Skip to contentSkip to conversation

SCSI block devices larger then 2TB

SCSI Michal Feix 4 participants 7 messages Dec 7, 2005 – Dec 19, 2005

lore
Share & tools
lore
#0 lore raw

Greetings!

Current aic79xxx driver doesn't see SCSI devices larger, then 2TB. It fails with READ CAPACITY(16) command. As far as I can understand, we already have LBD support in kernel for some time now. So it's only the drivers, that need to be fixed? LSI driver is the only one I found working with devices over 2TB; I couldn't test any other driver, as I don't have the hardware. Is it really so bad, that only LSI chipset and maybe few others are capable of seeng such devices?

My kernel output with aic79xxx follows:

scsi1 : Adaptec AIC79XX PCI-X SCSI HBA DRIVER, Rev 1.3.11
heracles kernel:         <Adaptec AIC7902 Ultra320 SCSI adapter>
heracles kernel:         aic7902: Ultra320 Wide Channel B, SCSI Id=7, PCI-X 101-133Mhz, 512 SCBs

scsi1:A:0:0: Tagged Queuing enabled. Depth 64
target1:0:0: Beginning Domain Validation
target1:0:0: wide asynchronous.
target1:0:0: FAST-80 WIDE SCSI 160.0 MB/s DT (12.5 ns, offset 127) target1:0:0: Ending Domain Validation
sde : very big device. try to use READ CAPACITY(16).
sde : READ CAPACITY(16) failed.
sde : status=0, message=00, host=5, driver=00
sde : use 0xffffffff as device size
SCSI device sde: 4294967296 512-byte hdwr sectors (2199023 MB) SCSI device sde: drive cache: write back
sde : very big device. try to use READ CAPACITY(16).
sde : READ CAPACITY(16) failed.
sde : status=0, message=00, host=5, driver=00
sde : use 0xffffffff as device size
SCSI device sde: 4294967296 512-byte hdwr sectors (2199023 MB) SCSI device sde: drive cache: write back
sde: unknown partition table
Attached scsi disk sde at scsi1, channel 0, id 0, lun 0

Signature · 1 line
Michal Feix
#1 lore raw
Quoted 9 lines · Michal Feix

Greetings!

Current aic79xxx driver doesn't see SCSI devices larger, then 2TB. It fails with READ CAPACITY(16) command. As far as I can understand, we already have LBD support in kernel for some time now. So it's only the drivers, that need to be fixed? LSI driver is the only one I found working with devices over 2TB; I couldn't test any other driver, as I don't have the hardware. Is it really so bad, that only LSI chipset and maybe few others are capable of seeng such devices?

I definitly works fine with Qlogic parallel scsi and fibrechannel and emulex fibre channel controllers aswell as lsi/engenio megaraid controllers.

It looks like aci79xx is just broken in that repsect. Unfortunately the driver doesn't have a proper maintainer, we scsi developers put in fixes and cleanups but we don't have the full documentation to fix such complicated issue. If you have a support contract with Adaptec complain to them.

#2 lore raw
Quoted 5 lines
Quoted 3 lines

Greetings!

Current aic79xxx driver doesn't see SCSI devices larger, then 2TB. It fails with READ CAPACITY(16) command. As far as I can understand, we already have LBD support in kernel for some time now. So it's only the drivers, that need to be fixed? LSI driver is the only one I found working with devices over 2TB; I couldn't test any other driver, as I don't have the hardware. Is it really so bad, that only LSI chipset and maybe few others are capable of seeng such devices?

Quoted 7 lines

I definitly works fine with Qlogic parallel scsi and fibrechannel and emulex fibre channel controllers aswell as lsi/engenio megaraid controllers.

It looks like aci79xx is just broken in that repsect. Unfortunately the driver doesn't have a proper maintainer, we scsi developers put in fixes and cleanups but we don't have the full documentation to fix such complicated issue. If you have a support contract with Adaptec complain to them.

As we do not have any special support contract with Adaptec, it's probably a dead end. I found some aic79xx driver on Adaptec website for 2.6 kernel. It detects full SCSI device capacity, but it hangs ocassionaly when that drive is beeing accessed, so it's unusable for every day use.

