Re: Changes for mpt* between NetBSD 10 and NetBSD 11

On 6/24/26 19:12, Jonathan A. Kollasch wrote:
> On Wed, Jun 24, 2026 at 09:09:25PM +0000, John Klos wrote:
>> Hi,
>>
>> I've finally gotten to get physical access to my AlphaServer DS25 :)
>>
>> My CPU fans seem fine, after the system previously shut down saying one of
>> them had failed.
>>
>> I decided to see if NetBSD 11.0_RC5 will run without issues... This was mpt0
>> from NetBSD 10.1, from 12-June-2025:
>>
>> [   1.0000000] mpt0 at pci3 dev 1 function 0: Symbios Logic SAS1068 (rev. 0x01)
>> [   1.0000000] mpt0: interrupting at dec 6600 irq 36
>> [   1.0000000] scsibus2 at mpt0: 112 targets, 8 luns per target
>> [   1.0000000] Compaq product a0f7 (PCI Hot-Plug system, revision 0x12) at
>> pci3 dev 6 function 0 not configured
>> [   1.0000000] tsciic0 at tsc0
>> [   1.0000000] iic0 at tsciic0: I2C bus
>> [   1.0000000] WARNING: system needs entropy for security; see entropy(7)
>> [   3.5241170] scsibus0: waiting 2 seconds for devices to settle...
>> [   3.5939651] scsibus1: waiting 2 seconds for devices to settle...
>> [   3.6657291] scsibus2: waiting 2 seconds for devices to settle...
>> [   5.7447517] sd0 at scsibus0 target 0 lun 0: <COMPAQ, BD0726536C, 3B02> disk fixed
>> [   5.8365462] sd0: 69464 MB, 28004 cyl, 20 head, 254 sec, 512 bytes/sect x
>> 142264000 sectors
>> [   5.9517776] sd0: sync (25.00ns offset 63), 16-bit (80.000MB/s) transfers, tagged queueing
>> [  20.9249817] sd1 at scsibus2 target 1 lun 0: <ATA, ST2000LM010-1RA1, SDM1> disk fixed
>> [  21.0216585] sd1: 1863 GB, 1907730 cyl, 16 head, 127 sec, 512 bytes/sect x 3907029168 sectors
>> [  21.4864900] sd1: tagged queueing
>> ...
>> (and more disks)
>>
>> With NetBSD 11.0_RC5:
>>
>> [   1.0000000] tsp3 at tsc0
>> [   1.0000000] pci3 at tsp3 bus 0
>> [   1.0000000] mpt0 at pci3 dev 1 function 0: Symbios Logic SAS1068 (rev. 0x01)
>> [   1.0000000] mpt0: interrupting at dec 6600 irq 36
>> [   1.0000000] scsibus2 at mpt0: 112 targets, 8 luns per target
>> [   1.0000000] Compaq product a0f7 (PCI Hot-Plug system, revision 0x12) at pci3 dev 6 function 0 not configured
>> [   1.0000000] tsciic0 at tsc0
>> [   1.0000000] iic0 at tsciic0: I2C bus
>> [   1.0000000] WARNING: system needs entropy for security; see entropy(7)
>> [   3.6707665] scsibus0: waiting 2 seconds for devices to settle...
>> [   3.7411845] scsibus1: waiting 2 seconds for devices to settle...
>> [   3.8129517] scsibus2: waiting 2 seconds for devices to settle...
>> [   5.8925025] sd0 at scsibus0 target 0 lun 0: <COMPAQ, BD0726536C, 3B02> disk fixed
>> [   5.9843003] sd0: 69464 MB, 28004 cyl, 20 head, 254 sec, 512 bytes/sect x 142264000 sectors
>> [   6.1034442] sd0: sync (25.00ns offset 63), 16-bit (80.000MB/s) transfers, tagged queueing
>> [  21.0784248] sd1 at scsibus2 target 1 lun 0: <ATA, ST2000LM010-1RA1, SDM1> disk fixed
>> [  21.1702231] sd1(mpt0:0:1:0):  Check Condition on CDB: 0x23 00 00 00 00 00 00 00 0c 00
>> [  21.2627331]     SENSE KEY:  Illegal Request
>> [  21.3105785]      ASC/ASCQ:  Invalid Command Operation Code
>> [  21.3781857] sd1: drive offline
>> [  21.4163227] sd1: tagged queueing
>> [  21.4671061] sd2 at scsibus2 target 2 lun 0: <ATA, ST2000LM010-1RA1, SDM1> disk fixed
>> [  21.5589045] sd2(mpt0:0:2:0):  Check Condition on CDB: 0x23 00 00 00 00 00 00 00 0c 00
>> [  21.6514203]     SENSE KEY:  Illegal Request
>> [  21.6992646]      ASC/ASCQ:  Invalid Command Operation Code
>> [  21.7668700] sd2: drive offline
>> [  21.8050045] sd2: tagged queueing
>> ...
>> (and more disks)
>>
>>
>> I'm now trying another controller (SATA instead of SAS), but I'm wondering
>> if anyone know what these "Check Condition on CDB" errors mean.
> 
> Looks like you're running into something that was fixed post branch but
> (probably) not pulled up to -11 already:
> 
> https://mail-index.netbsd.org/source-changes/2025/10/11/msg158519.html

I have not tried to boot ( or net boot ) my old DEC AlphaStation 600
lately but my guess is that I will see similar problems. I will take
a look at that later today and report back. It takes a while to move
that beast around and hook up the SCSI cables needed.

The really sad situation is that I have no way to use the "new" 1997
Digital SCSI controller I installed because the case of the machine
blocks access to the external scsi port. I may need a Dremel tool to
rip some steel away. That is a sad solution.


-- 
Dennis Clarke
RISC-V/SPARC/PPC/ARM/CISC
UNIX and Linux spoken