SCSI controller with CDB-16 (Was: sym53c8xx and Large disk (DCB-16))

SCSI controller with CDB-16 (Was: sym53c8xx and Large disk (DCB-16))

[Alexander Voropay]

Hello, linux-scsi!

I've re replaced sym53c8xx SCSI controller with
Adaptec ASC-29320ALP (chip AIC-7901).
- Ultra320 bus does not work physically, too many errors.
- Ultra160 bus works relable, no errors.

Unfortunately Adaptec does not work with CDB-16 too :(

scsi11 : Adaptec AIC79XX PCI-X SCSI HBA DRIVER, Rev 3.0
        <Adaptec 29320ALP PCIx Ultra320 SCSI adapter>
        aic7901: Ultra320 Wide Channel A, SCSI Id=7, PCI-X 101-133MHz, 512 SCBs
scsi 11:0:0:0: Direct-Access     IFT      A24U-G2421-1     347R PQ: 0 ANSI: 5
scsi target11:0:0: asynchronous
scsi11:A:0:0: Tagged Queuing enabled.  Depth 4
scsi target11:0:0: Beginning Domain Validation
scsi target11:0:0: wide asynchronous
scsi target11:0:0: FAST-80 WIDE SCSI 160.0 MB/s DT (12.5 ns, offset 127)
scsi target11:0:0: Ending Domain Validation
sd 11:0:0:0: [sde] Very big device. Trying to use READ CAPACITY(16).
sd 11:0:0:0: [sde] READ CAPACITY(16) failed
sd 11:0:0:0: [sde]  Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK
sd 11:0:0:0: [sde] Sense not available.
sd 11:0:0:0: [sde] Using 0xffffffff as device size
sd 11:0:0:0: [sde] 4294967296 512-byte logical blocks: (2.19 TB/2.00 TiB)
sd 11:0:0:0: [sde] Write Protect is off
sd 11:0:0:0: [sde] Mode Sense: 9b 00 00 08
sd 11:0:0:0: [sde] Write cache: enabled, read cache: enabled, doesn't
support DPO or FUA
sd 11:0:0:0: [sde] Very big device. Trying to use READ CAPACITY(16).
sd 11:0:0:0: [sde] READ CAPACITY(16) failed
sd 11:0:0:0: [sde]  Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK
sd 11:0:0:0: [sde] Sense not available.
sd 11:0:0:0: [sde] Using 0xffffffff as device size
sd 11:0:0:0: Attached scsi generic sg5 type 0
sde: unknown partition table
sd 11:0:0:0: [sde] Very big device. Trying to use READ CAPACITY(16).
sd 11:0:0:0: [sde] READ CAPACITY(16) failed
sd 11:0:0:0: [sde]  Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK
sd 11:0:0:0: [sde] Sense not available.
sd 11:0:0:0: [sde] Using 0xffffffff as device size
sd 11:0:0:0: [sde] Attached SCSI disk


CAN ANYONE PLEASE recommend a working Ultra320 SCSI controller
with CDB-16 support ?