Anyway, thanks for the info. And to everyone else, beware of Adaptec SCSI host adapters when using large SCSI arrays... :(

#3 lore raw
In reply to Christoph Hellwig
Quoted 17 lines · Christoph Hellwig
Quoted 9 lines · Michal Feix

Greetings!

Current aic79xxx driver doesn't see SCSI devices larger, then 2TB. It fails with READ CAPACITY(16) command. As far as I can understand, we already have LBD support in kernel for some time now. So it's only the

drivers, that need to be fixed? LSI driver is the only one I found working with devices over 2TB; I couldn't test any other driver, as I don't have the hardware. Is it really so bad, that only LSI chipset

and

Quoted 1 line

maybe few others are capable of seeng such devices?

I definitly works fine with Qlogic parallel scsi and fibrechannel and emulex
fibre channel controllers aswell as lsi/engenio megaraid controllers.

Quoted 5 lines

It looks like aci79xx is just broken in that repsect. Unfortunately the driver doesn't have a proper maintainer, we scsi developers put in fixes and cleanups but we don't have the full documentation to fix such complicated
issue. If you have a support contract with Adaptec complain to them.

I was at a BOF at LISA last week on this subject, the guy running it said that the common ultra320 chip used for parallel scsi doesn't implment READ CAPACITY(16), but instead implemnets a propriatary READ CAPACITY(12) which allows you to break the 2TB limit.

I asked him to send the patch that he's been maintaining seperatly (and providing to his customers, he's a storage hardware vendor) to the list to get integrated.

I'll see if I have any notes with his address on them, or you could check the BOF schedule online to see if it got listed there.

David Lang

Signature · 2 lines
There are two ways of constructing a software design. One way is to make it so simple that there are obviously no deficiencies. And the other way is to make it so complicated that there are no obvious deficiencies.
 -- C.A.R. Hoare
#4 lore raw
Quoted 20 lines
Quoted 14 lines
Quoted 6 lines

Current aic79xxx driver doesn't see SCSI devices larger, then 2TB. It fails with READ CAPACITY(16) command. As far as I can understand, we already have LBD support in kernel for some time now. So it's only the drivers, that need to be fixed? LSI driver is the only one I found working with devices over 2TB; I couldn't test any other driver, as I don't have the hardware. Is it really so bad, that only LSI chipset

and

Quoted 1 line

maybe few others are capable of seeng such devices?

I definitly works fine with Qlogic parallel scsi and fibrechannel and emulex
fibre channel controllers aswell as lsi/engenio megaraid controllers.

Quoted 5 lines

It looks like aci79xx is just broken in that repsect. Unfortunately the driver doesn't have a proper maintainer, we scsi developers put in fixes and cleanups but we don't have the full documentation to fix such complicated
issue. If you have a support contract with Adaptec complain to them.

Quoted 5 lines

I was at a BOF at LISA last week on this subject, the guy running it said that the common ultra320 chip used for parallel scsi doesn't implment READ CAPACITY(16), but instead implemnets a propriatary READ CAPACITY(12) which allows you to break the 2TB limit.

I asked him to send the patch that he's been maintaining seperatly (and providing to his customers, he's a storage hardware vendor) to the list to get integrated.

I'll see if I have any notes with his address on them, or you could check the BOF schedule online to see if it got listed there.

Looks like there's more work to be done inside aic79xx driver. After I gave up on making my Adaptec host adapter sees my 6TB SCSI array as a whole, I hit another bug. When the array was sliced into 3 smaller blocks which fits the 2TB limit, I can see the array as 3 separate SCSI blokc devices, but it works only with Ultra160 speed and lower. When I choose Ultra320 on my SCSI array, the driver starts dumping following errors:

