sil3112 on alpha

sil3112 on alpha

[Jure Peèar]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=UTF-8, Size: 3910 bytes --]


Hi all,

I have an alphaserver 1000a with four ibm IC35 drives sitting on two cmd680
cards in it. On them are a couple of evms volumes exported via nfs. Today i
started to put some load on them and got this:

alphabox kernel: hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error
}
alphabox kernel: hdg: dma_intr: error=0x01 { AddrMarkNotFound },
LBAsect=849603, sector=849600
alphabox kernel: hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error
}
alphabox kernel: hdg: dma_intr: error=0x01 { AddrMarkNotFound },
LBAsect=849600, sector=849600
alphabox kernel: hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error
}
alphabox kernel: hdg: dma_intr: error=0x01 { AddrMarkNotFound },
LBAsect=849600, sector=849600
alphabox kernel: hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error
}
alphabox kernel: hdg: dma_intr: error=0x01 { AddrMarkNotFound },
LBAsect=849600, sector=849600
alphabox kernel: hdg: DMA disabled
alphabox kernel: ide3: reset: success

and then a whole lot of:

CIA machine check: vector=0x670 pc=0xfffffc0000463534 code=0x98
machine check type: processor detected hard error
pc = [ide_mm_insl+36/80]  ra = [ata_input_data+128/224] ps = 0007    Not
tainted
v0 = 0000000000000020  t0 = ffffffff8a17e9fd  t1 = 0000000000000001
t2 = 00000000ffffffff  t3 = 000000000005d775  t4 = 0000000000000000
t5 = fffffc00006b0000  t6 = 0000000100f65566  t7 = fffffc00005d8000
a0 = fffffc86022f20c0  a1 = fffffc00005dbcb8  a2 = 0000000000000001
a3 = 0000000000000000  a4 = 0000000000000000  a5 = 0000000000000005
t8 = 000000000000001f  t9 = fffffc0000683540  t10= 0000000000000000
t11= 000000000000000a  pv = fffffc0000463510  at = 0000000000000000
gp = fffffc000066c400  sp = fffffc00005dbc30

I guess there's something in the recovery code not properly aligned the way
alpha CPU wants it. So maybe the times will help here:

dma_intr errors showed at 20:03:28, followed by machine check errors at
20:03:45, 20:04:59, 20:05:3, 20:05:32, 20:05:41, 20:05:42, 20:05:52,
20:06:10, 20:06:20, 20:06:21, 20:06:33, 20:09:01, 20:09:10, 20:09:11,
20:09:20, 20:09:21, 20:09:30, 20:09:31, 20:09:40, 20:09:41 and ended at
20:09:47 with 

alphabox kernel: hdg: task_in_intr: status=0x59 { DriveReady SeekComplete
DataRequest Error }
alphabox kernel: hdg: task_in_intr: error=0x01 { AddrMarkNotFound },
LBAsect=859808, sector=859808
alphabox kernel: hdg: task_in_intr: status=0x59 { DriveReady SeekComplete
DataRequest Error }
alphabox kernel: hdg: task_in_intr: error=0x01 { AddrMarkNotFound },
LBAsect=859808, sector=859808
alphabox kernel: hdg: task_in_intr: status=0x59 { DriveReady SeekComplete
DataRequest Error }
alphabox kernel: hdg: task_in_intr: error=0x10 { SectorIdNotFound },
LBAsect=859808, sector=859808
alphabox kernel: hdg: task_in_intr: status=0x59 { DriveReady SeekComplete
DataRequest Error }
alphabox kernel: hdg: task_in_intr: error=0x01 { AddrMarkNotFound },
LBAsect=862096, sector=862096
alphabox kernel: hdg: task_in_intr: status=0x59 { DriveReady SeekComplete
DataRequest Error }
alphabox kernel: hdg: task_in_intr: error=0x01 { AddrMarkNotFound },
LBAsect=862096, sector=862096
alphabox kernel: hdg: task_in_intr: status=0x59 { DriveReady SeekComplete
DataRequest Error }
alphabox kernel: hdg: task_in_intr: error=0x01 { AddrMarkNotFound },
LBAsect=862096, sector=862096
alphabox kernel: hdg: task_in_intr: status=0x59 { DriveReady SeekComplete
DataRequest Error }
alphabox kernel: hdg: task_in_intr: error=0x01 { AddrMarkNotFound },
LBAsect=862096, sector=862096
alphabox kernel: ide3: reset: success

Apart from the fact, that my hdg is obviously not all that healty, it looks
there's something in the dma recovery code that alphas do not really like.
Does the fact that it gets tipped about every 10 seconds for a second rings
any bells?

