PMP 3726 problem?

PMP 3726 problem?

[Phill Jolliffe]

Hi,

I'm having probelms with a DAS401 edge store disk enclosure. I have
4x Hitachi HDS72202 loaded into it. Wonder is this is a PMP driver
problem?

Unfortunately I have less than 50% chance on any given system startup
that the disks are detected. As best I can tell the the DAS401 uses a
SiI3726 Port-Multiplier built into the enclosure. I connect this to
the mobo single esata builtin port.

My mobo is a Gigabyte H55N-USB3

[root@JollyBox ~]# uname -a
Linux JollyBox 2.6.18-194.26.1.el5xen #1 SMP Tue Nov 9 13:35:30 EST
2010 x86_64 x86_64 x86_64 GNU/Linux

In the below log messages I see what looks like the system struggling
to detect the disk available over the esata. Successful attempt look
similar but seem to suceed after a few SError.N (errno=-5). Very
occasionally it boot without any of these errors and /dev/sd[b-e] are
all detected.

When I'm lucky and the system is able to detect the disks at boot it
seems to run fine without problems till the next reboot. But the
system is not in production so no real load is on it.

Dec 26 10:28:12 JollyBox kernel: ata6: SATA link up 3.0 Gbps (SStatus
123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.15: Port Multiplier 1.1,
0x1095:0x3726 r23, 6 ports, feat 0x1/0x9
Dec 26 10:28:12 JollyBox kernel: ata6.00: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.00: SATA link up 3.0 Gbps
(SStatus 123 SControl 320)
Dec 26 10:28:12 JollyBox kernel: ata6.01: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.01: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.02: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.02: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.03: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.03: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.04: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.04: SATA link down (SStatus 0
SControl 320)
Dec 26 10:28:12 JollyBox kernel: ata6.05: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.05: SATA link up 1.5 Gbps
(SStatus 113 SControl 320)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to IDENTIFY (I/O
error, err_mask=0x100)
Dec 26 10:28:12 JollyBox kernel: ata6.15: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.15: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.00: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.00: SATA link up 3.0 Gbps
(SStatus 123 SControl 320)
Dec 26 10:28:12 JollyBox kernel: ata6.01: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.01: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.02: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.02: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.03: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.03: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.05: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.05: SATA link up 1.5 Gbps
(SStatus 113 SControl 320)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to IDENTIFY (I/O
error, err_mask=0x100)
Dec 26 10:28:12 JollyBox kernel: ata6.15: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.15: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.00: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.00: SATA link up 3.0 Gbps
(SStatus 123 SControl 320)
Dec 26 10:28:12 JollyBox kernel: ata6.01: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.01: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.02: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.02: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.03: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.03: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.04: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.04: SATA link down (SStatus 0
SControl 320)
Dec 26 10:28:12 JollyBox kernel: ata6.05: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.05: SATA link up 1.5 Gbps
(SStatus 113 SControl 320)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to IDENTIFY (I/O
error, err_mask=0x100)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to recover link after
3 tries, disabling
Dec 26 10:28:12 JollyBox kernel: ata6: failed to recover PMP, retrying in 5 secs
Dec 26 10:28:12 JollyBox kernel: ata6.15: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.15: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to read SCR 0 (Emask=0x40)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to write SCR 1 (Emask=0x40)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to clear SError.N (errno=-5)
Dec 26 10:28:12 JollyBox kernel: ata6: failed to recover PMP, retrying in 5 secs
Dec 26 10:28:12 JollyBox kernel: ata6.15: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.15: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to read SCR 0 (Emask=0x40)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to write SCR 1 (Emask=0x40)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to clear SError.N (errno=-5)
Dec 26 10:28:12 JollyBox kernel: ata6: failed to recover PMP, retrying in 5 secs
Dec 26 10:28:12 JollyBox kernel: ata6.15: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.15: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to read SCR 0 (Emask=0x40)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to write SCR 1 (Emask=0x40)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to clear SError.N (errno=-5)
Dec 26 10:28:12 JollyBox kernel: ata6: failed to recover PMP, retrying in 5 secs
Dec 26 10:28:12 JollyBox kernel: ata6.15: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6.15: SATA link up 3.0 Gbps
(SStatus 123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to read SCR 0 (Emask=0x40)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to write SCR 1 (Emask=0x40)
Dec 26 10:28:12 JollyBox kernel: ata6.00: failed to clear SError.N (errno=-5)
Dec 26 10:28:12 JollyBox kernel: ata6: failed to recover PMP after 5
tries, giving up
Dec 26 10:28:12 JollyBox kernel: ata6.15: Port Multiplier detaching
Dec 26 10:28:12 JollyBox kernel: ata6.00: disabled
Dec 26 10:28:12 JollyBox kernel: ata6: exception Emask 0x2 SAct 0x0
SErr 0x0 action 0x6 frozen t4
Dec 26 10:28:12 JollyBox kernel: ata6: irq_stat 0x00800000, incorrect PMP
Dec 26 10:28:12 JollyBox kernel: ata6: hard resetting link
Dec 26 10:28:12 JollyBox kernel: ata6: SATA link up 3.0 Gbps (SStatus
123 SControl 300)
Dec 26 10:28:12 JollyBox kernel: ata6: EH complete