scsi1: Dumping Card State at program address 0x24 Mode 0x0 Card was paused
HS_MAILBOX[0x0] INTCTL[0x80]:(SWTMINTMASK) SEQINTSTAT[0x0] SAVED_MODE[0x11] DFFSTAT[0x33]:(CURRFIFO_NONE|FIFO0FREE|FIFO1FREE) SCSISIGI[0x24]:(P_DATAOUT_DT|BSYI) SCSIPHASE[0x0] SCSIBUS[0x0] LASTPHASE[0x1]:(P_DATAOUT|P_BUSFREE) SCSISEQ0[0x0] SCSISEQ1[0x12]:(ENAUTOATNP|ENRSELI) SEQCTL0[0x0] SEQINTCTL[0x0] SEQ_FLAGS[0x0] SEQ_FLAGS2[0x0] SSTAT0[0x0] SSTAT1[0x8]:(BUSFREE) SSTAT2[0x0] SSTAT3[0x0] PERRDIAG[0x0] SIMODE1[0xa4]:(ENSCSIPERR|ENSCSIRST|ENSELTIMO) LQISTAT0[0x0] LQISTAT1[0x0] LQISTAT2[0x80]:(PACKETIZED) LQOSTAT0[0x0] LQOSTAT1[0x0] LQOSTAT2[0xe1]:(LQOSTOP0|LQOPKT)

SCB Count = 4 CMDS_PENDING = 1 LASTSCB 0xffff CURRSCB 0x3 NEXTSCB 0xff00 qinstart = 58 qinfifonext = 58
QINFIFO:
WAITING_TID_QUEUES:
Pending list:
3 FIFO_USE[0x0] SCB_CONTROL[0x60]:(TAG_ENB|DISCENB) SCB_SCSIID[0x7]
Total 1
Kernel Free SCB list: 2 1 0
Sequencer Complete DMA-inprog list:
Sequencer Complete list:
Sequencer DMA-Up and Complete list:

scsi1: FIFO0 Free, LONGJMP == 0x8251, SCB 0x3 SEQIMODE[0x3f]:(ENCFG4TCMD|ENCFG4ICMD|ENCFG4TSTAT|ENCFG4ISTAT|ENCFG4DATA|ENSAVEPTRS) SEQINTSRC[0x0] DFCNTRL[0x0] DFSTATUS[0x89]:(FIFOEMP|HDONE|PRELOAD_AVAIL) SG_CACHE_SHADOW[0x2]:(LAST_SEG) SG_STATE[0x0] DFFSXFRCTL[0x0] SOFFCNT[0x0] MDFFSTAT[0x5]:(FIFOFREE|DLZERO) SHADDR = 0x00, SHCNT = 0x0 HADDR = 0x00, HCNT = 0x0 CCSGCTL[0x10]:(SG_CACHE_AVAIL)
scsi1: FIFO1 Free, LONGJMP == 0x8063, SCB 0x3 SEQIMODE[0x3f]:(ENCFG4TCMD|ENCFG4ICMD|ENCFG4TSTAT|ENCFG4ISTAT|ENCFG4DATA|ENSAVEPTRS) SEQINTSRC[0x0] DFCNTRL[0x0] DFSTATUS[0x89]:(FIFOEMP|HDONE|PRELOAD_AVAIL) SG_CACHE_SHADOW[0x2]:(LAST_SEG) SG_STATE[0x0] DFFSXFRCTL[0x0] SOFFCNT[0x0] MDFFSTAT[0x5]:(FIFOFREE|DLZERO) SHADDR = 0x00, SHCNT = 0x0 HADDR = 0x00, HCNT = 0x0 CCSGCTL[0x10]:(SG_CACHE_AVAIL)
LQIN: 0x8 0x0 0x0 0x3 0x0 0x1 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0
scsi1: LQISTATE = 0x1, LQOSTATE = 0x0, OPTIONMODE = 0x52
scsi1: OS_SPACE_CNT = 0x20 MAXCMDCNT = 0x1

SIMODE0[0xc]:(ENOVERRUN|ENIOERR)
CCSCBCTL[0x4]:(CCSCBDIR)
scsi1: REG0 == 0x7960, SINDEX = 0x102, DINDEX = 0x102
scsi1: SCBPTR == 0x3, SCB_NEXT == 0xff00, SCB_NEXT2 == 0xffdf
CDB 0 0 0 0 0 0
STACK: 0x13 0x0 0x0 0x0 0x0 0x0 0x0 0x0
<<<<<<<<<<<<<<<<< Dump Card State Ends >>>>>>>>>>>>>>>>>>
scsi1:0:0:2: Unable to deliver message
aic79xx_abort returns 0x2003
scsi1:0:0:2: Attempting to queue a TARGET RESET message:CDB: 0x0 0x0 0x0 0x0 0x0 0x0
scsi1: At time of recovery, card was not paused