2017-04-27 18:33 GMT+03:00 Alexander Voropay <alexander.voropay@gmail.com>:
> Hi!
>
> I have a nasty bug with ‘sym53c8xx_2’ SCSI driver. I’m trying to
> attach a large (>2Tb) storage to this controller but it does not work.
> Seems nobody tested such combination before...
>
>
> My SCSI controller is noname OEM for Intel, PCI-X card with SYN53C1010 chip.
> Storage: Infortrend A24U-G2421-1 (upto 24 SATA disks to Ulra-320 SCSI channel).
> I have 11x1Tb  RAID5 as one large disk with 10Tb in size.
> My Linux is RHEL 6, kernel "kernel-2.6.32-696.1.1.el6.x86_64".
>
> Physical SCSI link is OK, full Ultra-160, no physical/parity errors
> e.t.c. BUT IT DOES NOT WORK due to a software error.
>
> As it is known, Linux’s ‘sd’ uses CDB-16 command set for large SCSI
> disks. On the other hand ‘sym53c8xx_2’ driver “Utilizes SCRIPTS
> Load/Store command” and “Handles Phase Mismatch from SCRIPTS”. Seems
> there is a software error between driver and “sd”. As a result ALL
> disk operations are damn slow and gives a phase error in the “dmesg”:
> sd 4:0:0:0: phase change 2-7 16@37a9af60 resid=10.
> sd 4:0:0:0: phase change 2-7 16@37a9af60 resid=10.
> sd 4:0:0:0: phase change 2-7 16@37a9af60 resid=10.
> e.t.c.
>
> Another problem: “sd” incorrectly determines a disk size. Note a
> “phase change” errors after and before “capacity change”. As a result
> “sd” switches to CDB-10 and maximum disk size limited by 2Tb
>
> sym53c8xx 0000: 05:01.0: PCI INT A disabled
> sym53c8xx 0000: 05:01.0: PCI INT A -> GSI 24 (level, low) -> IRQ 24
> sym0: <1010-66> rev 0x1 at pci 0000: 05:01.0 irq 24
> sym0: Symbios NVRAM, ID 7, Fast-80, LVD, parity checking
> sym0: open drain IRQ line driver, using on-chip SRAM
> sym0: using LOAD/STORE-based firmware.
> sym0: handling phase mismatch from SCRIPTS.
> sym0: SCSI BUS has been reset.
> scsi4 : sym-2.2.3
> scsi 4:0:0:0: Direct-Access     IFT      A24U-G2421-1     347R PQ: 0 ANSI: 5
> scsi target4:0:0: tagged command queuing enabled, command queue depth 16.
> scsi target4:0:0: Beginning Domain Validation
> scsi target4:0:0: FAST-80 WIDE SCSI 160.0 MB/s DT (12.5 ns, offset 62)
> scsi target4:0:0: Ending Domain Validation
> sd 4:0:0:0: phase change 2-7 16@37a9af60 resid=10.
> sd 4:0:0:0: [sde] 4294967296 512-byte logical blocks: (2.19 TB/2.00 TiB)
> sd 4:0:0:0: [sde] Write Protect is off
> sd 4:0:0:0: [sde] Mode Sense: 9b 00 00 08
> sd 4:0:0:0: [sde] Write cache: enabled, read cache: enabled, doesn't
> support DPO or FUA
> sd 4:0:0:0: phase change 2-7 16@37a9af60 resid=10.
> sd 4:0:0:0: Attached scsi generic sg5 type 0
> sd 4:0:0:0: [sde] 19529912320 512-byte logical blocks: (9.99 TB/9.09 TiB)
> sde: detected capacity change from 2199023255552 to 9999315107840
> sde: unknown partition table
> sd 4:0:0:0: phase change 2-7 16@37a9af60 resid=10.
> sd 4:0:0:0: [sde] 4294967296 512-byte logical blocks: (2.19 TB/2.00 TiB)
> sd 4:0:0:0: [sde] Attached SCSI disk
>
> As a result ‘parted’ does not work too, it hangs on “mklabel gpt” and
> kernel log filled with tons of “phase change” errors.
>
> Interestingly, most of “sg_*” utilities works fine even with --16 option.
>
> [root@stora ~]# sg_readcap -v /dev/sg5
>    read capacity (10) cdb: 25 00 00 00 00 00 00 00 00 00
> READ CAPACITY (10) indicates device capacity too large
>  now trying 16 byte cdb variant