Some time I see this in dmesg as well:

Dec 26 10:55:44 JollyBox kernel: ata6.15: exception Emask 0x2 SAct 0x0
SErr 0x0 action 0x6 t3
Dec 26 10:55:44 JollyBox kernel: ata6.15: irq_stat 0x00800000, incorrect PMP


-- 
Phill Jolliffe


--
Phill Jolliffe

Re: PMP 3726 problem?

[Stan Hoeppner]

Phill Jolliffe put forth on 12/29/2010 6:44 PM:
> Hi,
> 
> I'm having probelms with a DAS401 edge store disk enclosure. I have
> 4x Hitachi HDS72202 loaded into it. Wonder is this is a PMP driver
> problem?

> Unfortunately I have less than 50% chance on any given system startup
> that the disks are detected. As best I can tell the the DAS401 uses a
> SiI3726 Port-Multiplier built into the enclosure. I connect this to
> the mobo single esata builtin port.

1.  Have you tried the PCIe x1 eSATA card that came bundled with your
unit?  Does it work properly using that card?

1.  Have you tried more than 1 eSATA cable?

3.  Do you have any equipment of any kind, that generates EMI or RFI, in
close proximity (1 meter) to the PC, disk box, or the cable?  This would
include high voltage transformers (florescent lights), low voltage
transformers (wall warts--cell charger, cordless drill charger, etc),
electric pencil sharpeners, paper shredders, anything with an electric
motor, a cordless phone/base unit, anything electronic that could be
injecting noise into the eSATA cable?

Having these answers make troubleshooting easier, especially the answer
to #1 above.

-- 
Stan

Re: PMP 3726 problem?

[Phill Jolliffe]

> 1.  Have you tried the PCIe x1 eSATA card that came bundled with your
> unit?  Does it work properly using that card?

Switched from the mobo's built in esata port to bundled card. All 4
sata drives are detected without fault on each of 5 trial reboots.

[root@JollyBox ~]# dmesg | grep  -i sil
sata_sil24 0000:01:00.0: version 1.1
scsi6 : sata_sil24
scsi7 : sata_sil24

Here is the lspci from the bundled card.

http://pastebin.com/1TuM99Q4

I had avoided using the bundled card and used the mobo builtin esata +
md sw raid as I need the single PCIe slot on my mobo for a graphics
card.

> 1.  Have you tried more than 1 eSATA cable?

Sorry, should of mentioned I already tried a new cable before.

> 3.  Do you have any equipment of any kind, that generates EMI or RFI, in

Nothing out of the norm but re juggled everything on the desk to
maximise the esata cables distance from even the power strip for the
PC.  Only electric items within a meter are the disk enclosure and PC
itself.

Re: PMP 3726 problem?