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Dump Card State Begins <<<<<<<<<<<<<<<<<

scsi1: Dumping Card State at program address 0x4 Mode 0x22 Card was paused
HS_MAILBOX[0x0] INTCTL[0x80]:(SWTMINTMASK) SEQINTSTAT[0x0] SAVED_MODE[0x11] DFFSTAT[0x33]:(CURRFIFO_NONE|FIFO0FREE|FIFO1FREE) SCSISIGI[0x24]:(P_DATAOUT_DT|BSYI) SCSIPHASE[0x0] SCSIBUS[0x0] LASTPHASE[0x1]:(P_DATAOUT|P_BUSFREE) SCSISEQ0[0x0] SCSISEQ1[0x12]:(ENAUTOATNP|ENRSELI) SEQCTL0[0x0] SEQINTCTL[0x0] SEQ_FLAGS[0x0] SEQ_FLAGS2[0x0] SSTAT0[0x0] SSTAT1[0x8]:(BUSFREE) SSTAT2[0x0] SSTAT3[0x0] PERRDIAG[0x0] SIMODE1[0xa4]:(ENSCSIPERR|ENSCSIRST|ENSELTIMO) LQISTAT0[0x0] LQISTAT1[0x0] LQISTAT2[0x80]:(PACKETIZED) LQOSTAT0[0x0] LQOSTAT1[0x0] LQOSTAT2[0xe1]:(LQOSTOP0|LQOPKT)

SCB Count = 4 CMDS_PENDING = 1 LASTSCB 0xffff CURRSCB 0x3 NEXTSCB 0xff00 qinstart = 58 qinfifonext = 58
QINFIFO:
WAITING_TID_QUEUES:
Pending list:
3 FIFO_USE[0x0] SCB_CONTROL[0x60]:(TAG_ENB|DISCENB) SCB_SCSIID[0x7]
Total 1
Kernel Free SCB list: 2 1 0
Sequencer Complete DMA-inprog list:
Sequencer Complete list:
Sequencer DMA-Up and Complete list:

scsi1: FIFO0 Free, LONGJMP == 0x8251, SCB 0x3 SEQIMODE[0x3f]:(ENCFG4TCMD|ENCFG4ICMD|ENCFG4TSTAT|ENCFG4ISTAT|ENCFG4DATA|ENSAVEPTRS) SEQINTSRC[0x0] DFCNTRL[0x0] DFSTATUS[0x89]:(FIFOEMP|HDONE|PRELOAD_AVAIL) SG_CACHE_SHADOW[0x2]:(LAST_SEG) SG_STATE[0x0] DFFSXFRCTL[0x0] SOFFCNT[0x0] MDFFSTAT[0x5]:(FIFOFREE|DLZERO) SHADDR = 0x00, SHCNT = 0x0 HADDR = 0x00, HCNT = 0x0 CCSGCTL[0x10]:(SG_CACHE_AVAIL)
scsi1: FIFO1 Free, LONGJMP == 0x8063, SCB 0x3 SEQIMODE[0x3f]:(ENCFG4TCMD|ENCFG4ICMD|ENCFG4TSTAT|ENCFG4ISTAT|ENCFG4DATA|ENSAVEPTRS) SEQINTSRC[0x0] DFCNTRL[0x0] DFSTATUS[0x89]:(FIFOEMP|HDONE|PRELOAD_AVAIL) SG_CACHE_SHADOW[0x2]:(LAST_SEG) SG_STATE[0x0] DFFSXFRCTL[0x0] SOFFCNT[0x0] MDFFSTAT[0x5]:(FIFOFREE|DLZERO) SHADDR = 0x00, SHCNT = 0x0 HADDR = 0x00, HCNT = 0x0 CCSGCTL[0x10]:(SG_CACHE_AVAIL)
LQIN: 0x8 0x0 0x0 0x3 0x0 0x1 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0
scsi1: LQISTATE = 0x1, LQOSTATE = 0x0, OPTIONMODE = 0x52
scsi1: OS_SPACE_CNT = 0x20 MAXCMDCNT = 0x1