>    read capacity (16) cdb: 9e 10 00 00 00 00 00 00 00 00 00 00 00 20 00 00
>    read capacity (16): requested 32 bytes but got 12 bytes
> Read Capacity results:
>   Protection: prot_en=1, p_type=7, p_i_exponent=15
>   Thin provisioning: tpe=1, tprz=1
>   Last logical block address=19529912319 (0x48c12cfff), Number of
> logical blocks=19529912320
>   Logical block length=512 bytes
>   Logical blocks per physical block exponent=15
>   Lowest aligned logical block address=16383
> Hence:
>   Device size: 9999315107840 bytes, 9536090.0 MiB, 9999.32 GB
>
> UPD: I’ve found another PCI-X card from the old SUN server on the
> NCR53C897 (it has no PC-style BIOS but recognized). IT DOES NOT WORK
> TOO with the SAME error.
>
> sym53c8xx 0000:06:02.0: PCI INT A disabled
> sym53c8xx 0000:06:02.0: PCI INT A -> GSI 26 (level, low) -> IRQ 26
> sym0: <896> rev 0x7 at pci 0000:06:02.0 irq 26
> sym0: No NVRAM, ID 7, Fast-40, LVD, parity checking
> sym0: SCSI BUS has been reset.
> scsi9 : sym-2.2.3
> sym53c8xx 0000:06:02.1: PCI INT B -> GSI 27 (level, low) -> IRQ 27
> sym1: <896> rev 0x7 at pci 0000:06:02.1 irq 27
> sym1: No NVRAM, ID 7, Fast-40, LVD, parity checking
> sym1: SCSI BUS has been reset.
> scsi10 : sym-2.2.3
> scsi 9:0:0:0: Direct-Access     IFT      A24U-G2421-1     347R PQ: 0 ANSI: 5
> scsi target9:0:0: tagged command queuing enabled, command queue depth 16.
> scsi target9:0:0: Beginning Domain Validation
> scsi target9:0:0: FAST-40 WIDE SCSI 80.0 MB/s ST (25 ns, offset 31)
> scsi target9:0:0: Domain Validation skipping write tests
> scsi target9:0:0: Ending Domain Validation
> sd 9:0:0:0: phase change 2-7 16@37aaff60 resid=10.
> sd 9:0:0:0: [sde] 4294967296 512-byte logical blocks: (2.19 TB/2.00 TiB)
> sd 9:0:0:0: Attached scsi generic sg5 type 0
> sd 9:0:0:0: [sde] Write Protect is off
> sd 9:0:0:0: [sde] Mode Sense: 9b 00 00 08
> sd 9:0:0:0: [sde] Write cache: enabled, read cache: enabled, doesn't
> support DPO or FUA
> sd 9:0:0:0: phase change 2-7 16@37aaff60 resid=10.
> sd 9:0:0:0: [sde] 19529912320 512-byte logical blocks: (9.99 TB/9.09 TiB)
> sde: detected capacity change from 2199023255552 to 9999315107840
> sde: unknown partition table
> sd 9:0:0:0: phase change 2-7 16@37aaff60 resid=10.
> sd 9:0:0:0: [sde] 4294967296 512-byte logical blocks: (2.19 TB/2.00 TiB)
> sd 9:0:0:0: [sde] Attached SCSI disk
>
>
> --
> -=AV=-

Re: SCSI controller with CDB-16

[Martin K. Petersen]

Alexander,

> - Ultra320 bus does not work physically, too many errors.
> - Ultra160 bus works relable, no errors.

Ultra320 was generally problematic. I worked on two different hardware
platforms that both gave up on the original plan of using 320 and went
back to an Ultra160 ASIC between prototype and GA. Mainly due to
signaling problems. Even internal controller to disk backplane cabling
was a problem.

So make sure your cabling and termination is in good working order and
rated for 320.

> Unfortunately Adaptec does not work with CDB-16 too :(

I don't have any SPI hardware anymore. But my suggestion would be to
look for an LSI 53C1030-based card. It's a different ASIC than the 1010
you tried (has more in common with the first gen SAS controllers). It's
the final generation of SPI products and would stand the best chance of
working.

-- 
Martin K. Petersen	Oracle Linux Engineering