DMA state: IDLE
Current buffer: 0x00000000/0x00000000
SyncOffset: 0
SyncPeriod: 0

DBG ------ OUT: 0x00
DBG -- BUS_FREE

No valid CDB is received. The failure occurs before INQUIRY, TEST UNIT READY, REQUEST SENSE, or any disk read command.

Additional behavior seen with 2026.04.27-rel

On some boots, communication progressed past MESSAGE_OUT, but every command byte was decoded as 0x80:

DBG ---- MESSAGE_OUT
DBG ------ OUT: 0x80

DBG ---- COMMAND: 0x80 (Unknown)
DBG ------ OUT: 0x80 0x80 0x80 0x80 0x80 0x80

DBG ---- STATUS: 2 CHECK_CONDITION
sense code 0x05
ASC 0x20

This repeated several times and eventually ended with:

scsi_accel_rp2040_finishRead timeout
SCSI DMA state: READ
SyncOffset: 0
SyncPeriod: 0

The repeated 0x80 bytes do not appear to be a valid CDB from the host.

Tests already performed

• Changed the PC host adapter from AHA-1542B to AHA-2940UW.
• Changed many SCSI cables.
• Tested both a fixed-disk image and a CD-ROM image.
• Used the Generic system preset.
• Confirmed that Apple quirks are disabled.
• Rolled firmware back from 2026.04.27-rel to 2024.04.02-release.
• Powered the Pico from an external USB 5 V supply.
• The same BlueSCSI unit is detected on a Macintosh.

Changing the host adapter and rolling back the firmware did not change the failure phase.

Expected behavior

After selection, BlueSCSI should receive the IDENTIFY message and a valid command such as INQUIRY or TEST UNIT READY, allowing the Adaptec BIOS or SCSI utility to detect the device.

Actual behavior

The target enters MESSAGE_OUT, but no complete message or command is transferred. It either:

1. remains in MESSAGE_OUT until the watchdog resets the bus; or
2. receives repeated 0x80 bytes as an unknown command.

P.S.

If needed, i can provide:

• Complete log.txt from 2026.04.27-rel
• Complete log.txt from 2024.04.02-release

[erichelgeson]

Please attach full debug logs.

[mika-sc]

log.txt
log_260427_1.txt
log_260427_2.txt

[erichelgeson]

Nothing sticks out - I use the 2940UW for testing all the time.

On the latest nightly firmware can you try 3 items/boots with full log:

; boot 1
[SCSI]
System=Generic
Debug=1
EnableSCSI2 = 0

; boot 2
[SCSI]
System=Generic
Debug=1
EnableSelLatch = 1

; boot 3
[SCSI]
System=Generic
Debug=1
PhaseChangeDelay = 100

And provide the logs and observed behavior. 2 years of changes is a lot to try to understand where it occurred.