SIMODE0[0xc]:(ENOVERRUN|ENIOERR)
CCSCBCTL[0x4]:(CCSCBDIR)
scsi1: REG0 == 0x3, SINDEX = 0x102, DINDEX = 0x102
scsi1: SCBPTR == 0xff03, SCB_NEXT == 0xff00, SCB_NEXT2 == 0x0
CDB 3 1 0 0 0 0
STACK: 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0
<<<<<<<<<<<<<<<<< Dump Card State Ends >>>>>>>>>>>>>>>>>>
scsi1:0:0:2: Unable to deliver message
aic79xx_dev_reset returns 0x2003
scsi: Device offlined - not ready after error recovery: host 1 channel 0 id 0 lun 2 scsi1 (0:2): rejecting I/O to offline device

This happens randomly after booting on various slices on the array.

After forcing Ultra160 as maximum speed on the SCSI array, driver works fine. If somebody is interested - the Array is Axus YI-16SAEU4 Ultra320 SCSI to SATA II RAID Subsystem. Host is SuperServer 5015P-8R with integrated Adaptec AIC7902 Ultra320 SCSI adapter.

Signature · 2 lines
Michal Feix
[email protected]
#5 lore raw

Michal,

This is a known problem with the Yotta, I already reported it to Axus, and they have confirmed the problem as of 2-3 months ago.

It will work with 320 LSI cards, it fails with the Adaptec cards, supposely it fails with Adaptec cards with all but a new "windows" driver also, so one would think that there is something in the driver doing it.

Roger

Quoted 195 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Michal Feix
Sent: Tuesday, December 13, 2005 2:43 AM
To: David Lang
Cc: Christoph Hellwig; [email protected]
Subject: Re: [SCSI] SCSI block devices larger then 2TB

Quoted 1 line
Quoted 1 line
Quoted 1 line

Current aic79xxx driver doesn't see SCSI devices larger,

then 2TB.

Quoted 1 line
Quoted 1 line
Quoted 1 line

It fails with READ CAPACITY(16) command. As far as I can

understand,

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

we already have LBD support in kernel for some time now. So it's only the drivers, that need to be fixed? LSI driver is

the only one

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I found working with devices over 2TB; I couldn't test any other driver, as I don't have the hardware. Is it really so

bad, that only

Quoted 6 lines
Quoted 6 lines
Quoted 1 line

LSI chipset

and

Quoted 1 line

maybe few others are capable of seeng such devices?

I definitly works fine with Qlogic parallel scsi and

fibrechannel and

Quoted 4 lines
Quoted 4 lines

emulex fibre channel controllers aswell as lsi/engenio megaraid controllers.

It looks like aci79xx is just broken in that repsect.

Unfortunately

Quoted 1 line
Quoted 1 line

the driver doesn't have a proper maintainer, we scsi

developers put

Quoted 1 line
Quoted 1 line

in fixes and cleanups but we don't have the full

documentation to fix

Quoted 1 line
Quoted 1 line

such complicated issue. If you have a support contract

with Adaptec

Quoted 3 lines
Quoted 1 line

complain to them.

I was at a BOF at LISA last week on this subject, the guy

running it

Quoted 2 lines

said that the common ultra320 chip used for parallel scsi doesn't implment READ CAPACITY(16), but instead implemnets a

propriatary READ

Quoted 8 lines

CAPACITY(12) which allows you to break the 2TB limit.

I asked him to send the patch that he's been maintaining seperatly (and providing to his customers, he's a storage hardware vendor) to the list to get integrated.

I'll see if I have any notes with his address on them, or you could check the BOF schedule online to see if it got listed there.