Kernel 2.6.7-rc3-mm1.


-- 

Jure Peèar

Re: sil3112 on alpha

[Jeff Garzik]

Jure Peèar wrote:
> Hi all,
> 
> I have an alphaserver 1000a with four ibm IC35 drives sitting on two cmd680
> cards in it. On them are a couple of evms volumes exported via nfs. Today i
> started to put some load on them and got this:


What does sata_sil do?

	Jeff

Re: sil3112 on alpha

[Jure Peèar]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=UTF-8, Size: 458 bytes --]

On Mon, 23 Aug 2004 15:07:55 -0400
Jeff Garzik <jgarzik@pobox.com> wrote:

> Jure Peèar wrote:
> > Hi all,
> > 
> > I have an alphaserver 1000a with four ibm IC35 drives sitting on two
> > cmd680 cards in it. On them are a couple of evms volumes exported via
> > nfs. Today i started to put some load on them and got this:
> 
> 
> What does sata_sil do?

Havent really tried, since cards & disks are all pata. Should I?


-- 

Jure Peèar

Re: sil3112 on alpha

[Jeff Garzik]

Jure Peèar wrote:
> On Mon, 23 Aug 2004 15:07:55 -0400
> Jeff Garzik <jgarzik@pobox.com> wrote:
> 
> 
>>Jure Peèar wrote:
>>
>>>Hi all,
>>>
>>>I have an alphaserver 1000a with four ibm IC35 drives sitting on two
>>>cmd680 cards in it. On them are a couple of evms volumes exported via
>>>nfs. Today i started to put some load on them and got this:
>>
>>
>>What does sata_sil do?
> 
> 
> Havent really tried, since cards & disks are all pata. Should I?


Your message said sil3112, which is SATA-only.

I guess you have a CMD680...

	Jeff

Re: sil3112 on alpha

[Jure Peèar]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=UTF-8, Size: 1725 bytes --]

On Mon, 23 Aug 2004 16:23:32 -0400
Jeff Garzik <jgarzik@pobox.com> wrote:

 
> Your message said sil3112, which is SATA-only.
> 
> I guess you have a CMD680...

Sorry, i forgot to fix the subject too :)
Yes, they are cmd680 cards, 'medley' or what are they called ... 

0000:01:02.0 RAID bus controller: Silicon Image, Inc. (formerly CMD
Technology Inc) PCI0680 Ultra ATA-133 Host Controller (rev 02)
	Subsystem: Silicon Image, Inc. (formerly CMD Technology Inc) Winic W-680
(Silicon Image 680 based)

used with CONFIG_BLK_DEV_SIIMAGE=y

and recognized as:

Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
SiI680: IDE controller at PCI slot 0000:00:0d.0
SiI680: chipset revision 2
SiI680: BASE CLOCK == 133
SiI680: 100% native mode on irq 22
    ide0: MMIO-DMA , BIOS settings: hda:pio, hdb:pio
    ide1: MMIO-DMA , BIOS settings: hdc:pio, hdd:pio
hda: IC35L060AVVA07-0, ATA DISK drive
Using anticipatory io scheduler
ide0 at 0xfffffc86023a1080-0xfffffc86023a1087,0xfffffc86023a108a on irq 22
hdc: IC35L060AVVA07-0, ATA DISK drive
ide1 at 0xfffffc86023a10c0-0xfffffc86023a10c7,0xfffffc86023a10ca on irq 22
SiI680: IDE controller at PCI slot 0000:01:02.0
SiI680: chipset revision 2
SiI680: BASE CLOCK == 133
SiI680: 100% native mode on irq 26
    ide2: MMIO-DMA , BIOS settings: hde:pio, hdf:pio
    ide3: MMIO-DMA , BIOS settings: hdg:pio, hdh:pio
hde: IC35L060AVVA07-0, ATA DISK drive
ide2 at 0xfffffc86022f2080-0xfffffc86022f2087,0xfffffc86022f208a on irq 26
hdg: IC35L060AVVA07-0, ATA DISK drive
ide3 at 0xfffffc86022f20c0-0xfffffc86022f20c7,0xfffffc86022f20ca on irq 26


-- 

Jure Peèar