[Stan Hoeppner]

Phill Jolliffe put forth on 12/30/2010 3:40 PM:

>> 1.  Have you tried the PCIe x1 eSATA card that came bundled with your
>> unit?  Does it work properly using that card?
> 
> Switched from the mobo's built in esata port to bundled card. All 4
> sata drives are detected without fault on each of 5 trial reboots.

That's a good.  Tells us the enclosure works.

> I had avoided using the bundled card and used the mobo builtin esata +
> md sw raid as I need the single PCIe slot on my mobo for a graphics
> card.

That's bad.  I hate to be the bearer of bad news, but I'm pretty sure
you are screwed.  According to Google results, which are slim pickens,
it seems the Intel H55 chipset does not support FIS based switching,
which means none of your mobo e/SATA  ports will work with a PMP.

If you really _need_ a PCIe GPU (which baffles me as you're running
Linux, and you have VGA, DVI, *and* HDMI outputs on the mobo), then it
appears a USB 3 external enclosure is most certainly in your future.
Frankly I'm surprised you didn't purchase one with both eSATA and USB 3
ports to begin with, just in case you ran into what you just did.

-- 
Stan

Re: PMP 3726 problem?

[Jan Ceuleers]

On 31/12/10 13:56, Stan Hoeppner wrote:
> Phill Jolliffe put forth on 12/30/2010 3:40 PM:
>> I had avoided using the bundled card and used the mobo builtin esata +
>> md sw raid as I need the single PCIe slot on my mobo for a graphics
>> card.
>
> That's bad.  I hate to be the bearer of bad news, but I'm pretty sure
> you are screwed.  According to Google results, which are slim pickens,
> it seems the Intel H55 chipset does not support FIS based switching,
> which means none of your mobo e/SATA  ports will work with a PMP.

But the H55 chipset supports six SATA ports. I've got the Zotac H55 
mini-ITX mobo and have four SATA disks connected to it. So an option is 
to connect the individual disks straight to the motherboard.

But not all H55 mobos make all six SATA ports directly available. And 
running multiple cables to the enclosure is ugly. And you'll probably 
want to get SATA to eSATA adapter assemblies and back again. This gets 
messy and expensive, and the cable management issues can contribute 
negatively to reliability as well. I speak from experience: my previous 
server had six disks in a separate enclosure, which were individually 
connected to the computer across two controllers.

That's why I now switched to the Zotac H55-based mobo, which is housed 
in an enclosure that can also accommodate six disks. Much better from 
multiple viewpoints: cheaper, less messy, more reliable, etc.

Jan

Re: PMP 3726 problem?

[Phill Jolliffe]

> That's bad.  I hate to be the bearer of bad news, but I'm pretty sure
> you are screwed.  According to Google results, which are slim pickens,
> it seems the Intel H55 chipset does not support FIS based switching,
> which means none of your mobo e/SATA  ports will work with a PMP.

That doesn't sound so good :-)

Can you expand on not work? It seems to be working perfectly *if* can
get through a boot and detect the drives. This is about a 50% chance.

> Frankly I'm surprised you didn't purchase one with both eSATA and USB 3
> ports to begin with, just in case you ran into what you just did.

The mobo HDMI / DVI / VGA are no good with the processor I have as
there is no integrated graphics in the i5-760. Unless you know someway
round that with Linux??

I'm running xen as well and dom0 will make use of the PCIe graphics card.

Re: PMP 3726 problem?

[Tejun Heo]

On Fri, Dec 31, 2010 at 05:53:50PM +0000, Phill Jolliffe wrote:
> > That's bad.  I hate to be the bearer of bad news, but I'm pretty sure
> > you are screwed.  According to Google results, which are slim pickens,
> > it seems the Intel H55 chipset does not support FIS based switching,
> > which means none of your mobo e/SATA  ports will work with a PMP.
> 
> That doesn't sound so good :-)
> 
> Can you expand on not work? It seems to be working perfectly *if* can
> get through a boot and detect the drives. This is about a 50% chance.

Does libata.force=1.5Gbps make any difference?

-- 
tejun