Looks like there's more work to be done inside aic79xx driver. After I gave up on making my Adaptec host adapter sees my 6TB SCSI array as a whole, I hit another bug. When the array was sliced into 3 smaller blocks which fits the 2TB limit, I can see the array as 3 separate SCSI blokc devices, but it works only with Ultra160 speed and lower. When I choose Ultra320 on my SCSI array, the driver starts dumping following errors:

scsi1: Dumping Card State at program address 0x24 Mode 0x0 Card was paused HS_MAILBOX[0x0] INTCTL[0x80]:(SWTMINTMASK) SEQINTSTAT[0x0] SAVED_MODE[0x11] DFFSTAT[0x33]:(CURRFIFO_NONE|FIFO0FREE|FIFO1FREE) SCSISIGI[0x24]:(P_DATAOUT_DT|BSYI) SCSIPHASE[0x0] SCSIBUS[0x0] LASTPHASE[0x1]:(P_DATAOUT|P_BUSFREE) SCSISEQ0[0x0] SCSISEQ1[0x12]:(ENAUTOATNP|ENRSELI) SEQCTL0[0x0] SEQINTCTL[0x0] SEQ_FLAGS[0x0] SEQ_FLAGS2[0x0] SSTAT0[0x0] SSTAT1[0x8]:(BUSFREE) SSTAT2[0x0] SSTAT3[0x0] PERRDIAG[0x0] SIMODE1[0xa4]:(ENSCSIPERR|ENSCSIRST|ENSELTIMO) LQISTAT0[0x0] LQISTAT1[0x0] LQISTAT2[0x80]:(PACKETIZED) LQOSTAT0[0x0] LQOSTAT1[0x0] LQOSTAT2[0xe1]:(LQOSTOP0|LQOPKT)

SCB Count = 4 CMDS_PENDING = 1 LASTSCB 0xffff CURRSCB 0x3 NEXTSCB 0xff00 qinstart = 58 qinfifonext = 58 QINFIFO:
WAITING_TID_QUEUES:
Pending list:

   3 FIFO_USE[0x0] SCB_CONTROL[0x60]:(TAG_ENB|DISCENB) 

SCB_SCSIID[0x7] Total 1 Kernel Free SCB list: 2 1 0 Sequencer Complete DMA-inprog list:
Sequencer Complete list:
Sequencer DMA-Up and Complete list:

scsi1: FIFO0 Free, LONGJMP == 0x8251, SCB 0x3 SEQIMODE[0x3f]:(ENCFG4TCMD|ENCFG4ICMD|ENCFG4TSTAT|ENCFG4ISTAT| ENCFG4DATA|ENSAVEPTRS)
SEQINTSRC[0x0] DFCNTRL[0x0] DFSTATUS[0x89]:(FIFOEMP|HDONE|PRELOAD_AVAIL) SG_CACHE_SHADOW[0x2]:(LAST_SEG) SG_STATE[0x0] DFFSXFRCTL[0x0] SOFFCNT[0x0] MDFFSTAT[0x5]:(FIFOFREE|DLZERO) SHADDR = 0x00, SHCNT = 0x0 HADDR = 0x00, HCNT = 0x0 CCSGCTL[0x10]:(SG_CACHE_AVAIL)
scsi1: FIFO1 Free, LONGJMP == 0x8063, SCB 0x3 SEQIMODE[0x3f]:(ENCFG4TCMD|ENCFG4ICMD|ENCFG4TSTAT|ENCFG4ISTAT| ENCFG4DATA|ENSAVEPTRS)
SEQINTSRC[0x0] DFCNTRL[0x0] DFSTATUS[0x89]:(FIFOEMP|HDONE|PRELOAD_AVAIL) SG_CACHE_SHADOW[0x2]:(LAST_SEG) SG_STATE[0x0] DFFSXFRCTL[0x0] SOFFCNT[0x0] MDFFSTAT[0x5]:(FIFOFREE|DLZERO) SHADDR = 0x00, SHCNT = 0x0 HADDR = 0x00, HCNT = 0x0 CCSGCTL[0x10]:(SG_CACHE_AVAIL)
LQIN: 0x8 0x0 0x0 0x3 0x0 0x1 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0
scsi1: LQISTATE = 0x1, LQOSTATE = 0x0, OPTIONMODE = 0x52
scsi1: OS_SPACE_CNT = 0x20 MAXCMDCNT = 0x1

SIMODE0[0xc]:(ENOVERRUN|ENIOERR) CCSCBCTL[0x4]:(CCSCBDIR)
scsi1: REG0 == 0x7960, SINDEX = 0x102, DINDEX = 0x102
scsi1: SCBPTR == 0x3, SCB_NEXT == 0xff00, SCB_NEXT2 == 0xffdf CDB 0 0 0 0 0 0
STACK: 0x13 0x0 0x0 0x0 0x0 0x0 0x0 0x0 <<<<<<<<<<<<<<<<< Dump Card State Ends >>>>>>>>>>>>>>>>>> scsi1:0:0:2: Unable to deliver message aic79xx_abort returns 0x2003
scsi1:0:0:2: Attempting to queue a TARGET RESET message:CDB:
0x0 0x0 0x0 0x0 0x0 0x0
scsi1: At time of recovery, card was not paused

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Dump Card State Begins <<<<<<<<<<<<<<<<<

scsi1: Dumping Card State at program address 0x4 Mode 0x22 Card was paused HS_MAILBOX[0x0] INTCTL[0x80]:(SWTMINTMASK) SEQINTSTAT[0x0] SAVED_MODE[0x11] DFFSTAT[0x33]:(CURRFIFO_NONE|FIFO0FREE|FIFO1FREE) SCSISIGI[0x24]:(P_DATAOUT_DT|BSYI) SCSIPHASE[0x0] SCSIBUS[0x0] LASTPHASE[0x1]:(P_DATAOUT|P_BUSFREE) SCSISEQ0[0x0] SCSISEQ1[0x12]:(ENAUTOATNP|ENRSELI) SEQCTL0[0x0] SEQINTCTL[0x0] SEQ_FLAGS[0x0] SEQ_FLAGS2[0x0] SSTAT0[0x0] SSTAT1[0x8]:(BUSFREE) SSTAT2[0x0] SSTAT3[0x0] PERRDIAG[0x0] SIMODE1[0xa4]:(ENSCSIPERR|ENSCSIRST|ENSELTIMO) LQISTAT0[0x0] LQISTAT1[0x0] LQISTAT2[0x80]:(PACKETIZED) LQOSTAT0[0x0] LQOSTAT1[0x0] LQOSTAT2[0xe1]:(LQOSTOP0|LQOPKT)

SCB Count = 4 CMDS_PENDING = 1 LASTSCB 0xffff CURRSCB 0x3 NEXTSCB 0xff00 qinstart = 58 qinfifonext = 58 QINFIFO:
WAITING_TID_QUEUES:
Pending list:

   3 FIFO_USE[0x0] SCB_CONTROL[0x60]:(TAG_ENB|DISCENB) 

SCB_SCSIID[0x7] Total 1 Kernel Free SCB list: 2 1 0 Sequencer Complete DMA-inprog list:
Sequencer Complete list:
Sequencer DMA-Up and Complete list:

scsi1: FIFO0 Free, LONGJMP == 0x8251, SCB 0x3 SEQIMODE[0x3f]:(ENCFG4TCMD|ENCFG4ICMD|ENCFG4TSTAT|ENCFG4ISTAT| ENCFG4DATA|ENSAVEPTRS)
SEQINTSRC[0x0] DFCNTRL[0x0] DFSTATUS[0x89]:(FIFOEMP|HDONE|PRELOAD_AVAIL) SG_CACHE_SHADOW[0x2]:(LAST_SEG) SG_STATE[0x0] DFFSXFRCTL[0x0] SOFFCNT[0x0] MDFFSTAT[0x5]:(FIFOFREE|DLZERO) SHADDR = 0x00, SHCNT = 0x0 HADDR = 0x00, HCNT = 0x0 CCSGCTL[0x10]:(SG_CACHE_AVAIL)
scsi1: FIFO1 Free, LONGJMP == 0x8063, SCB 0x3 SEQIMODE[0x3f]:(ENCFG4TCMD|ENCFG4ICMD|ENCFG4TSTAT|ENCFG4ISTAT| ENCFG4DATA|ENSAVEPTRS)
SEQINTSRC[0x0] DFCNTRL[0x0] DFSTATUS[0x89]:(FIFOEMP|HDONE|PRELOAD_AVAIL) SG_CACHE_SHADOW[0x2]:(LAST_SEG) SG_STATE[0x0] DFFSXFRCTL[0x0] SOFFCNT[0x0] MDFFSTAT[0x5]:(FIFOFREE|DLZERO) SHADDR = 0x00, SHCNT = 0x0 HADDR = 0x00, HCNT = 0x0 CCSGCTL[0x10]:(SG_CACHE_AVAIL)
LQIN: 0x8 0x0 0x0 0x3 0x0 0x1 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0
scsi1: LQISTATE = 0x1, LQOSTATE = 0x0, OPTIONMODE = 0x52
scsi1: OS_SPACE_CNT = 0x20 MAXCMDCNT = 0x1

SIMODE0[0xc]:(ENOVERRUN|ENIOERR) CCSCBCTL[0x4]:(CCSCBDIR)
scsi1: REG0 == 0x3, SINDEX = 0x102, DINDEX = 0x102
scsi1: SCBPTR == 0xff03, SCB_NEXT == 0xff00, SCB_NEXT2 == 0x0 CDB 3 1 0 0 0 0
STACK: 0x0 0x0 0x0 0x0 0x0 0x0 0x0 0x0 <<<<<<<<<<<<<<<<< Dump Card State Ends >>>>>>>>>>>>>>>>>> scsi1:0:0:2: Unable to deliver message aic79xx_dev_reset returns 0x2003
scsi: Device offlined - not ready after error recovery: host 1 channel 0 id 0 lun 2
scsi1 (0:2): rejecting I/O to offline device

This happens randomly after booting on various slices on the array.

After forcing Ultra160 as maximum speed on the SCSI array, driver works fine. If somebody is interested - the Array is Axus YI-16SAEU4 Ultra320 SCSI to SATA II RAID Subsystem. Host is SuperServer 5015P-8R with integrated Adaptec AIC7902 Ultra320 SCSI adapter.

--
Michal Feix
[email protected]
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at
http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#6 lore raw
In reply to David Lang
Quoted 36 lines · David Lang
Quoted 6 lines · Christoph Hellwig
Quoted 5 lines · Michal Feix

Greetings!

Current aic79xxx driver doesn't see SCSI devices larger, then 2TB. It fails with READ CAPACITY(16) command. As far as I can understand, we already have LBD support in kernel for some time now. So it's only

the

Quoted 4 lines
Quoted 4 lines

drivers, that need to be fixed? LSI driver is the only one I found working with devices over 2TB; I couldn't test any other driver, as I don't have the hardware. Is it really so bad, that only LSI chipset

Quoted 8 lines

and

Quoted 1 line

maybe few others are capable of seeng such devices?

I definitly works fine with Qlogic parallel scsi and fibrechannel and emulex
fibre channel controllers aswell as lsi/engenio megaraid controllers.

It looks like aci79xx is just broken in that repsect. Unfortunately

the

Quoted 1 line

driver doesn't have a proper maintainer, we scsi developers put in

fixes

Quoted 3 lines

and cleanups but we don't have the full documentation to fix such complicated
issue. If you have a support contract with Adaptec complain to them.

I was at a BOF at LISA last week on this subject, the guy running it said
that the common ultra320 chip used for parallel scsi doesn't implment READ
CAPACITY(16), but instead implemnets a propriatary READ CAPACITY(12) which
allows you to break the 2TB limit.

Quoted 8 lines

I asked him to send the patch that he's been maintaining seperatly (and providing to his customers, he's a storage hardware vendor) to the list to
get integrated.

I'll see if I have any notes with his address on them, or you could check
the BOF schedule online to see if it got listed there.

here is the BOF listing, hopefully someone will recongnise the names and be able to contact them directly

Large Filesystems: Breaking 2TB Limitation
Organizer: Sergey Sviridov, AC&NC Wednesday, December 7, 9:00 p.m.10:00 p.m., Hampton

How to address more than 2TB Storage Volume as a single LUN. Windows and Linux experience. Doug Hughes will talk about Solaris and Veritas Volume Manager.

David Lang